Академический Документы
Профессиональный Документы
Культура Документы
Prosa
Robert M.Ulfig Brian P.Geiser
Thomas F.Kelly
Local Electrode
Atom Probe
Tomography
A Users Guide
Foreword by
Professor Sir Colin J. Humphreys
Local Electrode Atom Probe Tomography
David J. Larson Ty J. Prosa Robert M. Ulfig
Brian P. Geiser Thomas F. Kelly
Foreword by
Professor Sir Colin J. Humphreys, CBE, FRS, FREng.
David J. Larson Ty J. Prosa
CAMECA Instruments, Inc. CAMECA Instruments, Inc.
Madison, WI, USA Madison, WI, USA
Thomas F. Kelly
CAMECA Instruments, Inc.
Madison, WI, USA
Over the years a number of atom probe books have been written by various authors
[110], so why write another one you might ask. We believe that this book is unique
in specifically targeting atom probe adopters who are new to the technique. The
book introduces new users to the process of performing all of the aspects of a Local
Electrode Atom Probe experiment. It includes the fundamentals of preparing
specimens for the microscope from a variety of materials, details of the instrumen-
tation used in data collection, parameters under which optimal data are collected,
current methods of data reconstruction, and selected methods of data analysis. In
addition, certain topics are explained specifically from a user perspective and
include details that are often learned only through trial and error, allowing users
to succeed more quickly in the challenging areas of specimen preparation and data
collection.
This book is meant to be a useful reference for the conventional wisdom type
of information that is not always found in academic books and is usually gained
only through experience. It is not meant to be a comprehensive treatment of atom
probe tomography but rather an everyday reference for data collection on the local
electrode atom probe and for the specimen preparation and data analysis that go
along with such experiments. For the most part, we have dealt with more advanced
topics, such as the details of the spatial reconstruction equations, by including the
information in appendices or by simply referring the reader to other textbooks or
journal articles. In this way we hope to have produced a very usable reference for
both novice users and experienced scientists. The future of atom probe tomography
is bright, and we hope that the path to adoption will be clearer with the availability
of this book.
v
vi Preface
References
1. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principles and Applications. Elsevier,
New York (1969)
2. Hren, J.J., Ranganathan, S. (eds.): Field-Ion Microscopy. Plenum Press, New York (1968)
3. Bowkett, K.M., Smith, D.A.: Field-Ion Microscopy. North-Holland, Amsterdam (1970)
4. Muller, E.W., Tsong, T.T.: Field Ion Microscopy, Field Ionization and Field Evaporation, vol.
4. Progress in Surface Science (1973)
5. Miller, M.K., Smith, G.D.W.: Atom Probe Microanalysis: Principles and Applications to
Materials Problems. Materials Research Society, Pittsburgh (1989)
6. Sakurai, T., Sakai, A., Pickering, H.W.: Atom probe field ion microscopy and its applications.
Adv. Electron. Electron. Phys. 20, 1299 (1989)
7. Tsong, T.T.: Atom-Probe Field Ion Microscopy: Field Ion Emission and Surfaces and
Interfaces at Atomic Resolution. Cambridge University Press, Cambridge, Great Britain
(1990)
8. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion Micros-
copy. Oxford University Press, Oxford (1996)
9. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York (2000)
10. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer Series
in Materials Science, vol. 160. Springer (2012)
Acknowledgment
Numerous people have contributed to the development and success of the LEAP1
over the years. We would certainly like to include everyone who participated in this
venture, but we are certain to miss a few names, and so we apologize in advance for
any omissions.
We would like to start with a special acknowledgement to the family of Tom
Kellywithout their encouragement and support over the years, the current LEAP
would not exist. We would like to acknowledge all of the past employees of
CAMECA1 (formerly Imago Scientific Instruments) who played critical roles
in the development of the LEAP; most notably we would like to thank Dr. Tye T.
Gribb for all of his contributions toward making the LEAP what it is today. We
would also like to thank all of the current employees of CAMECA with a special
acknowledgment to the engineering team for creating a marvelous instrument: Joe
Bunton, Dan Lenz, Jesse Olson, Ed Oltman, Tim Payne, Dan Rauls, Jeff Shepard,
Eric Strennen, Mark Lamers, and Dave Sund.
We would especially like to thank our colleagues who have assisted in the
preparation of this book by reviewing individual chapters. Professional colleagues
include Frederic Danoix, Baptiste Gault, Stephan Gerstl, Michael Miller, Sergej
Mutas, John Panitz, Danny Perea, Amanda Petford-Long, Krishna Rajan, C.J.
Humphreys, and Dan Schreiber. We will particularly like to thank Richard Forbes
for assistance with one of the appendices. Likewise, our CAMECA colleagues who
have reviewed chapters include Peter Clifton, Devin Giddings, Daniel Lawrence,
Dan Lenz, Mark Levesque, David Olson, Jesse Olson, Ed Oltman, David Reinhard,
Hugues Francois Saint Cyr, Michel Schumacher, Eric Strennen, Savanna Strennen,
Ludovic Renaud, and John Watson.
We would also like to thank the scientists and publishers who have granted us
permission to include portions of their work in this book: D. Abou-Ras,
H.O. Andren, O.H. Auciello, P.A.J. Bagot, J. Ballard, R. Banerjee, G. Brennecka,
R. Caballero, J. Cairney, A.J. Cavosie, A. Cerezo, V. Chitry, P.P. Choi,
O. Cojocaru-Miredin, X.Y. Cui, T.L. Daulton, A.M. Davis, F. DeGeuser,
S.P. DenBaars, A. Deschamps, D.R. Diercks, D.M. Dimiduk, J.W. Elam,
P. Felfer, C. Floss, H.L. Fraser, B. Gault, M. Gilbert, H. Gnaser, L.M. Gordon,
vii
viii Acknowledgment
B.P. Gorman, T.T. Gribb, G.H. Gu, E.E. Haller, M. Hasegawa, P.R. Heck,
M.A. Hekmaty, A. Henjered, K. Henry, J. Hiller, A.C. Hillier, C. Hombourger,
Y.L. Hu, J.Y. Hwang, K. Inoue, D. Isheim, K.M. Itoh, D. Jager, M.J. Jercinovic,
D. Joester, A.K. Kambham, Y. Kawamura, K.F. Kelton, M. Kimoto, R. Kirchhofer,
S. Koelling, N. Kruse, L.J. Lauhon, W. Lefebvre, J.L. Lensch, J. Lewis, T. Li,
P. Liddicoat, L. Longstreth-Spoor, A. Mane, E.A. Marquis, I. Martin, S.J. May,
B. Mazumder, A.J. Melmed, T.P. Melo, M.K. Miller, J. Mody, M.P. Moody,
D. Moser, C.K.S. Moy, E.W. Muller, Y. Nagai, S. Nakamura, A. Nishida,
O. Nishikawa, H. Norden, J.A. Panitz, C.G. Park, M.J. Pellin, D.E. Perea,
T.C. Petersen, D. Raabe, J.N. Randall, G. Ranzi, R.F. Reidy, L. Renaud,
S.P. Ringer, M. Romanes, M.R. Savina, D.W. Saxey, D. Schreiber,
D.N. Seidman, Y. Shimizu, R. Shivaraman, C. Sigli, V.S. Smentkowski,
G.D.W. Smith, D. Snoeyenbos, J.S. Speck, R. Srinivasan, F.J. Stadermann,
E. Steel, T. Stephan, A. Strickland, G. Sundell, H. Takamizawa, M.C. Teague,
M. Thuvander, J. Tiley, T. Toyama, S.C.E. Tsang, T. Tsunomura, A. Tyagi,
M. Uematsu, T. Ushikubo, J.W. Valley, W. Vandervorst, F. Vurpillot,
G.B. Viswanathan, B.W. Wessels, S. Wilde, C.A. Williams, M.H. Wong,
R. Wuerz, Y. Xiang, F. Yano, Y. Zhang, and H. Zhong. We also thank the
companies who granted us permission to display images, graphs, or schematics or
who aided in some implementation of LEAP analysis, including Evans Analytical
Group1, Dune Sciences, Gatan, Fischione, Omniprobe, Hummingbird Scientific,
and XEI Scientific Inc.
Foreword
Atom probe tomography has had explosive growth in recent years. The technique
was born way back in 1935 when Edwin Muller invented the field emission electron
microscope. This original atom probe instrument was difficult to use and could only
be applied to metals. Modern-day atom probe tomography is straightforward to use
and can be applied not only to metals but also to semiconductors, ceramics,
geological materials, organic materials, biological materials, and composites.
This is a wonderful book which should be in every science library and on the
bookshelves of scientists and engineers interested in the characterization of mate-
rials at the atomic level. The book requires no prior knowledge of atom probe
tomography and takes readers through from the history of the local electrode atom
probe to the latest developments in the field. Although atom probe tomography is
now an easy technique to use, the interpretation of the results requires considerable
care and understanding. This book takes readers step by step through data collec-
tion, data reconstruction, and data analysis. As with all microscopies, good speci-
men preparation is the key to obtaining world-class results. Various methods of
specimen preparation are described in detail, together with clear instructions on
how to optimize specimen preparation using, for example, focused ion beam
methods.
It is a great pleasure for me to recommend this comprehensive yet very readable
book on the local electrode atom probe, which complements other methods, such as
atomic resolution electron microscopy. I believe that atom probe tomography will
increasingly be used by scientists and engineers involved with the characterization
of materials and devices at the atomic level. I have no doubt that this excellent book
will be widely used by the atom probe community.
ix
Contents
xi
xii Contents
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Abbreviations: Variable and Function
Descriptions
a Acceleration
A Detected surface area
a 0, a 1, a 2 Voltage fitting constants
Ai Actual count of type i ion
B Background, broadening due to the adsorption process
Bgd Background level in ppm/ns
c Bowl fitting constant
C( ) Gaussian response function
Ci Concentration of ith range
d Separation distance
Dmax Maximum cluster ion separation
DR Detection rate
dR/dZ Radial change as a function of depth change
e Fundamental unit of charge
E Electric field, cluster detection erosion distance
F Force, evaporation field
f Frequency
F0 Field necessary for onset of field evaporation
FB ( ) Bowl correction function
fij Isotopic abundance of isotope type i in range j
fS/C Specimen sphere-to-cone ratio
FV ( ) Voltage correction function
FEV Field evaporation
h Specimen height
In nth ionization energy
k Electric field factor, confidence level factor
Ke Evaporation rate
xv
xvi Abbreviations: Variable and Function Descriptions
1.1 Introduction
Atom probe tomography (APT) is one of the most spectacular microscopies that
exist. A three-dimensional (3D) image at near atomic scale is produced with single-
atom sensitivity where each atom (actually each isotope) in the image is identified.
Because the fundamental data format is the 3D position and identity of individual
atoms in a volume containing potentially hundreds of millions of atoms, many types
of information may be gleaned. Elemental concentration may be determined in any
subvolume size or shape simply by counting atoms. Concentration profiles may be
calculated in any direction, even radially through a spherical feature or normal to
any defined surface. Isoconcentration surfaces can be set to delineate and measure
interfaces. Interatomic distribution functions can be determined for studying order-
ing, dopant interactions, cluster formation, crystal structure, diffusion, and early
stages of precipitation. Once the dataset is obtained, the quality and quantity of
results derived are limited principally by the intrinsic quality of the data and the
microscopists ability to direct a computer to extract the information.
APT is used to study an increasingly wide variety of inorganic materials and is
gradually demonstrating capabilities for analysis of organic materials as well.
The unique power of APT lies in its ability to tie compositional information
to structure. Other characterization techniques can determine physical structure at
sub-nanometer levels, but only APT has the capacity for broad compositional
identification (and quantification) at these length scales. Furthermore, APT accom-
plishes this with high analytical sensitivity which can reach one atomic part per
million (appm).
The atom probe was invented in 1967 at the Pennsylvania State University by
Muller, Panitz, and McLane [1] as a 1D analytical instrument that collected
hundreds of atoms/day with approximately a one-nanometer field of view (FOV).
By 1973, the origins of tomographic analysis were present in Panitzs 10 cm Atom
Probe [2]. A revolution in atom probe technology came when an operational, fully
3D imaging instrument was developed by Cerezo, Godfrey, and Smith [3] that
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 1
DOI 10.1007/978-1-4614-8721-0_1, Springer Science+Business Media New York 2013
2 1 History of APT and LEAP
could record hundreds of atoms/min in a ~15 nm wide FOV. In the past decade or
so, the local electrode atom probe (LEAP) [4, 5] and related contributions like the
scanning atom probe [6] and focused ion beam (FIB)-based specimen preparation
[7] have brought about another revolution in the development of APT [8]. With a
LEAP, millions of atoms/min can be collected with up to ~200 nm wide FOV with
high mass resolution (<1/1,000) on many types of materials.
It is useful to review the APT history so that users can understand the LEAP and
its evolution from its predecessors. This chapter is intended to be historical in nature
and focuses on the people and machines that have led to the current generation of
atom probe instruments. In Fig. 1.1, a flow chart of the predecessor instruments in
the lineage of the modern atom probe is presented. There have been dramatic
improvements in atom probe instrumentation throughout its history. Recent
advances have been driven by the availability of vastly superior commercial
instruments, most notably the LEAP.
First, to clarify some nomenclature: APT is the name of the technique. There are
many types of instruments which may be 1D atom probes (1DAP) or 3D atom
probes (3DAP). Some instruments that are 3DAPs were given names such as the
1.2 Ancestry of the Local Electrode Atom Probe 3
position-sensitive atom probe (PoSAP) [3] or the tomographic atom probe (TAP)
[9]. Instruments that feature a local electrode are LEAPs, which is a name used
by CAMECA Instruments, Inc.
The history of APT begins in the early nineteenth century. The electron was
discovered by Thomson [10] in 1897, and the modern view of the structure of the
atom was developed by Rutherford [11] and Bohr [12] in 1913. Experimental
confirmation of the atomic nature of materials and packing of atoms in crystals
was deduced by Bragg and Bragg [13] in 1913. Charged-particle quantum optics
theory was being developed in the late 1920s, and Fowler and Nordheim [14]
predicted the magnitude of electric field that would be required to strip an electron
from an atom, and physicists of the day could readily show that fields of about
109 V/m (1 V/nm) should suffice.
Physicists next considered how to create an electric field to form a beam of
emitted electrons from a material. Since the electric field in the region between
parallel plates is simply the voltage between them divided by their separation,
109 V/m is achieved by applying 10,000 V1 across a micrometer separation. This
was not, however, a practical scenario, and physicists of the day certainly under-
stood that. It was also known that with a concentric spherical geometry for the
electrodes, the field on the inner spheres surface would vary as the voltage between
the surfaces divided by the radius of the inner sphere. So for 10,000 V it is necessary
only to make the inner radius small enough. If the sphere (or hemisphere) is on
the end of a needle and is opposed by a flat plate normal to the axis of the needle,
one can achieve almost the same result. Thus for 10,000 V, a tip apex radius on a
needle of about 1 m will create these fields of 109 V/m and remove electrons from
the surface. This is a field emission electron source and brings us to about the end of
the 1920s when Professor Gustav Hertz of the Technical University of Berlin hired
Erwin W. Muller as a student to build a field electron emission microscope (FEEM)
which they used to demonstrate this effect and study the phenomenon. Erwin
Muller, Fig. 1.2, eventually became the father of this scientific discipline known
as high-field nanoscience. Incidentally, at the same institute at the same time,
Professor Max Knoll had a student, Ernst Ruska, working on a new instrument
called an electron microscope.
1
10,000 V is a convenient high voltage to produce experimentally in the laboratory without
needing any extreme measures.
4 1 History of APT and LEAP
Muller developed the FEEM [15] around 1935. By creating a field electron
emission source and projecting the emission pattern of electrons onto a fluorescent
screen, an image like that in Fig. 1.3b was formed. This was very exciting in the
1940s, when this work was starting to mature, and researchers started to ponder
the ultimate spatial resolution of this technique. Effectively, they were asking the
following question: Could we see atoms? By the mid-1940s, it was clear that the
answer was no. The FEEM achieved about two-nanometer resolution at best and
this has not improved significantly [16]. However, this work did spur some thinking
that led to the next development, which is the field ion microscope (FIM).
It became common practice to reverse the polarity on an FEEM tip to get rid of
adsorbed hydrogen on the specimen. Instead of simply cleaning the tip, they found a
steadily emitted stream of positive ions which formed a field ion emission pattern
on a fluorescent screen [17]. Higher fields of ~1010 V/m were required, so the
1.2 Ancestry of the Local Electrode Atom Probe 5
Fig. 1.3 (a) Schematic of the field electron emission microscope. (b) FEEM image of <110>
orientation of tungsten. The twofold symmetry of this pattern is evident. (c) Schematic of the field
ion microscope. (d) Field ion image of <110> tungsten. The first images of atoms were obtained
on October 11, 1955, by Bahadur and Muller. ((b) Reprinted with the permission from Springer-
Verlag GmbH York [17]. (d) Reprinted with permission from Elsevier [20])
Fig. 1.4 The original atom probe field ion microscope circa 1967. (a) Schematic drawing and (b)
photograph of the apparatus with John Panitz at the helm (figures courtesy of J. A. Panitz, The
University of New Mexico, Department of Physics and Astronomy)
After looking at images of atoms for half a dozen years one begins to ask, Cant we
figure out the identity of any or all of those atoms? This was first attempted using a
magnetic-sector mass spectrometer instrument by Barofsky and Muller [21], but
eventually Barofsky became convinced that time-of-flight mass spectroscopy was
the better approach [22]. Muller, Panitz, and McLane [1] thus settled on measuring
time of flight as the way to identify the ions visible in the FIM. This required a
pulsed field evaporation event in order to know when the atoms left the tip. John
Panitz (a Ph.D. student working with Muller), S. Brooks McLane, and Gerald
L. Fowler (technicians in Mullers lab) were responsible for building this instru-
ment under Mullers watchful eye. It was called the atom probe field ion micro-
scope (APFIM), Fig. 1.4, and was a 1D instrument. The operator would press a
button that sent a voltage pulse to the specimen, and the oscilloscope screen would
show a trace when the detector was hit by an ion, Fig. 1.5.
The technique progressed quickly as the groups instrument could collect tens of
atoms/h and had about a 1 nm FOV. The mass resolving power (MRP) was just
sufficient to resolve 56Fe+ from 57Fe+. Often working into the night, Panitz
would record the oscilloscope trace on Polaroid Type 57 film, typically exposing
1.2 Ancestry of the Local Electrode Atom Probe 7
a half-dozen traces per sheet to save film. A sine wave from a crystal oscillator was
recorded on the same sheet of film as a time reference independent of the oscillo-
scopes time base generator. He then projected each film positive 10 using an Aero
Ektar lens and, for each ion, measured the distance from a reference object with a
meter stick and converted that distance on the sine wave to a flight time. By morning,
he would have a list of about 100 atoms that had been recorded overnight. Groups at
the University of Pittsburgh under Sid Brenner and at the University of Oxford under
George Smith began developing atom probes. In addition to the group at Oxford,
groups at Chalmers (Hans-Olof Andren) and Cornell (David Seidman) started
computerizing control of the instruments. Soon commercial instruments by Vacuum
Generators (the VG APFIM 100 in 1980) and Applied Microscopy (the APFIM
200 in 1986) appeared and offered data collection rates of 100,000 atoms/day and
MRP (FWHM) of 1 part in 2,000 with voltage pulsing using a Poschenrieder lens
[23] and reflectron optic [24, 25], respectively. These were 1D instruments with a
probe hole for the atoms to pass through and a ~1 nm FOV.
The 3D atom probe seems a natural progression from the 1D atom probe, yet it was
a revolutionary departure from the APFIM when it was first pursued. John Panitz
developed what he called the imaging atom probe (IAP) [2] in 1972, Fig. 1.6. It was
a short flight path instrument with a large FOV, a design that we shall see below still
exists because of its fundamental merit. Time of flight could be measured anywhere
on the full detector because it used spherical microchannel plate amplifiers centered
on the specimen apex, keeping the distance to the channel plate amplifier the same
in all directions. The IAP would record the hit positions of a single atom type
(or isotope) at a time from any given voltage pulse by time gating the channel
plates. With this approach, an iron map in steel could be recorded, then a carbon
8 1 History of APT and LEAP
map could be recorded from the next part of the specimen, and so on. The IAP
collected 3D data by manually examining a series of 2D maps of a single
preselected species as a function of depth. Since the field desorption process is
destructive only a single species could be examined in each section, unlike a
modern 3DAP where a 2D map of all species can be obtained as a function of
depth. Nonetheless, the IAP produced tomographic data from an atom probe, and
Panitz should be credited with the initial inventive step in the development of APT.
The advent of a true 3DAP that recorded all atom species had to wait until the 1980s
when position-sensitive detector concepts were adapted to what essentially
amounted to IAP-like instruments. Miller conceived of such an instrument and
presented his concepts in the mid-1980s [26, 27]. The first fully operational 3D
atom probe was developed in 1988 at the University of Oxford by Cerezo, Godfrey,
and Smith and was dubbed the PoSAP [3], Fig. 1.7. These authors adapted a
wedge-and-strip detector, which was prevalent in astronomy to a commercial VG
APFIM 100 which was in their lab. By adding a 2D detector to a 1D atom probe,
1.2 Ancestry of the Local Electrode Atom Probe 9
Fig. 1.7 Photograph of the position-sensitive atom probe (PoSAP) developed by Cerezo,
Godfrey, and Smith at the University of Oxford in 1988 (figure courtesy A. Cerezo)
they created a working 3D atom probe. The images were spectacular for the time
period. They contained about a million atoms and showed precipitates and inter-
faces in materials. Analytical electron microscopists at the time appreciated how
challenging it would be to do similar work with an electron microscope.
After a postdoctoral experience at Cambridge in the early 1970s, Jean Gallot started
an activity based on field ion microscopy on his return to Rouen. The group
progressively grew from a handful of individuals to more than 150 today. This
growth is partly due to the excellence of their instrumental development of the atom
probe under the successive leadership of Alain Menand and Didier Blavette in what
became the Groupe de Physique des Materiaux (GPM), UMR CNRS 6634. Building
on a unique local expertise, Blavette played a key role in the evolution of the atom
probe from a one-dimensional technique into the tomography technique that we
know today. He supervised the development of the multi-anode position-sensitive
detector in the early 1990s and the methodology that enabled three-dimensional
images to be built, atom-by-atom. The instrumentation development is now in the
hands of a highly specialized team at the GPM, who actively works with CAMECA,
a company that has been working alongside the GPM for 20 years on the commer-
cialization and further advances of the technique.
10 1 History of APT and LEAP
The next step in the development of the atom probe was the SAP [6] invented by
Osamu Nishikawa (who also came from Mullers lab) in 1993, Fig. 1.8. The idea
was to use a funnel-shaped electrode positioned close to a surface with a high
applied potential. Any sharp points present in the region the aperture passed over
would cause ions to start evaporating, a fraction of which would pass through the
aperture. The geometry theoretically enabled a radically different approach to
making specimens. Specimens had always been needle shaped, but now Nishikawa
was suggesting microtip arrays. This concept had a major impact on how one could
think about specimens and instruments.
At this time, Kelly was still trying to address the basic conundrum of the 3DAP; it
was a spectacularly successful analytical tool, but it was too slow and had too
narrow of a FOV to be practical for many applications. Also about this time, field
electron emitter arrays were being developed that had field enhancement factors of
up to 10 or more due to proximal electrodes [31]. This field enhancement
1.2 Ancestry of the Local Electrode Atom Probe 11
Fig. 1.8 Schematic of the scanning atom probe (reprinted with permission from Elsevier [6])
(producing the same field using a lower applied voltage) provided a means to
achieve higher data collection rates. This all came together in two key thoughts:
(1) a local electrode could be used to lower voltages and attain high data collection
rates without an electron beam and (2) a relative reduction in the energy spread
(due to the voltage reduction) made it possible to achieve high mass resolution over
the entire FOV [4]. So, a local-electrode geometry, like that in Fig. 1.9, actually
mitigated or solved the three biggest issues with the atom probe of the day:
specimen geometry, data collection rate, and mass resolution over a large FOV.
Based on these ideas, in 1994 Thomas Kelly, David Larson, Patrick Camus, and
Louis Holzman incorporated Millennium Instruments. Kelly and Larson carried this
forward, but it was not until 1998 that the first full-time employee, Tye Gribb, was
hired. From the outset, Millennium intended to develop laser pulsing to provide an
instrument that could be used for the analysis of semiconducting materials.
The group raised capital and hired additional people. It took about 2 years to build
the voltage pulsed alpha prototype, Fig. 1.10a. The alpha prototype had two
ultra-high vacuum SEMs mounted on the side for positioning the tip relative to the
local electrode. After the turn of the millennium, the company became Imago
Scientific Instruments. In 2002, Michael Miller at Oak Ridge National Laboratory
purchased the voltage pulsed beta prototype LEAP, which was delivered in 2003.
Imago started working on laser pulsed LEAP in earnest in 2002 and received a grant
in 2003 from the US National Science Foundation [32] to develop laser pulsing.
Eventually Imago shipped the first commercial laser pulsed atom probe in 2006.
12 1 History of APT and LEAP
Also in 2006, Imago acquired Oxford Nanoscience which had developed a novel
curved reflectron [33] for large-FOV imaging in voltage pulsing and laser pulsing
modes. In 2010, Imago was acquired by AMETEK1 and was made part of
CAMECA SAS. A photograph of todays LEAP 4000X is shown in Fig. 1.10b for
comparison with the original alpha prototype.
In order for all of these developments to be realized, many things needed to
happen at the right time. If any one of them did not happen, the APT reality today
might be very different. The Imago team was contemplating generating terabytes of
data in 1998, which was a daunting thought at the time. The CPUs and the graphical
processing units back then were not capable of processing, visualizing, and ana-
lyzing a hundred-million-atom dataset. The only known way to cool a specimen
to 20 K without vibration was using liquid helium, which was a less than desirable
option. Fortunately, commercial vibration-isolated mechanical refrigeration sys-
tems appeared in 2000 just in time for the needs of this instrument. The lasers in the
early 2000s were not stable enough for a commercial atom probe offering. The first
suitable laser for small-spot laser pulsing became commercially available about
1 year prior to adoption and commercial FIB tools did not exist 20 years ago. Given
all of this, we now identify three crucial technologies that contributed to the current
status of APT: the local-electrode geometry, FIB-based specimen preparation, and
thermal pulsing with lasers.
1.2 Ancestry of the Local Electrode Atom Probe 13
Fig. 1.10 (a) Photograph of the alpha prototype of the commercial LEAP at Imago Scientific
Instruments (2001). (b) Photograph of a current LEAP 4000X system for comparison with the
alpha prototype (reprinted with the permission of Cambridge University Press [8])
Historically, voltage pulses have been used to generate pulsed field evaporation
[1]. This requires that the specimen has high electrical conductivity (>102 S/cm),
which limits the technique to metals and heavily doped semiconductors [37].
Alternatively, short laser pulses (<1 ns) directed at the specimen apex can be
used to induce a pulse in the evaporation rate for any material regardless of
electrical conductivity [38]. For proper quantitative operation, the DC voltage on
the specimen should be low enough to prevent evaporation between laser pulses.
Unlike the case of laser ionization mass analysis (LIMA), the temperature rise
required for laser pulsed APT can be small: typically 100300 K [39, 40]. Data
from Kellogg [41] show that for a given evaporation rate, the evaporation field
decreases as temperature is increased. There have also been suggestions that laser
pulsing may invoke athermal field evaporation pulsing mechanisms [42], but
experimental evidence has shown that thermal pulsing dominates the process
[4345].
Experiments performed in the 1980s by Kellogg and Tsong demonstrated that
semiconductor materials and metals could be analyzed with laser pulsing [38, 39,
4648]. The technique was not widely applied because lasers in 1980 were slow,
expensive, and unstable. The first 30 years of atom probe applications were
dominated by metals research using voltage pulsing. Eventually, given the
compelling reasons for using laser pulsing to analyze insulating materials coupled
with advances in laser technology, laser thermal pulsing caught on. The first
commercial laser atom probes were offered in 2005 by all of the commercial
16 1 History of APT and LEAP
vendors at the time. Now nearly every commercial atom probe being sold
incorporates a laser. The advent of faster laser systems and well-engineered com-
mercial atom probe instruments has made laser pulsed operation similar to that of
voltage pulsing from a users point of view. Laser pulsing opens a broad universe of
applications available for study by atom probe; examples include silicon-based
semiconductors and compound semiconductors [7, 49], multilayer thin films [50],
ceramics [5154], and organics [55, 56] (note that the authors have not attempted to
be complete here in referencing the various applications; a more complete review of
recent applications of APT is provided in Chap. 7). In addition, laser pulsing
produces a very low energy spread from the ions. The result is very high MRP
for long flight times. Indeed, Liu and Tsong [57] built an eight meter pulsed laser
atom probe that achieved an MRP of 4,000.
A main challenge in developing a laser pulsed atom probe was to create thermal
pulses of sub-nanosecond duration at greater than 100 kHz repetition rate that
would work with poor thermal conductors. By focusing the laser energy into a
diffraction-limited spot of less than 3 m diameter (4), a very small volume of the
specimen is heated. This makes it possible for the specimen to cool quickly and
results in an evaporation window in time (known as time-of-departure spread) of
less than 200 ps. It also makes it possible to analyze materials with poor thermal
diffusivity with relatively good MRP.
Another recent development is the migration to shorter wavelengths in commer-
cial atom probe systems. Early systems used a laser fundamental wavelength of
either 800 nm (titanium sapphire or Ti:sapphire) or 1,064 nm (neodymium-doped
yttrium aluminum garnet or Nd:YAG). Systems progressed to a frequency-doubled
green wavelength of 532 nm (Nd:YAG), and in 2009 frequency-tripled ultraviolet
wavelengths of 355 nm [58] or 343 nm [44] (Nd:YAG) were introduced commer-
cially. One reason for utilizing the shorter wavelengths is that the absorption depth
(anomalous skin depth) decreases as the cube of the wavelength, so in addition to a
smaller diffraction-limited spot, a shallower heated depth is also produced. These
advances result in smaller heated volumes and more rapid cooling in the apex of the
specimen. In all these cases, rapid cooling is the key to achieving the small time-of-
departure spread, which translates into high MRP [43].
Because it uses a relatively small, focused laser beam and a local electrode,
CAMECAs LEAP 4000X and its predecessor, the LEAP 3000X, are currently
the only laser pulsed atom probes with the capability to analyze microtip arrays.
The importance of this ability lies in the way specimens may be configured. As
shown below in Fig. 1.12a, arrays of microtips may be fabricated, for example, by
micro-electro-mechanical system (MEMS) processing methods for silicon. Such
microtip arrays can serve many purposes: (1) they can be analyzed directly after
some form of processing (thermal, implantation, etc.) (Fig. 1.12b); (2) they can be
coated with thin films of inorganic or organic materials (Fig. 1.12b); and (3) they
can serve as holders for extractions from wafers or other bulk materials (Fig. 1.12c).
Material extraction methods (lift-out) make it possible to prepare specimens from a
very small volume (<1 m3) and from a wide variety of sample types with excellent
reproducibility. As discussed in Chap. 2, small material wedge extractions can be
1.3 The State of Instrumentation 17
Fig. 1.12 SEM images of a microtip array manufactured by MEMS processing methods. (a)
Microtip array coupon, (b) a single microtip, (c) a flat-top microtip with lift-out specimen in
position for bonding, and (d) the apex of a pre-sharpened microtip
mounted on a microtip and an entire array of microtips populated for analysis. Most
of the applications described in this book were prepared via lift-out of material and
subsequent transfer to microtip carrier arrays.
Given that energy spread may be greater than a few electron volts, even in laser
pulsing, the options for improving peak discrimination are energy compensation
and energy determination. Energy compensation seeks to equalize ion flight time by
causing ions of higher energy to take longer pathways to the detector. When done
correctly, all ions with the same mass-to-charge-state ratio have the same flight
time. This approach makes it possible to achieve high MRP in voltage pulsing
mode as well as in laser pulsing mode even when large energy spreads exist.
Mamyrin et al. [25] designed the first reflectrons for general mass spectrometry
following a concept proposed by Alikhanov [24]. Waugh et al. [62] first used
reflectrons in APT for 1D atom probe instruments. Cerezo et al. [63, 64] adapted
reflectrons to an imaging application in 3D atom probes within a few years. All of
these reflectrons used planar electric fields, which became a problem in imaging
instruments because, although they create an output surface with reduced timing
spread, this comes at a cost of introducing spatial aberrations for ions of different
energychromatic aberration. The FOV was therefore restricted in reflectron-
based instruments to small values (~25 nm diameter) to limit this effect. Panayi
[33] solved this problem by designing a curved reflectron in which the output
surfaces for time-focused ions and spatial-focused ions are made to nearly coincide.
As a result, the FOV of curved reflectrons can be made approximately an order of
magnitude larger (>150 nm). These curved reflectrons were adapted to a commer-
cial remote-electrode atom probe [65] and are currently used in the LEAP 4000 HR
instrument [66].
1.5 Concluding Remarks 19
The second notable component of the significant recent advances in the field of APT
is fabricating appropriate samples from disparate materials (which is discussed in
detail in Chap. 2). Historically, researchers have always desired site-specific sample
preparation so that they could know, in advance, what was contained within the
specimen apex. About 20 years ago Alexander and colleagues at Oak Ridge National
Laboratory used a Gatan precision ion milling system to modify specimens
for APT [67]. The early instruments did not have the performance to meet
APT requirements, but they were the precursor to the FIB work that came in the
late 1990s.
Two FIB-based methods were proposed by Larson et al. [68, 69] in 1998 for
APT specimen preparation. In the first method, ions are scanned in an annular
pattern and the pattern size is subsequently made smaller and smaller until a sharp
needle is formed. A second method involves a sort of whittling away at the sample
until a needle is formed. The first method has become the standard method and is
used by most of the groups around the world; however, neither of these methods
allowed for site-specific sample preparation. About a half-dozen years later, in the
mid-2000s, there began a capitalization on the transmission electron microscopy
(TEM) preparation techniques that were being developed in the FIB [70]. These
became known as lift-out methods. Lift-out methods required a micromanipulator,
either external or internal to the FIB, which was used to pick up and attach extracted
specimen pieces onto sample receptacles (carriers).
There are early patent applications from Imago in 2001 which propose removing
a piece from an arbitrary bulk material and then using the FIB for shaping purposes
[71]. In 2005, Miller and colleagues [72, 73] at Oak Ridge National Laboratory
showed that this was feasible, and then in 2006 work by Thompson and colleagues
[74] laid the basic groundwork for the techniques being used today. It was found
early on in the efforts that finishing specimens with a low-ion-energy milling step
was very critical to preparing quality samples [75]. Next came the dual-beam FIB,
which is the current state of the art and enables very-high-quality imaging of the
specimen with electrons (no ion damage), while the FIB is used to shape the sample.
The results of successful application of FIB-based specimen preparation methods
may be seen in the majority of the application examples presented in Chap. 7.
The intent of this brief overview and history of APT is to prepare the reader for
exploration of more intimate details of the technique, especially as they pertain to
operation of the LEAP. The remainder of the book is dedicated to that task. Chapter 2
covers specimen preparation. Electropolishing as well as FIB-based methods
are discussed in sufficient detail to aid novice and advanced practitioners alike.
20 1 History of APT and LEAP
A section discussing considerations for hybrid TEM/APT is also included since this
is an area of active interest. Chapter 3 covers instrumentation specifically as it
pertains to the LEAP platform. First, the basic physics that governs the atom probe
technique is described followed by discussions related to performance and instru-
ment design philosophy. Then, specific components and systems on the LEAP are
described in detail. Chapter 4 deals with data collection. Topics and specific
checklists have been included to enhance APT experimental design and guide
LEAP users through the steps necessary to operate the LEAP. Chapter 5 details
the current reconstruction algorithm and its options (conversion of raw data into 3D
position and chemical information), provides a discussion to aid users in parameter
selection, and includes a case study to bring some substance to the reconstruction
discussion. Chapter 6 delves into the details of the analysis software Integrated
Visualization and Analysis Software (IVAS). Various features are discussed, and
some of the more complicated functions are described in detail. Finally, Chapter 7
includes summaries of published APT applications highlighting the strength of this
technique to provide nanoscale composition for a variety of material systems.
Our hope is that by gaining an appreciation for the evolution of APT and the
many contributions made by the forbearers of the technique, the present-day
practitioners will be better prepared to apply APT to new material systems and
continue the successful advance of the technique in the decades to come.
References
1. Muller, E.W., Panitz, J.A., McLane, S.B.: The atom-probe field ion microscope. Rev. Sci.
Instrum. 39, 8386 (1968)
2. Panitz, J.A.: The 10 cm atom probe. Rev. Sci. Instrum. 44(8), 10341038 (1973)
3. Cerezo, A., Godfrey, T.J., Smith, G.D.W.: Application of a position-sensitive detector to atom
probe microanalysis. Rev. Sci. Instrum. 59(6), 862866 (1988)
4. Kelly, T.F., Camus, P.P., Larson, D.J., Holzman, L.M., Bajikar, S.S.: On the many advantages
of local-electrode atom probes. Ultramicroscopy 62, 2942 (1996)
5. Kelly, T.F., Larson, D.J.: Local electrode atom probes. Mater. Char. 44(12), 5985 (2000)
6. Nishikawa, O., Kimoto, M.: Toward a scanning atom probe - computer simulation of electric
field. Appl. Surf. Sci. 76/77, 424430 (1994)
7. Larson, D.J., Prosa, T.J., Lawrence, D., Geiser, B.P., Jones, C.M., Kelly, T.F.: Atom probe
tomography for microelectronics. In: Haight, R., Ross, F., Hannon, J. (eds.) Handbook of
Instrumentation and Techniques for Semiconductor Nanostructure Characterization, vol. 2, pp.
407477. World Scientific Publishing, London (2011)
8. Kelly, T.F., Larson, D.J.: The second revolution in atom probe tomography. MRS Bull. 37,
150158 (2012)
9. Bostel, A., Blavette, D., Menand, A., Sarrau, J.M.: Toward a tomographic atom-probe. Colloq.
Phys. 50(C8), 501506 (1989)
10. Thomson, J.J.: Phil. Mag. 44, 293 (1897)
11. Rutherford, E.: The scattering of and particles by matter and the structure of the atom. Phil.
Mag. 21(6), 669678 (1911)
12. Bohr, N.: Phil. Mag. 26, 125 (1913)
13. Bragg, W.L.: The diffraction of short electromagnetic waves by a crystal. Proc. Camb. Phil.
Soc. 17, 4357 (1913)
References 21
14. Fowler, R.H., Nordheim, L.: Electron emission in intense electric fields. Proc. Roy. Soc. Lond.
A 119, 173181 (1928)
15. Muller, E.W.: Z. Tech. Phys. 17, 412 (1936)
16. Gomer, R.: Field Emission and Field Ionization. Harvard University Press, Cambridge, MA
(1961)
17. Muller, E.W.: Z. Phys. 120, 270 (1943)
18. Muller, E.W.: Z. Phys. 131(1), 136142 (1951)
19. Muller, E.W.: Resolution of the atomic structure of a metal surface by the field ion microscope.
J. Appl. Phys. 27(5), 474476 (1956)
20. Melmed, A.J.: Recollections of Erwin Mullers laboratory: the development of FIM
(19511956). Appl. Surf. Sci. 94/95, 1725 (1996)
21. Barofsky, D.F., Muller, E.W.: Mass spectrometric analysis of low temperature field evapora-
tion. Surf. Sci. 10(2), 177196 (1968). doi:10.1016/0039-6028(68)90018-6
22. Panitz, J.A.: Anecdotes from an atom-probe original. Microsc. Microanal. 4(S2), 7475 (1998)
23. Poschenrieder, W.P.: Multiple-focusing time-of-flight mass spectrometers. Part II. TOFMS
with equal energy acceleration. Int. J. Mass. Spectrom. Ion Phys. 9(4), 357373 (1972)
24. Falcone, R., Tirrell, M., Adesida, I., Bonnell, D., Broholm, C., Crabtree, G., Gibson, M.,
Hawker, C., Idzerda, Y., Jeelani, S., Murray, C., Phillips, J., Sinclair, R., Spence, J.: Devel-
oping a vision for the infrastructure and facility needs of the materials community: report of
NSF materials 2022(A Subcommittee of the Mathematical and Physical Sciences Advisory
Committee). In. US National Science Foundation (2012)
25. Mamyrin, B.A., Karataev, V.I., Shmikk, D.V., Zagulin, V.A.: The mass-reflectron, a new
nonmagnetic time-of-flight mass spectrometer with high resolution. Sov. Phys. JETP 37,
45 (1973)
26. Miller, M.K.: Presentation at the Microbeam Analysis Society Annual Meeting, Albuquerque,
NM (1986)
27. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York, NY (2000)
28. Kelly, T.F., Mancini, D.C., McCarthy, J.J., Zreiba, N.A.: Proposed configurations for a high-
repetition-rate position-sensitive atom probe. Surf. Sci. 246(13), 396407 (1991)
29. Kelly, T.F., Zreiba, N.A., Howell, B.D., Bradley, F.G.: Energy deposition and heat transfer in a
pulse-heated field emission tip at high repetition rates. Surf. Sci. 246(13), 377385 (1991)
30. Larson, D.J., Camus, P.P., Kelly, T.F.: Simulated electron beam trajectories toward a field ion
microscopy specimen. Appl. Surf. Sci. 67(14), 473480 (1993)
31. Spindt, C., Brody, I., Humphrey, L., Westerberg, E.R.: J. Appl. Phys. 47, 5248 (1976)
32. Bunton, J.H., Olson, J.D., Kelly, T.F.: Phase I technical report for grant 0340351, US National
Science Foundation (2005)
33. Panayi, P.: Curved reflectron. Patent UK 0509638.3 and US 60/682,863
34. Nishikawa, O., Akimoto, T., Tsuchiya, T., Yoshimura, T., Ishikawa, Y.: Field ion microscope
with photostimulable phosphor screennew approach for quantitative analysis of FIM
images. Appl. Surf. Sci. 7677, 359366 (1994)
35. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.C.,
Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data from a
commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383 (2004)
36. Bajikar, S.S., Larson, D.J., Kelly, T.F., Camus, P.P.: Magnification and mass resolution in
local-electrode atom probes. Ultramicroscopy 65, 119129 (1996)
37. Sijbrandij, S., Larson, D.J., Miller, M.: Atom probe field ion microscopy of high resistivity
materials. Microsc. Microanal. 4(2), 9091 (1998)
38. Kellogg, G.L., Tsong, T.T.: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys. 51
(2), 11841194 (1980)
22 1 History of APT and LEAP
39. Kellogg, G.L.: Determining the field emitter temperature during laser irradiation in the pulsed
laser atom probe. J. Appl. Phys. 52(6), 53205326 (1981)
40. Cerezo, A., Grovenor, C.R.M., Smith, G.D.W.: Pulsed laser atom probe analysis of semicon-
ductor materials. J. Microsc. 141(2), 155170 (1986)
41. Kellogg, G.L.: Temperature dependence on the silicon field evaporation voltage. Surf. Sci.
124, L55L59 (1982)
42. Gault, B., Vurpillot, F., Vella, A., Gilbert, M., Menand, A., Blavette, D., Deconihout, B.:
Design of a femtosecond laser assisted tomographic atom probe. Rev. Sci. Instrum. 77, 043705
(2006)
43. Bunton, J.H., Olson, J.D., Lenz, D.R., Kelly, T.F.: Advances in pulsed-laser atom probe:
instrument and specimen design for optimum performance. Microsc. Microanal. 13, 418427
(2007)
44. Vurpillot, F., Houard, J., Vella, A., Deconihout, B.: Thermal response of a field emitter
subjected to ultra-fast laser illumination. J. Phys. D Appl. Phys. 42, 125502 (2009)
45. Vella, A., Vurpillot, F., Gault, B., Menand, A., Deconihout, B.: Evidence of field evaporation
assisted by nonlinear optical rectification induced by ultrafast laser. Phys. Rev. B 73, 165416
(2006)
46. Liddle, J.A., Norman, A., Cerezo, A., Grovenor, C.R.M.: Application of position sensitive
atom probe to the study of the microchemistry and morphology of quantum well interfaces.
Appl. Phys. Lett. 54, 1555 (1989)
47. Grovenor, C.R.M., Cerezo, A.: Atom probe studies of the composition of low-temperature
oxides on (100) silicon and gallium arsenide surfaces. J. Appl. Phys. 65, 5089 (1989)
48. Mackenzie, R.A.D., Liddle, J.A., Grovenor, C.R.M.: Compositional homogeneity of
metalorganic chemical vapor deposition grown III-V compound semiconductor epilayers.
J. Appl. Phys. 69, 250 (1991)
49. Lauhon, L.J., Adusumilli, P., Ronsheim, P., Flaitz, P.L., Lawrence, D.: Atom-probe tomogra-
phy of semiconductor materials and device structures. MRS Bull. 34(10), 738743 (2009)
50. Larson, D.J., Cerezo, A., Juraszek, J., Hono, K., Schmitz, G.: Atom-probe tomographic studies
of thin films and multilayers. MRS Bull. 34(10), 732737 (2009)
51. Larson, D.J., Alvis, R.A., Lawrence, D.F., Prosa, T.J., Ulfig, R.M., Reinhard, D.A.,
Clifton, P.H., Gerstl, S.S.A., Bunton, J.H., Lenz, D.R., Kelly, T.F., Stiller, K.: Analysis of
bulk dielectrics with atom probe tomography. Microsc. Microanal. 14(S2), 12541255 (2008)
52. Chen, Y.M., Ohkubo, T., Kodzuka, M., Morita, K., Hono, K.: Laser-assisted atom probe
analysis of zirconia/spinel nanocomposite ceramics. Scr. Mater. 61, 693696 (2009)
53. Ohkubo, T., Chen, Y.M., Nogiwa, K., Nishimura, A., Hono, K.: Quantitative femtosecond
laser atom probe analysis of rare earth magnets. Microsc. Microanal. 15(S2), 310311 (2009)
54. Marquis, E.A., Yahya, N.A., Larson, D.J., Miller, M.K., Todd, R.I.: Probing the improbable:
imaging carbon atoms in alumina. Mater. Today 13(10), 4244 (2010)
55. Kelly, T.F., Nishikawa, O., Panitz, J.A., Prosa, T.J.: Prospects for nanobiology with atom-
probe tomography. MRS Bull. 34(10), 744749 (2009)
56. Prosa, T.J., Greene, M., Kelly, T.F., Fu, J., Narayan, K., Subramaniam, S.: Atom probe
tomography of mammalian cells: advances in specimen preparation. Microsc. Microanal. 16,
482483 (2010)
57. Liu, J., Wu, C.-w., Tsong, T.T.: Measurement of the atomic site specific binding energy of
surface atoms of metals and alloys. Surf. Sci. 246(13), 157162 (1991)
58. Bunton, J.H., Olson, J.D., Lenz, D.R., Larson, D.J., Kelly, T.F.: Optimized laser thermal
pulsing of atom probe tomography: LEAP 4000X. Microsc. Microanal. 16(S2), 1011 (2010)
59. Blavette, D., Al Kassab, T., Cadel, E., Mackel, A., Vurpillot, F., Gilbert, M., Cojocaru, O.,
Deconihout, B.: Laser-assisted atom probe tomography and nanosciences. Int. J. Mater. Res.
99(5), 454460 (2008). doi:10.3139/146.101672
60. Bostel, A., Renaud, L., Yavor, M., Deconihout, B.: Sonde tomographique grand angle a haute
resolution. Cameca Soc Par Actions Simplif April 2009: FR2922350
References 23
61. Yavor, M., Bostel, A.: Sonde atomique a haute acceptance. Cameca May 2009: WO 2009/
065938
62. Waugh, A.R., Richardson, C.H., Jenkins, R.: APFIM 200 - a reflectron-based atom probe. Surf.
Sci. 266, 501505 (1992)
63. Cerezo, A., Godfrey, T.J., Sijbrandij, S., Smith, G.D.W., Warren, P.J.: Performance of an
energy compensated three-dimensional atom probe. Rev. Sci. Instrum. 1(69), 4958 (1998)
64. Sijbrandij, S., Cerezo, A., Godfrey, T.J., Smith, G.D.W.: Improvements in the mass resolution
of the three-dimensional atom probe. Appl. Surf. Sci. 94/95, 428433 (1996)
65. Cerezo, A., Clifton, P.H., Lozano-Perez, S., Panayi, P., Sha, G.: Overview: recent progress in
three-dimensional atom probe instruments and applications. Microsc. Microanal. 13, 408417
(2007)
66. Clifton, P.H., Gribb, T.J., Gerstl, S.S.A., Ulfig, R.M., Larson, D.J.: Performance advantages of
a modern, ultra-high mass resolution atom probe. Microsc. Microanal. 14(S2), 454455 (2008)
67. Alexander, K.B., Angelini, P., Miller, M.K.: Precision Ion Milling of Field-Ion Specimens.
J. Phys. Colloq. 50(C8), 549554 (1989)
68. Larson, D.J., Foord, D.T., Petford-Long, A.K., Anthony, T.C., Rozdilsky, I.M., Cerezo, A.,
Smith, G.D.W.: Focused ion-beam milling for field-ion specimen preparation: preliminary
investigations. Ultramicroscopy 75, 147159 (1998)
69. Larson, D.J., Foord, D.T., Petford-Long, A.K., Liew, H., Blamire, M.G., Cerezo, A., Smith,
G.D.W.: Field-ion specimen preparation using focused ion-beam milling. Ultramicroscopy 79,
287293 (1999)
70. Giannuzzi, L.A., Drown, J.I., Brown, S.R., Irwin, R.B., Stevie, F.A.: Focused ion beam milling
and micromanipulation lift-out for site specific cross-section TEM specimen preparation.
Mater. Res. Soc. Symp. Proc. 480, 1927 (1997)
71. Kelly, T.F., Martens, R.L., Goodman, S.L.: Methods of sampling specimens for microanalysis.
United States Patent 6,576,900, June 10, 2003
72. Miller, M.K., Russell, K.F., Thompson, G.B.: Strategies for fabricating atom probe specimens
with a dual beam FIB. Ultramicroscopy 102, 287298 (2005)
73. Miller, M.K.: Sculpting needle-shaped atom probe specimens with a dual beam FIB. Microsc.
Microanal. 11(S2), 808 (2005)
74. Thompson, K., Lawrence, D.J., Larson, D.J., Olson, J.D., Kelly, T.F., Gorman, B.: In-situ site-
specific specimen preparation for atom probe tomography. Ultramicroscopy 107, 131139
(2007)
75. Thompson, K., Gorman, B.P., Larson, D.J., van Leer, B., Hong, L.: Minimization of Ga
induced FIB damage using low energy clean-up. Microsc. Microanal. 12(S2), 1736CD (2006)
Chapter 2
Specimen Preparation
2.1 Introduction
In 1965 Muller, the father of the atom probe, once observed, . . . the specimen
itself is also the image-forming lens and the imaging ion beams originate at the
specimen surface [1]. In other words, the specimen is the primary optic of the atom
probe microscope, and as such, specimen preparation forms a critical step in a
successful atom probe tomography (APT) analysis. Controlling the tip size and its
shape is essential for understanding and manipulating the ionic trajectories from
specimen surface to detector.
The motivation for nearly every specimen prepared for APT analysis is funda-
mentally the same: (1) the specimen must be sharp enough to allow field evapora-
tion, (2) it must be robust enough to allow for significant evaporation events to
occur, and (3) user-defined features of interest must be present in the near-apex
region to ensure that such features are included in the collected data. Two main
methods of specimen preparation are prevalent for APT: electropolishing and
focused ion beam (FIB) milling, which are the focus of this chapter.
2.2 Electropolishing
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 25
DOI 10.1007/978-1-4614-8721-0_2, Springer Science+Business Media New York 2013
26 2 Specimen Preparation
inception of high field work in the 1930s. General information on the techniques
can be found in any reference book on the atom probe technique [39], and a large
number of options for electropolishing and chemical polishing solutions may be
found [2, 10].
If bulk samples do not already exist as wires, then blanks must first be cut from
the bulk into appropriately sized pieces (~0.5 0.5 1020 mm3 is common).
This can be accomplished by using a diamond or a rotating wire saw or by electro-
discharge machining. An example of two finished electropolishing blanks is shown
in Fig. 2.1. It is important that the cross section is square or circular in nature as
asymmetric cross sections will lead to blade-shaped specimens that distort the ion
trajectories toward the detector.
The first electropolishing step involves removing material from the middle of a
blank in order to thin that region for separation and to form two roughly needle-
shaped specimens. This is done using a simple setup as shown schematically in
Fig. 2.2a. The blank is connected to a positive DC voltage (although AC is also
often employed), and a metal electrode is connected to the negative polarity and
placed inside a container with a dense inert liquid (such as Galden PFPE heat
transfer fluid) and a layer of electrolyte as shown in the schematic. The electrolyte
layer should be just thick enough to be continuous. The inert liquid is optional,
but dramatically reduces the volume of electrolyte required, and thus lowers storage
and disposal costs. For many ferrous or nickel-based metals and alloys, the standard
electropolishing solution used in this step consists of 10 % perchloric acid in
acetic acid by volume (the user is referred to standard reference books for other
electropolishing and chemical polishing solutions for various materials) [3].
The voltage is then switched on, and the specimen is manually moved up and
down as indicated by the dashed arrow in the figure. During this stage, the sample
may be easily held in locking tweezers or held by other mechanical means, while
the sample blank thins down from its original diameter (the largest amount of
thinning occurs in the center of the blank) and eventually separates into two pieces
2.2 Electropolishing 27
Fig. 2.2 Schematic of two types of electropolishing setups. (a) The suspended electrolyte setup
consists of a relatively narrow film of active electrolyte floating on an inert liquid. A counter
electrode is connected to the specimen blank via a potential difference. The specimen can be
manipulated to control the electropolishing action (dashed vertical arrow). The process ends when
the center of the blank necks until breaking (two-piece image immediately right of panel). This
setup is typically used for rough polishing. (b) An alternative setup that provides the user with finer
control includes a polishing loop that holds the electrolyte and is connected to the specimen via a
potential difference which can be manipulated by motion of the specimen (dashed horizontal
arrow) and gating of the electrical potential (switch)
as shown in Fig. 2.2a, right. At the completion of this process, both samples should
be rinsed in distilled water and ethanol to remove any residual electrolyte and
prevent deposition of any residue that might be dissolved in water remaining on the
tip surface. The blank material can then be mounted into a small copper (or other
metal) tube by crimping to enable easier handling and transfer for APT analysis.
It should be noted however that after completion of this coarse polishing process,
the specimens may not be sufficiently sharp for APT. This is accomplished in a
second stage (often referred to as micropolishing), which is shown schematically in
Fig. 2.2b.
The micropolishing stage forms the specimen into its final shape and may be
accomplished in a variety of ways. The application of this step requires a dynamic
application of voltage between electrolyte and specimen based on observed pro-
gression of the tip shape by the user [10]. The necking method, already described in
the preceding paragraph, can also be used during the micropolishing stage but at a
finer scale and with a smaller electrical potential. This procedure is used with the
aid of an optical microscope, as shown in Fig. 2.3. The specimen, which has a very
small shank angle, is positioned within a metal loop (negative polarity) filled with
electrolyte as shown schematically in Fig. 2.4. A very fine neck is created in the
specimen by moving the metal loop or the specimen back and forth (or up and
down, depending on the orientation of the apparatus) in the loop and leaving a small
region of the specimen outside of the loop (arrowed), Fig. 2.4b. For the final step of
the procedure, the specimen is positioned as shown in Fig. 2.4c (with the loop well
away from the necked region), the potential is switched on, and the specimen is
28 2 Specimen Preparation
Fig. 2.3 Photograph of an actual electropolishing setup (also see Fig. 2.2b). The principal
components include microscope, power supply, and switch (specimen and electropolishing loop
are located beneath the microscope optic)
Fig. 2.4 Schematic illustrating evolution in the micropolishing procedure. (a) Initial placement of
the electropolishing loop on a rough-polished specimen. (b) Typical motion of the loop (dashed
arrows) which produces necking near the end of the tip (arrow). (c) Continued motion of the loop
during application of the electric potential (dashed arrow) causing further necking near the end of
the tip. (d) Final tip shape after the process has caused breaking at the neck
2.3 Needles Versus Microtips 29
quickly withdrawn through the loop (dashed arrow in Fig. 2.4c). If the neck is not
removed, the potential is switched off, the specimen is returned to the position
shown in Fig. 2.4c, and the procedure is repeated until the neck is removed, leaving
a completed specimen shape, Fig. 2.4d. Generally, the specimen is sharpest just
after the neck has been removed, but this can be sharper than ideal for some
low-evaporation-field materials. Additional iterations, similar to those described
in Fig. 2.4c, can result in controlled blunting of the apex until the desired curvature
is attained. Again, the specimen should be rinsed in distilled water and ethanol after
the final shape is achieved. Following the end of APT data collection (either by
blunting or fracture), needle specimens can often be repolished many times using
the steps outlined in this micropolishing procedure.
In addition to the standard manual techniques, automated commercial instru-
ments have emerged [11]. Simplex Scientific (Middleton, WI, USA) offers an
instrument called the Electropointer that automatically prepares needle-shaped
specimens with minimal user input. Once the starting wire or blank is inserted
into the apparatus, the instrument automatically controls the voltage and current
is applied to a small wire loop surrounding the specimen normal to its long axis.
The loop is moved longitudinally by a stepper motor. All motions and electrical
signals are controlled and monitored by computer. There is also an integrated light
optical microscope that displays the magnified specimen shape during operation.
The intent of this instrument is to take as input a coarse specimen blank and output a
specimen that is ready for APT analysis. Feasibility for this approach has been
shown for simple metals [11].
The implementation of a local counter electrode in APT enables the use of microtip
arrays or coupons as a means to quickly manufacture extracted material
into APT-compatible specimens, an example of which is shown in Fig. 2.5.
Microtip coupons offer numerous advantages over traditional needle geometry
specimens [12]. The presence of many microtips per coupon makes it easier to
collect a statistically significant number of datasets in a short amount of time, under
very similar processing conditions, and without the need to cycle new specimens
into the vacuum chamber. This improvement of speed and precision not only
improves time to knowledge for typical applications, but it also allows implemen-
tation of a design-of experiment methodology.
The microtip coupon shown in Fig. 2.5 contains an array of microtip specimen
carriers oriented normal to the wafer surface. This provides a common height for all
the tips and offers two additional advantages: (1) APT analysis through multilayer
films is normal to the wafer surface (each carrier tip has exactly the same orienta-
tion, whereas a group of needles are not guaranteed to be identically straight), and
(2) it minimizes ion beam realignment when moving among specimens when FIB
30 2 Specimen Preparation
Fig. 2.5 SEM image of a microtip array (coupon). The coupons are 3 mm 7 mm with Eleven
carrier microtips positioned along two of the edges (22 carrier tips total). Three additional large
markers are visible more toward the middle of the coupon. These indicate sides 1 and 2 for
reference purpose
sharpening multiple specimens (required when carrier tips are not identical in
orientation and height).
Pre-sharpened microtips [13], with a tip radius of curvature 50 nm, offer an
additional advantage in that they do not require FIB sharpening. This array type is
designed to be run as a calibration standard for evaluating instrument performance
or to serve as a substrate for deposition of films for APT analysis. Elimination of the
need for any FIB processing also alleviates the detriments associated with gallium
contamination and knock-on or ion-mixing damage of the stack during FIB sharp-
ening [14] but with the caveat that the films are deposited on a highly curved
surface. Furthermore, the samples are ready for analysis immediately upon com-
pletion of film deposition.
The sharp tip geometry that makes it possible to produce electric fields sufficient to
field evaporate material from APT specimen surfaces also makes the tips suscep-
tible to undesirable electric field-induced issues during handling and transfer.
Electrostatic discharge (ESD) is a single-event, rapid transfer of electrostatic charge
between two objects, usually resulting when two objects at different potentials
come into direct contact with each other. ESD can also occur when a high
electrostatic field develops between two objects in close proximity. ESD is a
common problem in the microelectronics industry, and well-known steps can be
taken to minimize the potential for transfer of static charges to an APT specimen
and to protect insulating specimens from natural fields and voltage differences that
2.4 Electrostatic Discharge Considerations 31
Fig. 2.6 TEM images of a tungsten tip that has been modified due to electrostatic discharge.
(a) STEM-HAADF image showing amorphous WOx layer surrounding W core material deposited
through an ESD event (reprinted with permission from Cambridge University Press [16]).
(b) Stainless-steel specimen tip as polished (upper) and after ESD (lower). The outermost part
of the tip has disappeared and an amorphous outgrowth formed on the truncated tip (reprinted with
permission from IOP [17])
may cause damage [15]. Besides destruction of the APT specimen from a discharge
event between a specimen tip and its surroundings, ESD has also been observed to
initiate blunting, sometimes in conjunction with the growth of amorphous material
[1618], Fig. 2.6.
Figure 2.7 shows TEM images of specimens that have not been carefully
handled to minimize the potential for ESD. In Fig. 2.7a, a thick (>50 nm) WOx
layer is seen to have formed on the specimen surface, which is inconsistent with
normal growth expectations based on reactions of clean tungsten with air. In
Fig. 2.7b, we see an example where a substantial region of oxide has formed at
the end of the tip, consuming the previously FIB-targeted region of interest (ROI) in
the specimen. Since most ESD events that result in tip morphology changes are not
observable visually, good ESD tip handling protocols must be used. Continuous
grounding of specimen holders and tweezers during transfer is common sense to
help prevent ESD. When implemented, good practices can have a profound effect
on the ability to manufacture and image APT specimens.
Some consideration must also be taken when shipping and transferring speci-
mens. Non-ESD-compatible plastics can collect large amounts of static charge on
the surface, which is difficult to dissipate and should be avoided. These charges
may promote ESD by being transferred by non-grounded human touch to the
specimen. Conducting specimen carrier covers and tweezers that are grounded
during transfer provide the best protection. Conductive plastic carriers with anti-
static plastic covers have also proven adequate to protect STM tips and APT
microtip arrays during transport.
32 2 Specimen Preparation
Fig. 2.7 Examples of ESD damage. (a) TEM image of a W tip that has been modified due to
electrostatic discharge that results in the growth of WOx. Regions of pure W (dark) and WOx
(light) are observed (reprinted with permission from Cambridge University Press [16]).
(b) Substantial oxide grown on a ferritic alloy after site-specific specimen preparation had been
performed (courtesy D. W. Saxey, University of Western Australia)
Using ions to assist in the preparation of specimens for APT goes back nearly
40 years [19]. Broad ion beams have been used by a variety of researchers [2025]
to assist in specimen preparation, and early efforts were also made using FIB
[26, 27], although the capability of these early instruments was not good enough
for them to be broadly applied. The current generation of FIB instruments devel-
oped in the 1990s, however, has brought an entirely new capability to APT
specimen preparation methodology. The combination of the capability to image
specimens (using ions as the probe and secondary or backscattered electrons as the
detected signal) with the capability to remove substantial amounts of material using
ions provided something never before available: the ability to observe the features
of interest in specimens at high magnification while they are being sharpened. Over
the last decade or so, this capability has drastically modified the volume of
applications to which APT may be successfully applied.
The earliest FIB-based methods used to sharpen specimens for APT relied on
attaching a volume of material to the end of a needle [24, 28, 29] (using non-FIB
methods) and shaping the end form for APT analysis with a FIB [14, 3032]. Opti-
mal annular milling methods were gradually developed that provided for improved
tip shapes [12, 3336]. These methods not only allow the user to see the majority of
the specimen apex volume during preparation but are also generally applicable to
almost all materials. If care is taken during the final steps of preparation, minimal
2.5 Focused Ion Beam Methods 33
ion-induced damage will be present in the specimen ROI [37]. APT lift-out
methods [36, 3841] were adapted from transmission electron microscopy tech-
niques [42]. These methods enabled removal of a small region of material, while
being viewed in the FIB, from nearly any starting structure. This small region of
material eventually forms the apex region of the specimen.
Today, many applications and variations of the standard lift-out and sharpening
methods have been reported [36, 4348]. These include variations enabling analysis
parallel to the original specimen surface (cross-section orientation) [49] or inverted
relative to the original specimen surface (backside orientation) [50, 51]. Details of
these methods are described below.
Fig. 2.8 Typical gallium implantation caused by FIB milling at 5 and 30 keV. (a) Atom map
obtained from APT of a 5 keV-shaped silicon specimen. The dark dots represent individual
gallium atoms. The dashed lines represent the analysis volume used to produce the atomic gallium
composition as a function of depth shown in (b) (reprinted with permission from Elsevier [39]). (b)
A comparison of the gallium composition profile with depth for a specimen shaped with 5 and
30 keV gallium ions (reprinted with permission from Elsevier [39]). (c) Gallium implantation into
SiN following a 5 keV final milling step
and only undamaged volumes are typically retained for analysis. It is important to
minimize gallium implantation as much as possible as damage may lead to
intermixing of phases and regions of different compositions or even turn crystalline
regions amorphous [30, 31, 36]. The damage may also promote the creation of
regions with decreased structural integrity, increasing the chances that a specimen
will fail prematurely under application of high electric field. In either case, mini-
mizing gallium damage in a finished APT specimen is desirable.
For specimens where the original surface material is part of the ROI, it is
desirable to retain some of the capping material in the final specimen shape for
analysis to ensure damage-free analysis volumes and a buried interface. This
requires a cap with strong interfacial adhesion to the native surface [53]. For
microelectronics applications where silicon is the native surface material, common
seed layer materials (which naturally have good adhesion) are good sources as a
protective cap and include nickel, chromium, titanium, and silicon. Other criteria
should be considered as well and are discussed below, but if the specimen
cap/surface interface does not survive application of high electric field, then other
considerations are moot.
Controlled ion milling properties are important for controlling the shape of the
specimen/optic. Both the grain size and relative ion sputtering rates affect the tip
shaping process. Ideally, a capping material possesses a very large grain size (or is
amorphous) so that differential ion milling rates at grain boundaries do not induce
undesired topography into the forming specimen. Extremely fine grains (much
smaller than the tip dimension) produce a similar effect. Capping materials with
sputtering rates similar to or lower than those of the specimen are also desirable.
Evaporation field [54] is another consideration. Evaporation field differences
between layers are known to cause spatial reconstruction aberrations at interfacial
regions [55, 56]. Considering a microelectronics application again, sputtered nickel
2.5 Focused Ion Beam Methods 35
Fig. 2.9 SRIM [52] simulation of gallium ion implantation into silicon using (a) 30 keV and
(b) 5 keV ions. (c) Concentration profile normal to the original silicon surface showing the
implanted gallium ion depth
and polysilicon both have good adhesion and similar evaporation fields (less than
~10 % different) when compared to crystalline silicon and thus make reasonable
choices as capping materials. However, the gallium ion stopping power of
polysilicon is significantly less than that of nickel, and so a thicker cap of
polysilicon would be necessary to provide equivalent protection.
Ion mass peak overlap is another capping consideration. In order to easily
separate out the capping layer from the ROI, overlap of ion peaks from materials
in adjacent layers should be avoided. Materials with a small number of isotopes that
occupy unique regions of the mass spectrum are often considered for this reason
alone. Unfortunately, they commonly suffer serious limitations when considered
against the other capping considerations mentioned above. Gold has a single
36 2 Specimen Preparation
isotope far separated from nearly any expected ROI mass peak but has poor
adhesion and high sputtering yield compared to most materials. Cobalt likewise
has a single isotope and is uniquely placed, especially with regard to other transition
metals (no isotopic overlaps with iron or nickel), but it does not adhere particularly
well to other materials. FIB-deposited materials such as tungsten and platinum (and
even carbon) may adhere relatively well and have mass peaks that are often beyond
the mass-to-charge range of most commonly analyzed materials, but they often
contain high levels of carbon and gallium as a consequence of the carrier gas/ion
deposition process [57] and have an extreme evaporation field, resulting in prema-
ture specimen failure.
Finally, very low deposition rates for a capping material can lead to undesirably
long deposition times. For sputter-based deposition, it is the high sputtering yield
that enables fast deposition, but during gallium milling, the same property limits tip
shaping control as discussed earlier.
Multiple capping materials may also be used to manufacture a multilayer cap
with a combination of desired properties [58, 59]. Again using the microelectronics
example, a thin nickel or chromium layer may be selected for good adhesion to a
crystal silicon surface, a layer of gold or platinum might be included for a highly
visible end-point layer (with secondary electron imaging), gold or silver may be
used because of the high deposition rate, and, finally, platinum has durability under
ion milling (sputtering rate and gallium protection).
For material structures where (1) the ROI can be deduced from surface features,
(2) the ROI resides near the surface, and (3) the analysis direction is intended to
proceed from the top surface down into the material (typically referred to as
top-down or normal orientation), the standard lift-out preparation is appropriate
[36, 39, 40]. Although a number of variations of the general method exist [4346,
48, 6062], the specific steps illustrated below describe the method most commonly
used by the current authors.
Figure 2.10 illustrates the general steps that the user should follow during a
standard lift-out procedure. An optional protective capping layer may be applied
over the entire specimen surface. No additional (i.e., non-FIB deposited) capping
layer is used in the current example because the ROI is sufficiently far below the
original surface of the sample. A FIB-deposited platinum strip is added to protect
the surface and to mark the region to be extracted (arrowed in Fig. 2.10a). The
platinum layer is typically 23 m wide and ~100 nm thick with a length that
depends on the geometry of the ROI and the number of desired specimens to be
made from the extracted sample region.
Material is then removed around the specified ROI to create a wedge-shaped
mass of material to be removed from the surface (Fig. 2.10b, arrows). In this case,
trench 1 is created by tilting the stage to 22 (with respect to the electron beam) and
2.5 Focused Ion Beam Methods 37
Fig. 2.10 The steps involved in a standard FIB lift-out procedure. (a) A FIB-deposited protective
strip is placed over the region of interest. This protective material is often platinum or tungsten.
(b) The material is then removed by ion milling around three sides of the region (arrows) as well as
underneath to produce a long cantilevered wedge of material. (c) The wedge is removed by using
an in situ micromanipulator (attached to the left end of the wedge) and then cutting the wedge free
from the substrate (dashed line). (d) The micromanipulator is used to position the wedge above
the carrier microtip (plan-view). (e) The wedge is attached to the carrier tip surface with
FIB-deposited platinum (arrow) and then cut free from the carrier tip (dashed line) for transfer
to additional microtips. Once propagation of the wedge is complete, the FIB stage is rotated 180
so that a second platinum deposition can be applied to the opposite wedgepost interface of each
mounted post. (f) The final mounted wedge section is then ready for sharpening: The line of the
targeted FinFET is clearly visible along the line of the arrows (reprinted with permission from
Annual Reviews [77])
Once the wedge and manipulator are lifted free of the bulk, the wedge can be
repositioned above a carrier tip (Fig. 2.10d). In this image, the wedge is placed
above a ~2 m diameter, flat-topped, silicon microtip post (the post is centered
within the dashed circle). The microtip posts are manufactured as a 450 m array
with each post having an overall height of ~100 m above a planar surface [13]. This
height and spacing combination is sufficient to allow for rapid transfer of wedge
material to multiple posts and to allow independent LEAP analysis of each post
without field evaporating any neighboring specimens. The wedge is carefully
lowered until it comes into direct contact with the flat surface on the top of the
post. An ~1 m-by-1 m-by 0.5 m platinum patch is deposited at the wedgepost
interface to secure the wedge to the post (Fig. 2.10e, arrow). The region of the wedge
secured to the post is then cut free, as indicated by the dashed line in Fig. 2.10e,
and the wedge is moved to the next post and the procedure is repeated until the
wedge is gone.
Once propagation of the wedge to the microposts is finished, the stage is rotated
180 so that a second platinum patch can be applied to the opposite side of the
wedgepost interface for each mounted post. The final mounted wedge is shown in
Fig. 2.10f, with the ROI located along the line indicated by the arrows. Subsequent
cleaning of the micromanipulator to remove any remaining platinum or wedge
material is recommended before reuse.
The required specimen dimensions for APT are material and instrument dependent,
and the user must consider this fact, as well as the voltage-range limitation at which
an atom probe can operate, when preparing specimens. If the user is planning to
analyze materials that have particularly high evaporation fields (40 V/nm), this
should be taken into account when considering the final specimen radius. In
addition, features of interest which are relatively large need to be positioned
below the initial apex of a specimen. A general rule of thumb is to prepare a
specimen which has a lateral diameter at the ROI depth position of about two times
the lateral dimension of the ROI. Clearly this limits the size of features which may
be totally contained in the field of view of a LEAP analysis.
The process of converting a lift-out wedge into a sub-200 nm diameter sharp
needle is accomplished through a series of annular milling steps [31, 33] followed
by a low-energy FIB cleanup step [37, 39], as shown in Fig. 2.11. Illustrated in
Fig. 2.11a is a linear fin-shaped field-effect transistor (FinFET) structure centered
within the wedge, shown by the arrow. Tip shaping is accomplished by applying an
annular milling pattern (shown schematically at the top of Fig. 2.11b) with constant
outer diameter (~4 m) and a decreasing inner diameter (Fig. 2.11bd) with a beam
current of ~0.28 nA. The milling proceeds from the outer diameter of the pattern to
the inner diameter of the pattern to best control any potential for redeposition of
removed material. The first pattern (Fig. 2.11b) has an inner diameter of ~1.6 m
2.5 Focused Ion Beam Methods 39
Fig. 2.11 The steps involved in FIB sharpening process converting the wedge section in Fig. 2.10
into a specimen for APT analysis. (a) The region of interest is shown by the arrow. (b) The first
annular milling pattern shapes the tip into a cylinder, with the FinFET region clearly visible near
the (c) upper-center portion of the cylinder. The second and third milling patterns produce (c) a
tapered end and (d) a narrowed end. The final sharpening step is performed at low ion energy
(either 2 or 5 keV) and is simply a circular pattern that images the specimen end-on. During this
stage, the specimen is carefully monitored to ensure that the region of interest is positioned near
the apex of the final specimen. The (e) low- and (f) high-magnification images of the specimen that
is then ready for APT (reprinted with permission from Annual Reviews [77])
and produces a long cylindrical shape that proceeds well beyond the platinum-weld
region of the wedge (not shown). The second and third patterns (Fig. 2.11c, d) have
inner diameters of ~0.6 and ~0.3 m, respectively. Each pattern is applied for a
length of time sufficient to achieve the desired tip diameter to a location just beyond
the length of the ROI. The final tip shaping is accomplished with the low-energy
(5 keV) FIB cleanup step described below.
As mentioned previously, the implant and damage region created in silicon by a
30 keV gallium ion beam has been shown to extend significantly into the surface of
the sample, while 25 keV ions limit damage to less than ~5 nm [37]. The goal of
the low-energy (25 keV) milling step is to remove the 30 keV damaged region,
position the apex of the specimen at or slightly above the ROI, and narrow the
bottom section of the ROI to 200 nm or less in width. An annular milling pattern of
>4 m diameter is centered over the tip, and milling proceeds at a reduced beam
current (48 pA in the current example). The diameter of the pattern and beam
current can be adjusted to slow the rate of milling so that the user is able to carefully
control the stopping point. Most current FIB instruments provide live viewing of
ion milling which assists in real-time end-point control. After the final step, the
specimen shape is shown in Fig. 2.11e, f. The total milling time for the final low ion
40 2 Specimen Preparation
energy step is of the order of 60 s. Small variations in total milling time can quickly
change the dimensions of the final tip shape. For example, an additional 510 s of
milling in this case would move the tip apex position below the ROI.
Performing analysis along directions other than from the top surface into the bulk of
the sample can have advantages. For example, analyzing thin films parallel to the
film interface orientation (termed cross-section orientation) serves to both increase
2.5 Focused Ion Beam Methods 41
Fig. 2.12 An example of FIB deprocessing. (a) A rotated image of a transistor region within the
wedge. (b) The original extracted wedge after it has been transferred to a manipulator that has been
rotated by 90 . (c) Specimen wedge after it has been deprocessed to a point near the silicide region
above the transistor. (d) Specimen wedge after it has been further deprocessed to a point just above
the gate oxide region (reprinted with permission from World Scientific Publishing [60])
the volume of the ROI (the film interfaces) and improve the analysis yield by
changing the orientation of the applied stress relative to the interface(s) (see Chap. 4
for a description of the stresses applied to an APT specimen). Figure 2.13 illustrates
a cross-section specimen preparation procedure where the ROI consists of layers of
material positioned at the original sample surface.
In such cases, as mentioned previously, a thick sacrificial cap is first added to the
surface of the sample to protect the ROI and add additional material to the sample
so that the surface or the near-surface ROI can be centered in the final tip [49]. In
this case, ~500 nm of nickel has been deposited on the top surface of the sample
before a standard lift-out procedure is used to create a large lift-out wedge
(~5 m wide).
After attaching the manipulator needle and extracting the wedge from the bulk
sample, the wedge is transferred to a horizontal manipulator with a rotational
degree of freedom (Fig. 2.13a). The wedge is transferred by first attaching it to
the second manipulator with FIB-deposited platinum and then cutting the original
manipulator from the wedge. While attached to the horizontal manipulator, one side
42 2 Specimen Preparation
Fig. 2.13 The steps for performing site-specific lift-out and manipulation for cross-section APT
specimens. (a) The original capped specimen wedge is transferred to a manipulation probe which
allows rotation of the wedge. (b) Original wedge still in normal orientation is FIB cut to expose
cross-section ROI (see arrow) (c) Wedge is rotated by 90 , and the cross-section ROI surface is
capped ready for final preparation (ROI indicated by arrow). (d) A wedge section is mounted to a
post. (e) The mount is trimmed prior to annular milling. (f) The final tip with cross-section ROI
captured at the tip (reprinted with permission from World Scientific Publishing [60])
of the wedge is milled flat, creating a new surface that is at 90 to the original
surface, as shown in Fig. 2.13b (this will become the top surface of the final tip).
The arrow highlights a dark layer below the 500 nm nickel cap which is the location
of the ROI for this example. For site-specific cross-section preparation, this new
surface also needs to be positioned directly adjacent to (above in the final tip) the
feature of interest.
2.5 Focused Ion Beam Methods 43
After this new surface is created, the wedge is rotated back by 90 and an
additional capping layer (~50 nm nickel) is added to the new surface, Fig. 2.13c.
Sufficient capping material on the new surface ensures that the specimen can be
shaped with 30 kV ions and finished with 5 kV ions without exposing any of the
original sample material to gallium ions. Should this second cap be removed before
final specimen preparation is complete, differential milling of the different layers
can cause the tip to form into multiple protrusions.
Once the second capping material has been deposited, the wedge is transferred
back to the original micromanipulator and propagated to microtips in a manner
similar to the standard lift-out method described above (Fig. 2.10). Figure 2.13d
shows a portion of the wedge after it has been successfully propagated to a carrier
tip. Before tip shaping can begin, an effort is made to make the tip more cylindrical.
The portion of the wedge that extended beyond the edge of the microtip is removed
with a rectangular milling pattern before sharpening (Fig. 2.13e). The final speci-
men shape after the standard annular milling and low-energy cleanup steps
described above is shown in Fig. 2.13f.
The standard annular milling process is not quite as straightforward for
cross-section preparation because the ion milling rate of the specimen is likely to
have non-cylindrical symmetry. Any differential ion sputtering between layers will
cause asymmetry in the final shape. Often the center of the milling pattern will need
to be positioned away from the center of the region in anticipation of one side of
the specimen milling at a higher rate. The proper amount of this pattern shift is
usually determined by trial and error. A capping material used on the original
surface chosen to have similar milling properties of the ROI will assist in the
mitigation of this phenomenon.
Analyzing materials starting from within the bulk and proceeding toward the
original sample surface (termed backside) [50, 51] can serve multiple purposes.
Should weak or otherwise problematic materials or interfaces exist between the
sample surface and the ROI, analysis from the backside may allow for these regions
to be avoided. If there is a substantial thickness of insulating material (thermal or
electrical) present beneath the ROI, backside preparation may also prove useful, as
this material can be removed from the final specimen. In addition, specimen shape
distortions, and the resultant effects on data reconstruction, can be affected by the
order of evaporation of regions with different evaporation fields. Performing
analysis both in the top-down and backside orientations can provide meaningful
information to help separate real compositional trends from artifacts [64, 65].
Depending on the location of the ROI, backside specimen preparation may
require application of significant material to the top surface of the sample so that
the overall depth of the wedge remains ~23 m after it has been rotated 180 and
formed into a sharp specimen. As in previous examples, adhesion of the added
material to the original sample surface is very important because that interface will
44 2 Specimen Preparation
Fig. 2.14 An example of backside specimen preparation. The target ROI is the near-surface
region of a primarily silicon material. An ~2 m layer of sacrificial material (silver) has been added
to bury the original surface. (a) The FIB is used to cut a trench into this composite material to
expose the original surface within the depth of the sample. The interface is observed ~2 m below
the surface. (b, c) A larger-than-normal wedge is extracted from the host material with a
micromanipulator. (d) After transfer to a second micromanipulator with a rotational degree of
freedom, the wedge is rotated by 90 and the material is removed to expose the new uppermost
surface (arrow). (e) The wedge is rotated another 90 so that the uppermost surface is properly
oriented. (f) The wedge is reshaped so that the former top part of the original wedge has a
triangular shape to enable attachment via FIB-deposited platinum. (g) An image of the upside-
down, reshaped wedge after attachment to a microtip post using FIB-deposited platinum. (h) Final
tip shape after all annular milling steps have been completed
reside near the apex of the final specimen. One successful recipe used a palladium
seed layer followed by a silver layer ~3 m thick to provide the necessary thickness
for the backside procedure [51, 65]. The palladium provides good adhesion to the
top metallic surface and to the silver [66], while the use of silver as a capping layer
allows for very fast deposition of the 3 m of material and good electrical and
thermal conduction. The actual steps of the backside specimen preparation proce-
dure are similar to those for a cross-section specimen.
Figure 2.14 illustrates the backside specimen preparation process for a ROI that
resides near the sample surface. First the top surface is sufficiently buried (~2 m
silver with a palladium seed). Then, a trench is cut into the surface of the sample to
expose the depth of the material for SEM imaging. The buried interface is clearly
visible in Fig. 2.14a, confirming that the target ROI resides some 2 m below the
surface. Similar to cross-section specimen preparation, a larger-than-normal wedge
(~5 m wide) of material is extracted and transferred to a micromanipulator with
a rotational degree of freedom (Fig. 2.14b, c). The goal of the next set of steps is
to place the ROI near the top of a new wedge with triangular cross section.
To accomplish this, the wedge (Fig. 2.14c) is rotated 90 and material is removed
(arrow in Fig. 2.14d) with the FIB so that the ROI is near the new surface. Then,
the wedge is rotated another 90 so that the original wedge is upside down
(Fig. 2.14e). At this point an optional second capping layer may be added
to the newly created top surface to protect it during the annular milling process.
2.6 Hybrid Transmission Electron Microscopy/Atom Probe Tomography 45
Fig. 2.15 An example of the mortis and tenon method for wedge attachment to a carrier post.
(a) First a keyhole is milled through the object that is being attached to the carrier post (arrow).
This is commonly performed while the wedge is still attached to the micromanipulator (not
illustrated here). (b) Next the keyhole is filled with FIB-deposited platinum using a pattern of
the same size used in the milling step. (c) Because this wedge surface is oriented 90 to the carrier
surface, it is possible to make an additional FIB-deposited platinum weld to confidently secure the
lift-out to the carrier. Other wedge surface/carrier orientations may only allow for attachment
through the keyhole
To complete the wedge-shaping process, the stage is tilted to 22 with respect to the
electron column and each side of the wedge is trimmed until the triangular cross
section is formed (Fig. 2.14f). This shaping is necessary so that the FIB-deposited
platinum can access the intersection of the wedge and carrier tip with zero stage tilt.
Alternative adhesion strategies, such as the notch or mortise and tenon weld (see
Fig. 2.15) [40], can be used which do not require a triangular wedge cross section to
enable FIB adhesion. Finally, the wedge is transferred back to the mounting
micromanipulator, and the wedge material is transferred to carrier posts
(Fig. 2.14g) and annularly milled to its final tip shape (Fig. 2.14h).
Transmission electron microscopy (TEM) and APT each has different advantages
and limitations, but together they provide complementary information enabling
more complete characterization of the microstructure and chemistry of a materials
system. Although FIB/SEM images provide some information about the structure
of an APT specimen after the specimen fabrication process, higher resolution
46 2 Specimen Preparation
Fig. 2.17 Fixtures available to handle transfer of wire-shaped specimens. (a) Wire secured in a
1.8 mm copper sleeve. (b) SEM/FIB-compatible holder that can accommodate multiple 1.8 mm
objects. (c) 1.8 mm copper sleeve mounted within a Fischione TEM holder. (d) APT holder that
can accommodate multiple 1.8 mm objects
Wire-shaped specimen geometries are common in APT and are very useful for TEM
because the approximately cylindrical specimen cross section presents a constant
sample thickness for all angles in a TEM tomography acquisition [75]. A wire itself
may be difficult to handle and subsequently mount in a holder, so the wire can
be secured in a thicker fixture such as the crimpable 1.8 mm copper sleeve
shown in Fig. 2.17a. Wires are typically electropolished to a sharp tip so that
they can be analyzed as-polished or taken into the FIB for further processing
either as a specimen or a specimen carrier. Multiple wire specimens can be
loaded into the FIB using a carrier similar to that shown in Fig. 2.17b. A set
screw is used to secure Cu stubs into each hole of the holder. After FIB
processing is complete, wires can be placed into appropriate holders for TEM
analysis. For example, the Fischione (Export, PA, USA) 2050 TEM specimen
holder and LEAP wire specimen puck are shown in Fig. 2.17c, d, respectively.
These examples demonstrate how single-wire specimens can be successfully
shared between FIB, TEM, and APT instruments with minimal difficulty.
Half-grid-shaped specimen geometries have become quite commonplace in
TEM and can be accommodated in the LEAP as well. Advantages include a
geometry where the specimen is protected by the superstructure of the grid, the
fact that multiple tips may be carried on a single grid, and the common availability
of TEM holders compatible with grids in non-APT laboratories. In Fig. 2.18, three
different half-grid-format specimen carriers are shown: (a) a tungsten needle
embedded via the Short-Cut into an Omniprobe (Dallas, TX, USA) copper
half-grid [51]; (b) a wire grid that has been cut in half and electropolished suffi-
ciently sharp for analysis or mounting of lift-out specimens; and (c) a Dune
Sciences (Eugene, OR, USA) silicon half-grid with multiple specimen carrier
regions that are compatible for mounting lift-out specimens. Examples of grid-
compatible TEM holders and their LEAP-compatible specimen pucks are shown in
Fig. 2.19. Figure 2.19a, top, shows a generic TEM grid holder and LEAP puck with
48 2 Specimen Preparation
Fig. 2.18 Three half-grid-format specimen carrier options: (a) needle embedded via the Short-
Cut into an Omniprobe copper half-grid, (b) a wire grid that has been cut in half and
electropolished so that the ends are sharp, and (c) a Dune Sciences silicon half-grid with multiple
carriers (image courtesy of Dune Sciences, Eugene, OR, USA)
holder designed to hold a half-grid. The grid holder shown in the LEAP puck
(Fig. 2.19b, top) serves double duty as a grid carrier for the LEAP, but it can also be
carried into the FIB. With this approach, the user must handle the grid during some
of the transfers between FIB, TEM, and LEAP instruments. The Hummingbird
Scientific grid holder [50], Fig. 2.19a, bottom, consists of a removable end effector
that can be placed in a TEM holder or a LEAP puck. The advantage is that once a
half-grid is mounted in the end effector, it can be moved from FIB to transmission
electron microscope to LEAP without the need for handling the grid directly with
tweezers. Variations on the half-grid theme exist as well. Gorman et al. have used
the Short-Cut to host a micromanipulator needle, which serves as a specimen carrier
with the sharpened lift-out material at its apex [50]. Similar advances have been
made recently by Felfer et al. [76].
2.7 Summary 49
Fig. 2.19 TEM/APT specimen holder options. (a) TEM grid holders (top reprinted with permis-
sion from www.gatan.com; bottom reprinted with permission from Cambridge University Press
[50]) and (b) their equivalent LEAP-compatible specimen carriers. A generic half-grid handling
strategy is shown at the top of the figure, while the Hummingbird removable end effector strategy
is shown at the bottom of the figure
2.7 Summary
References
24. Larson, D.J., Petford-Long, A.K., Cerezo, A.: Field ion microscopy of multilayer film devices.
Microsc. Microanal. 4(S2), 112113 (1998)
25. Larson, D.J., Russel, K., Cerezo, A.: Sharpening of field-ion specimens and positioning of
features of interest by ion-beam milling. J. Vac. Sci. Tech. 18(1), 328333 (2000)
26. Waugh, A.R., Payne, S., Worrall, G.M., Smith, G.D.W.: In situ ion milling of field ion
specimens using a liquid metal ion source. J. Phys. 45(C9), 207209 (1984)
27. Alexander, K.B., Angelini, P., Miller, M.K.: Precision ion milling of field-ion specimens.
J. Phys. Colloq. 50(C8), 549554 (1989)
28. Hasegawa, N., Hono, K., Okano, R., Fujimori, H., Sakurai, T.: A method for preparing atom
probe specimens for nanoscale compositional analysis of metallic thin films. Appl. Surf. Sci.
67(14), 407412 (1993)
29. Martens, R.L., Larson, D.J., Kelly, T.F., Cerezo, A., Clifton, P.H., Tabat, N.: Preparation of
three-dimensional atom probe samples containing multilayer thin film structures using a
focused ion-beam. Microsc. Microanal. 6, 522523 (2000)
30. Larson, D.J., Foord, D.T., Petford-Long, A.K., Cerezo, A., Smith, G.D.W.: Focused ion-beam
specimen preparation for atom probe field-ion microscopy characterization of multilayer film
structures. Nanotechnology 10, 4550 (1999)
31. Larson, D.J., Foord, D.T., Petford-Long, A.K., Anthony, T.C., Rozdilsky, I.M., Cerezo, A.,
Smith, G.D.W.: Focused ion-beam milling for field-ion specimen preparation: preliminary
investigations. Ultramicroscopy 75, 147159 (1998)
32. Larson, D.J., Petford-Long, A.K., Cerezo, A., Smith, G.D.W.: Three-dimensional atom probe
studies of metallic multilayers. Acta Mater. 47(15), 40194024 (1999)
33. Larson, D.J., Foord, D.T., Petford-Long, A.K., Liew, H., Blamire, M.G., Cerezo, A., Smith,
G.D.W.: Field-ion specimen preparation using focused ion-beam milling. Ultramicroscopy 79,
287293 (1999)
34. Larson, D.J., Petford-Long, A.K., Ma, Y.Q., Cerezo, A.: Information storage materials:
nanoscale characterization by three-dimensional atom probe analysis. Acta Mater. 52(10),
28472862 (2004)
35. Miller, M.K.: Sculpting needle-shaped atom probe specimens with a dual beam FIB. Microsc.
Microanal. 11(S2), 808 (2005)
36. Miller, M.K., Russell, K.F., Thompson, G.B.: Strategies for fabricating atom probe specimens
with a dual beam FIB. Ultramicroscopy 102, 287298 (2005)
37. Thompson, K., Gorman, B.P., Larson, D.J., van Leer, B., Hong, L.: Minimization of Ga
induced FIB damage using low energy clean-up. Microsc. Microanal. 12(S2), 1736CD (2006)
38. Kelly, T.F., Martens, R.L., Goodman, S.L.: Methods of sampling specimens for microanalysis.
United States Patent 6,576,900, June 10, 2003
39. Thompson, K., Lawrence, D.J., Larson, D.J., Olson, J.D., Kelly, T.F., Gorman, B.: In-situ site-
specific specimen preparation for atom probe tomography. Ultramicroscopy 107, 131139
(2007)
40. Miller, M.K., Russell, K.F., Thompson, K., Alvis, R., Larson, D.J.: Review of atom probe
fib-based specimen preparation methods. Microsc. Microanal. 13, 428436 (2007)
41. Miller, M.K., Russell, K.F.: Atom probe specimen preparation with a dual beam SEM/FIB
miller. Ultramicroscopy 107(9), 761766 (2007)
42. Giannuzzi, L.A., Drown, J.I., Brown, S.R., Irwin, R.B., Stevie, F.A.: Focused ion beam milling
and micromanipulation lift-out for site specific cross-section TEM specimen preparation.
Mater. Res. Soc. Symp. Proc. 480, 1927 (1997)
43. Cairney, J.M., Saxey, D.W., McGrouther, D., Ringer, S.P.: Site-specific specimen preparation
for atom probe tomography of grain boundaries. Phys. B Condens. Mater 394, 267269 (2007)
44. Takahashi, J., Kawakami, K., Yamaguchi, Y., Sugiyama, M.: Development of atom probe
specimen preparation techniques for specific regions in steel materials. Ultramicroscopy 107
(9), 744749 (2007)
45. Saxey, D.W., Cairney, J.M., McGrouther, D., Honma, T., Ringer, S.P.: Atom probe specimen
fabrication methods using a dual FIB/SEM. Ultramicroscopy 107(9), 756760 (2007)
52 2 Specimen Preparation
46. Perez-Willard, F., Wolde-Giorgis, D., Al Kassab, T., Lopez, G.A., Mittemeijer, E.J.,
Kirchheim, R., Gerthsen, D.: Focused ion beam preparation of atom probe specimens
containing a single crystallographically well-defined grain boundary. Micron 39, 4552 (2008)
47. Rachbauer, R., Massl, S., Stergar, E., Felfer, P., Mayrhofer, P.H.: Atom probe specimen
preparation and 3D interfacial study of Ti-Al-N thin films. Surf. Coat. Tech. 204(11),
18111816 (2010). doi:10.1016/j.surfcoat.2009.11.020
48. Felfer, P., Ringer, S.P., Cairney, J.M.: Shaping the lens of the atom probe: fabrication of
site specific, oriented specimens and application to grain boundary analysis. Ultramicroscopy
111(6), 435439 (2011)
49. Lawrence, D., Alvis, R., Olson, D.: Specimen preparation for cross section atom probe
analysis. Microsc. Microanal. 14(S2), 10041005 (2008)
50. Gorman, B.P., Diercks, D., Salmon, N., Stach, E., Amador, G., Hartfield, C.: Hardware and
techniques for cross-correlative TEM and atom probe analysis. Microsc. Today 16(4), 4247
(2008)
51. Prosa, T.J., Lawrence, D., Olson, D., Larson, D.J., Marquis, E.A.: Backside lift-out specimen
preparation: reversing the analysis direction in atom probe tomography. Microsc. Microanal.
15(S2), 298299 (2009)
52. Ziegler, J.: The stopping and range of ions in matter. http://www.srim.org/ (2011).
53. Ohring, M.: The Materials Science of Thin Films. Academic, New York, NY (1992)
54. Tsong, T.T.: Field ion image formation. Surf. Sci. 70, 211233 (1978)
55. Vurpillot, F., Cerezo, A., Blavette, D., Larson, D.J.: Modeling image distortions in 3DAP.
Microsc. Microanal. 10, 384390 (2004)
56. Larson, D.J., Geiser, B.P., Prosa, T.J., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F.: Improve-
ments in planar feature reconstructions in atom probe tomography. J. Microsc. 243, 15 (2011)
57. Gerstl, S.S.A., Morrone, A., Kvitek, R.: Three-dimensional nanoscale characterization of pt
deposition from an organometallic precursor induced by a focused ion beam. Microsc.
Microanal. 12(S2), 12521253 (2006)
58. Mutas, S., Klein, C., Gerstl, S.S.A.: Investigation of the analysis parameters and background
subtraction for high-k materials with atom probe tomography. Ultramicroscopy 111, 546551
(2011)
59. Francois-Saint-Cyr, H.G., Mutas, S., Prosa, T.J., Ulfig, R.M., Klein, C., Lawrence, D., Olson,
D., Larson, D.J.: Assessment of protective coatings used in focused-ion-beam specimen
preparation for atom probe tomography. Abstracts of 53rd International Field Emission
Symposium, 216 (2012)
60. Larson, D.J., Prosa, T.J., Lawrence, D., Geiser, B.P., Jones, C.M., Kelly, T.F.: Atom probe
tomography for microelectronics. In: Haight, R., Ross, F., Hannon, J. (eds.) Handbook of
Instrumentation and Techniques for Semiconductor Nanostructure Characterization, vol. 2, pp.
407477. World Scientific Publishing, London (2011)
61. Larson, D.J., Lawrence, D., Olson, D., Prosa, T.J., Ulfig, R.M., Reinhard, D.A., Clifton, P.C.,
Kelly, T.F., Lefebvre, W.: From the store shelf to device-level atom probe analysis: an exercise
in feasibility. In: 36th International Symposium for Testing and Failuer Analysis, San Jose,
CA, 2011, pp. 189197. ASM International
62. Marquis, E.A., Hu, R., Rousseau, T.: A systematic approach for the study of radiation-induced
segregation/depletion at grain boundaries in steels. J. Nucl. Mater. 413(1), 14 (2011).
doi:10.1016/j.jnucmat.2011.03.023
63. Giannuzzi, L.A., Stevie, F.: Introduction to Focused Ion Beams, Instrumentation, Theory,
Techniques and Practice. Springer, New York, NY (2005)
64. Marquis, E.A., Geiser, B.P., Prosa, T.J., Larson, D.J.: Evolution of tip shape during field
evaporation of complex multilayer structures. J. Microsc. 241(3), 225233 (2011)
65. Larson, D.J., Prosa, T.J., Geiser, B.P., Egelhoff Jr., W.L.: Effect of analysis direction on the
measurement of interfacial mixing in thin metal layers with atom probe tomography.
Ultramicroscopy 111, 506 (2011)
References 53
66. Smith, G.D.W.: Pd as an adhesion layer for Ni and Ag. In. Personal Communication, Oxford,
UK (2009)
67. Gorman, B.P.: Atom probe reconstruction refinements by pre- and post- analysis TEM
structure quantification. Microsc. Microanal. 13(S2), 16161617 (2007)
68. Arslan, I., Marquis, E.A., Homer, M., Hekmaty, M.A., Bartelt, N.C.: Towards better 3-D
reconstructions by combining electron tomography and atom-probe tomography.
Ultramicroscopy 108, 15791585 (2008)
69. Gorman, B.P., Diercks, D., Kaufman, M.J., Ulfig, R.M., Lawrence, D., Thompson, K., Larson,
D.J.: Atomic scale compositional and structural characterization of nanostructured materials
using combined FIB, STEM, and LEAP. Microsc. Microanal. 12(S2), 1720CD (2006)
70. Gorman, B.P., Diercks, D.: Cylindrical specimen geometries for sub-nm 3-D characterization
of semiconductor devices. Microsc. Microanal. 13(S2), 822 (2007)
71. Petersen, T.C., Ringer, S.P.: Electron tomography using a geometric surface-tangent algo-
rithm: application to atom probe specimen morphology. J. Appl. Phys. 105, 103518 (2009)
72. Antcheva, I., Ballintijn, M., Bellenot, B., Biskup, M., Brun, R., Buncic, N., Canal, P., Casadei,
D., Couet, O., Fine, V., Franco, L., Ganis, G., Gheata, A., Gonzalez Maline, D., Goto, M.:
ROOTA C++ framework for petabyte data storage, statistical analysis and visualization.
Comput. Phys. Commun. 180, 24992512 (2009)
73. Williams, C.A., Marquis, E.A., Cerezo, A., Smith, G.D.W.: Nanoscale characterisation of
ODSEurofer 97 steel: an atom-probe tomography study. J. Nucl. Mater. 400, 3745 (2010)
74. Petersen, T.C., Ringer, S.P.: An electron tomography algorithm for reconstructing 3D mor-
phology using surface tangents of projected scattering interfaces. Comput. Phys. Commun.
181, 676 (2010)
75. Giannuzzi, L.A., Schwarz, S.M.: FIB specimen preparation for STEM and EFTEM tomogra-
phy. Microsc. Microanal. 10(S2), 142143 (2004)
76. Felfer, P., Alam, T., Ringer, S.P., Cairney, J.M.: A reproducible method for damage-free site-
specific preparation of atom probe tips from interfaces. Microsc. Res. Tech. 75, 484491
(2012)
77. Kelly, T.F., Larson, D.J.: Atom probe tomography 2012. Annu. Rev. Mater. Res. 42, 131
(2012)
Chapter 3
Design and Instrumentation
3.1 Introduction
This chapter describes the LEAP instrument and its hardware components with
various levels of detail. Following an introductory section, items are grouped as
they are related to the local electrode, elements critical to detection (detection and
imaging), transfer of specimens and materials though the various vacuum chambers
(transfer and storage of consumables), voltage supplies and laser systems (field
evaporation systems), as well as ancillary systems.
The goal of this chapter is for users to (1) realize that there exist fundamental
limitations inherent in the design of an atom probe and (2) become familiar with
major hardware systems that make up the LEAP.
Atom probes work by a process known as field evaporation [1, 2] (see also
Appendix B) where a high electric field at the surface of a material induces
desorption and ionization of atoms even at cryogenic temperatures. At electric
field strengths of 1010 V/m or greater, electrons can be stripped from atoms on
the surface by a tunneling process. The necessary field is created by applying a
modest voltage (510 kV) to a very sharp specimen. The resultant positive ions are
accelerated away by the electric field toward the counter electrode. As discussed in
Chap. 1 and in this chapter, the electric field is proportional to the applied voltage
and inversely proportional to the radius of curvature of the tip of the specimen.
Thus, a specimen with a radius of curvature around 100 nm or less and a voltage
around 10 kV successfully accomplishes this task.
Minimally, an atom probe includes a specimen mount, a counter electrode with
an aperture, and a detector, Fig. 3.1. During analysis, a positive electrical potential
is applied to the specimen, while the counter electrode, located between the
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 55
DOI 10.1007/978-1-4614-8721-0_3, Springer Science+Business Media New York 2013
56 3 Design and Instrumentation
specimen and the detector, is either grounded or negatively biased. A pulse, such as
a positive electrical pulse on the specimen (or negative electrical pulse on the
counter electrode) or a thermal pulse from laser irradiation, is periodically applied
to the specimen to initiate ionization of surface atoms at a known time.
A position-sensitive detector records the hit position of individual ions, the
accumulation of which forms an image of the tip surface and ultimately enables a
direct mapping of the atoms position on the original curved surface of the speci-
men apex. This image is a projection, and because the tip apex is about 100 nm
across while the detector is about 100 mm across, the projected image of atoms on
the specimen surface has a projection magnification of about one million times. At
this magnification, interatomic distances of 0.2 nm become 0.2 mm on the detector,
which is easy to resolve with modern detector technology. Because only surface
atoms are removed, the sequence of evaporation events is used to construct the third
dimension of the dataset. Thus, as the specimen is evaporated, the position1 of the
atom in xy is determined from its hit position on the detector and its z-position
(depth into the specimen along its long axis) is determined from the sequence of
1
This book uses the convention that x, y, and z are the coordinates in the real space structure of a
material, X and Y are coordinates of a position on a detector, and N is the sequence number of an
ion in the evaporated dataset.
3.2 How Atom Probes Work 57
conductivity (>102 S/cm) and work only with metals or other highly conductive
materials. Thermal pulsing has been accomplished by delivering energy with a
pulsed laser [3] and works with a broad range of materials independent of their
electrical conductivity. Semiconductors and insulators are the obvious materials that
benefit from laser pulsing, but metals can also benefit from this form of pulsing.
As a method of mass identification, TOF mass spectroscopy has many advan-
tages including the following:
All atoms and isotopes are detected.
It is not necessary to anticipate which elements to study in advance.
All species are detected with uniform efficiency.
It is possible to achieve high MRP.
The instrumental design parameters that are of highest interest are field of view
(FOV), MRP, and data collection rate. The adoption of a local-electrode geometry
and the availability of reflectron energy compensation provide certain important
advantages and limitations, and these influence other primary design parameters
(such as flight path length) which directly relate back to FOV, MRP, and data
collection rate. It is very difficult to avoid direct trade-offs inherent in choice of
flight path length (MRP versus data collection rate), although systems with and
without reflectron energy compensation provide some positive options. Ultimately,
the straight-flight-path LEAP 3000/4000X Si system is designed to promote the
highest laser mode data collection rates and the highest detection efficiency, while
the energy-compensated LEAP 3000/4000X HR (high resolution) system provides
superior spectral quality (MRP and background) especially under field-pulsing
acquisition conditions.
A large FOV makes it possible to capture larger material volumes with, ideally, no
loss in spatial fidelity. Larger volumes provide an advantage from a statistical
perspective when determining concentrations. In addition, they allow characteriza-
tion of larger scale features and enhance the ease of locating small features (e.g., in
discrete device structures). The first 3D atom probes had limited FOV. Though their
maximum FOV of about 30 nm was a huge improvement over 1D atom probes with
about a 1 nm field, it was insufficient for many material analysis applications. When
the goal is to access as large a FOV as possible, Fig. 3.1 suggests making the
detector size large and the flight path length small. However, the spatial focusing
provided by the reflectron enables longer flight paths without losses in FOV.
3.3 LEAP Performance Parameters 59
In a LEAP 3000/4000 Si, the detector subtends greater than 0.84 steradians
(>60 full angle). This corresponds to a FOV for an aluminum specimen at 10 kV
of greater than 200 nm diameter.
the electric fields converts the divergent ion ray paths to convergent ray paths. The
flight path length of such instruments is relatively long which has two main
ramifications: (1) the MRP in pulsed-laser mode can be very high due to the long
flight times and (2) the maximum pulse repetition rate may have to be limited
(<500 kHz) due to the long flight time.
With laser pulsing, the energy spread will not be dominated by the thermal
energy of the ions (0.020 eV at 250 K) but rather the energy spread inherent in field
evaporation processes. For most materials, this latter term has been found to be 3 eV
or less. Assuming that the energy spread is low (e.g., E/E V/V 3 eV/
10,000 eV 3 104), then tP for laser pulsing (see Appendix D) will be
small. Longer flight times in laser pulsing can therefore lead to greater MRP
[11]. Whether it is a long straight flight path in a LEAP XSi or a long curved flight
path in a LEAP XHR, high MRP may be achieved with laser pulsing.
The data collection rate of LEAPs is high compared with historical norms for atom
probes. Since the data collection rate is the product of the detection rate and the
pulse repetition rate we can look at these two parameters to understand how a high
rate is achieved. The detection rate is typically about 0.01 ions per pulse. This
seemingly low rate is necessary to keep the number of multiple evaporation events
low since most position-sensitive detectors have difficulties encoding multiple hits.
The ability to properly detect multiple hits occurring during a single pulse enables
acquisition at higher average hits per pulse.
Unless new detector technology comes along, the real opportunity to achieve
high data collection rates is to maximize the pulse repetition rate. The maximum
pulse repetition rate is dictated by the longest ion flight times expected in an
experiment, i.e., for the lowest operating voltage and the heaviest ions for a given
flight path. Short flight paths such as in the LEAP Si thus enable high data collection
rates. The presence of a reflectron constrains flight path lengths to larger values with
correspondingly smaller pulse repetition rates.
Laser pulse repetition rates are not a limitation since they can be much higher
than the physical flight times require. However, voltage pulse repetition rates are
limited by the technology of the pulser. The field enhancement made possible by
the local-electrode geometry means that lower voltages are needed for a given field.
Since much higher pulse repetition rates may be achieved at lower voltages, high
data collection rates under field pulsing conditions are achieved in a LEAP.
3.4 Instrumentation of the LEAP 61
The trade-offs inherent in the straight flight path LEAP 3000/4000X Si and curved
reflectron LEAP 3000/4000X HR are briefly summarized in Table 3.1. When high
MRP in field pulsing or for general applications is important (e.g., for detection of
dilute species and identification of chemical species), the reflectron instrument is
the best choice. Note that the detection efficiency of the HR model is about 37 %,
whereas the Si model is about 58 %. In some circumstances, this difference in
detection efficiency can be significant. The most common circumstance is analysis
of small clusters of atoms where higher detection efficiency makes it possible to
detect clusters at smaller size. When high data collection rate is premium, the
straight-flight-path instrument is the best choice. For general-purpose laboratories,
either configuration may prove suitable depending on the specific application
needs.
In the LEAP, an ultra-high vacuum (UHV) analysis chamber houses the detector,
specimen stage, ion pump, laser optics (optional), and reflectron (optional) as well
as the cryogenic and electronic pathways necessary to enable tool operation.
Ancillary chambers and hardware systems enable safe and clean operation. The
following sections provide descriptions of important hardware and systems present
in the LEAP.
The local electrode is the distinguishing component of the LEAP and the source of
its name. The funnel-shaped counter electrode [1216] is used to create the electric
field as illustrated in Fig. 3.1. With a 40 10 m diameter aperture at the apex of
the local electrode, the electric field can be confined or localized to very small
62 3 Design and Instrumentation
Fig. 3.3 Schematic illustration of (a) the electric field, E, on a metal sphere of radius, Ro, opposed
by an outer sphere of radius, RS, with a voltage, V, between them and (b) the electric field of a
metal needle with an apex of radius, Ro, opposed by a counter electrode some distance away, Rc,
with a voltage, V, between them. For both cases, the distance, r, is measured from the center of the
sphere to the bounding counter electrode
features. Though in principle the cathode in an atom probe could be the detector, in
practice a separate counter electrode is always used.
The electric field on a surface depends upon whether the field can penetrate the
surface. In a metal, field penetration may be neglected. In dielectrics, field pene-
tration may largely depend on charge redistribution and available electron energy
states. For the present purposes, we will consider a metal surface with no field
penetration. The easiest geometry to consider is a sphere inside a sphere as shown in
Fig. 3.3a. The magnitude of the electric field vector, E, on the surface of the inner
sphere is simply
V
E 3:1
r
and the orientation is normal to the surface at all points. For a needle with a
spherical apex opposed by a plate, Fig. 3.3b, the electric field magnitude can be
computed from [1] when the distance to the counter electrode RC >> r:
V
E : 3:2
kr
Here k is a constant (greater than unity) which accounts for the fact that the electric
field is reduced in magnitude by the presence of the shank of the needle and the
needle/plane geometry. This latter expression is meant to give the electric field
magnitude normal to the surface at the central axis of the apex. For this example,
k has a magnitude of about 5 [1] and has been found to increase with shank angle,
[1719].
3.4 Instrumentation of the LEAP 63
Fig. 3.5 (a) Schematic of aperture diameter, and hence tip-to-aperture distance, on the field
enhancement at a tip relative to a remote electrode. (b) Parameters for the configuration modeled
are as follows: 50 nm tip radius, 10 shank angle, 50 m diameter wire, 1 mm long, specimen-to-
local-electrode distance is that required to permit a 1.5 steradian (70 ) geometric FOV. Calcula-
tions courtesy T. T. Gribb
Fig. 3.6 Model parameters for the configuration shown in (a) are 50 nm tip radius, 10 shank
angle, 40 m diameter aperture, and a specimen-to-local-electrode distance of 48 m. The base
width, w, relative to the microtip height, h, is such that h/w << 1. (b) Variation of the field
enhancement at a tip with increasing height of the needle. Calculations courtesy T. T. Gribb
Fig. 3.7 (a) SEM image of a microtip array used in the LEAP 3000/4000. There are 22 microtips
in the array located close to the top and bottom edges. (b) A single microtip. (c) The apex of a
pre-sharpened microtip which has a radius at the apex of about 75 nm. (d) A flat-top microtip
which has a 2 m diameter flat at the apex with a lift-out specimen positioned for attachment with
FIB-deposited platinum
The geometry of the microtip array was developed over many years, driven by need
and practical constraints to improve the speed and precision of specimen prepara-
tion. Some of the first methods to fabricate a large number of specimens took
advantage of existing photolithography and deep reactive ion etching to make
various geometries of specimens that could be mounted to a pin and sharpened
using a focused ion beam (FIB) [21, 22]. As the LEAP and FIB instruments
evolved, so did the design of the specimen carrier resulting in the current generation
of microtip arrays. Ease of use, cost, LEAP design, and performance considerations
all influenced the latest versions which consist of <100>-oriented antimony-doped
silicon coupons with posts that are approximately 150 m tall as shown in Fig. 3.7.
The placement of the tips and marker posts on the 3 mm by 7 mm coupon allows for
straightforward handling and easy alignment in the LEAP while minimizing the
electric field on adjacent tips during application of high field. Microtips come in
both pre-sharpened and flat-topped varieties [23], and both are doped to a conduc-
tivity >100 S/cm (antimony composition on the order of 100 ppm).
66 3 Design and Instrumentation
Microtip analysis made possible by the local electrode provides some unique
advantages, but there are also potential pitfalls. One notable pitfall is that the
microtip geometry provides a higher relative field at the local electrode compared
to the wire geometry which encourages electron emission and can causes reduced
local-electrode lifetime.
The delay line detector (DLD) assembly consists of the MCP assembly and a delay
line anode (DLA). When an ion impacts the MCP assembly and creates a cascade of
secondary electrons, the resulting electrons (on the order of 106 per ion) spread over
a few millimeters (FWHM) and impact several of the serpentine wire traces of the
DLA. Because the electron cascade is larger in width (or equivalently in time) than
the transition time across adjacent delay lines, a single broad electrical pulse is
created. Fortunately, the weighted center of the pulse results in a spatial resolution
finer than the spacing of the individual delay lines.
MCP detection efficiency is affected primarily by the open-pore area and the
incident angle of ion impacts. Because amplification of the single ion into an
electron cloud is largely insensitive to ion mass (for ion energies greater than
about 2 keV), no calibration is necessary for measuring different ion species.
There is however the possibility of variation of amplification across the FOV due
to ion impact angle. At short flight lengths in a straight-flight-path instrument, ion
trajectories may be parallel to the angle of the openings in the MCP causing the
secondary electron cascade to start at a significantly greater depth and resulting in
less amplification. This can be noticed at the very edge of a LEAP dataset at short
flight path lengths at a particular angular location (note that this effect does not
occur in the reflectron-based system because the ion trajectories are more closely
normal to the MCP surface). There are also detection losses due to ions that impact
the MCP between channels as shown in Fig. 3.8 (filled white circle) because they
cannot produce secondary electron cascades.
The electrical pulse from the electron cloud on the delay line is detected at both
ends of the wire (see Fig. 3.9). The time of the leading edge of the pulse is recorded
by the amplifier time-to-digital conversion (ATDC) system as discrete timing events.
A 2D detector is fabricated by wrapping the wire in a meander pattern as shown in
Fig. 3.10 for X and Y. The Y electrode is nominally identical to X but is rotated 90 .
The X anode allows about half of the electron cloud to pass through to be detected by
the Y anode. The resulting timing precision provides approximately 800 800
pixels across the detector face resulting in sub-nanometer spatial resolution in the
final image. This type of detector is typically referred to as a crossed DLA.
3.4 Instrumentation of the LEAP 67
Fig. 3.9 Schematic illustration of how a 1D DLA operates. Position (X) is determined from the
difference of the arrival times at the ends of the wire. Time of flight is determined from the average
of the two pulse arrival times. X 0 is the center of the wire. t0 is a constant of the system
A multi-hit occurs when more than one ion hit is detected during a single pulse.
DLD systems in early LEAPs use a two-anode delay line. The addition of a third
anode enables more accurate positioning of the multiple ion impacts that are close
in time and space (described within Fig. 3.11). Current LEAP systems use a
detector with three-anode delay lines, creating up to six timing events for each
ion impact. Occasionally one or more of the six channels may not record a signal
(typically called a partial hit). In theory, as few as three hits (one from each anode)
may be used to record the location of an event, but the more stringent selection
criteria currently employed in hit-finding algorithms require five hit events. Fewer
than five hit events result in significantly increased uncertainty in time (and
therefore in space), and so these hit types are ignored (called unrecoverable
partials).
A three-anode DLD does not, however, overcome the potential overlap of
multiple impacts that are very close in time and space. Multiple impacts that have
a specific spatial relationship can appear at the ends of the DLD as a single pulse
and are recorded as a single discrete timing event. When two ion impacts result in a
single overlapping pulse event, there are ways to extract both impacts from the
eleven recorded signals (six signals per event minus one overlapping event). When
more than one overlapping timing event exists, there may be multiple solutions for
the locations of the two events, so current hit finding typically only faithfully
identifies and records the first arriving event in these latter cases.
Data acquisition rates are typically restricted to minimize the number of multiple
field evaporation events likely to happen during any given pulse. However, multiple
ionization events resulting from a single pulse are much more frequent than random
3.4 Instrumentation of the LEAP 69
3.4.2.3 FIM
Field ion microscopy (FIM) is the precursor to APT and until recently was a critical
step in each APT experiment [25]. In previous generations of instruments, each
experiment was preceded by FIM to ensure specimen viability, remove oxidation
and damage from specimen preparation, evaporate the specimen until the region of
interest (ROI) was at the surface, and orient the specimen such that the ROI was
aligned with the detector. Previous generations of APT systems typically had a
separate phosphor screen for FIM from the TOF detector. In the LEAP, FIM is not
often used prior to the APT experiment, and the single-detector system has been
optimized for atom probe data collection.
Although the LEAP system has a wide FOV in both APT and FIM modes, there
are times when FIM can be useful as it provides complementary structural infor-
mation about the specimen. FIM has higher spatial resolution than APT due to the
fact that gas ion trajectories are less influenced by local bonding and local changes
to the radius of curvature, so the relationship and size of small phases or precipitates
may be more precisely imaged. In the LEAP, FIM experiments may be digitally
captured, and although the gain and bias of the MCPs cannot be changed after the
experiment, the brightness, contrast, and detector coefficients can be optimized
after data collection. For references on HD-eFIM experimental conditions, see
[2630].
In LEAP terminology, specimens are affixed to stubs, stubs and local electrodes are
positioned on pucks, pucks are loaded onto carousels, and carousels go into the
LEAP and reside on the vertical transfer rod. Various types of pucks and stubs are
shown in Fig. 3.12. Carousels have six positions, where each position can hold
either a specimen puck or a local-electrode puck. The pucks have a self-locating
wedge shape and are made with a high thermal and electrical conductivity alloy to
facilitate rapid cooling once placed into the cold stage. The local-electrode pucks
are designed for lubricity under UHV conditions.
70 3 Design and Instrumentation
Fig. 3.12 Solid models of specimen and local-electrode transport system for LEAP; (a) a carousel
has six positions and can be stacked in the loadlock and storage chamber, (b) a local-electrode
puck, (c) a four-needle specimen puck, (d) a single-needle specimen puck with a 3 mm TEM grid
holder, (e) a specimen puck with a Cambridge-style SEM specimen carrier, and (f) a single-needle
specimen puck with a quick-release holder for microtip arrays
Specimens for the LEAP can have nearly any geometry smaller than about
1 1 1 cm3 (due to stage travel and in-vacuum optics limitations). Specimens
made from a bulk sample in a wire geometry can be affixed directly into a puck or
crimped into a cylindrical stub. Specimen extracts can also be affixed to carriers.
The carriers can be wires, microtip coupons, TEM half grids, etc. Specimens of
unique geometries (not accommodated by standard stubs) can be affixed to stubs
with UHV-compatible silver-embedded epoxy to ensure acceptable electrical and
thermal conductivity.
As in all UHV applications, objects entering the vacuum system should only be
handled with clean tools [31]. Analytical sensitivity in a properly designed exper-
iment can be limited by the background noise level, and a clean vacuum will result
in the lowest possible spectral noise level. Of special consideration should be
handling of the local-electrode pucks. While the carousel never enters the analysis
chamber, and the specimen puck will become part of the cryogenically cooled
specimen stage, the local electrode resides in the analysis chamber of the UHV
system at room temperature, so its cleanliness is critical. Although the pumping
system is robust enough to enable a user to introduce a specimen and begin data
collection in only a few minutes, practically speaking one should only introduce
specimens into the analysis chamber after they have been pumped down to base
pressures in the storage chamber. Local-electrode pucks and other in-vacuum
components should be stored in a clean dry environment. Do not use desiccants
that may contaminate the parts with dust, and do not use any oil-based pumps.
Ideally these parts would be stored in a box with a clean, dry cover gas, or in an
oil-free vacuum environment.
3.4 Instrumentation of the LEAP 71
The purpose of the loadlock is to allow fast specimen loading from atmospheric
pressure to high-vacuum conditions. The loadlock chamber is pumped with a high-
speed turbomolecular pump and is vented to atmospheric pressure using dry
nitrogen. Minimizing the introduction of moisture and other contaminants into
the loadlock will maximize vacuum quality. The loadlock chamber allows the
introduction of three or four carousels (depending on LEAP configuration) and
thus the use of up to 24 specimens or local-electrode pucks at the same time.
Pumping is achieved with a turbomolecular pump coupled to an oil-free backing
pump. The transfer pressure (maximum pressure for opening the valve between the
loadlock and storage chambers) must be better than 103 Pa, which is obtained in
less than 5 min. When the required transfer pressure in the loadlock is attained, the
loadlock chamber valve may be opened and the specimen may be transferred to the
storage chamber in a few seconds using the UHV vertical translation/rotation rods.
The UHV loadlock chamber gate valve is provided with automatic valve control
and interlocks to prevent accidental venting of the storage chamber.
To maximize data quality, many groups choose to quarantine freshly introduced
specimen pucks in the loadlock chamber for a minimum period (i.e., overnight),
even though the pressure interlock will allow specimen transfer prior to that time.
There is a trade-off between introduction time and vacuum quality, so some
planning can help avoid the need to introduce a puck immediately before analysis.
For 4000 series instruments that are heavily used, an additional option can be
selected which increases loadlock capacity to include a fourth carousel which
includes a heating capability. Best practice suggests that local electrodes need to
be stored under UHV conditions for outgassing for 34 days prior to use in order to
ensure low-noise, high-quality data. If heating is used in the loadlock, this time for
low noise data can be reduced to less than 8 hours.
The purpose of the storage chamber is to maintain specimens and local electrodes
under UHV conditions before analysis. It allows a fast specimen transfer from the
storage chamber to the specimen stage of the analysis chamber. The transfer of the
specimen or the local electrode is achieved by using a translation/rotation rod and
72 3 Design and Instrumentation
The analysis chamber is a UHV stainless steel chamber pumped by an ion pump and
titanium sublimation pump. The cryogenically cooled specimen stage acts as a third
pumping device. The chamber is bakeable to 150 C. Base pressure is better than
108 Pa when the stage is at 50 K. The LEAP instrument can be equipped with a
FIM system whose operation requires that the analysis chamber be filled with
ultra-high purity (99.999 %) helium, neon, or argon up to 103 Pa which is achieved
through a precision leak valve.
In both laser pulsed and voltage pulsed modes, highly stable DC voltage power
supplies provide voltage to the specimen stage, the reflectron (on a LEAP 3000/
4000 HR system), and the MCP assembly. The gain and bias voltages for the MCPs
are set during the detector calibration process upon installation or major servicing
and they are held constant during all experiments. The reflectron voltage and the
pulse voltage (in voltage pulse mode) scale with the specimen voltage. The pulse
voltage is generated from a high-amplitude solid-state pulse generator and further
shaped with a combination of discrete components to ensure a 1090 rise time of
less than 1 ns and a very narrow pulse width.
The LEAP laser system is an option that can be installed at the time of manufacture
or as a system upgrade after shipment. As with the main emphasis of the design and
3.4 Instrumentation of the LEAP 73
construction of the field-pulsed-only LEAP unit (LEAP 3000/4000 HR), the laser
system has been designed to allow the user to conduct APT experiments without
requiring significant knowledge of the sub-components. There are no user-
serviceable parts in the laser enclosure, which is sealed to protect users and the
Class IV laser from damage. This allows the LEAP to be certified as a Class I
(no user training required) system due to its interlocks and lightproof, tamper-
resistant enclosure.
The laser pulse is created first with pump diodes that illuminate an oscillation
medium with an 80 MHz natural oscillation frequency for 1,064 nm light. This
cavity uses a semiconductor saturable absorber mirror (which is a consumable
medium). For the LEAP 4000X systems, this light is frequency tripled to 355 nm,
conditioned with various optics and power/centering/polarization components until
it is pulse-picked such that pulses with a repetition rate of 11,000 kHz enter the
targeting assembly. The beam is then focused and directed with piezoelectric
mirrors through a fused silica air/vacuum barrier.
3.4.5.1 Cryogenics
The atom probe includes a cryogenic cooling device attached to the specimen to
reduce thermal motion of atoms that can result in positional errors in the data
collected, limit preferential evaporation, and also function as part of the vacuum
pumping system. Specimen base temperatures on the order of 2050 K are typically
used. In laser pulsing, since the probability of ionization at the specimen tip is a
rapidly increasing function of both field and temperature, the temperature at the
peak of a thermal pulse needs to be only ~250 K for a very short time (<1 ns). Thus
with a properly chosen laser pulse energy, the time that the specimen spends at
elevated temperature is small (<1 ns) and the effect of atomic thermal motion on
the image resolution is also small. Nonetheless, when careful comparisons are made
between field pulsing and thermal pulsing, it is possible to both observe changes in
charge state ratios and discern a small degradation in spatial resolution in laser-
pulsed data due, presumably, to thermal motion [32, 33].
The cryogenic system is composed of two main parts: the compressor and the
cold head. The former compresses the helium which is then cooled in a heat
exchanger and expanded in the cold head. The cold head is suspended inside an
inverted stainless steel can and isolated from the LEAP frame and vacuum vessel.
The specimen stage is connected to the base of the inverted can for cooling.
Helium gas fills the interspace between the cold head and the can to allow for
natural-convection heat transfer. A rubber membrane seals the helium gas into the
interspace and is the only mechanical connection between the cold head and the rest
of the instrument. The helium gas is kept at a positive pressure to eliminate
74 3 Design and Instrumentation
the accumulation of air, which would liquefy and freeze between the cold head and
the LEAP vacuum vessel.
The expander portion of the cold head reaches temperatures as low as 10 K. The
minimum achievable temperature at the specimen puck is ~25 K, which varies
slightly with environment. The temperature of the specimen is controlled by a
resistive heating element that is attached to the vacuum side of the cold head can.
In any laboratory setting one must worry about environmental disturbances that can
affect data quality and yield. In general, the imaging capability of APT is insensi-
tive to vibration because micron-level motions of the tip translate to micron-level
blurring in the image (smaller than the resolution of the APT image); however,
alignment of the laser system, which is focused to a micron-sized spot, is greatly
affected by vibration. A site survey tests temperature stability and vibration over a
large range of frequencies. Typically, a room must meet the criteria similar to those
necessary for scanning electron microscopy systems. For LEAP systems that do not
have a laser, there is very little effect of the environment on the operation of the
instrument.
When operating in laser pulsed mode, even minor vibrations can affect the position
of the focused spot relative to the specimen apex. Any vibration visible in the laser
microscope view or any vibration that can be felt by touching the instrument can
affect the laser spot. A slight twinkle of the reflected laser light in the camera view
is normal, but visible vibration or blinking of the reflected light indicates excessive
vibration. This may result in a drop in detection rate, a subsequent rise in voltage, a
decrease in signal-to-noise ratio, and a higher probability of specimen fracture.
When your system has noticeable vibration and the room has been qualified to be
3.5 Summary 75
within specification, first check for any items leaning against the system, especially
items that can transmit vibration from pumps, etc. The most common vibration
source is from the cryogenic cooling system. The LEAP 4000 frame rests on an
integrated air support system, but it may still be sensitive to serious environmental
vibration. It is important to understand that vibration levels may deteriorate with
time as locations become more congested or experience a change of use.
3.5 Summary
The state of the hardware used in APT has continuously evolved since its inception
in the late 1960s. Over the past 1015 years however, this evolution has been more
substantial [34] and has culminated in the current-generation LEAP. The benefits of
the LEAP designlarge FOV, high MRP, high data collection rate, and microtip
analysis capabilityare mutually inclusive. It is difficult to envision any one
without the others because a 50 times improvement in FOV requires 50 times
longer acquisition time without a concomitant increase in data collection rate.
APT applications will always benefit from larger analysis volumes, which have
been found to be critical to resolving questions such as dopant distribution in
semiconductor materials [35].
There are undoubtedly directions to move with future atom probe instrument
research and development, including larger detectors, higher detection efficiencies,
improvement in multi-hit detection, improved laser, other thermal pulsing methods,
and more advanced analysis software. The atom probe community continues to
76 3 Design and Instrumentation
evaluate and develop new technologies to move the capability of APT forward, so
the future of the technology looks extremely promising. We refer the interested
reader to [36, 37] for further speculation as to what the future of APT may hold.
References
1. Gomer, R.: Field Emission and Field Ionization. Harvard University Press, Cambridge, MA
(1961)
2. Miller, M.K., Forbes, R.G.: Tutorial review: atom probe tomography. Mater. Charact. 60,
461469 (2009)
3. Kellogg, G.L., Tsong, T.T.: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys. 51
(2), 11841194 (1980)
4. Alikhanov, S.G.: A new pulse method of measuring the masses of ions. Sov. Phys. JETP 4,
452453 (1956)
5. Mamyrin, B.A., Karataev, V.I., Shmikk, D.V., Zagulin, V.A.: The mass-reflectron, a new
nonmagnetic time-of-flight mass spectrometer with high resolution. Sov. Phys. JETP 37,
4548 (1973)
6. Waugh, A.R., Richardson, C.H., Jenkins, R.: APFIM 200 a reflectron-based atom probe.
Surf. Sci. 266, 501505 (1992)
7. Scheinfein, M.R., Seidman, D.N.: Time aberrations of uniform fields: an improved reflectron
mass spectrometer for an atom-probe field-ion microscope. Rev. Sci. Instrum. 64(11),
31263131 (1993)
8. Cerezo, A., Godfrey, T.J., Sijbrandij, S., Smith, G.D.W., Warren, P.J.: Performance of an
energy compensated three-dimensional atom probe. Rev. Sci. Instrum. 1(69), 4958 (1998)
9. Sijbrandij, S., Cerezo, A., Godfrey, T.J., Smith, G.D.W.: Improvements in the mass resolution
of the three-dimensional atom probe. Appl. Surf. Sci. 94/95, 428433 (1996)
10. Panayi, P.: Curved reflectron. Patent UK 0509638.3 and US 60/682,863
11. Liu, J., Wu, C.-W., Tsong, T.T.: Measurement of the atomic site specific binding energy of
surface atoms of metals and alloys. Surf. Sci. 246(13), 157162 (1991)
12. Nishikawa, O., Kimoto, M.: Toward a scanning atom probe computer simulation of electric
field. Appl. Surf. Sci. 76/77, 424430 (1994)
13. Kelly, T.F., Camus, P.P., Larson, D.J., Holzman, L.M., Bajikar, S.S.: On the many advantages
of local-electrode atom probes. Ultramicroscopy 62, 2942 (1996)
14. Kelly, T.F., Larson, D.J.: Local electrode atom probes. Mat. Char. 44(12), 5985 (2000)
15. Cerezo, A., Godfrey, T.J., Huang, M., Smith, G.D.W.: Design of a scanning atom probe with
improved mass resolution. Rev. Sci. Instrum. 71(8), 30163023 (2000)
16. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.
C., Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data
from a commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383
(2004)
17. Larson, D.J., Russell, K.F., Miller, M.K.: Effect of specimen aspect ratio on the reconstruction
of atom probe tomography data. Microsc. Microanal. 5, 930931 (1999)
18. Humphry-Baker, S.A.: Atom Scale Studies of Fusion Materials: Atom Probe Tomography of
TungstenRhenium Alloys. University of Oxford, Part II Thesis (2008)
19. Loi, S.T., Gault, B., Ringer, S.P., Larson, D.J., Geiser, B.P.: Electrostatic simulations of a local
electrode atom probe: the dependence of tomographic reconstruction parameters on specimen
and microscope geometry. Ultramicroscopy 132, 107113 (2013)
20. Nishikawa, O., Kimoto, M., Iwatsuki, M., Ishikawa, Y.: Development of a scanning atom
probe. J. Vac. Sci. Technol. B 13(2), 599602 (1995)
References 77
21. Larson, D.J., Foord, D.T., Petford-Long, A.K., Liew, H., Blamire, M.G., Cerezo, A., Smith, G.
D.W.: Field-ion specimen preparation using focused ion-beam milling. Ultramicroscopy 79,
287293 (1999)
22. Larson, D., Wissman, B.D., Martens, R., Viellieux, R.J., Kelly, T.F., Gribb, T.T., Erskine, H.
F., Tabat, N.: Advances in atom probe specimen fabrication from planar multilayer thin film
structures. Microsc. Microanal. 7, 2431 (2001)
23. Thompson, K., Larson, D.J., Ulfig, R.: Pre-sharpened and flat-top microtip coupons: a quan-
titative comparison for atom-probe analysis studies. Microsc. Microanal. 11(S2), 882 (2005)
24. Thuvander, M., Weidow, J., Angseryd, J., Falk, L.K.L., Liu, F., Sonestedt, M., Stiller, K.,
Andren, H.O.: Quantitative atom probe analysis of carbides. Ultramicroscopy 111(6), 604608
(2011). doi:10.1016/j.ultramic.2010.12.024
25. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York (2000)
26. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principles and Applications. Elsevier, New
York (1969)
27. Muller, E.W., Tsong, T.T.: Field ion microscopy, field ionization and field evaporation, vol.
4. Progress in Surface Science 1139 (1973)
28. Bowkett, K.M., Smith, D.A.: Field-Ion Microscopy. North-Holland, Amsterdam (1970)
29. Miller, M.K., Smith, G.D.W.: Atom Probe Microanalysis: Principles and Applications to
Materials Problems. Materials Research Society, Pittsburgh (1989)
30. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field ion Micros-
copy. Oxford University Press, Oxford (1996)
31. Hoffman, D., Singh, B., Thomas, J.H.I.: Handbook of Vacuum Technology, 1st edn. Aca-
demic, London (1998)
32. Gault, B., Moody, M.P., De Geuser, F., Haley, D., Stephenson, L.T., Ringer, S.P.: Origin of the
spatial resolution in atom probe microscopy. Appl. Phys. Lett. 95(3), 034103-1 (2009)
33. Gault, B., Moody, M.P., De Geuser, F., La Fontaine, A., Stephenson, L.T., Haley, D., Ringer,
S.P.: Spatial resolution in atom probe tomography. Microsc. Microanal. 16, 99110 (2010)
34. Kelly, T.F., Larson, D.J.: The second revolution in atom probe tomography. MRS Bull. 37,
150158 (2012)
35. Inoue, K., Yano, F., Nishida, A., Tsunomura, T., Toyama, T., Nagai, Y., Hasegawa, M.:
Monolayer segregation of as atoms at the interface between gate-oxide and Si substrate in a
p-MOSFET by 3D atom-probe technique. Appl. Phys. Lett. 92, 103506-1 (2008)
36. Kelly, T.F.: Kinetic-energy discrimination for atom probe tomography. Microsc. Microanal.
17, 114 (2011)
37. Kelly, T.F., Miller, M.K., Rajan, K., Ringer, S.P.: Atomic-scale tomography: a 2020 vision.
Microsc. Microanal. 19, 652664 (2013)
Chapter 4
Data Collection
4.1 Introduction
Data collection for an atom probe tomography (APT) experiment is the process of
(1) introducing an electric field on a specimen in order to initiate ion emission by
field evaporation, (2) using the initial ion events to allow fine positioning of the
specimen, and (3) increasing and adjusting the voltage (and thus the electric field)
to obtain a steady rate of detected ion events. The details of how one can maneuver
through these steps safely and efficiently are the focus of this chapter. Because a
strong electric field is required to initiate field evaporation, the resulting forces
create large stresses that act on the specimen, placing it in continual jeopardy of
mechanical failure. Consequently, a successful atom probe experiment is a
balancing act of minimizing the chance for specimen failure while at the same
time extracting maximum quality from the collected data. After reading this
chapter, it becomes clear that we prioritize yield over data quality (one must have
data before judgments can be made on its quality), but the ultimate decision on this
issue rests with the user. Depending on the details of any particular experiment, the
user may decide that either specimen yield or data quality takes precedence.
The goal of this chapter is to guide the reader in the steps necessary for operation
of the local electrode atom probe (LEAP) in order to collect high-quality data. This
includes a discussion of how the primary experimental parameters (temperature,
ion detection rate, etc.) influence the chances for experimental success (analysis
yield) and affect data quality, especially when approaching a new sample type.
A minimal description of the instrument control software is included as it is this
interface through that the user selects parameters and operates the instrument.
Suggested operating procedures for voltage acquisition and laser acquisition,
including laser pre-alignment, are included to aid novice users through a check
list of activities to begin an analysis. Experienced users may find more value in the
general discussions regarding parameter selection, data quality, and data collection
goals. This chapter is not meant to replace the LEAP User Manual which describes
specimen transfer, alignment, and the acquisition software in detail; instead it is
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 79
DOI 10.1007/978-1-4614-8721-0_4, Springer Science+Business Media New York 2013
80 4 Data Collection
Before discussing the various experimental parameters and the effects they have on
data quality, it is important to identify the basic metrics used to assess data quality.
Some of these metrics describe the quality of the mass spectrum, but all impact the
ability to accurately quantify the composition of the specimen. These data quality
metrics include mass resolving power (MRP), background, multi-hit performance,
and complex ion generation.
MRP and background are general metrics used to describe any spectrum. MRP
was defined in Chap. 3 and is a measure of the width of the time-of-flight (TOF)
range (or m/n range) for each ion type. Background is loosely defined as (1) events
detected at arrival times that do not correspond to a specific mass-to-charge (m/n)
peak or (2) ion evaporation events that occur out of synchronization (out-of-synch)
with the voltage or laser pulse. Some of these events originate from the specimen,
while others originate from residual gasses in the analysis chamber that were
adsorbed on the specimen and then field evaporated at a random time. The adsorbed
gasses also field evaporate due to the evaporation pulse, but these events are
observable as families of peaks in the m/n spectrum (commonly H+, H2+, H3+,
OH+, H2O+, H3O+, and others) and can be selectively filtered out during analysis.
The out-of-synch events are evenly distributed in time, which is observed as a
decaying background level in the m/n spectrum.
The presence of a background directly impacts data quality and should be
minimized as much as possible [1, 2]. A background component resides in each
m/n peak and can be estimated and subtracted from each peak when determining
chemical composition; however, correctly accounting for this effect in visualization
(atom maps) is complicated. In principle, each ion in a mass spectrum peak has
some probability that it is a residual gas ion and not an assigned ion type. Due to the
limitation of using a single set of mass ranges, this causes background ions of each
ion type to be distributed throughout the atom map. For example, because of m/n
ranging limitations (see Chap. 6 and Appendix D) a bilayer specimen of pure A on
pure B will display background levels of B ions in A and vice versa in any
atom map.
The out-of-synch component can strongly affect quantification when all ionic
species do not have an equal probability of exhibiting out-of-synch evaporation.
Insufficient pulse fraction (see next section) can cause preferential evaporation
and/or retention of certain ion types resulting in incorrect compositions [3, 4]. Back-
grounds composed primarily of out-of-synch evaporation events of equal probabil-
ity for each ion type do not affect composition accuracy and any potential
reconstruction artifacts can also be corrected (see Chap. 5).
4.3 Analysis Yield Considerations 81
In an APT experiment, the specimen is subjected to a very high electric field which
produces a very high stress proportional to the electric field squared. Various authors
have calculated the stresses on APT specimens using different methods
[1218]. These stresses can produces elastic strains of 12 % [19] and may be
sufficient to induce motion of unpinned defects such as dislocations [20, 21] and
cause in-situ twinning [19, 22]. Field-induced stress is also a cause of fracture in
specimens, which require a certain amount of elastic strength to produce good yield,
and as such we are interested in practical ways to minimize the effects of stress on
APT specimens. Numerical and analytical estimates of the magnitude of such stress,
similar to calculations from Moy et al. [18], are shown in Fig. 4.1. Note that the
analytical results closely match the previous results from Birdseye and Smith [12].
For specimens with specific regions of interest, a reasonable assumption is that
any specimen will yield relative to the ratio of the adhesion in these regions and the
stress generated by the electric field required to field evaporate the material. It is
often the case that regions of specimen surface which deviate from uniformity can
act as stress concentrators, a notorious example of these being the intersection of an
interface with the specimen surface. In order to improve yield, one can either reduce
the applied stress or increase the effective adhesion.
82 4 Data Collection
specimen field (from F1 to F2). Increasing the base temperature (method 4) also
applies to both voltage pulsed and laser pulsed modes.
Method 5 is seldom employed in the current generation of atom probes, although
hydrogen and some noble gasses have been shown to decrease the evaporation field
of some metals and semiconductors by 10 % or more [24].
Yield may also be increased by improving adhesion within specimens or by
changing the interaction between the applied stress and the inherently weak regions
in the specimen. Specimens with inherently weak interfaces will very seldom run
successfully in APT analysis, as shown by the example in Fig. 4.4a where specimen
failure has resulted in fracture at an interface. In Fig. 4.4b the stresses applied to a
specimen during field evaporation have resulted in shearing along {111} planes.
84 4 Data Collection
Fig. 4.4 (a) An example of a failed APT specimen that appears to have delaminated at a film
interface without destroying the rest of the tip and (b) a Ni-4Al alloy specimen which has sheared
on the {111} plane due to field-induced stresses (image in (b) courtesy D.K. Schreiber and
M.J. Olszta)
Fig. 4.5 Reorienting interfaces or layers can lead to improved analysis yield. (a) Multi-layer films
can be prepared in cross-section (film surfaces parallel to the analysis axis) thereby changing the
relative orientation between the applied force and the interface. (b) Samples with differing
material regions can be simplified by removing undesired material and replacing it with more
APT-friendly materials leading to improved analysis yield. (c) An example of an engineered
coating which encapsulates the region of interest and changes the net electrical and thermal
properties of the specimen promoting positive effects on analysis yield
The philosophy behind the experimental control of an atom probe is to provide the
user with parameters that combine ease of operation with good data quality. Ideally
an APT experiment would extract data from a material with optimum quality and
minimum analysis time. Because each specimen may be unique and all material
types do not perform similarly under APT analysis conditions, the user is often
required to choose control parameters that properly balance data quality, analysis
86 4 Data Collection
time, and analysis yield. This section discusses the various acquisition parameters
and algorithms, both basic and advanced, and provides some pointers for the
general LEAP user.
The general experimental parameters are listed in Table 4.1 and include pulse
rate ( f ), base temperature (T ), detection rate (DR in % of pulses that result in a
detected field evaporation event), pulse fraction (PFvoltage APT only
amplitude of the voltage pulse expressed as a percentage of the current DC voltage),
laser pulse energy (LPElaser APT only), and voltage (V ). A number of things
should be considered when selecting specific parameters.
The pulse rate is selectable by the user and, for most materials, the best rate is the
maximum rate that enables the m/n range of ions to be captured within a single
pulse window for all acquisition voltages. This keeps acquisition time to a mini-
mum and reduces the opportunity for out-of-synch evaporation to occur within the
pulse window. Because the m/n range is determined by the maximum TOF of
the ions and the TOF for a particular ion species decreases as voltage increases, the
recorded m/n range is both pulse rate and voltage dependent as shown in Tables 4.2,
4.3, and 4.4. Very high pulse rates may result in long TOF ions being detected
during subsequent pulses. A wrap-around effect is then observed with the large m/n
signal from one pulse detected on the next pulse window and observed as a broad
peak in the lower m/n range of the mass spectrum. This signal is generally not
recoverable as identifiable ions for reconstruction and chemical quantification.
Some materials with poor thermal diffusivity analyzed under high-energy laser
pulsing conditions may not cool sufficiently between heating pulses. This causes the
base temperature of the near-apex region of the specimen to remain above the set
base temperature during analysis and can result in long tails for peaks in the mass
spectrum. This base temperature offset effect may also be a function of pulse rate, so
slower pulse rates should be tried for materials exhibiting this behavior.
result in maximum heat flow rates (maximum gradient) out of the apex of the tip,
resulting in higher MRP. Low specimen-stage temperatures can minimize the
mobility of residual gasses present in the analysis chamber, which improves the
vacuum levels and lowers spectral background. Many materials also have thermal
transport properties that improve near 20 K [30].
4.4 Experimental Parameters 89
Once significant field evaporation is achieved, only small changes in the applied
electric field are necessary to change the DR because of the power-law behavior of
the response [3134]. Large DRs generally promote higher data quality (increased
signal-to-noise [19]). For many materials, this benefit is offset by decreased spec-
imen yield. For some elements (e.g., boron and carbon), increased DR also
increases the potential for lower detection efficiency due to an increase in the
proportion of multi-hit events [7, 3539].
The LPE is a measure of the total photon energy carried by each pulse. This energy
is spread nonuniformly over the laser spot resulting in some unknown energy
density at the specimen tip. Focus quality, spot size, and laser-specimen alignment
will influence the energy density experienced by the specimen. Additional proper-
ties of the specimen (e.g., size, shank angle, composition, reflectivity, absorption
coefficient, thermal diffusivity, and band gap) influence how energy is absorbed,
converted to heat, and removed from the apex region of the tip [2]. Because the
evaporation field required to evaporate ions from the surface of a specimen is
temperature dependent, the thermal pulse created during laser pulsed acquisition
lowers the required applied field, and thus the stress on the specimen, leading to
significant improvement in analysis yield. High LPE generally results in superior
yield, improved MRP, and low background. This provides an incentive to operate at
the highest LPE that does not result in other negative effects (e.g., degraded spatial
resolution).
Insufficient LPE can result in negative effects similar to those resulting from
insufficient pulse fraction in voltage pulse mode. High rates of out-of-synch
evaporation or preferential retention can result when there is too small a field
difference for specimen elements at base temperature and maximum pulse temper-
ature. Also note that peak apex temperature (and resulting required evaporation
field) generally changes as a function of voltage because the absorption properties
and heated volume (which impact apex temperature) evolve with specimen size.
The positive impact on yield and spectral quality suggests higher LPE settings,
but extreme LPE raises concerns regarding surface migration, complex ion gener-
ation, and nonuniform spatial evaporation behavior, negatively impacting the
ability to reconstruct ion positions accurately. In addition, application of laser
pulsed APT to oxides and nitrides suggests that minimal laser energies may be
required in order to obtain correct stoichiometry [40, 41]. One must take this all into
consideration when designing experiments to analyze multiple specimens under
equivalent conditions.
Each LEAP platform has a unique set of hardware dependent parameter ranges.
Note that some performance metrics evolve as computer power steadily increases.
Hardware parameter ranges are summarized in Table 4.5.
4.5 How to Start Your Investigation of Any New Material 91
The remainder of this chapter operates under the assumption that the user is familiar
with LEAP calibration files, acquisition control algorithms, and local-electrode
maintenance. If this is not the case, it might be a good idea to review the informa-
tion contained in Appendix E.
When attempting APT analysis of a material, the analysis conditions must be
appropriately selected. For materials that are slight variations of previously ana-
lyzed materials analyzed under known conditions, these known conditions make for
a sensible starting point. When attempting analysis of a new material where little
knowledge is available regarding APT analysis, consider the following suggestions:
assume the material is difficult to field evaporate, minimize the necessary evapo-
ration field, vary analysis conditions to obtain data, be prepared to make more
specimens than you think will be adequate, and optimize conditions once minimal
yield has been achieved.
1. Assume that the new material type is difficult to field evaporate. The expectation
is to go through a number of specimens before the analysis objectives are
achieved. It is difficult to gain any insights from specimens that repeatedly fail
before yielding any data, thus conditions that improve yield at the expense of
data quality considerations are often considered as a starting point. Initially, high
data quality (step four) is secondary to analysis yield.
2. Minimize the field necessary to field evaporate the material. Using reasoning that
the lower the evaporation field, the lower the stress on the tip and the greater the
likelihood for successful data collection suggests a certain acquisition parameter
space. Start with a moderate base temperature (50100 K), low DR (0.2 %), and
relatively high LPE. The actual LPE depends on the energy density of that
particular instrument, wavelength, and previous experience with related mate-
rials. Be aware that relatively high LPE for one material is not necessarily the
same as for another (metals versus semiconductors versus oxides for example).
3. Vary conditions to obtain data. Do not get too limited in your parameter
selection. If the first tip fails immediately under a set of conditions, then try a
92 4 Data Collection
Once a specimen has been fabricated, loaded into the LEAP, transferred onto the
cryogenic specimen stage, and coarsely aligned to the local electrode, data collec-
tion can begin. For the uninitiated, the LEAP User Manual provides detailed
descriptions of specimen transfer, coarse alignment, tool operation, and the soft-
ware that is used to accomplish these tasks. Separate software applications are used
to control the instrument and display the critical live acquisition data displays: the
LEAP Control Center (LCC) and the Data Acquisition & Visualization
(DAVis) applications. This section focuses on data collection from the time a
voltage is applied and the detector begins to register hits, to the final storage of data
files (see Appendices A and E). Consequently, only relevant LCC and DAVis
features will be covered. This section is not meant to be comprehensive, but instead
to provide the user with the essentials for getting an experiment under way and
running smoothly. Because the steps and considerations are slightly different for
voltage pulse and laser pulse mode acquisitions, each will be covered in turn.
The general flow of a LEAP acquisition is as follows: (1) the specimen puck is
transferred to the specimen stage and the base temperature is selected; (2) during
the approximately 10 min it takes for the specimen to cool, the specimen is coarsely
aligned to the local electrode using the optical microscope cameras; (3) appropriate
setup parameters are selected;(4) once the specimen temperature has reached
equilibrium, acquisition may be started. During the initial stages of the acquisition,
fine adjustments of the specimen position are made to appropriately center and fill
the detector with ions emitted from the specimen surface. Some users will use the
cool down time to start acquisition and alignment to take advantage of evaporating
through any surface oxides while still at an elevated temperature and thus minimize
early fracture. If the Auto Detection Rate Control has been selected, the instrument
attempts to maintain a voltage such that the actual DR remains near the target rate.
The experiment is terminated based on criteria set by the user (discussed later) or
when the instrument determines that the specimen has failed and self-terminates.
Detailed procedures for voltage pulsed and laser pulsed acquisition follow.
4.6 Brief Overview of LEAP Operation: Data Collection 93
Consider a specimen that is ready for analysis in voltage pulse mode. Before
initiating the acquisition, a number of parameter fields should first be evaluated.
The Voltage Atom Probe tab in LCC contains sections labeled Setup Parameters,
Run Statistics (Fig. 4.6a), and Advanced Controls (Fig. 4.6b). The Setup Parameters
include Detection Rate, Specimen Voltage, Pulse Rate, Pulse Fraction, and Cali-
bration as well as a number of check-boxes which enable Auto Detection Rate
Control, stopping conditions (Stop Elapsed Time, Stop Total Events, Stop Speci-
men Voltage), and Acquisition eMail Notifications. The following set of steps for
selecting important parameters and initiating analysis:
1. Load specimen (see Chap. 3)
2. Visually align specimen to LE
3. Make adjustments to the Advanced Control settings.
(a) Algorithm: Choose an algorithm. The most notable impact of this choice will
be minimum DR and initial voltage ramp rate. The minimum DR should be
~0.2 % for fragile specimens. Otherwise select according to personal pref-
erence with the option to change during acquisition.
94 4 Data Collection
Fig. 4.6 The LCC voltage atom probe tab: (a) Setup Parameters and Run Statistics and (b)
Advanced Controls
(c) Pulse Rate (Hz): Choose the highest pulse rate that results in the heaviest
element arriving within a single pulse window (see Tables 4.2, 4.3, and 4.4).
(d) Pulse Fraction (%): Typical choices are 1030 %. Depending on the mate-
rial, this choice could impact compositional accuracy as insufficient pulse
fraction can lead to low-field species evaporating between pulses. Also,
there is a maximum pulser voltage (instrument dependent). Pulse fractions
greater than ~10 % may not be possible for the top end of the 15 kV range of
the instrument DC voltage. For blunt specimens or specimens with high
evaporation field requirements, the user should consider this limitation. This
parameter cannot be changed during acquisition.
(e) Auto Detection Rate Control check-box: Most users will leave this selected
and the voltage will immediately ramp according to the selected control
algorithm (see Appendix E) until the desired DR is achieved. Advanced
users may choose to leave this unchecked and manually increase the voltage
until some ion emission is observed allowing for fine adjustment of the
specimen position while the voltage remains constant. This control feature
may be toggled on and off during acquisition.
5. Start the acquisition by left-clicking on the Start button.
6. Fine specimen alignment.
(a) Detector Event Histogram: As the voltage ramps (either manually or auto-
matically) pay close attention to the live data plots in DAVis. Of particular
interest is the Detector Event Histogram/Detector Ion Map. Any
misalignment of the specimen to the local electrode becomes clearly evident
as events accumulate on the detector.
(b) Lateral misalignment: This is when the circular shadow of the local elec-
trode is not centered on the detector, Fig. 4.7a. Adjustments are made with
specimen motion control arrows (Fig. 4.7 inset) to move the pattern to the
center of the detector (Fig. 4.7b). Some caution is prudent here in that
motion of the tip while under high field may promote specimen failure.
Adjustments should be completed before the voltage is high enough to
initiate significant field emission (less than about 0.2 % DR).
(c) Specimen-to-local-electrode separation: When the specimen is too close to
the local electrode, the ion pattern overfills the detector so that the circular
shadow of the local electrode does not blank out the outer edge of the data as
shown in Fig. 4.7b. Again, while the voltage is relatively low, the specimen
can be moved away from or toward the local electrode by using the speci-
men motion control + (closer separation or larger ion pattern) and
(farther separation or smaller ion pattern) buttons. It is important to note that
the motions toward the local electrode create rapid increases in the field
experienced by the specimen, possibly promoting tip failure. For especially
delicate specimens (most specimens) it is prudent to manually reduce the
voltage by a few percent, move the specimen closer to the electrode, and
then increase the voltage slowly until minimal ion emission allows you to
reassess alignment using the ion pattern.
4.6 Brief Overview of LEAP Operation: Data Collection 97
(b) Detection Rate (%): The rate can be slowly increased to the desired final rate.
Although the voltage increment is relatively small for increases in target
rate, increasing the detection rate, and thus the electric field, can impact the
longevity of the analysis. Besides having a negative impact on yield, this
parameter positively affects data quality and total time for data collection, so
it is important to experimentally investigate these tradeoffs and optimize this
parameter when frequently analyzing a particular material system.
(c) Stopping conditions and email notification: Before leaving the instrument
unattended, it is advisable to decide on termination conditions. Choosing
reasonable Stop Total Events and Stop Specimen Voltage values is prudent.
Often the voltage stop value may be influenced by the current local-electrode
flat test value (see Appendix E). Likewise, setting up the electronic mail
notifications to make notification when a specimen has completed acquisi-
tion (or failed), or met some other threshold in the Advanced Controls tab
will promote maximum utility from the LEAP.
Some parameter ranges discussed above are hardware dependent and those
generally achievable for the LEAP 3000 and 4000 platforms are listed in Table 4.5.
Consider a specimen that is ready for analysis in laser pulse mode. In voltage pulse
mode, the only alignment concern is the specimen position with respect to the local
electrode. In laser pulse mode specimen alignment is required as well, but now the
laser must also be aligned to the specimen. Any movement of the specimen position
requires realignment of the laser to the new specimen position, so very careful
control and monitoring is required during fine alignment where specimen position
adjustments are followed by laser adjustments. After satisfactory positioning of the
specimen relative to the LE, the laser focus must be adjusted so that optimum laser-
pulse energy density is achieved. Moving the specimen and laser and adjusting the
focus while the specimen is under high field increases the chance for specimen
fracture. Unless there is some confidence that the current position and focus of the
laser is approximately correct (e.g., a specimen has just been analyzed in laser pulse
mode and it was known to be well positioned and focused and the LE has not been
changed), then a high yielding, calibration specimen (pre-sharpened silicon
microtip) should be used to pre-align the laser to the local electrode. Following
this recommendation, there are two procedures for laser pulse mode operation:
(1) laser pre-alignment, and (2) laser acquisition.
4.6 Brief Overview of LEAP Operation: Data Collection 99
Fig. 4.8 The LCC laser atom probe tab showing the Setup Parameters and Run Statistics displays
(b) Manual Alignment: This is accomplished by first moving to the Laser Atom
Probe tab in LCC and left-clicking on the Manual Alignment button. This
will bring up the alignment window. Align the laser to the specimen by
moving the laser until maximum reflected light is observed from the spec-
imen apex.
Camera Focus: The camera laser focus should not be modified unless
alignment is being performed on a newly exchanged LE. For a newly
exchanged LE, the apex of the aligned specimen viewed in the Laser
Targeting camera should be in sharp focus. Manipulate the camera focus
on the left of the Laser Targeting window until the tip apex is in sharp
focus. Remember, the laser passes through this camera, so changing the
camera focus also changes the focal position of the laser spot.
4.6 Brief Overview of LEAP Operation: Data Collection 101
(continued)
102 4 Data Collection
1. Load specimen.
2. Visually align specimen to LE.
3. Make adjustments to the Advanced Laser settings (this is a separate tab in the
Advanced Controls panel).
(a) Algorithm: Choose an algorithm. The most notable impact of this choice will
be minimum DR and initial voltage ramp rate. The minimum DR should
be ~0.2 % for fragile specimens suggesting either algorithm 4 or 5
(see Appendix E). Otherwise select according to personal preference.
4. Make adjustments to the initial Setup Parameters.
(a) Detection Rate (%): See voltage acquisition.
(b) Pulse Rate (Hz): See voltage acquisition.
(c) Specimen Voltage (V ): See voltage acquisition.
(d) Pulse Energy (pJ or nJ): Different material types require different pulse
energies. The heating pulse requirements for field evaporation and the
material properties that effect absorption all play a part here. Here are
some things to consider:
Specimens blunt as they are analyzed thus increasing their heated vol-
umes. Consequently, specimens often require lower pulse energy at the
beginning of an analysis.
4.6 Brief Overview of LEAP Operation: Data Collection 103
The lower the evaporation field needed to initiate field evaporation, the
less likely it is that the specimen will fail. A lower rate of failure may
often be achieved by utilizing higher pulse energies at startup.
Any out-of-focus situation will require extra pulse energy.
Overall it is probably safest to start a specimen with extra laser pulse
energy. Once steady state analysis is achieved, be prepared to make
adjustments.
(e) Auto Detection Rate Control check box: See voltage acquisition.
5. Check the Advanced Laser tab.
(a) Be aware of the Tip Scan Modes: Scout, Standard, and Drift Comp
(b) Be aware of the Focus Scan Modes: Scout, Standard, and Drift Comp
6. Align the specimen apex to the laser:
(a) This is accomplished by first moving to the Laser Atom Probe tab in LCC
and left-clicking on the Manual Alignment button at the lower right of the
window (see Fig. 4.6b). This will bring up the Specimen and Laser Beam
Alignment window.
(b) Left-click the Beam On/off button.
(c) Align the specimen to the laser by moving the specimen using the stage
control arrows at the left of the window until maximum reflected light is
observed from the specimen apex. The laser should be reflected from the
same location relative to the local electrode as is required of all specimens.
Any difficulty in locating the laser suggests something is wrong (specimen is
not centered in both camera views, laser beam is not on, poorly pre-aligned
laser).
(d) Left-click OK to return to Laser Atom Probe tab.
7. Start the acquisition by left-clicking on the start button.
8. Fine specimen alignment:
(a) Detector Event Histogram and Mass Spectrum: As the voltage ramps (either
manually or automatically) pay close attention to the live data plots in
DAVis. Of particular interest is the Detector Event Histogram.
Misalignment of the specimen to the local electrode becomes clearly evident
as events accumulate on the detector. New users should be cautious here
when attempting operation at higher frequencies. Voltages can ramp very
quickly and it is possible to attain voltages in excess of the voltage required
for the desired DR when the laser or specimen is misaligned. This can lead to
immediate specimen failure during tip scanning. In addition, monitoring of
the Mass Spectrum provides information regarding the number of real
ion-based events versus background noise or out-of-synch events.
(b) Laser misalignment: Perform scout scans regularly until the specimen emits
strongly enough so that the laser scan snaps the laser position to the
specimen apex (scan at the beginning and every 5001,000 V to ensure
104 4 Data Collection
that the alignment is achieved before the applied field is strong enough to
yield significant field evaporation).
Note that the scan should start beyond the specimen such that you cannot
easily see any reflection as the scan starts. At the end of the scan, there
should be strong reflections as the scan goes across the specimen. If this is
not the case, consider going back to the initial alignment steps.
Note that some emission from the specimen may be observed before the
scout laser scan is completed. If the emission is significant enough such
that you can judge the specimen to be misaligned, then align specimen as
described in the next step and immediately rescan the laser. It is possible
that the tip position is so misaligned that the scout scan only appears to
scan over the tip, but the specimen is actually outside the scan area.
(c) Lateral misalignment: This is when the circular shadow of the local elec-
trode is not centered on the detector, Fig. 4.7a. Make adjustments with
specimen motion control arrows (Fig. 4.7 inset) to move the pattern to the
center of the detector (Fig. 4.7b). Some caution is prudent here in that motion
of the tip while under high field may promote specimen fracture. Adjust-
ments should be completed before the voltage is high enough such that the
specimen has reached a fully developed field evaporated endform. In addi-
tion, only small adjustments in position should be made before realigning the
laser. Large motions may exceed the range of laser scans making laser
realignment difficult without some manual adjustment of the laser position.
(d) Realign laser: After any specimen motion, the laser will need to be realigned
to the specimen apex. Repeat steps 7a, b, and c until the specimen is properly
centered within the local electrode and the laser has properly snapped to the
specimen apex due to a strong, sharp emission spike in the tip scan.
(e) Specimen-to-local-electrode separation: When the specimen is too close to
the detector, the ion pattern overfills the detector so that the circular shadow
of the local electrode does not blank out the outer edge of the detector, as
shown in Fig. 4.7b. While the voltage is relatively low, the specimen can be
moved away from or toward the local electrode by using the specimen
motion control + (smaller separation or larger ion pattern) and (larger
separation or smaller ion pattern) buttons. It is important to note that the
motions toward the local electrode create rapid increases in the field expe-
rienced by the specimen, possibly promoting tip failure. For especially
delicate specimens (most specimens) it is prudent to manually reduce the
voltage by a few percent, move the specimen closer to the electrode, and
then increase the voltage slowly until minimal ion emission allows you to
reassess alignment by way of the ion pattern.
(f) Realign laser: After any specimen motion, the laser will need to be realigned
to the specimen apex. Repeat steps 7ad until the specimen is properly
centered within the local electrode and the laser has properly snapped to
the specimen apex due to a strong, sharp emission spike in the tip scan.
4.6 Brief Overview of LEAP Operation: Data Collection 105
(g) Fine laser alignment: Once the specimen is in the proper position and the
laser has successfully identified the apex via a scout scan, a standard tip scan
at higher scan resolution is necessary to properly place the laser on the
specimen apex. After a successful standard scan, a drift comp scan is used to
make fine adjustments to the laser position. In this mode, the laser never
fully leaves the specimen but only makes small excursions that allow the
control algorithm to make small corrections to the laser position.
(h) Focus Scanning: A good pre-alignment likely ensures that the laser is
focused to within a couple of micrometers of the current specimen position.
For the LEAP 3000X that is likely sufficient since the laser spot size does not
change significantly over a couple micrometers. This is not as true for the
LEAP 4000X. A couple of micrometers misalignment can significantly
reduce the energy density. Therefore, it is recommended that a focus scan
be performed on the 4000X (optional for the 3000X) to place the laser focus
at the specimen apex.
The scout focus scan should be unnecessary if an adequate pre-alignment
has been performed.
Start with a standard scan.
Finish with a drift scan.
9. Adjustments during acquisition: Now that fine alignment of the specimen and
laser is complete, final adjustments can be made by the user and fully automated
acquisition and termination can be implemented.
(a) Auto Detection Rate Control check box: See voltage acquisition.
(b) Auto Tip Scan Settings and Auto Scanning settings in the Advanced Laser tab:
See the Laser Positioning Control section in Appendix D for a discussion
regarding appropriate scan settings. Default scan settings are a good starting
point, but these should be customized depending on sample and laboratory
environment needs. Changes in room temperature should be avoided in
particular as this may cause significant drift of the laser focus position
with respect to the specimen. Monitor auto operation for many scans before
leaving the tool unsupervised in automated mode. Scans that regularly
require laser repositioning to the edge of the scan area suggest that more
frequent and/or wider scans are required.
(c) Detection Rate (%): See voltage acquisition.
(d) Stopping conditions and email notification: See voltage acquisition.
Once automated acquisition is enabled and the specimen is emitting ions in a stable
manner, the LEAP can be left unattended; however, there are a few things one
should keep in mind before proceeding:
106 4 Data Collection
References
1. Bevington, P.R., Robinson, K.D.: Data Reduction and Error Analysis for the Physical Sci-
ences, 3rd edn. McGraw-Hill Higher Education, New York, NY (2003)
2. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer series
in materials science, vol. 160. Springer, New York, NY (2012)
3. Miller, M.K., Smith, G.D.W.: An atom probe study of the anomalous field evaporation of
alloys containing silicon. J. Vac. Sci. Tech. 19(1), 5762 (1981)
4. Miller, M.K., Smith, G.D.W.: Atom Probe Microanalysis: Principles and Applications to
Materials Problems. Materials Research Society, Pittsburgh, NJ (1989)
5. Andren, H.O., Henjered, A., Norden, H.: Composition of MC precipitates in a titanium
stabilized austenitic stainless-steel. J. Mater. Sci. 15(9), 23652368 (1980). doi:10.1007/
bf00552329
6. Thuvander, M., Weidow, J., Angseryd, J., Falk, L.K.L., Liu, F., Sonestedt, M., Stiller, K.,
Andren, H.O.: Quantitative atom probe analysis of carbides. Ultramicroscopy 111(6), 604608
(2011). doi:10.1016/j.ultramic.2010.12.024
7. Saxey, D.W.: Correlated ion analysis and the interpretation of atom probe mass spectra.
Ultramicroscopy 111(6), 473479 (2011)
8. Tang, F., Gault, B., Ringer, S.P., Martin, P., Bendavid, A., Cairney, J.M.: Microstructural
investigation of TiSiN hard coatings. Scripta Mater. 63, 192195 (2010)
9. Yao, L., Gault, B., Cairney, J.M., Ringer, S.P.: On the multiplicity of field evaporation events
in atom probe: a new dimension to the analysis of mass spectra. Phil. Mag. Lett. 90(2),
121129 (2010)
10. Tsong, T.T., Liou, Y.: Cluster-ion formation in pulsed-laser-stimulated field desorption of
condensed materials. Phys. Rev. B 32(7), 43404357 (1985)
11. Prosa, T.J., Alvis, R.A., Kelly, T.F.: Observations of cluster ions originating from
non-traditional atom probe materials. Microsc. Microanal. 14(S2), 1236 (2008)
12. Birdseye, P.J., Smith, D.A.: The electric field and the stress on a field-ion specimen. Surf. Sci.
23, 198210 (1970)
13. Birdseye, P.J., Smith, D.A., Smith, G.D.W.: Analogue investigations of electric field distribu-
tion and ion trajectories in the field ion microscope. J. Phys. D 7(12), 16421651 (1974)
14. Eaton, H.C., Bayuzick, R.J.: Field-induced stresses in field emitters. Surf. Sci. 70, 408426
(1978)
15. Smith, P.J., Smith, D.A.: Preliminary calculations of the electric field and the stress on a field-
ion specimen. Phil. Mag. 21, 907912 (1970)
16. Rendulic, K.D., Muller, E.W.: Elastic deformation of field-ion-microscope tips. J. Appl. Phys.
38(5), 2070 (1967)
17. Bowkett, K.M., Smith, D.A.: Field-Ion Microscopy. North-Holland, Amsterdam (1970)
18. Moy, C.K.S., Ranzi, G., Petersen, T.C., Ringer, S.P.: Macroscopic electrical field distribution
and field-induced surface stresses of needle-shaped field emitters. Ultramicroscopy
111, 397404 (2011)
19. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York, NY (2000)
20. Smith, D.A., Birdseye, P.J., Goringe, M.J.: Forces on dislocations in field-ion specimens;
further analysis of some previous observations. Phil. Mag. 27(5), 11751181 (1972)
21. Loberg, B., Norden, H., Smith, D.A.: Observation of dislocations in a tungsten grain boundary
by combined electron and field-ion microscopy. Arkiv Fysik 40(38), 513519 (1969)
22. Larson, D.J., Miller, M.K.: In-situ micro-twinning of TiAl in the field ion microscope. Mater.
Sci. Eng. A 250(1), 7276 (1998)
23. Koelling, S.: Three dimensional compositional analysis of semiconductors with the atom
probe. Ph.D. Thesis, Katholieke Universiteit Leuven (2011)
24. Nishikawa, O.: Operation of the field ion microscope with neon. J. Appl. Phys. 35(10),
28062812 (1964)
108 4 Data Collection
25. Lawrence, D., Alvis, R., Olson, D.: Specimen preparation for cross section atom probe
analysis. Microsc. Microanal. 14(S2), 10041005 (2008)
26. Larson, D.J., Prosa, T.J., Lawrence, D., Geiser, B.P., Jones, C.M., Kelly, T.F.: Atom probe
tomography for microelectronics. In: Haight, R., Ross, F., Hannon, J. (eds.) Handbook of
Instrumentation and Techniques for Semiconductor Nanostructure Characterization,
vol. 2, pp. 407477. World Scientific Publishing, London (2011)
27. Larson, D.J., Lawrence, D., Olson, D., Prosa, T.J., Ulfig, R.M., Reinhard, D.A., Clifton, P.C.,
Kelly, T.F., Lefebvre, W.: From the store shelf to device-level atom probe analysis: an exercise
in feasibility. In: 36th International Symposium for Testing and Failuer Analysis, San Jose, CA
2011, pp 189197. ASM International
28. Prosa, T.J., Lawrence, D., Olson, D., Larson, D.J., Marquis, E.A.: Backside lift-out specimen
preparation: reversing the analysis direction in atom probe tomography. Microsc. Microanal.
15(S2), 298299 (2009)
29. Dieter, G.E.: Mechanical Metallurgy, 3rd edn. McGraw-Hill, New York, NY (1986)
30. Powell, R.W., Ho, C.Y., Liley, P.E.: Thermal Conductivity of Selected Materials. In: National
Standard Reference Data Series-National Bureau of Standards. U.S. Government Printing
Office, Washington, DC (1966)
31. Kellogg, G.L.: Determining the field emitter temperature during laser irradiation in the pulsed
laser atom probe. J. Appl. Phys. 52(6), 53205326 (1981)
32. Kellogg, G.L.: Measurement of activation energies for field evaporation of tungsten ions as a
function of electric field. Phys. Rev. B 29(8), 43044312 (1984)
33. Wada, M.: On the thermally activated field evaporation of surface atoms. Surf. Sci.
145, 451465 (1984)
34. Menand, A., Blavette, D.: Temperature dependence of iridium field evaporation rate. J. Phys.
47(C7), 1720 (1986)
35. Thompson, K., Bunton, J.H., Kelly, T.F., Larson, D.J.: Characterization of ultralow-energy
implants and towards the analysis of three-dimensional dopant distributions using three-
dimensional atom-probe tomography. J. Vac. Sci. Technol. B 24(1), 421 (2006). doi:10.
1116/1.2141621
36. Blavette, D., Duval, P., Letellier, L., Guttmann, M.: Atomic-scale APFIM and TEM investi-
gation of grain boundary microchemistry in astrology nickel base superalloys. Acta Mater.
44(12), 49955005 (1996)
37. Tsong, T.T., Ng, Y.S., Krishnaswamy, S.V.: Quantification of atom-probe FIM data and an
application to the investigation of surface segregation of alloys. Appl. Phys. Lett. 32(11),
778780 (1978)
38. Cerezo, A., Smith, G.D.W., Waugh, A.R.: The FIM100 - performance of a commercial atom
probe system. J. Phys. C9(45), 329335 (1984)
39. Menand, A., Al Kassab, T., Chambreland, S., Sarrau, J.M.: Atom-probe study of aluminum-
lithium alloys. J. Phys. C6(49), 353358 (1988)
40. Sha, G., Ringer, S.P.: Effect of laser pulsing on the composition measurement of an
AlMgSiCu alloy using three-dimensional atom probe. Ultramicroscopy 109(5), 580584
(2009)
41. Marquis, E.A., Yahya, N.A., Larson, D.J., Miller, M.K., Todd, R.I.: Probing the improbable:
imaging carbon atoms in alumina. Mater. Today 13(10), 4244 (2010)
Chapter 5
Data Processing and Reconstruction
5.1 Introduction
Data processing refers to taking raw atom probe tomography (APT) data and
preparing it to be reconstructed into real-space data, while reconstruction takes
processed experimental data and converts it into 3D spatial coordinates. Once this
has been successfully accomplished, the data can be interrogated for useful nano-
structural information.
The reconstruction algorithm (the act of projecting from detector hit positions
back to a reconstructed specimen surface) is powerful, yet very simple. Accurate
reconstruction of ion position requires an accurate description of the specimen
shape as it evolves during field evaporation; however, the simple assumptions at
the core of the reconstruction algorithms do not always capture all the detector-
to-specimen projection details. Some deviations occur because of the nature of the
bonding in crystallographic specimens, while others originate from evaporation-
field differences between material phases [14].
In this chapter, we will discuss the full data processing cycle for reconstruction.
Details of the voltage, shank, and tip profile reconstruction protocols are included,
as well as common aberrations that result from limitations to these protocols.
Finally, analysis of an arsenic-implanted silicon sample is included, for which the
available data-reconstruction options are discussed and evaluated with respect to
the impact on the expected spatial positioning accuracy.
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 109
DOI 10.1007/978-1-4614-8721-0_5, Springer Science+Business Media New York 2013
110 5 Data Processing and Reconstruction
After completion of each LEAP acquisition, STR and RHIT files are created. The
STR file replaces the RRAW file format in LEAP systems after the 4000 base model.
The STR file contains unfiltered instrumental data with an entry for every pulse
event. The STR is then processed to identify ion hits and stored as an RHIT file.
The IVAS Cal/Recon Wizard guides the user through a series of steps which
facilitates the conversion of raw RHIT data into reconstructed positions and
chemical identities. These steps include: Select Ion Sequence Range, Select
Detector ROI, TOF Corrections, Mass Calibration, Ranged-Ion Assignment, and
Reconstruction. The functions of these steps are discussed in the following sections.
First, the user is asked to choose an appropriate subset of the raw data for
processing. This is done by selecting an ion collection range and areal region of
the detector (Sect. 5.3.1). Then, the individual ion flight times are corrected for
variations due to actual flight path and voltage, resulting in the generation of a mass
5.3 Conversion from Detector Space to Specimen Space Coordinates 111
spectrum (Sect. 5.3.2). This allows the user to associate chemical identities and
volumes to each ion type (Sect. 5.3.3). Finally, each ion is reconstructed by
assigning it with a 3D position. The reconstruction process is covered by sections
that detail the basic reconstruction algorithm (Sect. 5.3.4), the available IVAS
reconstruction protocols (Sect. 5.3.5), details of how the specimen apex relates
to the specimen body (Sect. 5.3.6), and tools available to assess reconstruction
parameter selection (Sect. 5.3.7). Finally, at the end of the process, the ROOT and
POS files are created so that the reconstructed data are available for material
analysis.
during an experiment into a single mass (or time) spectrum which is usually
displayed as a histogram plot of ion count versus either m/n or TOF. This allows
for identification of particular ranges of events as belonging to a particular atomic or
molecular species (in the form of singly or multiply charged ions). Due to the physics
of field evaporation and instrument limitations, measured flight times must be
corrected for various effects. The corrections necessary to obtain the most accurate
TOF values vary from specimen to specimen and therefore require optimization
for each individual dataset. Although these specimen-to-specimen differences are
small, they can result in significant variations in mass spectral quality. The following
sections detail the procedures used to calibrate individually measured events into a
coherent and optimized mass spectrum. In IVAS these calculations are performed in
the TOF Corrections step.
Individual ions are identified by measuring their flight times to reach the detector
(time between voltage or laser pulse and ion arrival at detector). To first order, m/n
of the incoming ion is calculated by equating the kinetic and potential energies:
1 2 d m 2eV
mv neV and v ! 2 t2 5:1
2 t n L
where e is the electron charge, n is the charge state, V is the total voltage applied to the
specimen during evaporation, v is the speed, L is the flight distance, and t is the TOF.
The calculation of m/n is based on two assumptions: first, the potential energy of the
ion is entirely converted to kinetic energy by the end of the flight path, and second, the
acceleration of the ion up to its terminal velocity is instantaneous (see Appendix D for
a more detailed discussion). Correction terms have been suggested that can deal with
the incompleteness of the acceleration and with the amount of applied voltage
actually transmitted to the specimen [5], but the parameterized approach applied to
LEAP data (described below) generally achieves satisfactory results.
Since the data collected is a collection of times of flight, corrections are first
made to the raw TOF data (Fig. 5.3) and then those values are converted to m/n.
Without these corrections, unique m/n ions do not present a narrow TOF range, but
instead a wider distribution of TOFs that depend on ionization voltage and impact
location on the detector. As illustrated in Fig. 5.3a, the raw TOF spectrum here
consists of three broad peaks that originate from the groups of well-defined peaks
shown in the corrected TOF spectrum in Fig. 5.3b. Motivated by (5.1), a corrected
TOF may be expressed as a separable set of functions
where FV(VDC) accounts for TOF variations due to voltage (i.e., oversimplified
electrostatic assumptions, linear detector response, detector bias, etc.), FB(X,Y )
114 5 Data Processing and Reconstruction
Fig. 5.3 Example TOF spectrum (a) before and (b) after voltage and bowl corrections. Although
this dataset is composed of a number of well-defined ion species, variations in flight times, and
path length as a function of location on the detector and voltage cause the raw flight data in (a) to
appear broadly distributed. After application of appropriate corrections, the corrected TOF
spectrum in (b) exhibits well-defined flight times for each ion species which can be optimized
for each individual dataset
takes into account flight path length differences as a function of detector hit location
(i.e., an appropriately curved detector would not require this correction; hence this
is often referred to as the bowl correction), and (traw t0) provides a TOF offset
which is required due to various signal cables and electronic processing.
p
According to (5.1), the voltage correction to the TOF should follow a 1= V DC
dependence; however, a three-parameter fit of the following ad hoc form captures
the experimental variations more completely [6]:
a0
FV V DC q : 5:3
a1 V DC a2 V 2DC
Because the detector is flat and the detected ions effectively originate from a
point source, the flight distances (and consequently TOF) for a straight flight path
instrument increase outward from the center of the detector (see Fig. 5.4a). Much
like the voltage correction, a parameterized bowl shaped function, FB(X,Y ) is
defined and its shape is chosen to fit the 2D TOF distribution originating from a
single ion peak. A complex bowl shape is required because this function corrects for
more than simple geometrical variations in flight distances. It also accounts for
things such as detector nonlinearities which affect calculations of detected posi-
tions. Bowl asymmetry provides optimized TOF for each dataset and so the user is
allowed to choose the subdivision level of the bowl for each optimization effort.
An example of a typical bowl shape and the residual TOF values for an example
peak are illustrated in Fig. 5.4. Optimizations of the various corrections are applied
5.3 Conversion from Detector Space to Specimen Space Coordinates 115
Fig. 5.4 (a) An example bowl correction color coded to show the depth of the bowl (or TOF
in nanoseconds) as a function of location on the 2D detector. (b) The resulting flight time residuals
as a function of radial location on the detector. A properly applied bowl correction will result in
uniform and minimal residuals as a function of detector location
in step four of the IVAS Cal/Recon Wizard. The initial values for the various
parameterizations are based on previously optimized calibration files created
when a LEAP system is set up by a service engineer. The correction parameters
in the calibration files are also applied during live data collection on a LEAP
system.
As part of the TOF Corrections step, optimization of a given dataset proceeds by
first selecting an appropriate mass peak to maximize MRP (if the user does not
manually select a mass peak, the largest peak is selected by default). An appropriate
peak will include ions that are distributed across the entire FOV, else the optimi-
zation algorithm for the flight path (bowl) function may fail. Next, the parameters
defining the voltage correction FV(VDC) are optimized. Then, the parameters defin-
ing the bowl correction FB(x,y) are optimized. These voltage and bowl optimiza-
tions are iterated until MRP stops improving by a predefined threshold. During this
process, t0 remains fixed to the initial calibration value. Values of t0 have not been
found to change significantly from specimen to specimen, so fine adjustments to
this value are accounted for in the conversion from TOF to m/n (see next section).
After successful application of these corrections, the TOF spectrum should consist
of a collection of sharp peaks (Fig. 5.3b).
Once the voltage and bowl corrections have been satisfactorily completed, conver-
sion from TOF to m/n is performed (Mass Calibration step). Two strategies are used
for this conversion, the parametric fit method and the linearization method. The
parametric fit method is based on the relationship described by (5.1) and is usually
used only to create a calibration file. The linearization method, which assumes that
116 5 Data Processing and Reconstruction
the general relationship defined by the calibration file is reasonably accurate, applies
a piecewise continuous set of linear corrections for each two-point region defined by
the user-selected peaks and is the default method in IVAS.
is fit to the identified peaks (Fig. 5.5) to determine c and t0. This t0 is the same value
used during the voltage and bowl iterations when generating a calibration file, so it
may be necessary to go back to that step and re-optimize the voltage and bowl fits
given the new t0 value (especially if t0 changes by more than a few percent). Note
that when a single peak is chosen for fitting, t0 remains fixed and only c is adjusted
in (5.4). For two or more peaks, both parameters are adjusted. Because this is a
two-parameter fit, it is advisable to pick more than two peaks to adequately
transform TOF to m/n over the entire range of TOF values. As t0 tends to have
little variation from analysis to analysis, the parametric fit method is generally
suggested for calibration file generation only. However, when data have been
collected with an inappropriate calibration file (e.g. different operational mode,
tip geometry, or outside the calibrated voltage range), then this method is favored.
5.3 Conversion from Detector Space to Specimen Space Coordinates 117
This method uses the c and t0 values from the previously created calibration file.
Single or multiple peaks are selected by the user and corrected with the function
m=n ct t0 2 t, 5:5
where (t) is a piecewise linear interpolation function that ensures that all identified
peaks have the correct m/n and regions between identified peaks are interpolated.
This method, which applies a linearization on top of the parametrically optimized
correction, is usually superior for properly aligning spectra from multiple datasets or
when unknown or poorly resolved peaks dominate the mass spectrum. Choose three
or more high signal-to-noise, well-separated peaks, whenever possible to get the
maximum benefit from the mass spectral calibration. If comparing different datasets,
be sure to use the same calibration peaks in each one.
Peak identification in a mass spectrum can appear to be the most simple of tasks,
but ranging choices have a significant impact on analytical results. Chemical
identification comes in two steps. First, one must accurately associate a peak with
an ion type, and then an appropriate choice of lower and upper limits is made to
identify where the relevant peak signal is located. For appropriate range selections,
a mass calibration must have been successfully completed.
Peak identification may take only minutes, for a simple or familiar mass
spectrum, or may take hours for unknown, complicated mass spectra. Complications
in mass spectra originate from the variations in elemental species and the number of
isotopes for each element, the distribution of charge states, the frequency of molec-
ular ion production, the frequency of closely timed (or overlapping) events, and peak
shapes. IVAS provides several tools for the identification of peaks including an
elemental overlay tool, an ion overlay tool, and a peak decomposition algorithm
(refer to Chap. 6 and the IVAS User Manual for more information).
Peak identification is approached in terms of the level of complexity. Simple
identifications include those peaks that are isolated and can be identified by simply
looking up masses in a periodic table where relative peak intensities are consistent
with the known isotopic distributions (isotopic fingerprints). It is useful to recog-
nize that peaks appearing near or fractions of a Dalton must represent doubly
or quadruplely charged ions. This allows the user to simply multiply m/n values by
2 or 4 to find the true mass of the ion (e.g., 13.5 Da could be 27Al2+). Peaks found
near the 1/3 or 2/3 fractions of a Dalton represent triply ionized ions. Again,
multiplying by three then provides the mass of the ion (e.g., 10.33 Da could be
31 3+
P ). Reference [7] provides tabulation of the most likely charge states as a
function of field for a variety of elements.
118 5 Data Processing and Reconstruction
Fig. 5.6 This example mass spectrum from a sample containing Ti, Fe, Co, and Ni ions illustrates
typical mass ranging challenges. The five distinct Ti peaks pose no ranging issues as the ranges
necessary to identify these five isotopes do not conflict with any other elements. This is not the
case for the Fe, Co, and Ni peaks. The region near 58 Da contains both 58Fe and 58Ni isotopes.
Likewise, a significant fraction of the 59Co peak contains the tail from the 58 Da region. Therefore,
careful consideration is necessary for the ranging boundary of this particular element
Fig. 5.7 This schematic shows the geometrical relationship between the tip radius, R, and the
image compression factor, , which is directly related to the angular magnification of the tip image
APT data are collected as individual ions. Data reconstruction consists of calculating
3D positions in specimen coordinates (referred to as real space, and denoted as lower
case x, y, and z) from a time-series of 2D positions in detector coordinates (referred
to as detector space, and denoted in upper case X and Y). The standard reconstruc-
tion methodology [1016] involves calculating a magnification in order to convert
from detector hit positions (X and Y) into real-space coordinates (where x and y are
contained in the plane normal to the specimen axis at the current apex position). The
depth of the ion along the specimen axis (z) is based on the assigned ionic volume
and the estimated imaged area of the specimen surface.
In a point projection, the magnification (M ) is simply the distance (L ) between
the sample and the detector divided by the radius of curvature of the specimen (R).
Ion trajectories from a needle (not an isolated point in space) are compressed by the
Image Compression Factor (), Fig. 5.7 and (5.6), which reduces the magnification,
thus the projection magnification is:
L
M 5:6
R
Note that expressions accurate for wide-FOV acquisitions are used in the Recon-
struction step in the IVAS Cal/Recon Wizard and are derived in Appendix C;
however, for the sake of simplicity small FOV expressions are used for illustrating
the main points in this section.
5.3 Conversion from Detector Space to Specimen Space Coordinates 121
It is generally accepted that the nominal value of is between one and two.
Deviation from unity is attributed to the effects of the shank region of the specimen
compressing the electric field lines inwards toward the detector as shown in Fig. 5.7.
Before the z-coordinate value for each ion is calculated, an estimate of the imaged
specimen surface area is required. This is accomplished by converting the detected
surface area (A) to specimen surface area using a geometrical model of the apex
and assuming the magnification computed above. For each reconstructed ion, the
spherical model z-coordinate is adjusted by a small increment equal to the atomic
volume () divided by the image areathe volume of the reconstructed ion is
spread uniformly over the entire apex. The z increment is also a function of the
Detection Efficiency () [11, 12]:
L2
z : 5:7
A2 R2
Prior to the advent of wide-FOV instruments [17, 18], the geometric assumptions
described by Blavette [10], and later applied by Bas [11], were widely considered as
the standard reconstruction technique. This model assumes that the original shape of
an acquired volume is small enough in lateral extent to be considered cylindrical and
the radius of curvature of the specimen is determined ion-by-ion from the specimen
voltage. In early reconstruction algorithms, the shank angle was assumed to be zero
in the calculation of the volume increment, although Walck [19] published a
description of specimen geometry that could be used to compute depth increments
that are appropriate for a wide FOV and included shank angle effects. These effects
were later taken into account by Geiser et al. [14].
A complete review of the current state of APT reconstruction, which currently is
a topic of much research within the atom probe tomography community, is beyond
the scope of this chapter, but some relevant references are provided for the
interested reader [3, 4, 1214, 16, 2027].
Instrument Parameters
Figure 5.8 shows the controls available on the Expert Tab within the Reconstruction
section of the IVAS Cal/Recon Wizard. Many of these variables are functions of
each other. The parameters at the top of the figure are those that might generally be
expected to depend on the instrument configuration.
For example, Detector Efficiency is an estimate of the overall detection
efficiency due to the MCP efficiency (and reflectron mesh when present). The
122 5 Data Processing and Reconstruction
Fig. 5.8 The IVAS Cal/Recon Wizard Expert Tab contains all the parameter details necessary for
reconstruction
efficiency has an obvious impact on depth scaling as indicated by (5.7). The depth
scaling applied ion-by-ion as reconstruction occurs is driven by the accumulation
of atomic volumes spread out across the apex FOV. The efficiency correction is
applied by scaling each ionic volume by the system efficiency factor to account for
undetected atoms. When the specimen radius evolution is driven by a model like
fixed shank angle or tip profile, an incorrect efficiency can promote errors in tip
radius by evolving the radius too rapidly or too slowly with accumulated volume.
The k-factor (k) [28], is important in the estimation of the Initial Tip Radius, R0,
from the applied voltage difference, V, between the specimen and the local
electrode:
V
R0 5:8
kF
where F is the Evaporation Field of the material being analyzed. The k-factor
depends upon geometric considerations such as shank angle and radius of the
specimen and separation distance between the specimen and local electrode [29].
5.3 Conversion from Detector Space to Specimen Space Coordinates 123
Material Parameters
Based on ion count, the highest concentration element becomes the Primary
Element and it will be used to automatically select a default evaporation field as
shown in Fig. 5.8. This value is then used to calculate the Initial Tip Radius from the
initial voltage as determined by the starting point selected in the Select Ion
Sequence Range step. The Primary Element will also be used to suggest an average
Atomic Volume to use for reconstruction if element-specific atomic volumes are not
used. If the material is homogeneous over the length scales of interest and has a
known ionic volume, one should deselect the Element-Specific checkbox and use
the known value. An information field next to the Primary Element indicates the
temperature under which ions were collected while an information field immedi-
ately to the right of the Evaporation Field indicates the temperature at which the
estimate of the Evaporation Field is based. When there is a significant difference
between these two temperatures, or when the material is an alloy, or when data
are collected in laser pulsed mode, adjustment from the default field value is
warranted.
124 5 Data Processing and Reconstruction
Specimen Parameters
This section of the Reconstruction tab is focused on specifying the tip radius
throughout the reconstruction. There are three different modes of Radius Evolution
that can be chosen: Voltage, Shank, and Tip Profile.
The Voltage Evolution mode uses (5.8) above so that the tip radius used for
reconstruction is strictly based on the total voltage for a given ion. The Evaporation
Field and k Factor directly update the Initial Tip Radius as they are modified, but if
desired, the user can directly modify the Initial Tip Radius by simply entering a
numerical value.
The Shank Evolution mode uses the same Initial Tip Radius as the Voltage
Evolution reconstruction, but the evolution of the tip radius is determined by the
specific choice of geometric parameters, as described in Appendix C. An estimate of
average shank angle is available which is based on the evolution of the voltage curve
and assigned atomic volumes. The correspondence between the shank angle used
in the reconstruction and the physical shank angle (as measured from scanning or
transmission electron microscopy (SEM, TEM)) will depend on how the field
evaporated end form intersects the specimen shank. Historically, the reconstruction
geometry required a tangential continuity between the hemispherical apex and the
shank. This could require entering a shank angle which was significantly different
from the measured shank angle to achieve a reasonable reconstruction because many
specimens exhibit an apex that is significantly blunter than the diameter of the conic
shank on which it resides. Some reconstruction algorithms allow the specification of
this discontinuity in the form of the sphere-to-cone ratio. The ratio is calculated
from the ratio of the radius of the hemispherical end form to the radius of the shank
where the two intersect and results in a specified shank angle more closely
corresponding to the physical shank angle. This enables high magnification SEM
or TEM images to be used successfully to set shank angles and initial tip radii. The
use of the SphereCone Radius Ratio parameter does not change the reconstruction
in any way other than scaling the constant shank angle entered by the user (for more
on the SphereCone Radius Ratio see section (Sect. 5.3.6) below).
Although the voltage and shank angle reconstructions are simple and powerful,
there may be a need to deviate from voltage or single shank angle prescriptions for
radius evolution as a function of depth. For example, multi-layer materials
exhibiting multiple evaporation fields can require significant deviation from a
single evaporation field relationship. Similarly, a lengthy analysis with evolving
shank angle and multiple evaporation fields is frustrated by a single shank angle
reconstruction. By obtaining a high quality electron microscopy image prior to
LEAP analysis, a radius evolution function can be based on the actual tip profile as
shown in Figs. 5.9 and 5.10 [30]. This approach can take into account the evolving
shank and mitigate some of the issues regarding evolving evaporation field. How-
ever, reconstruction challenges caused by non-hemispherical and/or evolving tip
shapes still remain.
The Tip Profile reconstruction is implemented during the final step in the IVAS
Cal/Recon Wizard (Figs. 5.9 and 5.10 show the image and corresponding radial
5.3 Conversion from Detector Space to Specimen Space Coordinates 125
SEM, and subsequently analyzed in a LEAP system. This procedure allows for
simple identification of R0 for the subsequent reconstruction. Figure 5.10 is a more
typical example where only a post-analysis image is available and some uncertainty
exists for placement of the R0 contour. For any analysis that does not start with a
field evaporated endform and does not utilize all the ions in the analysis, this
uppermost tracer must be approximately placed to accommodate those realities.
In this case, the sample did not end in failure, so a post-LEAP-analysis image was
taken and RFinal was measured. Variations of the R0 tracer can then be made until
the resulting RFinal matches the measured value in the SEM.
5.3 Conversion from Detector Space to Specimen Space Coordinates 127
Fig. 5.11 (a) Tangential continuity is shown where the sphere that represents the tip
surface intersects the shank () so that the intersection is smooth fs/c 1. (b) This tangential
discontinuity better reflects physical field evaporated surfaces where the sphere that represents the
tip surfaces intersects the shank nonsmoothly fs/c 6 1
We note that the user is not constrained to the tip profile for defining the radius
evolution. An experienced user might prescribe deviations from the tip profile near
interfaces known to produce non-hemispherical tip shapes [26]. Such deviations
might better approximate the radius evolution as the tip shape becomes highly non-
hemispherical or even non-singly curved during analysis through these regions and
result in more accurate reconstructions. In this sense, the tip profile method might
also be considered a custom radius evolution method.
An important advantage of the tip profile method is that the variables, k-Factor
and Evaporation Field, which can have significant uncertainties, are not used. With
additional images, additional constraints (R0 and/or RFinal) and the extent of tan-
gential discontinuity (see next section) may be measured and utilized. High mag-
nification images enable more accurate rendering of the profile, but some caution
must be used because of possible negative interactions with imaging and APT
analysis yield [31].
Rsphere
f s=c 5:9
Rcone
Using this lessened geometric restriction the evolution of the specimen radius
with analysis depth is described in Appendix C (17). This more general geometry is
illustrated in Fig. 5.11b.
As shown in Fig. 5.8, the option exists for the user to disable the tangential
continuity condition and directly enter the SphereCone Radius Ratio value. The
assumption of tangential continuity biases reconstructions toward parameters with
unphysically large shank angles. The redefinition of the dR/dZ term to allow for a
tangential discontinuity enables APT reconstructions to take advantage of electron
microscopy-based information about the actual geometry of specimen shank angle
and the level of discontinuity between the apex and the shank, both of which may be
used to provide constraints on the selection of other reconstruction parameters.
Reference [25] explains the effect of relaxing the tangential continuity constraint on
the reconstruction evolution in further detail.
Range and Ion Sequence Center allow the user to specify a sequence number range
for reconstruction. As illustrated in Fig. 5.12, an Ion Sequence Range of 50 %
(25 %) around the Ion Sequence Center of 50 % results in a range of 2575 % of
the data to be included.
The same reconstruction parameters available on the Expert tab are available in
the Reconstruction Explorer, as shown in Fig. 5.8. When dragging the slider bar for
any of the parameters, new reconstructions are only triggered upon release of the
slider. Note that at least one parameter must have changed from the original settings
when entering the Reconstruction Explorer in order for any parameters to be
transferred to the Expert tab upon exit.
When generating spatial distribution maps (see Chap. 6 for further details), the
Center Element specifies the species of ions around which the average neighbor-
hood intensities are computed. The Neighbor Element specifies the species of ions
considered as nearest neighbors for the distributions plotted. The Neighborhood
Extent specifies the size of the neighborhood plots to be created. Z Cut in XY Map
describes how close in the Z neighborhood ions must be to the Z Plane Offset in
order to be included in the plot. For example, by default, the Z Plane Offset is 0.0
(neither above or below), so neighbor atoms must have a z coordinate within
0.01 nm of the center atom in order to be included in the TM. Changing the
Z Plane Offset allows the XY distribution of nearby layers of ions to be observed in
well-resolved crystalline materials.
In the visual display region of the subvolume reconstruction, Fig. 5.12b, various
orthogonal projections of the current reconstruction can be chosen. Additionally,
for real-space projections of the reconstruction, the rendered depth can be adjusted
by the application of cut-planes, controlled by the double-ended scroll bar. If an
SDM has been generated, the user can choose between the three 1D projections
and three 2D projections. The 1D X neighborhood, the 1D Y neighborhood, and the
2D XY neighborhood plots are all subject to the Z distance cut from the center
atoms. The Z, ZX, and ZY plots are not subject to any cuts on distance from the
center atoms. By default the reconstruction plot aspect ratio will be physically
correct, but by selecting Fit to Window under Plot Aspect Ratio, the display can be
automatically rescaled to more completely fill the display area.
At the Reconstruction step, Fig. 5.8, the user has the option to create multiple
reconstructions using the Create Preview button. No information is saved however,
until the Save Reconstruction button is used to create the ROOT file which will
preserve all of the work done until this point. This file contains all the necessary
information to create a POS file from the RHIT file. By default (Open Analysis
checkbox option), the POS file is also created and opened in IVAS for further
analysis.
5.4 Discussion of Spatial Resolution and Spatial Positioning 131
Ideally, spatial resolution is a local measurement that should express the perfor-
mance of the instrument and not limitations originating from specimen or recon-
struction imperfections. The reality for APT spatial resolution is that it is not really
possible to decouple specimen effects from instrument effects because, as stated in
Chap. 2, the specimen itself is the primary optic of the microscope. Known imper-
fections that degrade resolution include: (1) specimen temperature, (2) detector and
optical limitations, (3) reconstruction errors, (4) trajectory aberrations in monophase
materials, (5) local magnification in polyphase materials, as well as others that will
not be discussed in this section. The spatial resolution achieved in a given dataset
will vary within the image and may never reach the ultimate potential of the
technique. Nonetheless, it is instructive to consider the best demonstrated spatial
resolution as a starting point and then consider each imperfection individually.
Vurpillot et al. [36] defined spatial resolution in APT as the inverse of the width of
a damping function convoluted with the reciprocal-space structure of an atom probe
dataset. This parameter gives information about the uncertainty in interatomic
distances as determined by APT. Geiser et al. [37] introduced the spatial distribution
map as a method of averaging real space information about interatomic distances in a
dataset and measured interatomic separation variations in the data. Kelly et al. [38]
summarized these approaches and proposed a standard measurement of spatial
resolution using tungsten specimens. Gault et al. [39, 40] proposed an expression
for spatial resolution in crystalline systems based on statistical considerations of
planar spacing and temperature.
These definitions all utilize to the same essential information to describe spatial
resolution: real-space atom-location uncertainty or the reciprocal-space counter-
part. Furthermore, there are two physically different mechanisms involved with
resolving information in the longitudinal and lateral directions of APT, and mea-
surements for both are required. Each of the references above concludes that a
lateral resolution of ~0.15 nm has been demonstrated in tungsten at ~50 K. Simi-
larly, the longitudinal (depth) spatial resolution has consistently been found to be
about 0.06 nm.
As mentioned previously, there are many factors that can cause the observed spatial
resolution to be less than optimal. They will be discussed individually in the next
sections.
132 5 Data Processing and Reconstruction
The position-sensitive detector used in an atom probe must have sufficient pixel
resolution to realize the ultimate achievable resolution. For example, in a 100 nm
diameter FOV, there will be about 500 atoms on the diameter of the image. The
detector will need at least 1,000 pixels across the diameter to position atoms with
precision. Stated differently, 1,000 pixels in 100 nm means that 0.1 nm spatial
resolution is the approximate geometric limitation of the detector. As specimens
blunt during analyses, magnification decreases correspondingly, and a greater
number of atoms are imaged on the detector diameter. It is therefore possible that
in the early part of an analysis the optimal spatial resolution is achieved but in the
latter stages of an analysis, the spatial resolution is limited by detector pixel size.
Optically, the spatial resolution can be limited by voltage instabilities that cause
variations in magnification. These aberrations affect only the lateral resolution.
In instruments using reflectron energy compensators [4549], ions may pass
through a metallic mesh causing a measureable loss in spatial resolution.
5.4 Discussion of Spatial Resolution and Spatial Positioning 133
Fig. 5.13 Schematic diagrams of the projection for (a) a hemispherical model and (b) a simulated
field evaporated model
The next level of complexity that limits spatial resolution is an aberration found in
crystalline materials. Atoms generally field evaporate from the edges of closely
packed planes and not from the interior of a plane [41]. As a result, in areas of high
atomic packing density, planar facets form on the evaporating specimen. These
planar facets create deviations from a smooth surface and result in local nonuni-
formities in image compression as a function of the azimuthal angle about a central
axis of the specimen [60, 61].
In principle, not only is it necessary to take all of the above effects into account, but
also it must be done for phases within phases. Using a previously described method
[24, 26, 62, 63] we have simulated the cases of a high-field precipitate in a low-field
matrix (matrix evaporation field is half that of the precipitate evaporation field) as
well as the case of a low-field precipitate in a high-field matrix (matrix field is 15 %
higher than precipitate field) [64]. Consider the former case of a high-field spherical
precipitate as shown in Fig. 5.14. After some evaporation, the specimen will form
an apex structure as shown in the series Fig. 5.14ac. Once the second phase
emerges at the surface, the entire endform is altered (with the high field phase
protruding from the surface) and the trajectories of ions in both the matrix and the
precipitate are different from those in a monophase material. In general, the atoms
of a high-field phase will see greater angular magnification than the low-field phase
[1, 2, 65]. If the evaporation field of the second phase is less than the surrounding
material as shown in the series Fig. 5.14df, the situation is reversed, and the
precipitate region is flattened on the surface. For both cases, once sufficient material
is evaporated following the disappearance of the precipitate, the specimen resumes
its equilibrium electrostatic endform.
The errors that occur when using standard reconstruction methods may be
evaluated by plotting the original atom positions (from the unevaporated simulation
data) vs. the detector impact positions from the simulation [26]. These data are
shown in Fig. 5.15 for the precipitate shapes presented in Fig. 5.14b, e. The slope of
the curves in Fig. 5.15 provides an estimate of the magnification. The data in
5.4 Discussion of Spatial Resolution and Spatial Positioning 135
Fig. 5.14 Simulated evaporated specimen shapes for (ac) a precipitate with an evaporation field
twice that of the matrix and (df) a precipitate with an evaporation field 15 % lower than the matrix
Fig. 5.15a suggest a very high magnification for the region of the precipitate near
the edge of the matrix where there is a narrow range of original atom positions
spread over a wide range of detector impact positions and a narrow region of
potential ion crossing in the matrix near the precipitate (arrowed). The results in
Fig. 5.15b suggest a low magnification in the precipitate region with a narrow
region of higher magnification in the matrix near the precipitate. Note that the solid
lines in the curves in Fig. 5.15 denote data that would be obtained if there were no
precipitates present in the simulations.
In either case, the effects of the resulting shapes on ion trajectories from either
phase raise the possibility of ion trajectory crossing [66]. If ion paths do cross, the
task of correcting this aberration gets more challenging and may be uncorrectable in
a traditional ion-by-ion treatment.
Reconstruction of Precipitates
Fig. 5.16 An array of reconstructions for a dataset containing Mg precipitates in a GaN matrix
created using shank angle (10 ) reconstruction mode with 70 % efficiency and fs/c of 1.2 with a
variety of initial radii and values. The array is organized with fixed radii columns and fixed
rows (see text for details). Note that all reconstruction volumes have been scaled to ~80 nm in the
z dimension
of evaporated specimen shape (radius and shank angle), in order to minimize the
errors. In the case where no additional information is available for a starting
reconstruction, the user can make multiple reconstructions like those shown in
Fig. 5.16 and visually select the one which results in spherical precipitate shapes
and also maintains the image compression near the nominal mean value of the
instrument.
138 5 Data Processing and Reconstruction
In this section we present field evaporation simulation examples for thin films in two
different specimen orientations: plan-view and cross-section (see Chap. 2 for details
about specimen orientations) so that the user may understand how to interpret the
results of reconstructions obtained for such structures. While the cross-section
orientation may seem more simple due to its steady-state nature, for a zero shank
angle case, it has the more difficult condition of not possessing azimuthal symmetry.
Two cross-section examples consisting of ABA and BAB layered structures
(each layer is 5 nm) were simulated using an electrostatic simulation [3, 24, 62, 63]
similar to that developed by Vurpillot et al. [69, 70]. Each specimen was assumed to
be face-centered cubic and oriented in the <001> direction with a lattice parameter
equal to that of aluminum. The evaporation field for material A was chosen to be
20 % higher than that of material B and the initial shape of the specimen was a
hemisphere for each case. Figures 5.17a, b show the evaporated specimen surface
and the simulated detector field evaporation patterns for the ABA and BAB cases,
respectively [64]. In Fig. 5.7a, a flattening of the low field B region is evident in the
specimen center. This causes a lower magnification in this region (compared to
the A regions) and thus, in the hit-map, the B region is compressed compared to the
A regions. For the case of BAB, Fig. 5.17b, a protruding region of high field A
material in the specimen center is observed, with the same result in the hit-map;
that is, the higher density regions correspond to the B material, which has flattened
5.4 Discussion of Spatial Resolution and Spatial Positioning 139
at the surface in both cases. The important thing for the user to understand in these
cases is that, due to the anisotropic magnification across the surfaces, the B layer
in Fig. 5.17a will be reconstructed as narrower than 5 nm while the A layer in
Fig. 5.17b will be reconstructed as wider than 5 nm.
For the plan-view thin film example, a simple five layer system was simulated [3].
Image hump model evaporation fields [51] were used (30 V/nm for Cu, 37 V/nm for
Co, 27 V/nm for Cr and 33 V/nm for Si) for a simulation of a plan-view multilayer
structure containing layers of Si Substrate/Cr(5)/Co(5)/Cu(5)/Co(5)/Cr(50) (num-
bers in parentheses indicate layer thickness in nm), Fig. 5.18a. The images in
Fig. 5.18a show the shape of the specimen as it is evaporated through the layers.
An estimate of the radius of curvature (measured on the specimen axis) during the
evaporation of this structure is presented in Fig. 5.18b. A highly non-monotonic
result is observed as the specimen radius (on axis) varies from ~20 nm to nearly
140 nm during the course of the shape evolution [3]. Note the differences in the
behavior of the specimen radius when comparing the case of evaporation from
low-field material to high-field material with the opposite case. When a low-field-
on-high-field interface is encountered the specimen radius becomes very high, while
evaporation of a high-field-on-low-field interface produces a very low radius [24]
and a high magnification.
140 5 Data Processing and Reconstruction
It is often the case when reconstructing complex datasets that the user has some
known information regarding portions of the microstructure. Examples include an
interface that is nominally planar, a measured thickness of a region or phase, the
shape of precipitates, or combinations of any of the above. In this section we will
present and discuss the trends in interface curvature and layer thickness with
variation in the reconstruction variables: Initial Tip Radius, Image Compression
Factor, Shank Angle, and Sphere-Cone Radius Ratio (R, , , and fs/c). Our
reference sample is a repeating layered structure of Ni (57 nm) and Cr (56 nm)
layers from the National Institute of Standards and Technology Standard Reference
Material (NIST SRM 2135C). Figure 5.19 shows the LEAP data on which the
analysis below was performed.
5.4 Discussion of Spatial Resolution and Spatial Positioning 141
Fig. 5.20 Variable radius analysis of the NIST NiCr thin films. (a) R0=75nm, (b) R0=100nm,
and (c) R0=125nm
Figure 5.20 shows three reconstructions using increasing initial radii from 75 to
125 nm. The reconstructions shown in this figure were created using 15 ,
1.5, and fs/c 1.5. Varying the radius changes both the curvature of the upper
Ni/Cr interface as well as the Cr layer thickness. For the purposes of this discussion
we will define an interface curvature like that in Fig. 5.20a (concave up) as negative
(with a flat interface having a curvature of zero). The trends in the reconstruction
variables on the upper Ni/Cr interface curvature and the Cr layer thickness are
shown in Fig. 5.21 (for 15 ). Note that a smaller shank angle ( 5 ) has
the effect of shifting the reconstructed Cr layer thickness to greater values while
only slightly shifting curvature values to be more negative, Fig. 5.22. Both and
R have a significant effect on the curvature and thickness values in these examples
while the effect of fs/c is less evident. The best reconstruction parameters in this
situation both minimize the curvature and provide the correct Cr thickness (56 nm).
By fitting a polynomial model using all of the input parameters to the reconstruction
and investigating the (R, ) solution space to simultaneously optimize both Cr
thickness and Ni/Cr interface curvature, we can draw several conclusions:
1. Being constrained to a small shank angle of 5 with tangential continuity
( fs/c 1.0) forces the best fit (intersection of the minimum curvature and the
correct Cr thickness) toward low radius values (~75 nm) and high image com-
pression (~2.1).
2. Shifting to greater fs/c (1.5) slightly shifts the intersection (in R, space) of the
minimum curvature and the correct Cr thickness.
3. A greater has more impact than fs/c on shifting the minimum curvature and the
correct Cr thickness toward solutions that are parallel (in R, space).
4. Finally, greater fs/c and greater shank angle values shift the predicted lines of
minimum curvature and correct Cr thickness on top of each other, thus resulting
in a series of (R, ) solutions that both minimize curvature and result in the
correct Cr thickness.
142 5 Data Processing and Reconstruction
Fig. 5.21 Trends with specimen radius for curvature and Cr layer thickness using 15 for
(a and c) fs/c 1.0 and (b and d) fs/c 1.5. The dashed lines show the nominally correct values
for Cr thickness and curvature
In APT datasets, the reconstructed density is not uniform due to specimen endform
deviations away from the hemispherical shape that is assumed in the reconstruction
protocol. Improving the aberration by gridding the data and smoothing the density
variations (density relaxation, with the objective of density correction) has been
proposed by various authors as a way to rectify these errors [23, 67, 71, 72]. IVAS
enables the user to perform Density Relaxation on a ROOT or POS file. For the
former, the feature requires an active link to the corresponding RHIT file. The user
can perform the density relaxation only after a reconstruction has been completed
(see the IVAS User Manual for further details). The output from this procedure is a
new POS file. A brief explanation of the variables used in the density relaxation
procedure is included below:
Smooth Iterations specifies how many smoothing iterations the software
performs. In order to attempt to affect different scale features, different XY
gridding is used (both grid-size and grid-orientations are randomly varied) for
5.5 A Word on Density Relaxation 143
Fig. 5.22 Trends with specimen radius for curvature and Cr layer thickness using 5 for (a and
c) fs/c 1.0 and (b and d) fs/c 1.5. The dashed lines show the nominally correct values for Cr
thickness and curvature
each iteration. Note that processing time for the correction increases with
increased iterations. Artifacts can become visible if too many iterations are
performed. This parameter applies only to density relaxation using the XY
Relax option.
Ions Per Shell specifies how many ions to relax per depth shell. Smaller values
allow smoothing of smaller gaps, but can cause low-statistics artifacts. Larger
selected fiducial regions, from the Detector Event Histogram, require a higher
Ions Per Shell value than do smaller regions. Ion values of 5,000 for small fields
of view and 20,000 for larger fields of view detectors are typical. This parameter
applies only to density correction using the XY Relax option.
Z Extent is a slider bar for adjusting the percentage of the reconstruction data or
POS file volume to be density corrected. For ROOT files, the reconstructed apex
of the specimen is 0 %, and the end of the dataset is 100 %. This parameter limits
the density-corrected POS file output to a particular specimen-depth ROI.
Do MCP Correction is a toggle switch that allows the user to prohibit ions from
being relaxed into any low-efficiency microchannel plate region contained in the
data. By default, the density correction procedure relaxes ions into the MCP
144 5 Data Processing and Reconstruction
Fig. 5.23 Example of a transistor atom map (a) before and after (b) density relaxation
lateral relaxation. The light gray points are Si ions and the dark gray points are O
ions. It is believed that the oxide layers are artificially broadened due to local
magnification (as illustrated in Fig. 5.17). Correcting the resulting lower local
density restores the oxide layers to a more physically correct proportion while the
overall size of the structure is approximately preserved. Also note that the edge
atoms are locked in place and thus subject to more distortion as the rest of the ROI is
corrected.
The 3D reconstruction of ion positions is most often performed using one of the
three radius evolution approaches described previously: Voltage, Shank, or Tip
Profile. The number of user-selected reconstruction parameters varies depending
on the algorithm chosen, as summarized in Fig. 5.24. Extra measurements can be
taken to establish, or narrow the range of, the values of some of these parameters;
however, because the algorithms are approximations and do not capture the exact
field evaporated shapes for real materials, nonphysical parameter selections may
result in more accurate reconstructions. In this section, general considerations for
selection of these parameters are discussed in the context of a relatively simple
material system, the National Institute of Standards and Technology Standard
Reference Material (NIST SRM 2134), which is As-implanted silicon.
Fig. 5.24 Reconstruction process flow chart including the relevant reconstruction parameters
(highlighted in yellow)
146 5 Data Processing and Reconstruction
Fig. 5.25 Specimen, MCP detector, and resultant detector hit density map showing ion trajecto-
ries and their path into the MCP. The resulting efficiencies are reflected in variations in the hit
density map. A low-efficiency region of the detector is highlighted by the dashed-red circle where
ion trajectories are such that they pass through the top MCP plate without causing a significant
electron cascade
mesh in the flight path which further reduces the efficiency in proportion to the
transmissivity.
Hardware limitations are not the only factors which affect efficiency. For all
available reconstruction algorithms, efficiency modifies the z increment of each
ion (5.7 above) to compensate for the missing volume of undetected ions. For
analyses that have a significant number of events that are not contained in ranged
peaks, this is effectively a loss in detection efficiency. In such cases, the overall
efficiency should be modified to take this into account. The actual percentage of
unranged peaks can be measured, or the constant time-independent background
estimate provided by IVAS can be used to estimate both the unranged ions and
those missed while the detector is off between pulses:
f
0 5:10
f Bgd
Image compression has been defined previously and results from a combination of
flight path, specimen, and electrode geometry effects [74]. Figure 5.26 shows the
modeled variation in image compression for various tip dimensions and shank
angles for the straight flight path configuration [29] (note that the LEAP
reflectron configuration of reflectron and detector leads to differences in these
variations) as well as experimental estimates based on hit-map and pole projection
evolution [31]. IVAS implements a constant for the entire reconstructed volume,
148 5 Data Processing and Reconstruction
Fig. 5.26 (a) Modeled 4000X Si as a function of tip radius for various shank angles (Reprinted
with permission from Elsevier [29]). (b) Experimental 4000X Si as a function of voltage as
determined by variation in magnification of crystallographic features
which may limit reconstruction accuracy for analyses with large changes in voltage
(specimen radius). A reconstruction approach where a single dataset of large
voltage range is broken into multiple reconstruction regions, each with a limited
range and thus better approximated by a constant , may improve overall accuracy.
This limitation is mitigated in specimens that have narrow analysis voltage range.
Specimens with nearly parallel sides (limited radius evolution during analysis) are
more accurately reconstructed using to single value, although such geometries
may be more prone to premature failure or provide poor thermal/electrical proper-
ties due to their narrow cylindrical shape.
The most commonly adjusted parameter is R0, which is determined by using (5.8).
Unfortunately, the product (kF) is prone to large uncertainties. The k-factor may
vary substantially depending on the specific local electrode-specimen geometry.
Likewise, evaporation field tables exist [51], but they are typically model driven
and list values at a single temperature. Even in tables which include experimental
measurements, temperature variation is not normally reported and thus these values
can only be considered as estimates. Experienced users commonly choose values
based on past experience. If there are features in the ROI, such as a planar interface
or spherical precipitates, then R0 is often adjusted until these geometries have the
correct shape.
At this point, it is useful to summarize the effects of and R0 on reconstruction.
detector efficiency affects only the z-scale of the reconstruction. For a voltage
reconstruction, changing the efficiency expands or contracts the z-positions of the
ions and does nothing else. For shank angle and tip profile reconstructions, effi-
ciency changes the depth of the reconstruction and affects regions of the radius
evolution function it is more complicated than just skewing the z-axis, but the net
5.6 Reconstruction Case Study: NIST Standard Reference Material 2134 149
effect is similar. On the other hand, and R0 both tend to change the shapes of
geometrical features and their effects can overlap [29], so users commonly fix one
parameter () and adjust the other to achieve an accurate result as interpreted by the
user. To some degree, evaluation of validity of the reconstruction by the user is
generally an unavoidable requirement of current reconstruction algorithms.
Much of the anecdotal knowledge for the physical basis of is related to the
electrostatic relationship between the specimen and the aperture. It should be noted
however that other optical elements contribute to as well, in particular detector
bias and reflectron elements [47].
These final two parameters are only relevant for nonvoltage reconstructions and
are determined from the shape of the specimen. The shank angle or evolution of
the shank angle (see Appendix C for details of calculation of shank angle) and the
sphere-to-cone ratio can be used to dictate the radial evolution of the reconstruction.
Because apex curvature can become complicated (nonspherical) by nonuniform
evaporation fields within a single specimen, deviations from the actual tip profile
may be required to provide for more accurate reconstructions than the physical
profile would provide.
For these particular specimens, smooth variations in analysis voltage suggest that the
reconstruction algorithm plays only a minor role in final reconstruction accuracy.
Since voltage reconstruction assumes a constant relationship between voltage and
radius, let us consider the voltage evolution and silicon charge state ratio (CSR or Si2
+
/Si+), which provide some indication of changes in evaporation field [7]. Figure 5.27
shows the silicon CSR evolution and analysis voltage and as a function of detected
ion number (or depth). Except for the region near the native oxide, both curves
follow a similar trend. Ignoring the region near the native oxide, the CSR evolution
curve is easy to understand. When the overall specimen shape is small, the heated
volume is small. As the analysis proceeds, the specimen radius increases, the heat is
spread throughout larger and larger volumes providing for a lower peak apex
temperature (for fixed laser pulse energy and absorption). Under the constraint of
an approximately constant evaporation rate, a lower evaporation temperature
requires a higher evaporation field, which is exhibited by the increase in CSR.
Effectively, the evaporation field is changing as a function of analysis depth.
5.6 Reconstruction Case Study: NIST Standard Reference Material 2134 151
This suggests that voltage reconstruction may not be appropriate, and a tip profile or
constant shank angle reconstruction may be more accurate (especially given the tip
shape near the implant).
It should be noted that an opposite CSR trend is observed in voltage pulse mode.
As in the case above, as the specimen blunts during analysis, the FOV increases and
the constant detection rate causes the number of events per unit specimen area per
unit time to decrease. This results in a lowering of the evaporation field over time
and a decrease in the CSR. For the laser pulsed case, this effect is usually
outweighed by the required field increase due to a lower evaporation temperature.
In order to maximize precision (and secondarily accuracy), reconstruction
parameters should be independent of arbitrary judgments by the user. The following
analysis protocol was used:
1. Images were taken of specimens before analysis to make accurate measurements
of the shank angle in the vicinity of the implant region.
2. Only analyses which successfully obtained data from the entire implant and were
stopped in a controlled manner (without specimen failure) were considered.
152 5 Data Processing and Reconstruction
3. Images were taken of the specimens after analysis to make accurate measure-
ments of the final radius (RF) and sphere-cone radius ratio ( fS/C).
4. Specimens were fabricated to be of similar sizes, and specimen-electrode dis-
tances were chosen so that projected ion patterns filled similar regions of the
detector at similar magnification values (minimizing detector efficiency and
variations from analysis to analysis).
5. The default efficiency value of 0.5 was assumed (LEAP 4000X Si default).
6. Two doses were extracted based on differing constraints:
a. A single, fixed (1.45) was used for all analyses.
b. Each analysis utilized an independent based on a quantitative optimization
of some aspect of the reconstructed geometry (other than the known arsenic
distribution).
In this procedure each free reconstruction parameter is either held constant
(efficiency), determined by microscopy (shank angle, RF (which may be used in
place of R0), and fS/C), or determined by quantitative appraisal of some feature
in the reconstruction (). The ultimate result is a high precision and an acceptable
accuracy.
The reconstruction procedure benefits from selection of some internal feature, or
features, to judge the quality of the reconstruction as a function of . Figure 5.28
shows an atom map from an example implant to illustrate the various features
present in the sample. Potential internal calibration features within the sample
volume include: flatness of the various interfaces (nickel/poly-silicon and native
oxide), thickness of the poly-silicon region, and any atomic plane spacing that
might be observed in crystalline regions of the acquisition when the data quality is
sufficient to resolve such features.
Quantitatively assessing these features is time consuming, but relatively straight
forward. IVAS allows for chemical isoconcentration surfaces to be identified and
measures of roughness are provided for each surface (see Chap. 6). The roughness
values are calculated from a best-fit plane, so minimization of the roughness
parameters quantitatively determines the optimal flatness. Once interface curvature
is minimized, measurements of layer thickness can be performed by interpreting 1D
composition plots and enforcing critical concentration criteria as defining the
location of layer boundaries. Atomic plane spacing information can be extracted
from SDM calculations (see Chap. 6) also available in IVAS. We now consider the
merit of the measurable features available in this specimen type one by one.
The nickel/poly-silicon interface and the poly-silicon thickness are possible feature
choices. In principle, the interface should be approximately flat and the poly-silicon
thickness constant and easy to measure to high precision (194 2 nm). However,
5.6 Reconstruction Case Study: NIST Standard Reference Material 2134 153
Fig. 5.28 An example arsenic implant dataset which illustrates the various features present in the
capped sample. The colored atom map shows a sampled number of ions (Ni green, Si gray, O blue,
As magenta). A 3 % O isoconcentration surface is also shown. The nickel and poly-silicon caps are
above the blue native oxide region that formed on the silicon wafer surface prior to deposition of
the cap (near the center of the image). The changing arsenic composition is highlighted by the
varying density of magenta atoms below the native oxide. Plotting the arsenic composition as a
function of depth results in a 1D composition profile (red graphic to the right of atom map) and
integrating this curve results in the dose
this particular interface is not very well behaved and has a very small area (high
magnification), providing a roughness metric with a poorly defined minimum. Also,
the fact that the top portion of the tip has a larger shank angle, and that it differs
significantly from the shank angle in the implant region, makes it difficult to
reconstruct both the cap and implant regions accurately. In other words, choosing
a to give the correct cap thickness will provide a that produces a distorted
implant region. For Fig. 5.28, the thickness of the poly-silicon is ~170 nm while the
implant depth is reasonably well approximated when compared to SIMS.
By the time the analysis proceeds to the native oxide interfaces, microscopy
images reveal a shank angle that is nearly constant and small, while reconstruction
154 5 Data Processing and Reconstruction
trends show the native oxide layer is well defined with significant surface area. As
shown in Fig. 5.28, the native oxide actually consists of two separate surfacesthe
top and bottom chemical interfaces. A series of reconstructions were generated with
a variety of values using the constant shank angle option. Only the region
beginning 1020 nm above the oxide and continuing through the implant was
evaluated. Because RF was known from electron microscopy measurements, R0 was
adjusted until the reconstruction resulted in the correct RF, and fS/C was extracted
from the micrographs. The quality of the oxide interfaces was found to depend on
the definitions of the 3D gridding parameters (see Chap. 6) as well as the specific
isoconcentration surface. In order to produce two distinct surfaces that are not in
contact, smoothing (or delocalization) of the data was necessary to damp statistical
fluctuations while leaving enough detail to define the curvature of the surface.
Oxygen composition values between 0 and 5 at.% identified appropriate top and
bottom oxide surfaces. Smoothing greater than 2 nm often failed to produce
independent surfaces while no smoothing produced highly roughened surfaces
that again made it difficult to produce independent surfaces. Voxels smaller than
1 nm3 provided insufficient statistics to provide smooth composition values, while
larger volumes smeared the data and made it difficult to identify the interfaces.
Ultimately, 3 % oxygen surfaces with voxel sizes of (1 1 1) nm and (x,y,z)
delocalization of (2,2,1) nm provided well defined, independent surfaces without
holes as well as roughness values that followed understandable trends, as illustrated
in Fig. 5.29. Unfortunately, the top and bottom portions of the interface require very
different values to obtain flat interfaces. This is in part due to the different
specimen shapes that occur during field evaporation of the two different types of
interfaces. Using one of these values or some combination (average) may provide a
systematic way to choose an image compression yielding repeatable results, but
with some significant reduction in accuracy. To provide additional insights into this
behavior, a comparable model structure was field evaporated via simulation and
reconstructed.
5.6 Reconstruction Case Study: NIST Standard Reference Material 2134 155
Figure 5.30 shows the simulated shank angle reconstruction of a 2 shank angle
specimen created with a 2 nm thick region of material (light blue) having 1.5 times
the evaporation field of the surrounding material (dark blue). The reconstruction of
the simulated data results in obvious distortions surrounding the oxide region
(in agreement with previously observed trends [26]). These distortions result
from the non-hemispherical surfaces that occur as the specimen is field evaporated
through the differing evaporation field materials. A positive feature of the
reconstructed simulation is that while the reconstructed atoms above the oxide
are aberrated for 5 nm of depth (note the curved atomic planes and nonuniform
reconstruction atomic densities between the dashed red lines), the atoms immedi-
ately below the oxide appear unaberrated. Thus it should be possible to reconstruct
the entire region below the oxide accurately.
156 5 Data Processing and Reconstruction
The final available feature is the atomic plane spacing exhibited by the crystalline
silicon bulk. Figure 5.31 shows the spatial distribution map calculated from the
{100} planes of an example implant dataset. Creation of relatively good quality
spatial distribution maps allow the user to vary the image compression value until
the atomic plane spacing matches the ~0.543 nm {100} spacing known for this
material. Of course, in order to create an SDM, there must be an initial value used
for the image compression. As long as this initial guess for the image compression
was less than ~20 % of the optimal value determined at the end of the process, good
quality SDMs were produced and this method is relatively straightforward.
The atomic plane spacing does not suffer from the same limitations as features
discussed above. The thickness is known to an extremely high accuracy and
precision from x-ray diffraction methods. The layers do not suffer from evapora-
tion field differences as do almost any other layer-based metric, and thus the
specimen shape evolution is uniform (outside of crystallographic effects) and
resides within the ROI. Therefore, this feature is the most appropriate for
5.6 Reconstruction Case Study: NIST Standard Reference Material 2134 157
Table 5.1 Arsenic dose values calculated for a constant of 1.45 as well as a variable
Measured dose Peak as
(1e14 cm2) 1.45 Measured dose (1e14 cm2) composition
f R (nm)
Analysis Instrument Lift-out Local electrode S-C F 0.5 Variable 0.5 (1e20 cm2)
1 41 A 1.16 99.5 6.95 NA NA 1.051
2 41 A 1.16 99.5 7.51 7.28 1.470 1.062
3 41 B 1.14 86 8.17 7.06 1.545 1.048
4 41 B 1.14 85.5 7.86 7.26 1.510 1.03
5 41 C 1.15 101.5 7.38 6.69 1.420 1.086
6 44 D 1.15 93 7.65 7.75 1.435 1.088
7 44 E 1.15 92.5 7.07 7.61 1.410 1.111
8 44 E 1.14 90 7.03 7.59 1.405 1.035
9 44 E 1.14 90 7.39 7.35 1.455 1.069
All analyses Average 7.45 7.32 1.46 1.064
Standard deviation 5.4 % 4.7 % 3.4 % 2.5 %
Instrument 44 Average 7.29 7.58 1.43 1.076
Standard deviation 4.0 % 2.2 % 1.6 % 3.0 %
5 Data Processing and Reconstruction
References 159
Noting that the counting statistics limit the ultimate expected fluctuations to ~1 %,
the reconstruction limitations here conspire to provide a similar level of uncertainty
under the best of conditions (limited acquisition and specimen variability).
The process of taking raw LEAP data and transforming it into 3D real space
coordinates appears deceptively simple in commercially available software like
IVAS but, as we have shown, to take full advantage of the impressive capabilities of
the resultant APT data one must be aware of the subtleties of the interrelated
parameters to maximize accuracy and precision of conclusions taken from the data.
References
1. Miller, M.K., Hetherington, M.G.: Local magnification effects in the atom probe. Surf. Sci.
246(13), 442449 (1991)
2. Sha, G., Cerezo, A.: Field ion microscopy and 3-D atom probe analysis of Al3Zr particles in
7050 Al alloy. Ultramicroscopy 102, 151159 (2005)
3. Larson, D.J., Prosa, T.J., Geiser, B.P., Egelhoff Jr., W.L.: Effect of analysis direction on the
measurement of interfacial mixing in thin metal layers with atom probe tomography.
Ultramicroscopy 111, 506 (2011)
4. Marquis, E.A., Geiser, B.P., Prosa, T.J., Larson, D.J.: Evolution of tip shape during field
evaporation of complex multilayer structures. J. Microsc. 241(3), 225233 (2011)
5. Miller, M.K., Smith, G.D.W.: Atom Probe Microanalysis: Principles and Applications to
Materials Problems. Materials Research Society, Pittsburgh (1989)
6. Sebastian, J., Hellman, O., Seidman, D.: New method for the calibration of three-dimensional
atom-probe mass spectra. Rev. Sci. Instrum. 72(7), 29842988 (2001)
7. Kingham, D.R.: The post-ionization of field evaporated ions: a theoretical explanation of
multiple charge states. Surf. Sci. 116, 273301 (1982)
8. Kelly, T.F.: Kinetic-energy discrimination for atom probe tomography. Microsc. Microanal.
17, 114 (2011)
9. Ulfig, R.M., Kelly, T.F., Gault, B.: Promoting standards in quantitative atom probe tomogra-
phy analysis. Microsc. Microanal. 15(S2), 260261 (2009)
10. Blavette, D., Sarrau, J.M., Bostel, A., Gallot, J.: Direction et distance danalyse a la sonde
atomique. Rev. Phys. Appl. 17, 435440 (1982)
11. Bas, P., Bostel, A., Deconihout, B., Blavette, D.: A general protocol for the reconstruction of
3D atom probe data. Appl. Surf. Sci. 87/88, 298304 (1995)
12. Gault, B., de Geuser, F., Stephenson, L.T., Moody, M.P., Muddle, B.C., Ringer, S.P.: Estima-
tion of the reconstruction parameters for atom probe tomography. Microsc. Microanal. 14,
296305 (2008)
13. Gault, B., Moody, M.P., De Geuser, F., Tsafnat, G., La Fontaine, A., Stephenson, L.T., Haley,
D., Ringer, S.P.: Advances in the calibration of atom probe tomographic reconstruction.
J. Appl. Phys. 105, 034913/034911034919 (2009)
14. Geiser, B.P., Larson, D.J., Oltman, E., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F., Prosa, T.J.:
Wide-field-of-view atom probe reconstruction. Microsc. Microanal. 15(S2), 292293 (2009)
15. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York (2000)
16. Gault, B., Haley, D., De Geuser, F., Moody, M.P., Marquis, E.A., Larson, D.J., Geiser, B.P.:
Advances in the reconstruction of atom probe tomography data. Ultramicroscopy 111(6),
448457 (2011)
17. Kelly, T., Larson, D.J.: Local electrode atom probes. Mat. Char. 44(12), 5985 (2000)
160 5 Data Processing and Reconstruction
18. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.C.,
Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data from a
commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383 (2004)
19. Walck, S.D., Buyuklimanli, T., Hren, J.J.: Extended depth profiling with the IAP. J. Phys.
47-C2(3), 451458 (1986)
20. Gault, B., Moody, M.P., Marquis, E.A., De Geuser, F., Geiser, B.P., Larson, D.J., Kelly, T.F.,
Ringer, S.P., Smith, G.D.W.: Tomographic reconstruction in atom probe microscopy: past,
present. . . future? Microsc. Microanal. 15(S2), 1011 (2009)
21. Petersen, T.C., Ringer, S.P.: Electron tomography using a geometric surface-tangent algo-
rithm: application to atom probe specimen morphology. J. Appl. Phys. 105, 103518 (2009)
22. Gault, B., Haley, D., De Geuser, F., Moody, M.P., Marquis, E.A., Larson, D.J., Geiser, B.P.:
Advances in the reconstruction of atom probe tomography data. Ultramicroscopy 111,
448 (2010)
23. De Geuser, F., Lefebvre, W., Danoix, F., Vurpillot, F., Blavette, D., Forbord, B.: An improved
reconstruction procedure for the correction of local magnification effects in three-dimensional
atom-probe. Surf. Interface Anal. 39(23), 268272 (2007)
24. Larson, D.J., Geiser, B.P., Prosa, T.J., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F.: Improve-
ments in planar feature reconstructions in atom probe tomography. J. Microsc. 243, 15 (2011)
25. Larson, D.J., Geiser, B.P., Prosa, T.J., Ulfig, R., Kelly, T.F.: Non-tangential continuity
reconstruction in atom probe tomography data. Microsc. Microanal. 17(S2), 740741 (2011)
26. Larson, D.J., Geiser, B.P., Prosa, T.J., Kelly, T.F.: On the use of simulated field-evaporated
specimen apex shapes in atom probe tomography data reconstruction. Microsc. Microanal. 18
(5), 953963 (2012)
27. Vurpillot, F., Gault, B., Geiser, B.P., Larson, D.J.: Reconstructing atom probe data: a review.
Ultramicroscopy 132, 1930 (2013)
28. Gomer, R.: Field Emission and Field Ionization. Harvard University Press, Cambridge, MA
(1961)
29. Loi, S.T., Gault, B., Ringer, S.P., Larson, D.J., Geiser, B.P.: Electrostatic simulations of a local
electrode atom probe: the dependence of tomographic reconstruction parameters on specimen
and microscope geometry. Ultramicroscopy 132, 107113 (2013)
30. Vurpillot, F., Da Costa, G.: unpublished
31. Prosa, T.J., Olson, D., Geiser, B., Larson, D.J., Henry, K., Steel, E.: Analysis of implanted
silicon dopant profiles. Ultramicroscopy 132, 179185 (2012). doi:10.1016/j.ultramic.2012.
10.005
32. Loberg, B., Norden, H.: Observations of the field-evaporation end form of tungsten. Arkiv.
For. Fysik. 39(25), 383395 (1968)
33. Wilkes, T.D., Smith, G.D.W., Smith, D.A.: On the quantitative analysis of field-ion micro-
graphs. Metallography 7, 403430 (1974)
34. Gerstl, S.S.A., Geiser, B.P., Kelly, T.F., Larson, D.J.: Evaluation of local radii of atom-probe-
tomography specimens. Microsc. Microanal. 15(S2), 248249 (2009)
35. Shariq, A., Mutas, S., Wedderhoff, K., Klein, C., Hortenbach, H., Teichert, S., Kucher, P.,
Gerstl, S.S.A.: Investigations of field-evaporated end forms in voltage- and laser-pulsed atom
probe tomography. Ultramicroscopy 109(5), 472479 (2009)
36. Vurpillot, F., Da Costa, G., Menand, A., Blavette, D.: Structural analyses in three-dimensional
atom probe: a Fourier transform approach. J. Microsc. 203(3), 295302 (2001)
37. Geiser, B.P., Kelly, T.F., Larson, D.J., Schneir, J., Roberts, J.P.: Spatial distribution maps for
atom probe tomography. Microsc. Microanal. 13, 437447 (2007)
38. Kelly, T.F., Voelkl, E., Geiser, B.P.: Practical determination of spatial resolution in atom probe
tomography. Microsc. Microanal. 15, 1213 (2009)
39. Gault, B., Moody, M.P., De Geuser, F., Haley, D., Stephenson, L.T., Ringer, S.P.: Origin of the
spatial resolution in atom probe microscopy. Appl. Phys. Lett. 95(3), 034103 (2009)
40. Gault, B., Moody, M.P., De Geuser, F., La Fontaine, A., Stephenson, L.T., Haley, D., Ringer,
S.P.: Spatial resolution in atom probe tomography. Microsc. Microanal. 16, 99110 (2010)
References 161
41. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principals and Applications. Elsevier, New
York (1969)
42. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion Micros-
copy. Oxford University Press, Oxford (1996)
43. Gault, B., Muller, M., La Fontaine, A., Moody, M.P., Shariq, A., Cerezo, A., Ringer, S.P.,
Smith, G.D.W.: Influence of surface migration on the spatial resolution of pulsed laser atom
probe tomography. J. Appl. Phys. 108, 044904/044901044906 (2010)
44. Cerezo, A., Clifton, P.H., Gomberg, A., Smith, G.D.W.: Aspects of the performance of a
femtosecond laser-pulsed 3-dimensional atom probe. Ultramicroscopy 107(9), 720725
(2007)
45. Mamyrin, B.A., Karataev, V.I., Shmikk, D.V., Zagulin, V.A.: The mass-reflectron, a new
nonmagnetic time-of-flight mass spectrometer with high resolution. Sov. Phys. JETP 37,
4548 (1973)
46. Scheinfein, M.R., Seidman, D.N.: Time aberrations of uniform fields: an improved reflectron
mass spectrometer for an atom-probe field-ion microscope. Rev. Sci. Instrum. 64(11),
31263131 (1993)
47. Cerezo, A., Godfrey, T.J., Sijbrandij, S., Smith, G.D.W., Warren, P.J.: Performance of an
energy compensated three-dimensional atom probe. Rev. Sci. Instrum. 1(69), 4958 (1998)
48. Sijbrandij, S., Cerezo, A., Godfrey, T.J., Smith, G.D.W.: Improvements in the mass resolution
of the three-dimensional atom probe. Appl. Surf. Sci. 94/95, 428433 (1996)
49. Panayi, P.: Curved reflectron. Patent UK 0509638.3 and US 60/682,863
50. Wada, M.: On the thermally activated field evaporation of surface atoms. Surf. Sci. 145,
451465 (1984)
51. Tsong, T.T.: Field ion image formation. Surf. Sci. 70, 211233 (1978)
52. Michaelson, H.B.: The work function of the elements and its periodicity. J. Appl. Phys. 48(11),
47294733 (1977)
53. Larson, D.J., Russel, K.F., Miller, M.K.: Effect of specimen aspect ratio on the reconstruction
of atom probe tomography data. Microsc. Microanal. 5, 930931 (1999)
54. Gipson, G.S., Yannitell, D.W., Eaton, H.C.: On the electric field distribution within the field
ion microscope and near the surface of field emitters. J. Phys. D. Appl. Phys. 12, 987996
(1979)
55. Gipson, G.S., Eaton, H.C.: The electric field distribution in the field ion microscope as a
function of specimen shank. J. Appl. Phys. 51(10), 55375539 (1980)
56. Gipson, G.S.: An improved empirical formula for the electric field near the surface of field
emitters. J. Appl. Phys. 51(7), 38843889 (1980)
57. Smith, R., Walls, J.M.: Ion trajectories in the field-ion microscope. J. Phys. D. Appl. Phys. 11,
409419 (1978)
58. Waugh, A.R., Boyes, E.D., Southon, M.J.: Investigations of field evaporation with a field-
desorption microscope. Surf. Sci. 61, 109142 (1976)
59. Oberdorfer, C., Schmitz, G.: On the field evaporation behavior of dielectric materials in three-
dimensional atom probe: a numeric simulation. Microsc. Microanal. 17, 1525 (2011)
60. Moore, A.J.W.: The structure of atomically smooth spherical surfaces. J. Phys. Chem. Solids
23(7), 907912 (1962)
61. Moore, A.J.W., Ranganathan, S.: The interpretation of field ion images. Phil. Mag. 16,
723737 (1967)
62. Geiser, B.P., Larson, D.J., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F., Prosa, T.J., Olson, J.D.:
A system for simulation of tip evolution under field evaporation. Microsc. Microanal. 15(S2),
302303 (2009)
63. Marquis, E.A., Geiser, B.P., Prosa, T.J., Larson, D.J.: Evolution of tip shape during field
evaporation of complex multilayer structures. J. Microsc. 241, 225233 (2010)
64. Larson, D.J., Gault, B., Geiser, B.P., Vurpillot, F.: Atom probe tomography spatial reconstruc-
tion: status and directions. Curr. Opin. Solid. State. Mater. Sci. 17(5), 236247 (2013)
162 5 Data Processing and Reconstruction
65. Marquis, E.A., Vurpillot, F.: Chromatic aberrations in the field evaporation behaviour of small
precipitates. Microsc. Microanal. 14(6), 561570 (2008)
66. Birdseye, P.J., Smith, D.A., Smith, G.D.W.: Analogue investigations of electric field distribu-
tion and ion trajectories in the field ion microscope. J. Phys. D 7(12), 16421651 (1974)
67. Vurpillot, F., Gruber, M., Da Costa, G., Martin, I., Renaud, L., Bostel, A.: Pragmatic recon-
struction methods in atom probe tomography. Ultramicroscopy 111(8), 12861294 (2011)
68. Gault, B., Loi, S.H., Araullo-Peters, V., Stephenson, L.T., Moody, M.P., Shrestha, S.,
Marceau, R.K.W., Yao, L., Cairney, J.M., Ringer, S.P.: Dynamic reconstruction for atom
probe tomography. Ultramicroscopy 111(11), 16191624 (2011)
69. Vurpillot, F., Bostel, A., Blavette, D.: The shape of field emitters and the ion trajectories in
three-dimensional atom probes. J. Microsc. 196(3), 332336 (1999)
70. Vurpillot, F., Bostel, A., Blavette, D.: A new approach to the interpretation of atom probe field-
ion microscopy images. Ultramicroscopy 89, 137144 (2001)
71. Sauvage, X., Renaud, L., Deconihout, B., Blavette, D., Ping, D., Hono, K.: Solid state
amorphization in cold drawn Cu/Nb wires. Acta Mater. 49, 389394 (2001)
72. Vurpillot, F., Larson, D.J., Cerezo, A.: Improvement of multilayer analyses with a three-
dimensional atom probe. Surf. Interface Anal. 36, 552558 (2004)
73. Lauhon, L.J., Praneet, A., Ronsheim, P., Flaitz, P.L., Lawrence, D.: Atom-probe tomography
of semiconductor materials and device structures. MRS Bull. 34, 738743 (2009)
74. Simons, D., Kim, K., Benbalagh, R., Bennett, J., Chew, A., Gehre, D., Hasegawa, T., Hitzman,
C., Ko, J., Lindstrom, R., MacDonald, B., Magee, C., Montgomery, N., Peres, P., Ronsheim,
P., Yoshikawa, S., Schuhmacher, M., Stockwell, W., Sykes, D., Tomita, A., Toujou, F., Won,
J.: Round-robin study of arsenic implant dose measurement in silicon by SIMS. Appl. Surf.
Sci. 252(19), 72327235 (2006). doi:10.1016/j.apsusc.2006.02.152
75. Simons, D.S., Downing, R.G., Lamaze, G.P., Lindstrom, R.M., Greenberg, R.R., Paul, R.L.,
Schiller, S.B., Guthrie, W.F.: Development of certified reference materials of ion-implanted
dopants in silicon for calibration of secondary ion mass spectrometers. J. Vac. Sci. Technol. B
25, 13651375 (2007)
76. Depth profile data provided by Evans Analytical, Santa Clara, CA 95051
Chapter 6
Selected Analysis Topics
6.1 Introduction
LEAP data analysis is performed using the CAMECA Integrated Visualization and
Analysis Software (IVAS) package. IVAS was introduced in 2004 on a Windows
XP software platform based on Java and Netbeans and continues to require a
Windows operating system environment. CAMECA offers several levels of licens-
ing from free and short term use to permanent full versions. Through the years,
IVAS development has closely followed a guiding principle that prioritizes ease of
use and availability of open file formats so that analysis can be performed both with
IVAS as well as user-developed software. Discussing all the features and concepts
pertaining to IVAS would require a separate book (see the IVAS User Guide).
Consequently, this chapter describes only some key IVAS topics and features
useful for atom probe tomography (APT) practitioners. The features and topics
addressed are in no particular order, but each should have some relevance for
experienced and novice analysts alike.
The first topic considers spectral ranging. Because APT is fundamentally a time-
of-flight (TOF) mass spectrometry technique, ranging and identification of chem-
ical species are important for 3D reconstruction of detected ions (see Chap. 5), and
especially important for chemical quantification. Mass spectra from various mate-
rials are considered as well as ranging criteria. The impact on compositional
precision and accuracy is a key component of the discussion. These concepts are
applied to an example mass spectrum and the resulting composition, compositional
uncertainty, detection and quantification limitations of the data are highlighted.
The next topic considers key aspects related to the organization of the 3D data.
The quantity (many millions of ion positions) and 3D nature of the APT informa-
tion often requires some form of gridding or voxelization and delocalization to
enable reasonable calculation times and enable various forms of visualization
beyond simple atom maps. Understanding how these gridding and smoothing
choices affect extracted compositional trends is necessary for understanding the
accuracy of any analysis.
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 163
DOI 10.1007/978-1-4614-8721-0_6, Springer Science+Business Media New York 2013
164 6 Selected Analysis Topics
6.2.1 Ranging
(1) What is an appropriate histogram bin width? and (2) How should the range
boundaries be chosen if one desires both compositional accuracy and precision?
The width of the histogram binning determines the number of bins over which
T is spread. Coarse binning has the effect of smoothing features and helps smooth
background fluctuations to better define small amplitude signal peak. Overly fine
binning can make it difficult to identify a weak signal on a relatively high back-
ground as a signal peak. In terms of APT quantification, the goal is to choose a
histogram bin width such that measurement of peak location and T are insensitive to
the bin boundaries. A general rule of thumb is for the full width at half maximum
(FWHM) of all peaks to extend over 5 bins [2]. The peak in Fig. 6.1 follows this
rule with bins widths of 0.005 Da yielding 12 bins across the FWHM.
Range boundaries determine the fraction of the signal, S, that is measured. An
accurate measure of S requires sufficiently large ranges to ensure that ~100 % has
been captured within the range; however, large ranges also increase the
B component which, if extreme enough, can lower the overall precision of the
result (increases the spread in multiple measurements). Generally, for high intensity
non-overlapping peaks, it is relatively easy to capture all of the peak intensity with
optimum precision. For low intensity peaks (and overlapping peaks), overly narrow
or wide ranges can result in systematic loss in accuracy or precision respectively.
Ultimately, a good combination of accuracy and precision can be achieved by
capturing uniform signal fraction from each peak and also sufficient signal fraction
such that a small change in range width does not significantly change the measured
fraction.
LEAP mass spectra generally possess asymmetric (non-Gaussian-shaped)
mass spectrum peaks, but some insights can be gained from considering analytic
peak shapes. For a Gaussian peak shape, describes the location of the peak
166 6 Selected Analysis Topics
center and is related to the FWHM (FWHM 2.354) and describes the peak
width [2]. The integrated peak fractions or confidence levels (CL listed in %)
contained within a region related to the peak center are defined by a numerical
factor k and listed in Table 6.1. Commonly quoted 2-tailed regions of , 2,
3 capture 0.683, 0.954, and 0.997 of the total peak intensity, respectively. By
capturing >99 % of the total signal and requiring that that FWHM spans more than
5 bins, one can generally achieve the goal of capturing a uniform and sufficient
fraction of the signal.
In terms of application of the above discussion to real APT peak shapes, there is
significant added complexity. Generally, APT peaks have steep rise on the low m/n
side and an extended tail on the high m/n side. Real APT spectra may have peaks
with differing peak shapes, shapes that broaden as a function of m/n, significant
signal fraction contained within the peak tail that extends into neighboring peaks, or
other features that make it difficult to assign ranges that achieve the goal of uniform
and sufficient signal fraction. Unfortunately, application of custom range sizes may
be necessary in these cases, and again a good criterion is uniform and sufficient
signal fraction to maintain good accuracy and precision.
There are some basic questions one asks when considering detection of dilute
species and almost all of the answers to these questions depend on the background
level: (1) When is a net signal statistically significant? (2) When a net signal is not
statistically significant, then what is the maximum level that could exist without
detection? and (3) What signal level is necessary to achieve some minimal level of
measurement precision? For a more detailed statistical discussion of these questions
see [3].
Data collected from an APT experiment can be described as a Bernoulli
sequence and so follows a Binomial distribution [4]. For any viable LEAP mea-
surement, the Binomial distribution is in the limit where it can be approximated by a
Poisson distribution for weak peaks or a particular Gaussian distribution for strong
peaks [1, 5, 6]. For either case, this presents at least two options for estimating the
expected standard deviation, i, for many measurements estimated from a single
6.2 Spectral Analysis 167
because each range i has a i that can be properly propagated when performing
general calculations [1].
The second is useful for calculating concentrations where background contribu-
tions are negligible (Ci Ti/TTotal) [4, 7]
s
Ci 1 Ci
i (6.2)
T Total
where Ti represents the total number of counts for an atomic species i and TTotal is
the total number of counts for all the species. For accurate measurement of dilute
species, it is important to correct the measured signal for background effects (e.g.,
ions from adjacent peaks, out-of-time ion events, etc.). General approaches to these
corrections are illustrated in [8]. For APT data, approximating the background
signal by making use of adjacent range(s) with the same range width is useful,
although there are other approaches as well. With this approach in mind consider
the questions in the following sections.
This is achieved when the background fluctuations cannot account for the entire
measured signal to some desired CL. For the discussions that follow, the total
number of counts in any given range is composed of the signal plus the background
(T S + B). So consider the number of counts measured for a range defining a
peak as (T ) and the adjacent range(s) as a measurement estimating B. The
corresponding standard deviations for the counts in the peak and the background are
p p
T T and B B: (6.3)
This is achieved when the low-side tail of the statistical fluctuations in a normally
detected signal exceeds a critical level based on the background measurement. The
result again depends on the CL chosen (for a complete derivation see [3]). If the CL
for the critical level and the detection level are the same, then the minimum
detection level becomes
p p
k2 2k 2B or 2:71 4:65 B for a 95% CL: (6.5)
In other words, for the case where no signal was detected given the desired CL
(here 95 %), the actual signal level could be as high as this and not have been
detected at least 1-CL (here 5 %) of the time because of statistical fluctuations.
This is achieved when the estimated error of S, S, meets the criteria specified.
Summarizing from [3] the minimum number of signal counts must exceed the
values below for the relative precision in the net signal, k S/S, to reach 10 %
(kQ 1/0.1) minimum:
q
kkQ 2 kkQ 4 8BkkQ 2
S> (6.6)
2
p p
S > 200 + 40; 000 800B ! S 14.1k B for k S/S < 10 % for a 95 % CL.
6.2.3.3 Example
Consider the arsenic implant case study from Chap. 5. Assessing the background
levels in the tail of the arsenic implant can help guide decisions for generating 1D
composition profiles to meet analysis goals. Fine resolution of the implant profile
(z on the order of 0.1 nm resolution along the depth-dimension of the composition
profile) is possible when arsenic concentration is sufficiently high, but more coarse
resolution (z ~ 110 nm) may be necessary to allow quantification as concentra-
tions become more dilute. Assessing the background level allows one to determine
6.2 Spectral Analysis 169
Fig. 6.2 Relevant Si (light gray) and As (dark gray) mass ranges and their equivalent background
ranges used to illustrate best methods for determining composition and estimates of uncertainty
symmetrically on either side of peak T. The latter approach has the advantage of
estimating any linear variation in the background level. The former approach is
shown in Fig. 6.1.
For this example, backgrounds for each peak were estimated by use of an
equivalent sized range at the low m/n side of the peak as indicated in Fig. 6.2.
Because the silicon isotope peaks are somewhat closely spaced, the Si2+ and Si+
ranges utilize a single background range of width equal to the sum of the three
individual isotope range widths. The number of counts in each peak (T) and the
number of counts in each background range (B) are listed in Table 6.2.
In order to determine the background corrected composition, the uncertainty in
the number of counts in each range must be estimated using (6.1) and these errors
propagated to yield uncertainties for the reported concentration. Because the
background levels in this example are relatively low, the final results and uncer-
tainties should be similar to those predicted by (6.2) without background subtrac-
tion. It is important to note, however, that as the arsenic signal decreases and the
background level remains relatively constant, the calculated uncertainties with and
without background subtraction will diverge.
The relevant steps for determining the composition and estimate of uncertainty
for this binary system are summarized in Table 6.4. First, the net number of Si ions
is determined by pooling all the counts from the Si ranges together (576,552) and
subtracting from it all the counts from the pooled Si background ranges (327). The
uncertainty in this value becomes the square-root of the sum of the uncertainties
p
of the individual uncertainties added in quadrature ( 576, 552 327 760 ).
The same steps were performed for the As ions. The composition becomes CSi
TSi/TSi+As and the uncertainty is estimated by Si/TSi+As.
There is an apparent contradiction here in terms of the uncertainty estimates.
Because this is a binary composition, the uncertainties of the two compositions are
systematically linked and should be equal because any change in the composition of
one requires an equal change in the other. Using (6.2), which does not allow us to
consider the effects of background subtraction, the resulting uncertainty is close to
6.2 Spectral Analysis 171
that of the As fraction estimate using (6.1). The apparent discrepancy lies in the fact
that there are two ways to estimate the uncertainty. The table illustrates using
the counts of each ion type and the total ions. An alternative would be to use the
As result to determine the Si result (CSi 1 CAs) which results in an equal
uncertainty for both numbers. The best estimate of uncertainty requires the practi-
tioner to use a relationship that results in minimum uncertainty. For a system with
N chemical species, the N 1 species with minimal fractional uncertainty should
the Nth X
be calculated as in the table, while species should
iN1
be determined as the
balance from the N 1 species CN 1 i1
Ci .
The claims of the preceding paragraph can be verified via simulation. 100,000
randomly generated compositions were generated based on a 577,559 ion dataset
and average compositions for SiS+B, SiB, AsS+B, and AsB based on data from
Table 6.2. Simulated background corrected compositions of Si and As were deter-
mined by subtracting the simulated background counts for each ion type for each
simulation. Average Si and As compositions (Simulated Composition) and their
standard deviations (Simulated S) are reported in Table 6.3. There is excellent
agreement with the estimated uncertainty of the As listed in the table consistent
with the preceding discussion.
Estimating the background levels for the As ions allows one to consider how 1D
composition bin sizes impact the critical level, minimum detection level and
quantification levels for an analysis. Based on the background level measured
from the preceding example (98 counts), the three statistical thresholds previously
discussed have been calculated and listed in Table 6.4: (1) The net As counts must
be greater than 23 for an As signal to be detected at a 95 % CL. (2) When a signal is
not considered as detected, then an As level as high as 49 counts might be present in
the sample and not reach the detection threshold at least 5 % of the time. (3) The net
As counts must be greater than 544 to have 95 % CL error bars smaller than 10 % of
the reported value for this sample size.
It is important to emphasize that the preceding discussions on measurement
uncertainties are solely based on the counting statistics contained within a given
mass spectrum. They cannot account for other sources of error (e.g., systematic
measurement errors from multi-hit/pile-up effects, complex ion disassociation,
peak overlap issues, poor detector calibration, etc.).
IVAS has an ROI Simulation Tool that can be used to investigate quantification
outcomes based on variations in ranging, peak shape, and background level.
Consider a high-quality mass spectrum, like the Asimplaned Si example above,
but with some unknown level of As to be analyzed on a LEAP 4000HR. For the
same material system, the HR may provide a typical MRP of 1,100 FWHM and
lower background levels to go with lower detection efficiency. One may be
interested in the As detection limits. To address this issue, mass spectra were
simulated with various levels of background and three levels of As concentration
(atoms/cm3). The As was distributed at a ratio of 1.5:1 in the As2+ and As+ charge
states, respectively, and the background levels for each range were estimated
based on equivalent sized ranges as in the previous example. The results of the
simulations were minimum detection levels as plotted in Fig. 6.3. The y-axis is the
172 6 Selected Analysis Topics
Fig. 6.3 Results for minimum detection level based on simulated LEAP 4000 HR mass spectra.
The estimated analysis volume (plotted as the cube root of the analysis volume) necessary to
achieve detection levels of 1e18cm3, 1e19cm3, and 1e20cm3 are illustrated. These level
depend directly on the average background level in each spectrum
cube-root of the volume necessary to reach a minimum detection level (95 % CL)
given a background level in ppm/ns.
f ij Ai ij Mj (6.7)
Ai: actual count of type i ion (still unknown), Mj: measured counts in each range j,
fij: isotopic abundance that isotope type i has in range j, ij: (unknown) fractions of
the ions in range j that are type i.
Peak decomposition is accomplished by calculating the unknown fractions ij
and the Ai by reformulating (6.7) as an optimization problem with linear constraints
(i.e., sum of the counts from all peaks before and after decomposition must be
conserved). For example, in an ironchromiumnickel system (doubly charged
chromium isotopes at 25, 26, 26.5, and 27 Da) the overlap at 27 Da is constrained
by ACr M25 26.5 + / 27 Cr M27 and / Cr + / Fe 1. These types of constraints
27 27
of such features, and chemical intermixing between these features. Single elements
or combinations of elements may be used in the definition of the surface.
First, a concentration space grid is specified by the user by defining voxel (volume
pixel) size and delocalization. The user may individually set the x-, y-, and z-dimen-
sions of the voxels keeping in mind that very small voxel sizes retain a high resolution
in concentration space but suffer from high statistical noise (in the limit of small voxel
sizes, many of the voxels may have only a few or no ions at all). Larger voxel sizes
have the advantage of better counting statistics for the concentration calculation, but
result in degraded spatial resolution. A delocalization function is also applied to the
data following the concentration calculation, which distributes the contribution of
each ion over a range of voxels (an example 2D version is shown in Fig. 6.4).
Several of the spatial analyses performed by IVAS (isosurface, proxigram,
surface roughness, preview compositions) require that the acquired hit-data be
reformatted from an ionic point-list into a set of 3D voxels. The simplest way to
grid the data is to simply assign each ion entirely to the 3D voxel into which it falls.
This has the effect of relocating ions from their various locations within the voxel to
the centers of the voxel. If an ion is detected near the edges of a voxel, then it is
desirable to share it among the neighboring voxels according to a localized transfer
function to smooth or delocalize the count values in these grids. One 3D grid is
created per mass range. At least schematically, IVAS follows the procedure
described by Hellman [10].
Ideally, for any level of delocalization desired by the user, IVAS would perform
a delocalization by convolving the desired 3D kernel with the list of detected points.
It is frequently useful to use rather large kernels, so this procedure can be extremely
slow. However, grid-based convolutions can be much faster because of tools like
the Fast Fourier Transform (FFT).
Based on these two considerations, IVAS performs a two-step delocalization
process. The first step applies a mask-based convolution with an extremely localized
Gaussian kernel to move ions from the point-list into the grid. IVAS uses a kernel in
which only the tails beyond 3 of a Gaussian kernel are outside of a distance of 0.5
voxel lengths. This is only applied once during the initial creation of the grids.
6.3 Concentration Space Analyses 175
Fig. 6.5 The process of interpolating specific isoconcentration points. Here the location of 60 %
and 60.3 % are shown as linearly interpolated positions between voxels with average values of
59.9 % and 60.4 % respectively
roughness. See [12] for an example of the effect of voxel size and delocalization on
the resultant roughness measurements of a single interface between two layers.
The detriment to any smoothing is that some of the spatial information of the
original feature(s) is lost. This section examines the sensitivity of the grid delocal-
ization parameter on the capability to measure the composition and thickness of a
simple planar feature. As a test, consider a synthetic dataset where perfectly sharp
chemical interfaces exist on each side of a single thin film (denoted B in Fig. 6.6),
and all atoms are on a simple cubic lattice. The feature is characterized by a
thickness (t) and a concentration level ( p) and centered at z 0.
In this case of a Gaussian response characterized by delocalization sigma (), the
resulting concentration profile can simply shown to be the following:
p z 2t z 2t
Cz erf p erf p : (6.8)
2 2 2
Fig. 6.9 (a) Data showing two phases in a nickel-based alloy with the interfaces created at the
10 at.% aluminum level, (b) proxigram created using only the upper interface shown in (a), and (c)
proxigram created using all of the interfaces shown in (a)
proxigram calculation, each ion is placed into the distance bin which is the smallest
distance away from any point on the interface, thus effectively taking into account
any curvature in the interface while still maintaining good counting statistics.
An estimate of the standard deviation is calculated on each bin of the histogram,
and as discussed previously, the counting statistics are not the only source of error.
Proximity histograms are a very powerful tool for detecting variations in concen-
trations near buried interfaces, something atom probe tomography is uniquely
suited to do.
We will use the dataset shown in Fig. 6.9 as an example of a simple proxigram
analysis. A dataset containing two phases in a nickel-based alloy is shown in
Fig. 6.9a. A 10 at.% aluminum isoconcentration surface has been used to define
the interfaces shown in the figure. Creation of a proxigram with 0.1 nm bins and a
maximum distance of 5 nm from the interface arrowed in Fig. 6.9a is shown in
Fig. 6.9b. The position of 0 on the x-axis of the proxigram corresponds to the
interface from which the profile was calculated (10 at.% Al), although it may be
slightly offset in some cases depending on the bin size selected for the proxigram
because the surface concentration is calculated from the 3D grid and the proxigram
is based on the discrete data. Negative x-axis values correspond to lower concen-
tration side of the isosurface. From Fig. 6.9b, the user is immediately able
to determine the partitioning of the elements Ni, Al, Mo, Co, and Cr between
the phases and to observe that no obvious segregation is apparent at this interface.
180 6 Selected Analysis Topics
All six of the interfaces visible in Fig. 6.9a may be used to create a single
proxigram, Fig. 6.9c, which shows features very similar to those in Fig. 6.9b, but
with less statistical variation. In the case of a large number of secondary phase
regions of different sizes, the user may wish to consider separate proxigrams based
on the size of the phase (using either volume or surface area) to look for evidence of
size affecting observed composition.
Fig. 6.10 Atom diagrams illustrating the cluster-finding algorithm: (a) solute atoms (red) in a
matrix (white) suggest the presence of a cluster (dashed line), (b) the Dmax parameter shown as a
green circle positioned above two solute atoms (black) of interest, (c) results of application of the
Dmax parameter (atoms shown in blue are determined to be part of the cluster) with the size of
the envelope parameter (L ) used to determine potential matrix atoms contained in the cluster
illustrated in as the gray circle, (d) shows all matrix atoms (gray) which may be evaluated to
determine if they are contained in the cluster, (e) the erosion distance parameter (E) (shown as a
purple circle) is applied to all non-cluster atoms to remove atoms around the cluster edges (purple
spheres), and (f) the final estimate of the atoms (6.14) contained in the cluster
atoms which may be in the cluster. For the gray circle shown in Fig. 6.10c, we
identify six potential matrix atoms which may be part of the cluster (gray atoms in
Fig. 6.10c). Figure 6.10d shows the result of using this process to determine all of the
potential matrix atoms (solid gray circles) which may be contained in the cluster. We
now have cluster atoms defined as shown in blue (matrix) and gray (solute).
It is clear however, that while this method allows us to identify some obvious
matrix atoms contained in the cluster (arrow in Fig. 6.10d), it has also added matrix
atoms (usually around the outside of the cluster) which we do not wish to identify as
being part of the cluster. Thus we need to remove these atoms and this is done using
another parameter, which we will call E (also known as the erosion distance). E is
often taken to be slightly smaller than Dmax, and is shown by the purple circles in
Fig. 6.10e. A circle of radius E is positioned over all of the atoms in the dataset
which have been determined to not be part of the potential cluster and are also
within at least a distance E from the cluster. Any cluster atom found within this circle
182 6 Selected Analysis Topics
Fig. 6.11 A
20 20 2 nm3 slice
of data containing
approximately 3 at.%
solute (red atoms) which
appears to be nonrandomly
distributed
is removed from the potential cluster. The purple atoms shown in Fig. 6.10e are an
example of this erosion process. After it is completed, we have the final estimate of
the atoms contained in the cluster, Fig. 6.10f. We can see that our L process added
four matrix atoms (gray) to our initial determination of solute cluster atoms (blue).
The cluster contained 14 atoms and is approximately 71 % solute.
An additional variable often used to refine the cluster detection process is Nmin.
Clusters which contain a number of atoms less than Nmin are ignored because they
are assumed to be due to random fluctuations in the material [18]. Randomized
version of experimental data is often used to estimate random cluster distributions.
Fig. 6.12 Cluster count distribution for the data shown in Fig. 6.11 using parameters Order 1
and Nmin 10. The count distribution that results from randomizing all of the ions (using the
measured concentration of 3 at.% solute) is shown as the gray line
Fig. 6.13 Cluster size distribution using a Dmax value of 0.5 nm as determined from Fig. 6.12. The
size distribution that results from randomizing all of the ions (using the measured concentration
of 3 at.% solute) is shown as the gray line
Fig. 6.14 Results of the cluster detection algorithm with parameters Dmax 0.5 and Nmin 8
showing a distribution of 35 clusters having an average of 1112 solute atoms per cluster with a
mean volume of ~1 nm3 and a concentration of ~30 at.% solute. Note that this figure only shows a
slice of the original 20 20 20 nm3 dataset. Matrix atoms are shown in gray and solute atoms
are shown in black
Fig. 6.15 A summed concentration profile for all of 35 clusters showing a maximum average
cluster concentration of ~35 at.% solute. The scaling of the distance axis is performed by scaling
atom 3D displacements from the cluster center of mass to the 3D radius of gyration vector or to the
3D vector of maximum displacements of cluster atoms
6.5 Spatial Distribution Maps 185
demonstrated that a suitable value of Dmax is inversely proportional to the cube root
of the local density of solute atoms in the cluster. Hyde has recently simulated a
number of situations and made some recommendations with respect to the starting
point for the variables. He points out that (1) Dmax is the most important variable,
(2) Nmin is a function of Dmax and ion detection efficiency, and (3) both Dmax and
Nmin need to be increased as the ion detection efficiency decreases [25].
For crystalline materials, there is often significant order present in the atom
positions in APT reconstructions. This circumstance originates from the naturally
ordered progression over which atoms are removed from the curved atom probe
specimen surface. Evaporation is biased toward evaporation at terrace edges,
especially in pure materials. Consequently, atoms are removed from the terrace
edges where the electric field is highest (ignoring variations in surface evaporation
field from crystallographic orientation). After a single atom is removed from a step
edge, some subsequent step edge experiences the highest electric field and begins to
field evaporate which can be a very ordered process under certain conditions. Some
materials exhibit more ordered evaporation than others. Tungsten and aluminum
are two metals that are known to exhibit strongly ordered evaporation.
Ideally, reconstructions of the detected ions from all crystalline specimens
would result in perfect arrays of ion positions, but in reality, even the best recon-
structions have some inherent level of disorder. Some of the known sources of
imperfection include:
1. Out-of-sequence evaporation
2. Trajectory aberrations due to crystallography and local magnification effects
3. Over-simplified reconstruction algorithm
4. Nonspherical tip shape
5. Nonuniform reconstruction parameters (kF, , etc.)
When evaluating the quality of a reconstruction, it is desirable to validate the
periodicity of the reconstruction against known planar spacings within the speci-
men. Some quantitative measure of both short-range (nearest-neighbor arrange-
ments) and long-range order is necessary to allow assessment of the accuracy of
reconstructed image and any general limitations of the technique. The SDM [26]
was developed to enable validation of reconstructions for highly ordered systems as
well as look for types of chemical ordering with APT tomograms. The result of an
SDM may be applied to a variety of analyses including trajectory aberrations,
atomic ordering and clustering, element- or isotope-specific radial distribution
functions, quantum detection efficiency, and real-space crystal structure. A very
similar type of data analysis is known as the near atom vicinity method [27].
186 6 Selected Analysis Topics
1
SDMX; Y; Z x; y; zx X, y Y, z Z dxdydz NX, 0 Y , 0 Z, 0
1
N X
X N
xi , xj X yi , yj Y zi , zj Z
j6i i
(6.9)
XN
where x; y; z i
x xi y yi z zi is composed of N Dirac delta
functions that represent the N ion positions. A useful property of the SDM is that the
SDM of a full dataset and the SDM of the same dataset where only a random fraction
of the dataset is used to calculate the SDM are directly related by :
For simplicity, consider the 1D example shown in Fig. 6.16. A pair of ions is
represented by a pair of delta functions at positions A on the x-axis (Fig. 6.16a).
Consider a copy of this same function (two delta functions at A) translated some
arbitrary distance x along the axis. The autocorrelation for some value x is the
integral of the product of these two functions (the original and the copy translated a
distance x). In this case, the result will be zero for all values of x except when
x 2A, 0, and 2A. When x 2A, the product is a single delta function squared
and the integral over all space is 1. When x 0, the product is a pair of delta
functions squared and the integral over all space is 2. As a result, the autocorrelation
of the two points in Fig. 6.16a consists of the three points shown in Fig. 6.16b
where the middle point has a value of 2 while the adjacent points have values
of 1. Calculation of the SDM follows the same process except that the translation
represented by x 0 (the self-correlation) is excluded, so the result is two points
each with a value of 1 as shown in Fig. 6.16c. Alternatively, one can think about
(or calculate) the SDM from a nearest-neighbor point-of-view. Since the point at
188 6 Selected Analysis Topics
Fig. 6.17 A simple 2D example of how spatial distribution maps may be understood by compar-
ison to autocorrelation of ions represented by delta functions: (a) an arrangement of five delta
functions, (b) the autocorrelation of (a), (c) the SDM of (a)
x A has a single neighbor at +2A and the x A point has a single neighbor at
2A, the SDM has two histogram values of 1 at 2A respectively.
Next consider the 2D example shown in Fig. 6.17. Here we want to consider the
product of the five points as shown in Fig. 6.17a multiplied by all possible 2D
translations of a copy of that same function. For all possible translations of the copy
(except when the translation is zero), only a single point from the original and the
copy overlap resulting in a value of 1 for the autocorrelation for those 20 xy
translations. When the translation is zero, all five points overlap giving an integral
value of 5. So the autocorrelation plotted in Fig. 6.17b has 20 points each with a
value of 1 and a central point with a value of 5. The SDM, shown in Fig. 6.17c, is
the same pattern except that the central point is missing. Alternatively, one could
create the SDM by considering a histogram of all the neighbors for each of the 5 ion
sites and it would result in the same plot for the SDM as shown in Fig. 6.17c.
The SDM is a record of all inter-ionic position vectors in a dataset from zero to
some prescribed maximum vector length (Neighborhood Extent). For the ith ion in
an image at (xi,yi,zi), the relative position of the jth ion (xij, yij, zij) is recorded.
In principle, the relative position of every atom pair in the image can be recorded
(the infinite SDM). In practice, we have limited the information retrieved to a cell of
dimensions (xmax, ymax, zmax). Once the positions of all atoms in the cell
relative to the ith atom have been recorded, the process is repeated for the jth atom
and so on for all atoms in the image (or subset of the image). The result is a three-
dimensional histogram of all relative positions out to the dimensions of the cell.
This histogram maps the average near-neighbor environment in the image. Note
that there should be no data at (0,0,0) since there should be no atom pairs where
each atom occupies the same location. If we disregard edge effects for simplicity,
the SDM size grows as
6.5 Spatial Distribution Maps 189
N n ncell (6.11)
where Vcell is the cell volume and is the number of atoms per unit volume in the
material. For a cubic cell, this becomes
Thus, the SDM size increase is linear in n and cubic in xmax. This growth rate of
the SDM data size is modest and is much lower than other common techniques that
grow as n2 or n ln n especially as n gets large. For example, the complexity of
Fourier techniques mentioned above grows as n ln n, where n grows as the cube of
the linear spatial extent of the image. This small data file size is one of the
advantages of the SDM over other types of analyses.
Once the SDM cell is fully populated, it is informative to set a region of interest
around the principal signal so that periodicities corresponds to the structure may be
more readily visualized. This is illustrated in Fig. 6.18, which shows A 3D plot from
a <110>-oriented tungsten dataset where an iso-intensity surface has been imposed
on the SDM such that 15 % of the signal is contained within the surface. This
process effectively removes the background signal for greater visibility. The
enclosed surfaces in Fig. 6.18 are the atomic separations represented in the cell
that have a high repeat frequency in the data. The structure of these atomic
separations corresponds directly to the real space lattice.
IVAS allows the user to perform SDM calculations both during the reconstruc-
tion phase (within the Cal/Recon Wizard as part of the Reconstruction Explorer)
and after reconstruction. The types of 1D and 2D plots are illustrated in Fig. 6.19.
The range of the SDM is defined over a volume defined by Neighborhood Extent
(in nm) along all three axes. Once the SDM has been calculated, the user then
chooses how to visualize the data. Projections of the entire 3D range are possible as
ZX, ZY, and Z projections. Alternatively, single planes within the SDM range can be
selected by utilizing Z-Cut and Z-Plane-Offset (Reconstruction Explorer only) and
projected onto the XY-plane or along the x- or y-axes. Note that when visualizing the
SDM within the Reconstruction Explorer, the SDM is sampled, while SDMs
generated after reconstruction (on the POS file) are not sampled.
Besides calculating SDMs that utilize all the ion events, it is possible to choose the
chemical identities of the center elements as well as the neighbor elements. This
allows one to investigate neighborhood relationships such as site-occupancy [27, 37].
190 6 Selected Analysis Topics
Fig. 6.18 3D plot of the SDM data from a <110>-oriented tungsten dataset using a 15 %
iso-intensity surface has been imposed. The bounding box of the cell is a cube with xmax
ymax zmax 1 nm. (a) View at an angle to the {110} (x axis tilted up). (b) View along the
<100> which is 45 from the x and z directions. Reprinted from [26] with permission from
Cambridge University Press
Fig. 6.19 (a) A full 3D SDM is represented as an array of high intensity regions (red regions).
IVAS provides multiple ways of viewing this 3D volume. (b) A limited subvolume is first defined
by the user. This is equivalent to considering nearest-neighbors out to a maximum distance DNE.
(c) Users can select to view a 2D projection of this volume in the ZX or ZY plane, or as a 1D
projection along the Z axis. (d) Users can also select an individual Z plane of features using ZCUT
and displaying them as a 2D YX projection or a 1D projection along the X or Y axis
6.6 Application of Spatial Distribution Maps 191
the interatomic spacing along this direction in the tungsten specimen is clearly
resolved by the SDM in Fig. 6.21. The <110> zone axis of the tungsten specimen is
visible near the center of Fig. 6.20b and is oriented close to the z-axis in
Fig. 6.21. In the data, a {110} plane normal is near parallel to the z-axis, so the
signal in the SDM can be integrated in xy to reveal the planes, Fig. 6.22. Each peak
is the cumulative signal from the <110> planes in Fig. 6.20. Thin slices normal to
z and centered on the peaks in Fig. 6.21 are used to isolate the strongest signal and
obtain the images shown in Fig. 6.22. The z-oriented slices are created with
zslice 0.01 nm. These slices clearly reveal the presence of the periodicity in
the <110> planes which is consistent with the body-centered cubic (BCC) real-
space crystal lattice.
Fig. 6.23 (a) The portion of the XY SDM to be used for SDM-based efficiency calibration. (b) The
z cut is loosened to include an entire atomic layer. (c) This results in the much lower-resolution XY
SDM that is integrated to estimate the efficiency
a third scale factor. It should be noted that the application of these post-reconstruction
scale factors does not inherently preserve the reconstruction density implied by the
specified atomic volumes. Depending on the quality of the dataset, we will typically
need to smooth and threshold the SDMs in order to process them this way.
Once a dataset has been reconstructed and calibrated we can begin the efficiency
estimation procedure. The idea is that since we have calibrated the reconstruction
we should be able to investigate the occupancy of the average atoms neighbor-
hood. Specifically, for each atom we see and treat as a center, we will look at the
SDM volume that ought to occupied by some number of neighbors and ask how
many were actually seen.
We have chosen to focus on a method that counts the observed fraction of atoms
found in the central 8 in-plane spots closest to the analyzed center atom. In a
correctly calibrated <110> W reconstruction the region occupied by these 8 spots
is not square. So to facilitate our integration of the SDM histogram we apply a 2
stretch to the reconstructed x positions before filling the SDM. Now, the central
eight SDM intensity maxima are easily counted within a square area as shown in
Fig. 6.23a. (This adjustment is not necessary if calibrating on <100> FCC poles for
example.) We then loosen the tight Z cuts generally applied to make the XY SDM so
that they include the full width of the z-plane, Fig. 6.23b. This obviously degrades
the signal-to-noise ratio (SNR) of the resulting SDM calculation but is necessary to
ensure we count all the appropriate atoms, Fig. 6.23c. The actual efficiency
calculation is simply the observed occupancy of the very loosely cut SDM. In
this example we are interested in the spatial region defined by the 8 closest in-plane
spots. So our efficiency estimate is:
N SDMCounts
8 : (6.14)
8N Centers
We have found that the efficiency estimate obtained can strongly depend on the
projection settings used to perform the initial reconstruction. Simply scaling an
arbitrary reconstruction to correct the SDM spacings is not sufficient to specify
a unique efficiency estimate. We have found that even after calibration,
efficiency estimates decrease if the reconstruction radius is too large. A 20 % increase
in reconstruction radius can result in a factor of 2 decrease in estimated efficiency.
SDMs have multiple applications to ordered alloys. They may be used to demon-
strate that long range order is present in a dataset, investigate whether a particle or
phase is coherent with its matrix, and estimate site occupancy of solute atoms on a
sub-lattice. Examples of these three applications will be presented in this section.
Many heat resistant ferritic steels have superior thermal expansion and conductivity
properties compared to Ni-based alloys but are somewhat limited in their applica-
tions above ~600 C by their mechanical properties. Efforts are being made to use
precipitation strengthening to improve these properties and thus the usefulness of
such alloys. One such effort uses coherent (Ni,Fe)Al precipitates in a ferritic matrix
and has shown good initial results [40].
196 6 Selected Analysis Topics
Fig. 6.24 (a) LEAP data of two Al3Sc precipitates in an Al matrix with the <100> direction
nearly parallel to the analysis direction. A 1 at.% Sc isoconcentration surface has been used in
order to extract the precipitates only. Analysis of the planar spacings along the <100> direction is
shown by the z-SDMs of (b) the AlAl atoms and the (c) ScSc atoms within the Al3Sc pre-
cipitates. This comparison is quantified in (d)
A subset of LEAP data from a NiAlFeCr alloy [41] is shown in Fig. 6.25a, where
small FeCr precipitates are clearly visible in the microstructure. SDM analysis may
be performed on either single elements or combinations of elements, and in this
case the analyses were performed by grouping together the NiAl atoms and the
FeCr atoms. The results are shown in Fig. 6.25b where the gray line denotes the
NiAl SDM results and the purple line denotes the FeCr results. The curves match
well in the overlaid plot, indicating that the spacing between Fe and Cr atoms (in the
bcc phase) is the same as the spacing between the Ni and Al atoms (B2 phase).
Fig. 6.25 (a) LEAP data from a NiAlFeCr alloy where the microstructure consists of small FeCr
precipitates in a NiAl matrix. (b) NiAlNiAl and FeCrFeCr z-SDMs showing very similar planar
spacing between the matrix and the precipitates
Fig. 6.26 Site occupancy of Nb atoms in a 46 at.%Ti52 at.%Al2 at.%Nb alloy which exhibits
the L10 crystal structure (a) as determined by (b) z-SDM analysis of AlAl, AlTi, and AlNb [37]
References
1. Bevington, P.R., Robinson, K.D.: Data Reduction and Error Analysis for the Physical
Sciences, 3rd edn. McGraw-Hill Higher Education, New York, NY (2003)
2. Gedcke, D.A.: How histogramming and counting statistics affect peak position precision.
ORTEC Application Note AN58 (2005)
198 6 Selected Analysis Topics
3. Currie, L.A.: Limits for qualitative detection and quantitative determination: application to
radiochemistry. Anal. Chem. 40, 586592 (1968)
4. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion Micros-
copy. Oxford University Press, Oxford (1996)
5. Meier, P.C., Zund, R.E.: Statistical Methods in Analytical Chemistry. Wiley, New York, NY
(2005)
6. Boyd, R.K., Basic, C., Bethem, R.A.: Trace Quantitative Analysis by Mass Spectrometry.
Wiley, West Sussex (2008)
7. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer Series
in Materials Science, vol. 160. Springer, New York, NY (2012)
8. Gedcke, D.A.: How counting statistics controls detection limits and peak precision. ORTEC
Application Note AN59 (2001).
9. Kelly, T.F.: Kinetic-energy discrimination for atom probe tomography. Microsc. Microanal.
17, 114 (2011)
10. Hellman, O.C., Rivage, J.B., Seidman, D.N.: Efficient sampling for three-dimensional atom
probe microscopy data. Ultramicroscopy 95, 199205 (2003)
11. Lopes, A., Bordlie, K.: Interactive approaches to contouring and isosurfaces for geovisua-
lization. In: Dykes, J., MacEachren, A.M., Draak, M.J. (eds.) Exploring Geovisualization,
p. 345. Elsevier, London, UK (2005)
12. ONeill, R.W., Larson, D.J., Thompson, K., Kunicki, T.C., Geiser, B.P.: Measuring the
roughness of buried interfaces in nanostructures by local electrode atom probe analysis.
Microsc. Microanal. 12(S2), 1746CD1748CD (2006)
13. Hellman, O.C., Vandenbroucke, J.A., Rusing, J., Isheim, D., Seidman, D.N.: Analysis of three-
dimensional atom-probe data by the proximity histogram. Microsc. Microanal. 6, 437444
(2000)
14. Hyde, J.M., English, C.A.: An analysis of the structure of irradiation induced Cu-enriched
clusters in low and high nickel welds. MRS Symp. Proc. 650(R6.6), R6.6.1R6.6.12 (2000)
15. Vaumousse, D., Cerezo, A., Warren, P.J.: A procedure for quantification of precipitate
microstructures from three-dimensional atom probe data. Ultramicroscopy 95, 215221 (2003)
16. Heinrich, A., Al-Kassab, T., Kircheim, R.: Investigation of the early stages of decomposition
of Cu-0.7%at.% Fe with the tomographic atom probe. Mater. Sci. Eng. A 353(12), 9298
(2003)
17. Stephenson, L.T., Moody, M.P., Ringer, S.P.: Techniques for the analysis of clusters and
aggregations within atom probe tomography data. Microsc. Microanal. 12(S2),
1732CD1733CD (2006)
18. Cerezo, A., Davin, L.: Aspects of the observation of clusters in the 3-dimensional atom probe.
Surf. Interface Anal. 39(23), 184188 (2007)
19. Ceguerra, A.V., Moody, M.P., Stephenson, L.T., Ringer, S.P.: Analysis techniques for nano-
scale solute clustering in atom probe tomography. Microsc. Microanal. 13(S2), 16021603
(2007)
20. Serizawa, A., Miller, M.K.: Influence of analysis parameters on the microstructural character-
ization of nanoscale precipitates. MRS Symp. Proc. 1231, 1925 (2010)
21. Stephenson, L.T., Moody, M.P., Gault, B., Ringer, S.P.: Estimating the physical cluster-size
distribution within materials using atom-probe. Microsc. Res. Tech. 74(9), 799803 (2011).
doi:10.1002/jemt.20958
22. Stephenson, L.T., Moody, M.P., Ringer, S.P.: Theory of solute clustering in materials for atom
probe. Phil. Mag. 91, 2200 (2011). http://dx.doi.org/10.1080/14786435.2011.554909
23. Marquis, E.A., Hyde, J.M.: Applications of atom-probe tomography to the characterisation of
solute behaviours. Mater. Sci. Eng. R Rep. 69(45), 3762 (2010)
24. Stephenson, L.T., Moody, M.P., Liddicoat, P.V., Ringer, S.P.: New techniques for the analysis
of fine-scaled clustering phenomena within atom probe tomography (APT) data. Microsc.
Microanal. 13, 448463 (2007)
References 199
25. Hyde, J.M., Marquis, E.A., Wilford, K.B., Williams, T.J.: A sensitivity analysis of the
maximum separation method for the characterisation of solute clusters. Ultramicroscopy 111
(6), 440447 (2011). doi:10.1016/j.ultramic.2010.12.015
26. Geiser, B.P., Kelly, T.F., Larson, D.J., Schneir, J., Roberts, J.P.: Spatial distribution maps for
atom probe tomography. Microsc. Microanal. 13, 437447 (2007)
27. Boll, T., Al-Kassab, T., Yuan, Y., Liu, Z.G.: Investigation of the site occupation of atoms in
pure and doped TiAl/Ti3Al intermetallic. Ultramicroscopy 107(9), 796801 (2007)
28. Fultz, B., Howe, J.: Transmission Electron Microscopy and Diffractometry of Materials.
Springer, New York, NY (2013)
29. Camus, P.P., Larson, D.J., Kelly, T.F.: A method for reconstructing and locating atoms on the
crystal lattice in three-dimensional atom probe data. Appl. Surf. Sci. 87/88, 305310 (1995)
30. Cerezo, A., Abraham, M., Lane, H., Larson, D.J., Thuvander, M., Seto, K., Warren, P.J.,
Smith, G.D.W.: Three-dimensional atomic scale analysis of interfaces. Paper presented at the
Electron Microscopy and Analysis Group Conf. EMAG99, Sheffield, UK, 1999
31. Vurpillot, F., Da Costa, G., Menand, A., Blavette, D.: Structural analyses in three-dimensional
atom probe: a Fourier transform approach. J. Microsc. 203(3), 295302 (2001)
32. Vurpillot, F., Renaud, L., Blavette, D.: A new step towards the lattice reconstruction in 3DAP.
Ultramicroscopy 95, 223229 (2003)
33. Warren, P.J., Cerezo, A., Smith, G.: Towards 3D lattice reconstruction with the position
sensitive atom probe. Microsc. Microanal. 4(S2), 8687 (1998)
34. Vurpillot, F., Geuser, F.D., Costa, G.D., Blavette, D.: Application of Fourier transform and
autocorrelation to cluster identification in the three-dimensional atom probe. J. Microsc. 216,
234 (2004)
35. Moody, M.P., Gault, B., Stephenson, L.T., Ringer, S.P.: Applications of spatial distribution
maps for advanced atom probe reconstruction and data analysis. Microsc. Microanal. 15(S2),
246247 (2009)
36. Suram, S.K., Rajan, K.: Refining spatial distribution maps for atom probe tomography via data
dimensionality reduction methods. Microsc. Today 18(5), 941952 (2012)
37. Marquis, E.A., Geiser, B.P., Hekmaty, M.A., Larson, D.J.: Application of spatial distribution
mapping to the investigation of alloys. Microsc. Microanal. 13(S2), 196197 (2007)
38. Seidman, D.N., Marquis, E.A., Dunand, D.C.: Precipitation strengthening at ambient and
elevated temperatures of heat-treatable Al(Sc) alloys. Acta Mater. 50, 40214035 (2002)
39. Larson, D.J., Kelly, T.F.: Nanoscale analysis of materials using a local-electrode atom probe.
Microsc. Anal. 78, 5962 (2006)
40. Stallybrass, C., Sauthoff, G.: Ferritic FeAlNiCr alloys with coherent precipitates for high-
temperature applications. Mater. Sci. Eng. 387389, 985990 (2004)
41. Larson, D.J., Prosa, T.J., Kostrna, S.L.P., Ali, M., Kelly, T.F., Stallybrass, S., Schneider, A.,
Sauthoff, G., Degass, J.: Local electrode atom probe study of phase transformations in an
Fe-Al-Ni-Cr alloy. Microsc. Microanal. 12(S2), 968CD970CD (2006)
42. Al-Kassab, T., Yuan, Y., Kluthe, C., Boll, T., Liu, Z.-G.: Investigation of the ordering and
atomic site occupancies of Nb-Doped TiAl/Ti3Al intermetallics. Surf. Interface Anal. 39(23),
257261 (2007)
43. Yang, R., Hao, Y.L., Song, Y., Guo, Z.X.: Site occupancy of alloying additions in titanium
aluminides and its application to phase equilibrium evaluation. Zeitschrift Fur Metallkunde 91
(4), 296301 (2000)
44. Hao, Y.L., Yang, R., Cui, Y.Y., Li, D.: The influence of alloying on the alpha(2)/(alpha
(2) + gamma)/gamma phase boundaries in TiAl based systems. Acta Mater. 48(6),
13131324 (1999)
Chapter 7
Applications of the Local Electrode
Atom Probe
Over the past decade there has been a substantial expansion of the applicability of
atom probe tomography (APT) to materials of all types. This expansion has been
spurred on by major instrumentation developments resulting in the achievement of
high data collection rates, large fields of view, and renewed utilization of laser
pulsing (see Chap. 3). Furthermore, the advent of focused ion beam (FIB) methods
for specimen preparation (see Chap. 2) in APT has had an equally profound impact.
FIB-based methods have not only made it possible to create a LEAP specimen from
nearly any bulk material type, but have also provided the capability to create
specimens from specified regions of a sample and in nearly any orientation. At the
turn of the century, random site specimen creation was the norm and site-specific
specimen creation was a time-consuming process involving many handling steps in
conjunction with electropolishing and electron microscopy. Today, FIB-based site-
specific specimen creation is routine and electropolishing is often only used when it
is more practical. While the use of thermal pulsing is necessary for materials that
cannot be evaporated successfully with field pulsing, thermal pulsing also has been
found to improve yield for a range of metal applications as well. Certain metal
specimens, even high-strength steels, that will not run with adequate yield in voltage
pulsing have been found to run with much higher success rates in laser pulsing. With
these major advances, applications of APT have blossomed.
This final chapter is a survey of some of the most recent APT applications. The
content is organized by material type as opposed to structural type or industry.
Thus, there are sections on metals, ceramics and geological materials, semiconduc-
tors, organic and biological materials, and composite structures and devices for
combinations of the above. Note that there are literally thousands of examples
which could be used in this chapter (see [1] for a recent review of many additional
examples), and no attempt has been made to be comprehensive in any of these
categories. Rather, we have chosen examples that illustrate the different ways
that APT has been used or contributed to material characterization challenges. By
describing a wide array of examples, we hope that the reader may gain a broader
perspective on the many ways that atom probes are being used today and perhaps
this may spur users to develop new applications for the technique.
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 201
DOI 10.1007/978-1-4614-8721-0_7, Springer Science+Business Media New York 2013
202 7 Applications of the Local Electrode Atom Probe
7.1 Metals
Because atom probe analysis relied almost exclusively upon voltage pulsing in its
first three decades, virtually all of the work from the 1970s through 2000 was on
metals. As a result, these applications tend to be the most mature. However, new
and more sophisticated applications are constantly being pursued and reported in
the literature, as highlighted in the following examples.
Srinivasan et al. [2] have used high-angle annular dark-field scanning transmission
electron microscopy (HAADF STEM) and APT to study the orderdisorder transi-
tion between the disordered matrix and ordered 0 precipitates in a nickel-base
superalloy, Fig. 7.1. STEM was used to elicit the variation in order across the
coherent <001> hetero-interface through intensity ratios of the aluminum-rich/
aluminum-poor {002} plane signal. APT was used to measure the aluminum
composition profile and estimate the ordering transition width across this same
interface type. The two techniques agree very well. Furthermore, the experimental
results matched with high precision the atomistic modeling results of Mishin [3]
Fig. 7.1 Orderdisorder transition at a hetero-interface. (a) {002} plane intensity in HAADF
STEM at the 0 interface. (b) APT image with phase (Co atoms shown in blue) and 0 phase
(Al atoms shown in red). (c) Close-up of the lower interface in (b) showing the Al atoms in the
{002} where the interface transition based on atomic ordering is about 46 atomic layers
(0.71.0 nm) thick. (d) A proximity histogram across the same interface showing a compositional
interface width of about 2 nm. Reprinted from Srinivasan et al. [2]. Copyright (2009) American
Physical Society
7.1 Metals 203
and Ardell and Ozolins [4]. They found that the orderdisorder transition width
(about 0.71 nm) is smaller than the compositional width of the interface (about
2 nm), which raises fundamental questions concerning the definition of interface
width at orderdisorder interfaces. This work illustrates an important trend in APT
work today: there is a strong convergence between the increasing capabilities of
atomistic modeling and the experimental data available from APT (and TEM). This
synergistic relationship is expected to continue as it is a natural marriage of
predictive and characterization capabilities at the atomic scale.
As the scale and detail available in atom probe data are directly comparable with
that available in atomistic simulations, synergistic integration of the two techniques
is expected to grow rapidly. Gault et al. [5] have used APT to determine the
occupancies of various sites in an ordered phase, L12 (0 , Al3(Li,Mg)), in a model
aluminum alloy. In particular, it was not known whether magnesium would inhabit
the aluminum sites or the lithium sites in an alloy precipitate and APT appeared to
be the only experimental technique that might readily resolve the question. The
results are directly compared with atomistic simulations.
Figure 7.2 shows LEAP data for two heat treatments with precipitates present.
Subvolumes that contained closely spaced atomic planes were extracted from each
dataset. Spatial distribution maps (SDMs, see Chap. 6) were used to analyze the
subvolumes for atomic position correlations. The result is shown in Fig. 7.3. In the
disordered fcc matrix, aluminum, lithium, and magnesium each occupy any lattice
site with equal probability. In the precipitates, however, it is clear that the magne-
sium preferentially occupies the lithium sites in the Al3Li structure. The relative
occupancy fraction between lithium and magnesium on the lithium sites was not
determined precisely but is a subject for further investigation. Atomistic simula-
tions show that the lowest energy configuration for magnesium in the Al3Li pre-
cipitates is for magnesium to sit on the lithium sites; this is directly aligned with the
experimental LEAP findings.
Fig. 7.2 Typical three-dimensional reconstruction from atom probe analyses of the alloy after (a)
8 h at 150 C and (b) 24 h at 120 C shown at the same scale. (c) A close-up showing the {022}
atomic planes imaged in the dataset displayed in (a). (d) A close-up showing the {002} atomic
planes imaged in the dataset displayed in (b). (e, f) Model view of the fcc Al matrix and of the L12
structure. Reprinted from Gault et al. [5] with permission from Scripta Materialia
Miller et al. [6] have also investigated the use of iso-density surfaces as a way to
reveal the presence of small voids in a structure. They used a combination of
simulations and experimental results to determine that voids as small as 1 nm
diameter should be detectable in APT. Figure 7.4b shows an iso-density surface
that reveals three spheroidal nanovoids in an aluminumyttriumiron metallic
glass. Carbon and oxygen segregation to the nanovoids was also observed by
superimposing atom maps onto the iso-density surfaces. The data show that the
trace concentrations of carbon and oxygen in this material are preferentially
segregated to the voids.
Fig. 7.3 Normalized species-specific SDMs for the two heat treatments in Fig. 7.2 in the matrix
(top row) and the precipitates (bottom row). Note the change in the periodicity between the
precipitates and the matrix in both cases. The corresponding structure is shown in the inset for
each case. Reprinted from Gault et al. [5] with permission from Scripta Materialia
Fig. 7.4 (a) Nanovoids in nanocrystalline copper revealed by a thin section atom map. Oxygen
and sulfur are evidently segregated to the voids. (b) Nanovoids in an aluminumyttriumiron
metallic glass revealed by an isodensity surface. Note that carbon and oxygen are found to
segregate to the nanovoids in this material. Reprinted from Miller et al. [6] with permission
from Ultramicroscopy
Fig. 7.5 STEM X-ray map of the near surface region of alloy 600 after IGA. Chromium oxides
form a continuous film to depths of 200 nm. Sulfides form discontinuous precipitates at greater
depths up to 500 nm. Reprinted from Schreiber et al. [7] with permission from Microscopy and
Microanalysis
7.2 Catalytic Materials 207
Fig. 7.6 (a, b) LEAP reconstruction of the IGA in alloy 600 depicting isoconcentration surfaces
of oxides (blue), sulfides (red), and a large carbide (maroon) in crosssection and plan-view,
respectively. The surface of the specimen prior to FIB lift-out is noted by a yellow dashed line,
above which exists some oxide overgrowth. (cf) Atom maps (10 nm image depth) depicting Ni
(green), Cr (yellow), 16O+ (black), 32S2+ (red), and Cu (blue). Reprinted from Schreiber et al. [7]
with permission Microscopy and Microanalysis
example of the need for, and execution of, site-specific analysis with APT and its
complementary nature with SIMS and TEM.
Metal use in nanoparticles is industrially important, but offers further challenges
in preparation for APT analysis. Some examples of this application are given in the
next section.
Fig. 7.7 (a) Atom probe microscopy results from the CoCuMn catalyst nanoparticles. The images
provided here are sections through 3D tomographic reconstructions so as to reveal the
intragranular structure of the nanoparticles. The data are presented in the form of atom maps
where cobalt atoms are depicted as blue spheres, copper atoms as orange spheres, manganese
atoms as green spheres, and oxygen as white spheres. A faceted coreshell structure that
contained intracore clusters was observed. The data displayed are from a 10 nm thick slice through
one grain. (b) An enlarged view from a 5 nm thick slice of the data at the coreshell interface.
Oxygen (white) was distributed throughout the coreshell interface. (c) The intracore clusters
were observed to contain a high concentration of copper and manganese; 5 nm thick atom map. (d)
A 3D sectional view of the element distribution. Reprinted from Xiang et al. [8] with permission
from Journal of the American Chemical Society
After activating the catalyst precursor material, a catalyst powder was obtained
which had to be passivated prior to APT analysis. An APT specimen was prepared
using FIB-based techniques to encapsulate and then sharpen the specimen. This
preparation made it possible to analyze a single nanoparticle as shown in Fig. 7.7.
The particle appears faceted and a striking coreshell structure is observed,
Fig. 7.7a, b. The shell thickness is about 2 nm with all three elements present in
the shell. There is also a region of the nanoparticle that contains elevated levels of
copper and manganese relative to the cobalt, Fig. 7.7a, c, and which may be
considered a precipitate. Note also that there is very little oxygen in the shell
which indicates that metal oxides are not forming. The key finding in this work is
the metallic shell structure with high levels of intermixed cobalt, copper, and
manganese. The finding is consistent with expectations and explains the high
catalytic selectivity of these processed nanoparticles. This work is an illustration
of how APT may impact the development of real catalyst nanoparticles and can
help in the design of catalyst compositions that lead to large improvements in a
pervasive industrial process.
7.2 Catalytic Materials 209
Fig. 7.8 A serial expose of the evolution of the microstructure in a Pt-24 at.%Rh-10 at.%Ru alloy
after oxidation at 873 K with increasing time. After 10 min (second from left), a
rhodiumruthenium oxide phase forms on a grain boundary. After 30 min (middle), this oxide
volume fraction has increased markedly and small amounts of a ruthenium-rich oxide have
formed. After 90 min, distinct rhodium-rich and ruthenium-rich oxide phases have formed. The
oxide is now the majority of the specimen. Finally, after 300 min, the metallic phase has almost
been fully replaced by oxide and the volume fraction of the ruthenium-rich oxide has increased. At
the bottom, a specimen after 120 min reduction at 673 is shown rhodium-rich and ruthenium-rich
metallic phases have been produced. Reprinted from Li et al. [11] with permission from Journal of
Physical Chemistry C. Copyright (2012) American Chemical Society
metallic phase form. The authors suggest that this reduced microstructure, a
novel complex island structure, may be a candidate for an improved catalyst
with extremely small metallic nanophases on the order of 5 nm diameter.
Certainly, the atomic-scale characterization used in these studies is essential to
developing an understanding of the microstructures on a scale that is relevant for
catalysis work.
The reemergence of pulsed lasers in APT analysis has opened the way for the
analysis of insulating materials including oxides of various elements like lithium
[12] magnesium [13, 14], aluminum [1517], silicon [15, 1820], iron [21], nickel
[17, 22], zinc [13, 23, 24], zirconium [12, 25], indium [26], and cesium [12,
2729]. In addition, some theoretical studies have also been published [3034]. Sev-
eral examples of the application of APT to ceramic materials are presented in this
section.
Cerium oxide is a model surrogate material for the various nuclear fuel materials in
both light water reactors and fast breeder reactors. Kirchhofer et al. [29] have
studied CeO2 to establish appropriate handling and analysis conditions without
the complication of radioactivity. Their work is noteworthy because they made a
concerted effort at the outset to study a host of analysis conditions and in particular,
to establish the conditions under which LEAP tomography will produce the correct
composition in these materials. Figure 7.9 shows a mass spectrum from CeO2 taken
under the optimal conditions: 1 pJ energy per pulse, 20 K base specimen temper-
ature, and a charge state ratio (Cs++/Cs+) of 40. Under these conditions, the
composition is determined to be 67 at.% oxygen as shown in Fig. 7.10. Higher
laser pulse energies were postulated to lead to evaporation of neutral molecular
oxygen in increasing fraction of the total. Neutral atoms are not accelerated
following field evaporation and are not therefore detected. In addition, the authors
found that oxygen ion evaporation is not homogeneous across the apex of the
specimen and that the inhomogeneity gets worse with increasing laser energy,
Fig. 7.11.
In this work, higher laser pulse energy also produced lower mass resolving
power (MRP) in the spectra, which was attributed to greater heating and thus
greater cooling time required for the specimen. In many other materials, higher
laser power leads to greater MRP since the cooling rate from higher temperatures is
greater. However, if the specimen does not cool sufficiently to the base temperature
7.3 Ceramic and Geological Materials 211
Fig. 7.9 Mass spectrum of a cerium oxide used for the experiments obtained at 20 K specimen
base temperature and 1 pJ laser energy. Note that Hf was present in the sample at concentration
below 1 at.%, which is commonly found as an impurity in cerium oxide. Some minor peaks are not
labeled for clarity. Reprinted from Kirchhofer et al. [29] with permission from Journal of Nuclear
Materials
Fig. 7.10 Measured oxygen content for the cerium oxide on the laser side of the specimen, or
opposite from the laser side, shown for different laser energies and for a specimen base temper-
ature of 20 K. The oxygen peak at 16 Da was ranged as both O+ and O2++ for the composition
analysis; the best composition was obtained using the latter ionic ranges. Reprinted from
Kirchhofer et al. [29] with permission from Journal of Nuclear Materials
212 7 Applications of the Local Electrode Atom Probe
Fig. 7.11 2D concentration maps showing the relative distribution of ionic clusters (using a
thermal color gradient, where the red color indicates the highest frequency of evaporation events)
across the specimen for the conditions that yielded the best stoichiometric evaporation of ions. The
laser pulse was incident on the specimen at the lower left corner of each plot. Note that evaporation
of O+/O2++ ions was not even across the specimen, while field evaporation of CeO+/CeO++ clusters
was more homogeneous. Reprinted from Kirchhofer et al. [29] with permission from Journal of
Nuclear Materials
between pulses, then there is an effective rise in base temperature, resulting in data
quality reduction. Indeed, increasing thermal tails were observed in the spectra with
increasing laser energy. Another observed benefit of lower laser energy was higher
spatial resolution in the data which is attributed to diminished potential surface
diffusion at the lower temperatures associated with lower pulse energy. One
disadvantage of lower laser pulse energies was a higher fraction of multiple hits
which was explained as due to a lower fraction of atoms evaporating as
molecular ions.
7.3 Ceramic and Geological Materials 213
Fig. 7.12 (a) Yttrium L X-ray image from an SXFive electron probe microanalyzer of a 0.5 Ga
zircon. The yttrium signal is elevated at the fossil resorption front. (b) The image from (a)
superimposed on a backscatter image of the specimen after a FIB lift-out from the region of the
resorption front [35]
7.3.2 Zircons
Zircons (ZrSiO4) are valuable minerals scientifically due to their extreme stability
and hence longevity in solar (terrestrial and extraterrestrial) and extrasolar envi-
ronments. Like most materials, their microstructure contains a record of their
history. Two such examples illustrate how LEAP data have contributed to the
understanding of not just Earths history, but possibly the history of Mars as well.
Snoeyenbos et al. [35] were the first to demonstrate LEAP analysis of ancient
zircons. They used an electron probe microanalyzer (EPMA) to map the low
concentration element, yttrium, in a 0.5 Ga zircon (1 Ga109 years old),
Fig. 7.12, which guided the subsequent specimen preparation by FIB to a resorption
front in the zircon. The yttrium concentration at the resorption front is too low to
detect in standard energy-dispersive X-ray spectroscopy (EDX) analysis in an
SEM. The LEAP data show that a band of elevated uranium, thorium, yttrium,
phosphorous, and aluminum at the coreovergrowth boundary is consistent with a
fossil resorption front. Some yttrium, phosphorous, uranium and thorium from the
resorption front was remobilized up to several m into the overgrowth along
subparallel, sub-planar structures, Fig. 7.13. Despite considerable compositional
heterogeneity in the overgrowth, there is no evidence of migration of radiogenic
lead from its uranium parent. This research demonstrates that the decades of work
with SIMS instruments to use lead isotope ratios for dating was valid because lead
did not migrate away from its point of origin.
Valley et al. [36] were able to reaffirm that dating based on 206Pb207Pb isotope
ratios in LEAP data for 4.37 Ga zircons match SIMS-based dating within the
statistical uncertainty of the smaller number of atoms in LEAP datasets. Further-
more, yttrium-rich clusters were found, Fig. 7.14, that were interpreted as having
214 7 Applications of the Local Electrode Atom Probe
Fig. 7.13 Atom maps of the key elements of interest for a specimen taken near the resorption front
of Fig. 7.12. The lead does not appear to have migrated away from its parents uranium and thorium
[35]
Fig. 7.14 Atom map containing 260 million atoms (only yttrium shown) of a 4.4 Ga zircon.
Yttrium-rich clusters are clearly visible
formed, not during their original crystallization, but later during the magmatic
upheaval event at 3.4 Ga. Lead ratios inside the yttrium-rich clusters, Fig. 7.15,
match models for diffusive migration of the lead to the clusters during the upheaval
event [36]. Geological applications of LEAP have just begun, but they are already
having impact on the field. There is a bright and exciting future for geological
science that utilizes APT to acquire heretofore unattainable information about the
largest-scale events in our universe from the smallest-scale traces of their existence.
Fig. 7.15 Segment of the mass spectrum from the dataset shown in Fig. 7.14 that contains the
Pb++ isotopes. (a) Lead signal from the full dataset, (b) lead signal from inside the yttrium-rich
clusters, and (c) lead signal from outside the yttrium-rich clusters. It is clear that the lead is
segregated to the yttrium-rich clusters and that 207Pb is even more preferentially found in these
clusters
have been created in another star prior to our solar system. In truth, there are many
events that can lead to the formation of nanoscale diamonds, including: between
presolar supernovae, condensation in the interstellar medium, formation in the
protosolar nebula, and even formation in the early solar system. The relative
amounts of stable isotopes of elements such as carbon are known to be different
for each of these sources and can be used as a signature. However, any given
collection of nanodiamonds could have representatives from any or all of these
sources. Thus, though isotope ratios may be determined from a collective of
particles by any of several techniques, what is needed is analysis of isotope ratios
from many individual nanodiamonds. APT is the only technique that can deliver
such information. Heck et al. [37] have spent several years trying to employ APT to
discern this very information. They faced two primary obstacles: (a) specimen
preparation of sub-10 nm diamonds and (b) qualification of the accuracy of the
isotope ratios determined from APT.
Specimens were prepared from nanodiamonds using an acid-dissolution residue
Allende DM, a well-studied nanodiamond isolate [38]. The isolate was suspended
in alcohol, dispersed on a silicon wafer, and then covered with electron-beam-
deposited platinum in a FIB. A small wedge of material was removed and
microtips were fabricated (see Chap. 2). Data were acquired as shown in
Fig. 7.16. Great care was taken to ensure that the isotope ratios of 12C13C were
not influenced by such experimental parameters as multi-hit detection efficiency,
choice of region of interest, and hydride formation. Their preliminary results are
shown in Fig. 7.17. Note that the accepted terrestrial 12C13C ratio is 89 [39]
(shown by the solid line). Within the scope of this first effort, no statistically
significant difference was found between terrestrial detonation nanodiamonds
(DND), terrestrial ultrananocrystalline diamonds (UNCD), and meteoritic
nanodiamonds (Allende DM). Such careful analyses and attention to detail are
essential to developing a method and establishing the reliability of APT in deliv-
ering accurate isotopic ratios for forensic cosmology.
216 7 Applications of the Local Electrode Atom Probe
Fig. 7.17 Carbon isotope peak ratios of regions of interest (ROIs) defined from selected synthetic
(DND and UNCD) and meteoritic Allende nanodiamond tips. The horizontal line corresponds to
the terrestrial 12C/13C ratio of 89 [39]. Error bars are 2 and are based on counting statistics.
Reprinted from Heck et al. [37] with permission from Meteoritics & Planetary Science
7.3.4 Ferroelectrics/Piezoelectrics
Ferroelectric oxides are a class of technologically important ceramics that are used
in a wide range of electronic components, from capacitors and transistors to
piezoelectric transducers. Relaxor ferroelectrics, in particular, have extraordinarily
7.3 Ceramic and Geological Materials 217
high piezoelectric and electrostrictive coefficients which makes them valuable for
electromechanical applications [40]. Through substitutions of niobium or lantha-
num for some of the lead in lead zirconium titanate (PZT), PNZT and PLZT relaxor
ferroelectrics are created. It has been proposed that relaxor ferroelectrics achieve
their electrostrictive properties through nanoscale phase separation. APT, therefore,
is a logical tool to investigate these compositional variations. Since these complex
oxide materials have not been significantly studied using APT, Kirchhofer
et al. [40] have performed a series of test experiments to establish the optimal
data collection conditions. In their work, APT and TEM are used to complement
each other and ensure the accuracy of the APT results. APT instrumental parame-
ters such as laser pulse energy, acquisition rate, and specimen base temperature
were optimized to obtain accurate cation and anion stoichiometry from the oxides.
The mass spectrum for PZT in the optimized run condition is shown in Fig. 7.18.
The composition determined from this analysis closely matches the expected
composition of the material, Table 7.1. This research is at an early stage, but it
has already demonstrated that this important class of materials runs well in APT and
that such studies can lead to new understanding of piezoelectric behavior.
Analysis of ceramic-based materials and composites continues to develop,
taking the field of APT into nontraditional applications. Further examples of
novel applications are summarized in the sections Organics and Biological Mate-
rials and Composites Structures/Devices below.
218 7 Applications of the Local Electrode Atom Probe
Over the past few decades, characterization of semiconductor devices has been
primarily achieved through use of electron microscopy, ion microscopy, various
X-ray technologies, as well as some emission and light optical techniques. As the
industry has moved further into the nanoelectronics realm, the use of these methods
has become more challenging due to the atomistic scale and low dimensionality of
the devices of interest. One significant goal in the application of APT to silicon-
based structures has been the analysis of actual product devices instead of test
structures. Down-scaling of devices and the reduction of sourcedrain junction size
to avoid short-channel effects combined with the increased field of view (FOV)
now available in modern atom probes [41, 42] now make the analysis of entire
device structures a reality. Several researchers recently have used APT in the
analysis of basic configurations used in semiconductor devices or portions of
partially processed device structures and these are presented in this section. Anal-
ysis of more complete or entire devices [4355] is explored in the section below
titled Composite Structures/Devices.
Fig. 7.19 (a) Atom maps of the five Ge stable isotopes (70Ge, 72Ge, 73Ge, 74Ge, and 76Ge) in the
germanium isotopic multilayers. Doubly charged (Ge++) and singly charged (Ge+) peaks in the
mass spectra in the 1st pair of the (b) 70Ge and (c) natGe layers (selected volume: 20 20 3
nm3) are shown in (b) and (c), respectively. Reprinted from Shimizu et al. [58] with permission
from Journal of Applied Physics. Copyright (2013) American Institute of Physics
220 7 Applications of the Local Electrode Atom Probe
Fig. 7.20 (a) Depth profiles by APT (solid curves) of 70Ge and 74Ge in the germanium isotopic
multilayers obtained from an area of 10 10 nm2, overlaid with the SIMS profiles (broken
curves). (b) 70Ge and 74Ge mappings in a volume of 10 10 100 nm3 with the isoconcentration
surfaces of 70Ge at 50 % for highlighting the 70GenatGe interfaces. Reprinted from Shimizu
et al. [58] with permission from Journal of Applied Physics. Copyright (2013) American Institute
of Physics
multilayers and the same result was found for both studies to within experimental
error. Thus, where the field difference between two layers or sections of a structure
is not large, APT has been shown to deliver quite high spatial resolution across
hetero-interfaces.
Nanowires have been around now for well over a decade and they remain an
interesting prospect for nanoelectronic device components. Understanding and
developing doping methods for nanowires during growth are the aims of current
research. However, there has been little information about the effects of growth
conditions on dopant concentrations in nanowires. APT analyses of nanowires
would seem to be a logical undertaking to answer these questions.
Lauhons group at Northwestern University has pioneered many of the tech-
niques needed to perform APT on nanowires. Surprisingly, specimen preparation is
more difficult than one might guess given the near-perfect shape and size of most
7.4 Semiconductor Materials 221
Fig. 7.21 (a) End view through the first 20 nm of the dopant decay region in the phosphorus-
doped germanium nanowire. (b) Schematic of part a, illustrating the hexagonal cross section
of the catalyst-defined core and associated radial doping gradient. (c) End view of the first
50 nm of the dopant decay region of a phosphorus-doped germanium nanowire with a
chromium shell and (e) the first 60 nm of the dopant decay region of a boron-doped silicon
nanowire. Scale bars are 10 nm. (d, f) Radial composition profile derived from the region in
the red rectangle in (c, e), respectively. The trijunction location is indicated by a
dasheddotted line and serves as the reference point (r 0) for each graph. Error bars
represent the statistical uncertainty in the number of counts. Reprinted from Perea
et al. [60] with permission from Applied Physics A
nanowires for APT specimens. The biggest challenge has been handling the
nanowires such that a single wire may be positioned and analyzed in an atom
probe. One way that Lauhons group solved this challenge was by growing single
nanowires on top of flat-top microtips [60]. In some recent work, silicon and
germanium nanowires were grown with boron and phosphorus dopants, respec-
tively, and the spatial distribution of the dopants was assessed, Fig. 7.21. A rim of
elevated concentration of the dopants is found as observed in Fig. 7.21a and in the
radial composition profiles of Fig. 7.21d, f. The growth direction of the wires is
<111> and the symmetry of the phosphorus distribution in the cross section of the
wire appears in accordance with this symmetry. The authors posit that there is
faceting at the liquidsolid interface that leads to this distribution. The advances in
understanding of the vaporliquid-solid growth process derived from this study are
an important step toward developing practical processing approaches for fabrica-
tion of doped nanowires. This work is a good example of a characterization task for
which APT may be the only way to obtain the desired information.
222 7 Applications of the Local Electrode Atom Probe
Fig. 7.22 Atom map of a multilayer structure containing silicon nanocrystals. The red surface is
an isoconcentration surface at 70 % silicon. The nanocrystals are formed in layers by depositing
alternating layers of silicon dioxide with silicon-rich silicon oxide that is subsequently
decomposed into nanocrystals. Image courtesy H. Gnaser [62]
Gallium nitride (GaN) has achieved great success as the basis for a wide range of
optoelectronic devices. Updoped GaN emits in the ultra-violet (UV), and alloying it
with indium nitride (InN) or aluminum nitride (AlN) produces emission with
wavelengths ranging from the green to deep UV. GaN-based multiple quantum
well (MQW) structures are used as the active regions of commercial light-emitting
diodes (LEDs) and laser diodes. Such structures often exhibit bright-line emission
7.4 Semiconductor Materials 223
Fig. 7.23 A thin section of the atom map in Fig. 7.22 that contains a single layer of nanocrystals.
A cluster finding approach has been used to isolate the nanocrystals and determine their average
phosphorus concentration at 0.77 0.4 %. The nanocrystals are about 4 nm diameter. Image
courtesy H. Gnaser [62]
despite very high dislocation densities. Recently, APT has been used to investigate
the chemical and morphological microstructure of GaN- and GaAs-based structures
[6479].
It has been proposed that carrier localization can originate from structural and
compositional effects like quantum well width fluctuations and indium clustering
[80]. APT is ideally suited to describe compositional fluctuations and look for
nonrandom statistical fluctuations such as those proposed by Miller et al. [81]. An
LED device structure (50 nm p-GaN/30 nm GaN/3.5 nm In0.15Ga0.85N QW/n-
GaN) grown by metalorganic chemical vapor deposition on a semi-polar
10 1 1 GaN substrate has recently been analyzed with APT [71]. Figure 7.24a
shows the APT reconstruction of a section of the quantum well. A proxigram
analysis, Fig. 7.24b, was done relative to the 2 % indium composition surfaces
for the top (blue line in figure) and bottom (magenta line in figure) interfaces. The
bottom interface is abruptno spectral evidence for indium is present for regions
below 0.1 at.% indium, while the top interface has a ~2 nm tail of indium into the
GaN layer above. Frequency distribution analyses were also performed [71], and
for this particular device structure, no evidence was obtained to cause rejection of
the null hypothesis at the 95 % confidence level. In other words, the indium
fluctuations in this device were consistent with random fluctuations and could not
cause carrier localization.
224 7 Applications of the Local Electrode Atom Probe
Fig. 7.24 (a) APT reconstruction of a section of a quantum well. (b) Proxigram analysis relative
to the 2 % indium composition surfaces for the top (blue line) and bottom (magenta line)
interfaces. The bottom interface is abruptno spectral evidence for indium is present for regions
below 0.1 at.% indium, whereas the top interface has a 2-nm tail of indium into the GaN layer
above. Atom colors: In, magenta; Ga, orange; N2, green. Reprinted from Prosa et al. [71] with
permission from Applied Physics Letters. Copyright (2011) American Institute of Physics
APT analyses can be challenging enough when a researcher has thorough knowl-
edge of the global structure of a specimen. However, when the structure of a
macroscopic specimen is unknown, it can be a daunting task to find the region of
interest at the atomic scale. Reverse engineering of devices can pose just such a
challenge (also see section Commercial Devices Analysis; Intel i5-650 below for
another example of competitive analysis on a logic device).
In order to explore the feasibility of competitive analysis on whole devices,
LEAP analyses were obtained from individual commercial LEDs from fully pack-
aged devices. OSRAM 455 nm Golden Dragon Plus LEDs were purchased from a
retail supplier, depackaged, and fabricated into LEAP compatible specimens using
standard FIB methods ([82, 83] and Chap. 2), Fig. 7.25a, b. The LEDs were first
depackaged to expose the active component, Fig. 7.25a. FIB was used to cut into the
structure to expose the electrically active region, Fig. 7.25b. At this point the project
encountered difficulty. The light-emitting components of the device were not
obvious in the SEM image and the depth of this component was almost 10 m.
Without a priori knowledge it could have taken dozens of specimens prepared at
different depths to target the region of interest given that the average depth of these
LEAP analyses are on the order of a few hundred nanometers. A CAMECA
SXFiveFE EPMA was used to generate an X-ray map of indium and aluminum in
this region, Fig. 7.25c. It is clear from this X-ray map that an aluminum blocking
layer was present as evidenced by the bright aluminum band across the region. The
FIB lift-out effort was then able to focus on this region and produce a specimen that
led to the LEAP image in Fig. 7.26.
7.4 Semiconductor Materials 225
Fig. 7.25 (a) Image of a commercial white LED with the active region in the center. (b) Top
surface of white LED after FIB processing to expose the active component for lift-out. (c) X-ray
map of the active area cross section in (b) taken with a CAMECA field-emission SXFiveFE EPMA
which shows the location of the respective regions based on composition
Clearly identifiable in the structure are at least four distinct regions [84] (from
top to bottom in the atom map): (1) magnesium doping (to form p-type GaN)
near the base of the structure, (2) an aluminum-rich electron blocking layer,
(3) an indium-rich quantum well region, and (4) an indium-based superlattice
structure near the top of the dataset. A composition profile vertically through this
structure is shown in Fig. 7.27. This work illustrates the feasibility of using
LEAP for both competitive analysis and failure analysis within the commercial
LED market.
Fig. 7.27 Concentration profile from a 20 nm diameter cylindrical volume through the center of
the atom map in Fig. 7.26
7.5 Organics and Biological Materials 227
CdTe is a candidate thin-film solar cell material because it may be readily deposited
on large areas of a wide range of materials including metals, glass, and polymers.
APT data have been collected from a CdTe layer within a CdTe solar cell [85, 86]
that has been prepared with standard CdCl2 and copper treatments [87]. Figure 7.28a
shows the mass spectrum which is one of the most complex APT spectra published
to date. The spectrum contains a number of high mass molecular ions. Figure 7.28b
contains 2D atom maps of CdS, CdCl, and TeS molecular ions, with a color overlay
of CdCl (orange), Cd (black), and Te (dark green). These 2D renderings show
enrichment of sulfur and chlorine at the grain boundary which is suspected to cause
degradation of performance of the solar cells. Quantitative composition profiles
generated from the dashed-box region show peak chlorine concentrations at the
grain boundary to be well over 1 at.% and sulfur concentrations of about 0.5 at.%,
which are very large amounts for impurities. Despite the complexity of the mass
spectrum, the observation of grain boundary segregation was a fairly straightfor-
ward result that provided crucial insight into the processing as it may have affected
the performance of this device.
Thin film solar cells based on CuInSe2 (CIS) and Cu(InGa)Se2 (CIGS) offer
potentially high efficiency and stability combined with a low cost of production.
Improvements in efficiency have been held back by uncertainties in the mechanism
of charge carrier losses at the pn junction created with an overcoat layer of CdS.
Cojocaru-Miredin et al. [88, 89] have analyzed this interface with a CIS layer,
Fig. 7.29. Using a proxigram analysis, they found copper, indium, and selenium in
the CdS layer. More importantly, they found cadmium diffused into the CIS layer
on a 2-nm scale, Fig. 7.29b, in sufficient quantities to create a buried pn
homojunction rather than a heterojunction. This homojunction formation could be
responsible for the high photo-conversion efficiencies of these solar cells.
If APT can be applied to organic materials with similar success to that of its
application to inorganic materials, it would be a major development. To date,
there have been several attempts and progress is being made. However, there has
yet to be an analysis of a monolithic soft biological specimen in an atom probe. This
can be ascribed, in part, to the lack of necessary equipment: a complete cryogenic
preparation, transfer, and analysis capability has yet to be fully realized. Note,
therefore, that each of the results below have a key component in common: they are
228 7 Applications of the Local Electrode Atom Probe
Fig. 7.28 (a) APT mass spectrum and (b) 3D atom maps of CdS, CdCl, and TeS molecular ions,
with a color overlay of CdCl (orange), Cd (black), and Te (dark green). These two-dimensional
renderings show enrichment of sulfur and chlorine at the grain boundary. Quantitative composition
profiles that were generated from the dashed-box region show peak chlorine concentrations at the
grain boundary to be well over 1 at.% and sulfur concentrations of approximately 0.5 at.%, which
are very large amounts for impurities. Reprinted from Choi et al. [85] with permission from
Microscopy Today
7.5 Organics and Biological Materials 229
Fig. 7.29 (a) APT atom maps of copper, indium, selenium, cadmium, and sulfur. The volume size
shown is 18.5 16 45.5 nm3. (b) Proximity histogram with respect to the cadmium 26 at.%
isoconcentration surface and 0.3 nm bin size. Reprinted from Cojocaru-Miredin et al. [88] with
permission from Applied Physics Letters. Copyright (2011) American Institute of Physics
all projects where the specimen preparation was accomplished at room temperature.
Two universities, the Eidgenossische Technische Hochschule Zurich and the Uni-
versity of Michigan each have implemented cryogenic preparation and transfer
capabilities for the LEAP, including a cryostage for FIB. What is still missing is a
cryogenic micromanipulator for a FIB, although there is progress in this area for
electron microscopy [90]. When this technique is applied for atom probe specimen
preparation, we should see an expansion of biological applications for the LEAP.
Kelly et al. recounted the history of attempts to analyze organic materials
[91]. A couple of examples will be given here to illustrate the state of this art.
230 7 Applications of the Local Electrode Atom Probe
Fig. 7.30 Typical mass spectrum for C60-doped P3HT polymer, highlighting the rich ion frag-
ment distribution observed for polymer films deposited on sharp carrier tips (inset). Expanded
mass-to-charge-state ratio range showing the large number of C60 ions observed. Reprinted from
Joester et al. [97] with permission from Cambridge University Press
Fig. 7.31 Mass spectrum for PCBM doped P3HT sample prepared by FIB lift-out extraction from
a spin-coat deposited film. The ion fragment distribution has fewer peaks and chemical species like
individual carbon ions are unambiguously identifiable. The chemical structures for PCBM and
P3HT are shown as well (inset). Reprinted from Joester et al. [97] with permission from Cam-
bridge University Press
Among the most advanced analysis of a biological material comes from the recent
work of Gordon and Joester on chiton teeth [98]. Specimens were FIB-extracted
from the cutting edge of the magnetite-based teeth which contain 510 nm diameter
chitin protein fibers. The mass spectra were logical for the materials and showed
clearly distinct peaks from the organic and mineral components. Atom maps of two
analyses are shown in Fig. 7.32 (a and d). The organic-matter-derived ions are
shown to originate from fibers included within the mineral. The total amount of
232 7 Applications of the Local Electrode Atom Probe
Fig. 7.32 Atom maps of two analyses using chitin fibers. (a, d) 3D reconstructions and proxigrams.
Two representative samples containing organic fibers that exclusively bind Na+ (ac) or Mg++
(df). For clarity, only approximately 5 % of the Fe/O ions are rendered; the edge of the FOV is
marked (dashed line in panel b). (b, e) Overlay of Na+ (b, red spheres) and Mg++ (e, magenta
spheres) ion positions on carbon concentration maps integrated over the boxed regions indicated in
panels a and d. Some regions of the fibers appear devoid of sodium or magnesium (arrows). (c, f)
Proximity histograms (error bars, 1) of (c Na/C and f) Mg/C across the organicinorganic
interface of fibers indicated by arrows in panels a and d. Interfaces appear graded over 24 nm.
Images courtesy L. Gordon and D. Joester (Northwestern University)
carbon is less than what would be expected from crystalline chitin fibers which may
be a consequence of overlapping peaks in mass spectra, preferential evaporation, or
biological remodeling of fibers during mineralization. Nevertheless, the fiber diam-
eter (510 nm) closely resembles the organic fibers observed in STEM images of
the mineralized tooth [98]. Furthermore, sodium and magnesium clearly co-localize
with organic fibers, Fig. 7.32b, c and e, f, respectively. This work was the first to
show microstructural information from an APT image of a material of biological
origin. It will stand as a portent of great things to come.
7.5.3 Ferritin
The Joester group has also recently analyzed a portion of a single biomolecule:
ferritin. This molecule has been the object of efforts over the past three decades
[99102] to observe its structure in APT precisely because it is a model molecule
that contains about 4,500 iron atoms in an oxide core with a protein shell surround-
ing it. Gordon and Joester [103] were able to prepare isolated molecules by
depositing a sparse layer of ferritin molecules on goldpalladium-coated flat-top
microtip arrays, Fig. 7.33a. After deposition, the specimens were overcoated with
7.6 Composite Structures/Devices 233
Fig. 7.33 (a) Schematic illustration of the preparation of LEAP specimens of ferritin molecules.
A flat-top microtip is coated with goldpalladium (left), a sparse layer of ferritin molecules is
deposited (middle left), the microtip is then encapsulated (middle right), and the FIB is used to
sharpen the resultant microtip (right). (b) Atom map (left) and proxigram (right) for an Fe-rich
core in a ferritin molecule. An isoconcentration surface for 5 iron atoms/nm3 is shown. Note the
elevated iron and oxygen signals inside the isoconcentration surface and the elevated carbon signal
outside the surface. (a) courtesy L. Gordon and D. Joester (Northwestern University). (b) reprinted
from Gordon et al. [103] with permission from Microscopy and Microanalysis
more goldpalladium and the result was sharpened into a LEAP specimen using
FIB techniques. Though the LEAP analysis was not as precise as obtained with
most inorganics, it was consistent with expectations: iron and oxygen in a core with
carbon and nitrogen surrounding it, Fig. 7.33b. A proximity histogram composition
profile about the molecule center shows an expected iron-rich core with carbon
reaching a maximum just outside the core, Fig. 7.33c. It is likely that there is a
significant local magnification effect for the molecule since goldpalladium and the
ferritin molecule likely have different evaporation fields. Nonetheless, work like
this is starting to show that organic materials with observable microstructure may
be analyzed at the atomic scale with APT.
There are many types of complex structures that are candidates for APT analysis.
These structures may contain a variety of material types: metals, semiconductors,
dielectrics, and polymers, in various morphological configurations. Having
234 7 Applications of the Local Electrode Atom Probe
reviewed several examples of each type of material analysis above, we will now
look at combinations of these materials as found in such structures as microelec-
tronic devices and oxidized metals.
The simplest composite structure is a bilayer. Most metals are not inert to environ-
mental degradation which is an expensive constraint on their utilization. Oxides
grown on metals form a bilayer where the details of the interface between layers can
be critical to the performance of the part. APT of these interfaces has been fruitfully
applied to learn novel information in many cases.
Zirconium and its alloys have been investigated by a variety of researchers using
APT [104112], often due to the fact that zirconium alloys are used in nuclear
applications for fuel rod cladding. They are exposed to high temperature water or
steam in a radiation environment. The protective oxide coating that forms on
zirconium alloys is crucial to their utility in this application. Transition metal
elements are known to improve the performance of these alloys in the reactor
environment, but the details of the physical mechanism by which this occurs is
not exactly known. Sundell et al. [110] have studied corrosion of Zircaloy-2 in
steam at 400 C which formed a 1.3 m thick oxide in 3 days. APT images of a
metaloxide interface are shown in Fig. 7.34. Two planar features are visible in this
image from the atom distribution pattern of Fe and Ni. Since low-angle grain
boundaries are known to form during oxidation of this alloy, these features are
presumed to be evidence of segregation of iron and nickel to low angle grain
boundaries near this interface. The presence of iron and nickel at boundaries
associated with this interface may have profound effects on the corrosion behavior
of this alloy in practice because oxygen transport is assumed to occur through oxide
grain boundaries.
Inoue and co-authors have recently done substantial work investigating the distri-
bution of dopants in various parts of MOSFETs [4446, 55]. Knowledge of the
detailed dopant distributions in microelectronic devices is becoming more and more
important as the devices get smaller and are using increasingly complex 3D internal
structures. In the n-MOSFET gate region shown at the left in Fig. 7.35, phosphorus
atoms segregate to grain boundaries and polycrystalline siliconoxide interfaces. In
contrast, no segregation of boron atoms is observed in the p-MOSFET gate (right
side of Fig. 7.35) [46]. Arsenic and boron atoms were detected in the source/drain
regions and channel regions, respectively for p-type FET (and vice versa for the
n-type structure). Statistical evaluation of the data suggested that the fabrication
process increases the boron fluctuation but not the arsenic fluctuation, thus produc-
ing more variations from device to device in p-type MOSFETs [46].
7.6 Composite Structures/Devices 235
Fig. 7.34 LEAP images of the interface between the oxide and metal in a Zircalloy-2 alloy. (a, b)
are the same image rotated by 90 about the long axis. A presumed low-angle grain boundary is
visible in each of the metal and oxide layers. Both nickel and iron are segregated to these
boundaries. Fe is shown as purple spheres, Ni is shown as green spheres, O is shown as blue
dots, and Zr is shown as orange dots. The size of the box is 50 50 140 nm3. Reprinted from
Sundell et al. [110] with permission from Corrosion Science
In current generation devices the entire transistor channel may have on the order
of only a few hundreds of dopant atoms, which leads to potential sources for
variability in performance. To investigate this variability, Takamizawa et al. [55]
have made repeated observations of channel doping in 65 nm node n-MOSFETs by
measuring the dopant concentration ~30 nm below the gate oxide, Fig. 7.36a. Only
transistors having threshold voltage values ranging from 4 to +4 in one million
transistors were selected. This data was compared to the threshold voltage (1 M tran-
sistors) of transistors on with five different boron implant doses as measured with
SIMS, Fig. 7.36b. The APT boron concentrations (vs. measured threshold voltage)
show a correlation to electrical performance when compared to transistors with
known dopant concentration [55].
FinFETs [113] are field effect transistors that maximize the area of the gate oxide
by wrapping the oxide and gate structure around a fin-shaped feature, Fig. 7.37a.
Intel started using these structures in its 22 nm node products in 2012 [114]. This
type of structure is actually part of a larger group of devices called multi-gate
transistors. The international technology roadmap for semiconductors (ITRS) refers
to these types of structures as advanced nonclassical CMOS devices [113].
236 7 Applications of the Local Electrode Atom Probe
Fig. 7.35 Atom maps of MOSFET gate regions. Left: In the n-MOSFET gate region, phosphorus
atoms segregate to grain boundaries and polycrystalline siliconoxide interfaces. Right: In con-
trast, no segregation of boron atoms is observed in the p-MOSFET gate. Arsenic and boron atoms
were detected in the source/drain regions and channel regions, respectively, for the p-type FET
(and vice versa for the n-type structure). Statistical evaluation of the data suggested that the
fabrication process increases the boron fluctuation but not the arsenic fluctuation, thus producing
more variations from device to device in p-type MOSFETs [46]. Reprinted from Kelly and Larson
[1] with permission from Annual Review of Materials Research
While APT has been applied to transistor and FinFET-type structures as shown
above, these structures often have been stopped at some point in the fabrication
process in order to accommodate the analysis. There are many situations in semi-
conductor processing where APT analysis of a finished product is desirable;
competitive analysis and failure analysis are two good examples. Only recently
have APT results been obtained from fully processed, off-the-shelf transistor
structures that are part of a finished product [49, 51, 53]. This section presents
data showing the feasibility of APT analysis for fully packaged integrated-circuit
microelectronic devices.
238 7 Applications of the Local Electrode Atom Probe
Fig. 7.37 (a) A schematic and (b) an APT data reconstruction of an entire high- metal gate
FinFET structure (100 nm in height and 20 nm in width) starting with nickel silicide (near top) and
ending near the bottom of the fin (bottom). Reprinted from Kelly and Larson [1] with permission
from Annual Review of Materials Research
Fig. 7.38 Composition profiles across and along fins in FinFETs ion implanted at two different
impingement angles as shown in the figure. Note the relative homogeneity of the implanted
concentration at 45 compared with the large variations that resulted from implantation at 10 .
Reprinted from Kambham et al. [48] with permission from Ultramicroscopy
7.6 Composite Structures/Devices 239
Fig. 7.39 Specimens created from a commercially available 32 nm node CMOS logic device
that was depackaged and then prepared by FIB methods. (a) APT analysis shows a Hf-based
high- gate oxide near the top of the dataset. (b) Regions of SiGe are visible on either side of a
predominantly silicon region. (b, c) Spatial distributions of the Ge, As, B, and C atoms.
Whereas the arsenic is distributed relatively uniformly (with a concentration of 0.016 at.%),
the boron is correlated with the regions of high-germanium content [49], and the carbon atoms
appear to be clustered [53]. (d) A high-angle annular dark-field STEM image [51] (courtesy
W. Lefebvre, Universite de Rouen) of a structure similar to the one analyzed in panels ac. An
undercut of the gate oxide into the channel is apparent, matching the germanium distribution
shown in panel b. Reprinted from Kelly and Larson [1] with permission from Annual Review of
Materials Research
240 7 Applications of the Local Electrode Atom Probe
Fig. 7.40 Growth of the number of publications per year referencing 3DAP or atom probe
overlaid with the relative number of LEAP instruments in the field. The implication is that the
proliferation of LEAP instruments and their inherently high productivity rate are behind the upturn
in number of publications per year using APT
7.7 Conclusions
APT has entered a new stage in its maturation. When some of the authors of this
book were introduced to the technique, there were literally only a handful of people
in the world who could fluently describe the operation of an atom probe. We learned
through years of experience to recognize when an audience needed a primer
in the technique. As years have passed and both LEAP and FIB technology
have coevolved, we have found that fewer and fewer minutes of a presentation
were needed to introduce the technique. The number of users worldwide has now
expanded from dozens to many hundreds and will soon pass thousands. The number
of publications per year referencing APT may be taken as an indication of this
fact as illustrated in Fig. 7.40. Yet we may all consider that APT is still in its
formative years. As a reader of this book, you are part of this growth and the authors
want to thank you and your contributions to this rapidly evolving and very suc-
cessful community. The future will be truly exciting and it will be observed at the
atomic scale.
References
1. Kelly, T.F., Larson, D.J.: Atom probe tomography 2012. Ann. Rev. Mater. Res. 42, 131
(2012)
2. Srinivasan, R., Banerjee, R., Hwang, J.Y., Viswanathan, G.B., Tiley, J., Dimiduk, D.M.,
Fraser, H.L.: Atomic scale structure and chemical composition across orderdisorder inter-
faces. Phys. Rev. Lett. 102, 086101 (2009). doi: http://dx.doi.org/10.1103/PhysRevLett.102.
086101
3. Mishin, Y.: Atomistic modeling of the and 0 -phases of the NiAl system. Acta Mater. 52
(6), 14511467 (2004). doi:10.1016/j.actamat.2003.11.026
4. Ardell, A.J., Ozolins, V.: Trans-interface diffusion-controlled coarsening. Nat. Mater. 4(4),
309316 (2005)
5. Gault, B., Cui, X.Y., Moody, M.P., De Geuser, F., Sigli, C., Ringer, S.P., Deschamps, A.:
Atom probe microscopy investigation of Mg site occupancy within delta precipitates in an
AlMgLi alloy. Scripta Mat. 66, 903906 (2012)
6. Miller, M.K., Longstreth-Spoor, L., Kelton, K.F.: Detecting density variations and
nanovoids. Ultramicroscopy 111(6), 469472 (2011). doi:10.1016/j.ultramic.2011.01.027
7. Schreiber, D.K., Olszta, M.J., Saxey, D.W., Kruska, K., More, K.L., Lozano-Perez, S.,
Bruemmer, S.M.: Examinations of oxidation and sulfidation of grain boundaries in alloy
600 exposed to simulated pressurized water reactor primary water. Microsc. Microanal. 19,
676687 (2013)
8. Xiang, Y., Chitry, V., Liddicoat, P., Felfer, P., Cairney, J., Ringer, S., Kruse, N.: Long-chain
terminal alcohols through catalytic CO hydrogenation. J. Am. Chem. Soc. 135(19),
71147117 (2013). doi:10.1021/ja402512r
9. Bagot, P.A.J., de Bocarme, T.V., Cerezo, A., Smith, G.D.W.: 3D atom probe study of gas
adsorption and reaction on alloy catalyst surfaces I: instrumentation. Surf. Sci. 600,
30283035 (2006)
10. Li, T., Bagot, P.A.J., Marquis, E.A., Tsang, S.C.E., Smith, G.D.W.: Characterization of
oxidation and reduction of a palladiumrhodium alloy by atom-probe tomography. J. Phys.
Chem. C 116(7), 47604766 (2012). doi:10.1021/jp211687m
242 7 Applications of the Local Electrode Atom Probe
11. Li, T., Bagot, P.A.J., Marquis, E.A., Tsang, S.C.E., Smith, G.D.W.: Characterization of
oxidation and reduction of PtRu and PtRhRu alloys by atom probe tomography and
comparison with PtRh. J. Phys. Chem. C 116(33), 1763317640 (2012). doi:10.1021/
jp304359m
12. Ohkubo T, Chen YM, Kodzuka M, Li F, Oh-ishi K, Hono K (2009) Laser-assisted atom probe
analysis of bulk insulating ceramics. In: MRS Fall Meeting, Materials Research Society,
Boston, MA
13. Chen, Y.M., Ohkubo, T., Hono, K.: Laser assisted field evaporation of oxides in atom probe
analysis. Ultramicroscopy 111(5), 562 (2011)
14. Kuchibhatla, S.V.N.T., Shutthanandan, V., Arey, B.W., Kovarik, L., Wang, C.M., Prosa, T.J.,
Ulfig, R., Thevuthasan, S., Gorman, B.P.: Embedded nanoparticle analysis using atom probe
tomography and high-resolution electron microscopy. Microsc. Microanal. 17(S2), 760761
(2011)
15. Larson, D.J., Alvis, R.A., Lawrence, D.F., Prosa, T.J., Ulfig, R.M., Reinhard, D.A., Clifton, P.
H., Gerstl, S.S.A., Bunton, J.H., Lenz, D.R., Kelly, T.F., Stiller, K.: Analysis of bulk
dielectrics with atom probe tomography. Microsc. Microanal. 14(S2), 12541255 (2008)
16. Marquis, E.A., Yahya, N.A., Larson, D.J., Miller, M.K., Todd, R.I.: Probing the improble:
imaging carbon atoms in alumina. Mater. Today 13(10), 4244 (2010)
17. Stiller, K., Viskari, L., Sundell, G., Liu, F., Thuvander, M., Andren, H.O., Larson, D.J., Prosa,
T.J., Reinhard, D.: Atom probe tomography of oxide scales. Oxid. Met. 79, 227238 (2013)
18. Talbot, E., Larde, R., Gourbilleau, F., Dufour, C., Pareige, P.: Si nanoparticles in SiO2: an
atomic scale observation for optimization of optical devices. Eur. Phys. Lett. 87, 26004-
2600126004-26005 (2009)
19. Prosa, T.J., Lawrence, D., Olson, D., Lenz, D., Bunton, J.H., Larson, D.J.: Atom probe
tomography analysis of thick film SiO2 and oxide interfaces: conditions leading to improved
analysis yield. Microsc. Microanal. 17(S2), 750751 (2011)
20. Roussel, M., Chen, W.H., Talbot, E., Larde, R., Cadel, E., Gourbilleau, F., Grandidier, B.,
Stievenard, D., Pareige, P.: Atomic scale investigation of silicon nanowires and nanoclusters.
Nanoscale Res. Lett. 6, 271 (2011). 27110.1186/1556-276x-6-271
21. Bachhav, M., Danoix, R., Danoix, F., Hannoyer, B., Ogale, S., Vurpillot, F.: Investigation of
wustite (Fe1-xO) by femtosecond laser assisted atom probe tomography. Ultramicroscopy
111, 584588 (2011)
22. Chen, Y., Reed, R.C., Marquis, E.A.: As-coated thermal barrier coating: structure and
chemistry. Scr. Mater. 67(9), 779782 (2012)
23. Larde, R., Talbot, E., Vurpillot, F., Pareige, P., Schmerber, G., Beaurepaire, E., Dinia, A., Pierron-
Bohnes, V.: Investigation at the atomic scale of the Co spatial distribution in Zn(Co)O magnetic
semiconductor oxide. J. Appl. Phys. 105(12), 12610713 (2009) doi: 10.1063/1.3152579
24. Larde, R., Talbot, E., Pareige, P., Bieber, H., Schmerber, G., Colis, S., Pierron-Bohnes, V.,
Dinia, A.: Evidence of superparamagnetic Co clusters in pulsed laser deposition-grown Zn
(0.9)Co(0.1)O thin films using atom probe tomography. J. Am. Chem. Soc. 133(5),
14511458 (2011). doi:10.1021/ja108290u
25. Chen, Y.M., Ohkubo, T., Kodzuka, M., Morita, K., Hono, K.: Laser-assisted atom probe
analysis of zirconia/spinel nanocomposite ceramics. Scr. Mater. 61, 693696 (2009)
26. Payne, D.J., Marquis, E.A.: Three-dimensional spatial distribution of Cr atoms in doped
indium oxide. Chem. Mater. 23(5), 10851087 (2011)
27. Li, F., Ohkubo, T., Chen, Y.M., Kodzuka, M., Ye, F., Ou, D.R., Mori, T., Hono, K.: Laser-
assisted three-dimensional atom probe analysis of dopant distribution in Gd-doped CeO2. Scr.
Mater. 63, 332335 (2010)
28. Li, F., Ohkubo, T., Chen, Y.M., Kodzuka, M., Hono, K.: Quantitative atom probe analyses of
rare-earth-doped ceria by femtosecond pulsed laser. Ultramicroscopy 111(6), 589594
(2011)
References 243
29. Kirchhofer, R., Teague, M.C., Gorman, B.P.: Thermal effects on mass and spatial resolution
during laser pulse atom probe tomography of cerium oxide. J. Nucl. Mater. 436, 2328 (2013)
30. Oberdorfer, C., Schmitz, G.: On the field evaporation behavior of dielectric materials in three-
dimensional atom probe: a numeric simulation. Microsc. Microanal. 17, 1525 (2011)
31. Tsukada, M., Tamura, H., McKenna, K.P., Shluger, A.L., Chen, Y.M., Ohkubo, T., Hono, K.:
Mechanism of laser assisted field evaporation from insulating oxides. Ultramicroscopy 111
(6), 567570 (2011)
32. Mazumder, B., Vella, A., Deconihout, B., Al-Kassab, T.: Evaporation mechanisms of MgO in
laser assisted atom probe tomography. Ultramicroscopy 11(6), 571575 (2011)
33. Karahka, M., Kreuzer, H.J.: Field evaporation of oxides: a theoretical study. Ultramicroscopy
132, 5459 (2013) doi: http://dx.doi.org/10.1016/j.ultramic.2012.10.007
34. Tamura, H., Tsukada, M., McKenna, K.P., Shluger, A.L., Ohkubo, T., Hono, K.: Laser-
assisted field evaporation from insulators triggered by photoinduced hole accumulation.
Phys. Rev. B 86, 195430, 195431195436 (2012)
35. Snoeyenbos, D., Jercinovic, M.J., Reinhard, D.A., Hombourger, C.: Characterization of
minerals of geochronological interest by EPMA and atom probe tomography.
Am. Geophys. Union Abstr. V23C-2830 (2012)
36. Valley, J.W., Cavosie, A.J., Ushikubo, T., Reinhard, D.A., Snoeyenbos, D., Lawrence, D.,
Larson, D.J., Clifton, P.H., Kelly, T.F., Strickland, A., Wilde, S., Moser, D.: Radiogenic
dating at the atomic scale in 4.4Ga Zircons. Nature Geoscience in press (2013)
37. Heck, P.R., Stadermann, F.J., Isheim, D., Auciello, O.H., Daulton, T.L., Davis, A.M., Elam,
J.W., Floss, C., Hiller, J., Larson, D.J., Lewis, J., Mane, A., Pellin, M.J., Savina, M.R.,
Seidman, D.N., Stephan, T.: Atom-probe analysis of nanodiamonds from allende. Meteoritics
Planet. Sci. in press (2013)
38. Lewis, R.S., Anders, E., Draine, B.T.: Properties, detectability and origin of interstellar
diamonds in meteorites. Nature 339, 117121 (1989)
39. Coplen, T.B., Bohlke, J.K., De Bievre, P., Ding, T., Holden, N.E., Hopple, J.A., Krouse, H.R.,
Lamberty, A., Peiser, H.S., Revesz, K., Rieder, S.E., Rosman, K.J.R., Roth, E., Taylor, P.D.
P., Vocke, R.D., Xiao, Y.K.: Isotope-abundance variations of selected elements(IUPAC
Technical Report). Pure Appl. Chem. 74, 19872017 (2002)
40. Kirchhofer, R., Diercks, D.R., Gorman, B.P., Brennecka, G.: Atomic scale composition
profiling of ferroelectrics via laser-pulsed atom probe tomography and cross-correlative
transmission electron microscopy. Microsc. Microanal. 19(s2), 908981 (2013)
41. Kelly, T., Larson, D.J.: Local electrode atom probes. Mat. Char. 44(12), 5985 (2000)
42. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.
C., Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data
from a commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383
(2004)
43. Moore, J.S., Jones, K.S., Kennel, H., Corcoran, S.: 3-D Analysis of semiconductor dopant
distributions in a patterned structure using LEAP. Ultramicroscopy 108, 536539 (2008)
44. Inoue, K., Yano, F., Nishida, A., Tsunomura, T., Toyama, T., Nagai, Y., Hasegawa, M.:
Monolayer segregation of As atoms at the interface between gate-oxide and Si substrate in a
p-MOSFET by 3D atom-probe technique. Appl. Phys. Lett. 92, 103506/103501103503
(2008)
45. Inoue, K., Yano, F., Nishida, A., Takamizawa, H., Tsunomura, T., Nagai, Y., Hasegawa, M.:
Dopant distribution in gate electrode of n- and p-type metal-oxidesemiconductor field effect
transistor by laser-assisted atom probe. Appl. Phys. Lett. 95, 043502 (2009)
46. Inoue, K., Yano, F., Nishida, A., Takamizawa, H., Tsunomura, T., Nagai, Y., Hasegawa, M.:
Dopant distributions in n-MOSFET structure observed by three dimensional atom probe
microscopy. Ultramicroscopy 109(12), 14791484 (2009)
47. Lauhon, L.J., Adusumilli, P., Ronsheim, P., Flaitz, P.L., Lawrence, D.: Atom-probe tomog-
raphy of semiconductor materials and device structures. MRS Bull. 34(10), 738743 (2009)
244 7 Applications of the Local Electrode Atom Probe
48. Kambham, A.K., Mody, J., Gilbert, M., Koelling, S., Vandervorst, W.: Atom-probe for
FinFET dopant characterization. Ultramicroscopy 111(6), 535539 (2011)
49. Larson, D.J., Lawrence, D., Olson, D., Prosa, T.J., Reinhard, D.A., Ulfig, R.M., Clifton, P.H.,
Bunton, J.H., Lenz, D., Olson, J.D., Renaud, L., Martin, I., Kelly, T.F.: Prospects for atom
probe tomography of commercial semiconductor devices. Microsc. Microanal. 17(S2),
752753 (2011)
50. Larson, D.J., Prosa, T.J., Lawrence, D., Geiser, B.P., Jones, C.M., Kelly, T.F.: Atom probe
tomography for microelectronics. In: Haight, R., Ross, F., Hannon, J. (eds.) Handbook of
Instrumentation and Techniques for Semiconductor Nanostructure Characterization, vol.
2, pp. 407477. World Scientific Publishing, London (2011)
51. Larson, D.J., Lawrence, D., Lefebvre, W., Olson, D., Prosa, T.J., Reinhard, D.A., Ulfig, R.M.,
Clifton, P.H., Bunton, J.H., Lenz, D., Renaud, L., Martin, I., Kelly, T.F.: Toward atom probe
tomography of microelectronic devices. J. Phys. 326, 012030 (2011)
52. Panciera, F., Hoummada, K., Gregoire, M., Juhel, M., Bicais, N., Mangelinck, D.: Three
dimensional distributions of arsenic and platinum within NiSi contact and gate of an n-type
transistor. Appl. Phys. Lett. 99(5), 0519111 (2011). doi: 05191110.1063/1.3616150
53. Larson, D.J., Lawrence, D., Olson, D., Prosa, T.J., Ulfig, R.M., Reinhard, D.A., Clifton, P.C.,
Kelly, T.F., Lefebvre, W.: From the store shelf to device-level atom probe analysis: an
exercise in feasibility. In: 36th International Symposium for Testing and Failuer Analysis,
San Jose, CA, pp. 189197. ASM International (2011)
54. Gilbert, M., Vandervorst, W., Koelling, S., Kambham, A.K.: Atom probe analysis of a 3D
finFET with high-k metal gate. Ultramicroscopy 111, 530534 (2011)
55. Takamizawa, H., Shimizu, Y., Inoue, K., Toyama, T., Okada, N., Kato, M., Uchida, H., Yano,
F., Nishida, A., Mogami, T., Nagai, Y.: Origin of characteristic variability in MOSFET
revealed by three-dimensional atom imaging. Appl. Phys. Lett. 100, 253504/
253501253503 (2012)
56. Shimizu, Y., Kawamura, Y., Uematsu, M., Itoh, K.M., Tomita, M., Sasaki, M., Uchida, H.,
Takahashi, M.: Atom probe microscopy of three-dimensional distribution of silicon isotopes
in 28Si/30Si isotope superlattices with sub-nanometer spatial resolution. J. Appl. Phys. 106,
076102 (2009)
57. Shimizu, Y., Kawamura, Y., Uematsu, M., Tomita, M., Kinno, T., Okada, N., Kato, M.,
Uchida, H., Takahashi, M., Ito, H., Ishikawa, H., Ohji, Y., Takamizawa, H., Nagai, Y., Itoh,
K.M.: Depth and lateral resolution of laser-assisted atom probe microscopy of silicon
revealed by isotopic heterostructures. J. Appl. Phys. 109, 036102/036101036103 (2011)
58. Shimizu, Y., Takamizawa, H., Kawamura, Y., Uematsu, M., Toyama, T., Inoue, K., Haller, E.
E., Itoh, K.M., Nagai, Y.: Atomic-scale characterization of germanium isotopic multilayers
by atom probe tomography. J. Appl. Phys. 113, 3 (2013). doi:10.1063/1.4773675
59. Moutanabbir, O., Isheim, D., Seidman, D.N., Kawamura, Y., Itoh, K.M.: Ultraviolet-laser
atom-probe tomographic three-dimensional atom-by-atom mapping of isotopically modu-
lated Si nanoscopic layers. Appl. Phys. Lett. 98, 013111/013111010113/013113 (2011)
60. Perea, D.E., Lensch, J.L., May, S.J., Wessels, B.W., Lauhon, L.J.: Composition analysis of
single semiconductor nanowires using pulsed-laser atom probe tomography. Appl. Phys. A
85(3), 271275 (2006)
61. Canham, L.: Gaining light from silicon. Nature 408, 411412 (2000)
62. Gnaser, H.: Atom probe tomography of Si nanocrystals embedded in silicon oxide. Paper
presented at the SiSS-15, Seikei University, Tokyo
63. Roussel, M., Talbot, E., Gourbilleau, F., Pareige, P.: Atomic characterization of Si
nanoclusters embedded in SiO2 by atom probe tomography. Nanoscale Res. Lett. 6,
164 (2011)
64. Galtrey, M.J., Oliver, R.A., Kappers, M.J., Humphreys, C.J., Stokes, D.J., Clifton, P.H.,
Cerezo, A.: Three dimensional atom probe studies of an InxGa1xN/GaN multiple quantum
well structure: assessment of possible indium clustering. Appl. Phys. Lett. 90, 0619031
(2007)
References 245
65. Gorman, B.P., Norman, A.G., Yan, Y.: Atom probe analysis of IIIV and Si-based semicon-
ductor photovoltaic structures. Microsc. Microanal. 13, 493502 (2007)
66. Galtrey, M.J., Oliver, R.A., Kappers, M.J., McAleese, C., Zhu, D., Humphreys, C.J., Clifton,
P.H., Larson, D.J., Cerezo, A.: Compositional inhomogeneity of a high-efficiency
InxGa1xN based multiple quantum well ultraviolet emitter studied by three dimensional
atom probe. Appl. Phys. Lett. 92, 041904 (2008)
67. Galtrey, M.J., Oliver, R.A., Kappers, M.J., Humphreys, C.J., Clifton, P.H., Larson, D.J.,
Saxey, D.W., Cerezo, A.: Three-dimensional atom probe analysis of green- and blue-emitting
InxGa1xN/GaN multiple quantum well structures. J. Appl. Phys. 104, 013524 (2008)
68. Galtrey, M.J., Oliver, R.A., Kappers, M.J., McAleese, C., Zhu, D., Humphreys, C.J., Clifton,
P.H., Larson, D.J., Cerezo, A.: Atom probe reveals the structure of InxGa1xN based
quantum wells in three dimensions. Phys. Stat. Sol. B 245, 861867 (2008)
69. Muller, M., Cerezo, A., Smith, G.D.W., Chang, L., Gerstl, S.S.A.: Atomic scale character-
ization of buried InxGa1xAs quantum dots using pulsed laser atom probe tomography.
Appl. Phys. Lett. 92, 233115 (2008)
70. Kodzuka, M., Ohkubo, T., Hono, K., Matsukura, F., Ohno, H.: 3DAP analysis of (Ga, Mn)As
diluted magnetic semiconductor thin film. Ultramicroscopy 109(5), 644648 (2009)
71. Prosa, T.J., Clifton, P.H., Zhong, H., Tyagi, A., Shivaraman, R., DenBaars, S.P., Nakamura,
S., Speck, J.S.: Atom probe analysis of interfacial abruptness and clustering within a single In
(x)Ga(1x)N quantum well device on semipolar (1011) GaN substrate. Appl. Phys. Lett.
98(19), 191903191905 (2011). doi:19190310.1063/1.3589370
72. Tomiya, S., Kanitani, Y., Tanaka, S., Ohkubo, T., Hono, K.: Atomic scale characterization of
GaInN/GaN multiple quantum wells in V-shaped pits. Appl. Phys. Lett. 98(18), 181904
(2011). doi:10.1063/1.3585118
73. Bennett, S.E., Saxey, D.W., Kappers, M.J., Barnard, J.S., Humphreys, C.J., Smith, G.D.W.,
Oliver, R.A.: Atom probe tomography assessment of the impact of electron beam exposure on
In(x)Ga(1x)N/GaN quantum wells. Appl. Phys. Lett. 99(2), 0219061 (2011). doi:
02190610.1063/1.3610468
74. Tomiya, S., Kanitani, Y., Tanaka, S., Ohkubo, T., Hono, K.: Atomic scale characterization of
GaInN/GaN multiple quantum wells in V-shaped pits. Appl. Phys. Lett. 98(18),
1819041 (2011) doi: 18190410.1063/1.3585118
75. Agrawal, R., Bernal, R.A., Isheim, D., Espinosa, H.D.: Characterizing atomic composition
and dopant distribution in wide band gap semiconductor nanowires using laser-assisted atom
probe tomography. J. Phys. Chem. C 115(36), 1768817694 (2011). doi:10.1021/jp2047823
76. Bennett, S.E., Clifton, P.H., Ulfig, R.M., Kappers, M.J., Barnard, J.S., Humphreys, C.J.,
Oliver, R.A.: Mg dopant distribution in an AlGaN/GaN p-type superlattice assessed using
atom probe tomography, TEM and SIMS. In: Walther, T., Nellist, P.D., Hutchison, J.L.,
Cullis, A.G. (eds.) 16th International Conference on Microscopy of Semiconducting Mate-
rials, vol. 209. J. Phys. Conf. Ser. (2010)
77. Holzworth, M.R., Rudawski, N.G., Pearton, S.J., Jones, K.S., Lu, L., Kang, T.S., Ren, F.,
Johnson, J.W.: Characterization of the gate oxide of an AlGaN/GaN high electron mobility
transistor. Appl. Phys. Lett. 98(12), 122103 (2011). doi:12210310.1063/1.3569715
78. Bennett, S.E., Smeeton, T.M., Saxey, D.W., Smith, G.D.W., Hooper, S.E., Heffernan, J.,
Humphreys, C.J., Oliver, R.A.: Atom probe tomography characterisation of a laser diode
structure grown by molecular beam epitaxy. J. Appl. Phys. 111, 053508 (2012)
79. Giddings, A.D., Keizer, J.G., Hara, M., Hamhuis, G.J., Yuasa, H., Fukuzawa, H., Koenraad,
P.M.: Composition profiling of InAs quantum dots and wetting layers by atom probe
tomography and cross-sectional scanning tunneling microscopy. Phys. Rev. B 83, 205308
(2011)
80. Ali, S., Cooley, L.D., McCammon, D., Nelms, K.L., Peck, J., Prober, D., Swetz, D., Timbie,
P.T., Weide, D.V.D.: Planar antenna-coupled transition-edge hot electron microbolometer.
IEEE Trans. Appl. Supercond. 13(2), 184187 (2003)
246 7 Applications of the Local Electrode Atom Probe
81. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion
Microscopy. Oxford University Press, Oxford (1996)
82. Thompson, K., Lawrence, D.J., Larson, D.J., Olson, J.D., Kelly, T.F., Gorman, B.: In situ site-
specific specimen preparation for atom probe tomography. Ultramicroscopy 107, 131139
(2007)
83. Miller, M.K., Russell, K.F., Thompson, K., Alvis, R., Larson, D.J.: Review of atom probe
FIB-based specimen preparation methods. Microsc. Microanal. 13, 428436 (2007)
84. Larson, D.J., Prosa, T., Olson, D., Lefebvre, W., Lawrence, D., Clifton, P.H., Kelly, T.F.:
Atom probe tomography of a commercial light emitting diode. In: 18th International Con-
ference on Microscopy of Semiconducting Materials. J. Phys.: Inst. Phys. Conf. Ser. in press
(2013)
85. Choi, P.-P., Cojocaru-Miredin, O., Abou-Ras, D., Caballero, R., Raabe, D., Smentkowski, V.
S., Park, C.G., Gu, G.H., Mazumder, B., Wong, M.H., Hu, Y.-L., Melo, T.P., Speck, J.S.:
Atom probe tomography of compound semiconductors for photovoltaic and light-emitting
device applications. Microsc. Today 20(3), 1824 (2012). doi:10.1017/S1551929512000235
86. Larson, D.J., Smentkowski, V.S., Reinhard, D.A., Prosa, T.J., Olson, D., Lawrence, D.,
Clifton, P.H., Ulfig, R.M., Kelly, T.F.: Atom probe tomography analysis of grain boundaries
in CdTe. Microsc. Microanal. 18(S2), 928929 (2012)
87. Paudel, N.R., Kwon, D., Young, M., Wieland, K.A., Asher, S., Compaan, A.D.: In: 35th IEEE
PVSC, pp. 10091013. (2010)
88. Cojocaru-Miredin, O., Choi, P., Wuerz, R., Raabe, D.: Atomic-scale characterization of the
CdS/CuInSe2 interface in thin-film solar cells. Appl. Phys. Lett. 98, 103504/103501103503
(2011)
89. Cojocaru-Miredin, O., Choi, P., Wuerz, R., Raabe, D.: Atomic-scale distribution of impuri-
ties in CuInSe2-based thin-film solar cells. Ultramicroscopy 111(6), 552556 (2011). doi:10.
1016/j.ultramic.2010.12.034
90. Ditto, J.J.: Devices and methods for cryo lift-out with in situ probe. U.S. Patent Application
Patent 13/570,127, Accessed 14 Feb 2013
91. Kelly, T.F., Nishikawa, O., Panitz, J.A., Prosa, T.J.: Prospects for nanobiology with atom-
probe tomography. MRS Bull. 34, 744749 (2009)
92. Gault, B., Yang, W., Ratinac, K.R., Zheng, R., Braet, F., Ringer, S.P.: Investigation of self-
assembled monolayer by atom probe microscopy. Microsc. Microanal. 15(S2), 272273
(2009)
93. Prosa, T.J., Kostrna, S.L.P., Kelly, T.F.: Laser atom probe tomography: application to poly-
mers. 50th International Field Emission Symposium, pp. 533534. (2006)
94. Zhang, Y., Hillier, A.C.: Three-dimensional atom probe tomography of oxide, anion, and
alkanethiolate coatings on gold. Anal. Chem. 82(14), 61396147 (2010). doi:10.1021/
ac1009035
95. Prosa, T.J., Kostrna Keeney, S., Kelly, T.F.: Atom probe tomography analysis of poly
(3-alkylthiophene)s. J. Microsc. 237, 155 (2010)
96. Li, H.C., Rao, K.K., Jeng, J.Y., Hsiao, Y.J., Guo, T.F., Jeng, Y.R., Wen, T.C.: Nano-scale
mechanical properties of polymer/fullerene bulk hetero-junction films and their influence on
photovoltaic cells. Sol. Energy Mater. Sol. Cells 95(11), 29762980 (2011). doi:10.1016/j.
solmat.2011.05.039
97. Joester, D., Hillier, A.C., Zhang, Y., Prosa, T.J.: Organic materials and organic/inorganic
heterostructures in atom probe tomography. Microsc. Today 20, 2631 (2012)
98. Gordon, L.M., Joester, D.: Nanoscale chemical tomography of buried organicinorganic
interfaces in the chiton tooth. Nature 469, 194198 (2011)
99. Panitz, J.A., Giaver, I.: Ferritin deposition on field-emitter tips. Ultramicroscopy 6, 36
(1981)
100. Panitz, J.A.: Point-projection imaging of unstained ferritin clusters. Ultramicroscopy 7,
241248 (1982)
References 247
101. Panitz, J.A.: Biomolecular deposition on multiple field-emitter tips. Rev. Sci. Instrum. 56(4),
572574 (1985)
102. Greene, M.E., Kelly, T.F., Larson, D.J., Prosa, T.J.: Focused ion beam fabrication of
solidified ferritin into nanoscale volumes for compositional analysis using time-of-flight
mass spectrometry methods. J. Microsc. 247(3), 288299 (2012). doi:10.1111/j.1365-2818.
2012.03644.x
103. Gordon, L.M., Joester, D.: Towards atom probe tomography of hybrid organicinorganic
nanoparticles. Microsc. Microanal. 19(S2) 952952 (2013)
104. Andren, H.O., Mattsson, L., Rolander, U.: Atom-probe analysis of zircaloy. J. Phys. 47(C-2),
191196 (1986). doi:10.1051/jphyscol:1986228
105. Sano, N., Takeda, K.: Atom probe analysis of Sn in Zr-based alloys. J. Nucl. Mater. 252(12),
6370 (1998). doi:10.1016/s0022-3115(97)00302-4
106. Hudson, D., Cerezo, A., Smith, G.D.W.: Zirconium oxidation on the atomic scale.
Ultramicroscopy 109(5), 667671 (2009). doi:10.1016/j.ultramic.2008.10.020
107. Hudson, D., Smith, G.D.W.: Initial observation of grain boundary solute segregation in a
zirconium alloy (ZIRLO) by three-dimensional atom probe. Scr. Mater. 61(4), 411414
(2009). doi:10.1016/j.scriptamat.2009.04.032
108. Hudson, D., Smith, G.D.W., Gault, B.: Optimisation of mass ranging for atom probe
microanalysis and application to the corrosion processes in Zr alloys. Ultramicroscopy 111
(6), 480486 (2011). doi:10.1016/j.ultramic.2010.11.007
109. Thuvander, M., Andren, H.O.: Methods of quantitative matrix analysis of Zircaloy-2.
Ultramicroscopy 111(6), 711714 (2011). doi:10.1016/j.ultramic.2010.12.008
110. Sundell, G., Thuvander, M., Andren, H.O.: Enrichment of Fe and Ni at metal and oxide grain
boundaries in corroded Zircaloy-2. Corrosion Sci. 65(0), 1012 (2012). doi: http://dx.doi.org/
10.1016/j.corsci.2012.08.061
111. Sundell, G., Thuvander, M., Andren, H.O.: Enrichment of Fe and Ni at metal and oxide grain
boundaries in corroded Zircaloy-2. Corros. Sci. 65, 1012 (2012). doi:10.1016/j.corsci.2012.
08.061
112. Cockeram, B.V., Leonard, K.J., Snead, L.L., Miller, M.K.: The use of a laser-assisted Local
Electrode Atom Probe and TEM to examine the microstructure of Zircaloy and precipitate
structure following low dose neutron irradiation at nominally 358 C. J. Nucl. Mater. 433
(13), 460478 (2013). doi:10.1016/j.jnucmat.2012.10.006
113. Colinge, J.-P. (ed.): FinFET and Other Multi-Gate Transistors. Springer, New York (2008)
114. Markoff, J.: To Enhance Chip Speed, Intel Enters 3rd Dimension. New York Times, New
York (2011)
115. Marquis, E.A., Geiser, B.P., Prosa, T.J., Larson, D.J.: Evolution of tip shape during field
evaporation of complex multilayer structures. J. Microsc. 241(3), 225233 (2011)
116. Larson, D.J., Geiser, B.P., Prosa, T.J., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F.: Improve-
ments in planar feature reconstructions in atom probe tomography. J. Microsc. 243, 15 (2011)
117. Gault, B., Haley, D., de Gueser, F., Moody, M.P., Marquis, E.A., Larson, D.J., Geiser, B.P.:
Advances in the reconstruction of atom probe tomography data. Ultramicroscopy 111,
448457 (2011)
118. Larson, D.J., Geiser, B.P., Prosa, T.J., Kelly, T.F.: On the use of simulated field-evaporated
specimen apex shapes in atom probe tomography data reconstruction. Microsc. Microanal. 18
(5), 953963 (2012)
119. Vandervorst, W., Jurczak, M., Everaert, J., Pawlak, B.J., Duffy, R., Del-Agua-Bomiquel, J.,
Poon, T.: J. Vac. Sci. Technol. B 26, 396 (2008)
Appendix A: Data File Formats
A number of file formats have been utilized for APT over the years (see [1] for
additional examples). As the field continues to grow, new file and data standards
may be required [2]. This appendix describes the file formats pertinent to LEAP.
The RRAW, STR, RHIT, and ROOT files have formats based on the ROOT
software libraries provided by CERN [3]. These files contain binary objects that
represent either data structures native to the ROOT system (such as histograms and
graphs) or proprietary objects developed by CAMECA. The POS, ePOS, ATO,
and range files are more readily interpretable by users, and their formats are
described in detail below.
RRAW files contain the raw detector data and the order of ion collection from the
LEAP as well as the history of various parameters measured or controlled through-
out the acquisition. Raw detector data consists of measurements of the time interval
between the evaporation pulse (voltage or laser) and the detection of event pulses at
either ends of multiple delay lines. The RRAW data is the primary input to the ion
hit finding process and includes an entry for every evaporation pulse and any
associated entries for event pulses. STR files replaced RRAW files in LEAP 4000
Systems starting in 2013 and contain the same information but in a modified format.
General users do not interact with RRAW and STR files, but they can provide useful
diagnostics to service personnel. For additional information on detection and hit
finding see Chap. 3.
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 249
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
250 Appendix A: Data File Formats
RHIT files contain the nonzero entries of the RRAW or the STR file that meet
certain quality criteria. These criteria collectively make up the algorithm known as
hit finding. The RHIT file also contains other pertinent information for each event
such as specimen voltage, pulse number, and temperature. An RHIT file contains a
single interpretation of the hit finding of RRAW or STR data. If the hit finding
criteria change, then alternative RHIT files could be generated.
A ROOT file consists of a set of parameters used to convert the RHIT data into a
specimen-space spatial reconstruction. Parameters included in the ROOT file
include evaporation field, k factor, image compression factor, ion detection effi-
ciency, atomic density, specimen shank angle, apex continuity relationships, as well
as calibration factors used for conversion from time-of-flight spectra to mass-to-
charge-ratio (m/n) spectra (see Chap. 5 for a detailed discussion of these parame-
ters). In addition, histograms and graphs created as part of the calibration and
reconstruction processing are stored for later reference. Unlike the RHIT and
RRAW files which are generally created once per acquisition, multiple ROOT
files may be created as per acquisition based on the specific parameters selected
by the user: mass spectrum ranging definitions, the specific reconstruction algo-
rithm, and individual reconstruction parameters. A ROOT file contains all of the
parameters required to create POS or ATO files from an RHIT file.
A POS file consists of the reconstructed ion positions and the m/n for each
reconstructed ion as indicated in Table A.1. The POS file is created by applying
the reconstruction algorithms specified in the ROOT files to the RHIT data and is
the minimum set of information (with an appropriate range file described below)
required to perform subsequent data analysis. This file type is typically used for
Range files refer to files that describe a set of mass ranges associated with particular
ion types. A range file applied to the m/n information in a POS file defines the
elemental identity of each of the ions for 3D compositional analysis. Range files are
ASCII (1 byte per character) format files that can be opened, read, and edited with
many simple programs. At a minimum, they describe a lower limit and an upper
limit to a range of bins in the m/n histogram that are identified with a specific ion
type. Range files often include volume and color information as well. A hybrid
version of a range file, the LAWATAP ENV file, includes instrument and recon-
struction information [1].
An example RNG file is shown in Table A.2. The format originated from work at
Oak Ridge National Laboratory [4] and the latest version includes fields defining
ion color, lower/upper m/n limits, and composition (elemental or molecular). IVAS-
generated RNG files are used with software developed for the LEAP instrument and
may contain fields that are incompatible with the original format (ion descriptions
with more than two characters and polyatomic ion for example). These incompat-
ible items are stored after the last carriage return as a comments field that can be
reincorporated into an RRNG file used in IVAS.
Each RNG file (Table A.2) begins with two integers that indicate the number of
atom types and ion ranges (Line 1). Next, each 12-character atom type is followed
by that same 12-character atom type and three values indicating the RGB color of
the atom (Lines 210). Next, a table is included (Lines 1019) which associates
each ion m/n range (minimum and maximum value) with an integer combination of
elements from the atom list in the first section (Lines 19). For example, Line
19 indicates that an ion with m/n between 71.5600 and 72.6640 Da is FeO
(indicated by the entries of 1 under Fe and O and entries of 0 under Ni and
H). After these required fields are additional fields added by IVAS to allow for
proper translation into an RRNG format (see next section). In this example, the FeO
ion is provided with a defined color (Lines 2126).
252 Appendix A: Data File Formats
An example RRNG file is shown in Table A.3. First, an ion section is defined (Lines
59) followed by range definitions (Lines 1419). Each range includes the range
minimum and maximum, ion volume, ions atomic composition, and an optional
color. Further descriptions of the required and optional fields are included within
commented lines of the file itself.
The ePOS (or extended POS) file is the same headerless big-endian binary format as
the POS file, but it includes additional four-byte columns describing detector
information that enable more sophisticated analyses. The format detailed in
Table A.4 includes ion position, m/n, far (uncorrected) time of flight, voltage,
detector hit positions, and information on hit multiplicity. The extended informa-
tion increases the file size but also expands the users ability to analyze correlative
Appendix A: Data File Formats 253
The ATO file format evolved to support TAP instrumentation and includes four
major versions sharing the same ATO extension. The two contemporary versions
are the version-three (V3) ATO which is associated with the LAWATAP instru-
ments and the FlexTAP ATO format.
254
Table A.5 summarizes the LAWATAP V3 ATO file format. The first record is a
header with subsequent records containing information about individual ions. Each
record is stored in big-endian format. The header consists of eight 1-byte entries
where the fifth entry contains the value 0x03. The data section of the file consists of
56-byte fixed-length records recorded in acquisition order. The fourteen 4-byte
floating point columns include ion position, m/n, pulse number, voltage, time of
flight, and detector hit position. Additional fields exist but are not defined for use
with IVAS software. It is worthy to note that the pulse number is a 32-bit floating-
point binary, and unless it is reset to zero every 224 (~16.7 million) pulses (as is
typical for the LAWATAP system) it will lose precision as modern datasets
typically contain additional pulse events.
The FlexTAP version ATO file is similar to a tagged image file format (TIFF)
and contains a 16-byte header followed by a table describing the tags. FlexTAP
ATO files are converted automatically to RHIT files after acquisition of data from a
FlexTAP instrument.
References
1. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer Series
in Materials Science, vol. 160. Springer (2012)
2. Miller, M.K.: Proposed XML-based three-dimensional atom probe data standard. Surf. Inter-
face Anal. 36(56), 601605 (2004)
3. Brun, R., Rademakers, F.: ROOT - an object oriented data analysis framework. Nucl Instrum.
Meth. Phys. Res. A 389, 8186 (1997)
4. Miller, M.K.: The ORNL atom probe software package. J. Phys. 47-C2(3), 499504 (1986)
Appendix B: Field Evaporation
After Muller demonstrated experimental evidence of field ionization using the FIM
in 1951 [1], a theoretical explanation was initiated and developed by Inghram and
Gomer [2] and was continued by Muller and Bahadur [3]. In field-ion microscopy,
imaging gas atoms in the vicinity of the specimen undergo field ionization. Due to
an applied voltage on the specimen, the gas atoms are attracted to the sample
surface by an induced dipole force [4]. The gas atoms impact the specimen and
rebound from the surface, losing a portion of their kinetic energy. This process
occurs repeatedly until the gas atom is ionized by the high field directly above the
most protruding atoms. During ionization, the gas atoms topmost electron tunnels
into the sample surface, and the gas ion is then accelerated away from the specimen
by the electric field (typically of strength 20 V/nm).
Ionization occurs because the potential-energy barrier for the topmost electron is
reduced both by the applied field and by interactions with the specimen surface [5].
The highest probability of tunneling occurs at a distance of ~0.20.5 nm above the
specimen and is greatest in the high-field regions directly above the nuclei of
protruding specimen atoms. Because the gas atoms are preferentially ionized at
these positions, an image is obtained which shows the protruding surface atoms as
bright spots. Thus, an FIM image maps the high-electric-field locations above a
specimen surface. The emission current density is determined both by the tunneling
rate constant and by gas-distribution effects that influence the probability that a gas
atom is in the right place to be ionized [6, 7].
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 257
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
258 Appendix B: Field Evaporation
Field evaporation (FEV) is the removal of specimen atoms, in the form of ions, from
a surface, as result of applying a very high electric field. This is in contrast to
adsorbate field desorption, which is closely related to field evaporation and refers
to the removal of adsorbed foreign atoms from the specimen surface. The two
processes are basically the same from a theoretical standpoint.
FEV is sometimes a two-stage process, involving first the escape of an atom into
an escape charge state n (i.e., its charge is ne, where e is the elementary positive
charge), and then post-field ionization of the ion, in one or more steps, to a higher
final charge state. The escape step is normally taken to be thermal activation over a
barrier of height Qn, with FEV rate constant kn(F,T ) given by
where F is the atom-probe (AP) operating field (as applied to surface atoms that are
at high risk of FEV), T is the thermodynamic temperature, kB is the Boltzmann
constant, and A is the rate constant pre-factor (often taken as equal to the surface-
atom vibration frequency). The rate constant kn is measured in s1. If nhr denotes the
count of specimen surface atoms, within the AP detectors field of view, that are at
high risk of FEV, then the effective operating flux (or rate) R during FEV is given
by the FEV emission equation
The value of nhr is not well established, but is sometimes taken as 0.01 layers or
100 atoms. The flux or rate R is measured in layers/s, atoms/s, or ions/s.
In principle, two parameters in basic FEV theory are particularly important for
atom-probe operation: (a) the zero-barrier evaporation field (ZBEF) FE, which is
the operating field at which Qn becomes equal to zero, and (b) the field sensitivity
dQn/dF, for F-values near FE. Practical operating-field values depend both on the
chosen (but often not well known) operating flux and on emitter temperature (which
is usually not well known for laser pulsed AP). Consequently, practical F-values
are not themselves well known. However, in principle they must be slightly or
somewhat less than the ZBEF. Hence, a theoretical estimate of the ZBEF can be
used as a first estimate of the AP operating field. Theoretical ZBEF estimates can
be derived from an atomic-level version of the principle of conservation of energy,
as below.
Knowledge of the field sensitivity dQn/dF, near FE, is needed if it is wished to
predict how, for given flux R, the corresponding operating field FR varies with
temperature T, and consequently how the fractional field reduction (FR%) varies
with temperature. This information is important for correct AP operation. To derive
dQn/dF, it is necessary to know how Qn varies with F, near FE. This, in turn,
requires that a reliable physical model of the FEV escape process be available.
Appendix B: Field Evaporation 259
In the early days of field-ion microscopy, it was assumed that FEV could be
described by the classical image-hump model [8] (for reviews, see [9, 10]), but
tests have repeatedly shown that this model is unable to accurately describe exper-
imental data relating to field sensitivity (e.g., [11]). Hence, it came to be thought that
escape takes place via some form of charge-exchange mechanism [12, 13], in which
escape and ionization take place simultaneously at/near the top of an activation-
energy hump.
It is now believed [14] that FEV escape takes place by a particular form of
charge-exchange, termed the revised charge-draining mechanism. In this, FEV is
described as the evaporation of a partial-ion, in a partially charged state, over a
charge-draining hump. (A partial-ion is an atom that does not have an integral
number of electrons in the region of space associated with the atom. Gauss theorem
implies that atoms at highly charged surfaces must behave like this.)
Charge-draining is an intrinsically quantum-mechanical (QM) phenomenon.
Hence, Qn(F) can be reliably determined only by QM calculations. In the high-
field situation of the AP, it is intensely difficult to make such calculations fully
reliable. Useful FEV calculations have been performed only for tungsten (e.g., [15])
and (perhaps) for aluminum [16]. Consequently, for most materials, data relating to
fractional field reduction needs to be acquired via experiment.
Simple (approximate) ZBEF prediction can be carried out using the following more
general mathematics. First, obtain an expression for the potential energy (PE) of an
n+ ion, measured relative to the energy level of the base of the partial-ions bonding
well. This PE can be denoted by Wn(x, F), where x is the distance of the ion nucleus
from the specimens electrical surface. (The electrical surface is near the outer edge
of the surface-atom electron charge clouds.) The FEV activation energy is then
given by Qn(F) Wn(xe, F), where the escape distance xe is the distance of the
top of the activation-energy hump from the specimens electrical surface and is a
function of operating field F. The zero-activation-energy condition is then
expressed as
Qn FE Wn xe0 ; FE 0, B:3
where xe0 is the zero-barrier escape distance, i.e., the escape distance
corresponding to the condition Qn 0. The resulting expression, sometimes called
an energy-balance equation, contains three variables {n, xe0, FE}. This equation is
then solved for FE, using auxiliary methods to determine values for xe0 and n, as
described below.
The mathematics just set out is general: the main approximation comes in
finding a suitable simple expression for Wn(x, F). This can be done using three
260 Appendix B: Field Evaporation
physical steps. First, calculate the local zero-field activation energy (Q0n) needed
to create an n+ ion at a position x somewhat outside a defined part of the surface,
outside the reach of all short-range ion-surface interactions, in the absence of any
applied field. This is done by the following formal cycle. (1) Remove the bound
surface atom to become a neutral atom in remote field-free space: assume this
takes work 0, where 0 is the zero-field bonding energy of the atom, normally
taken equal to the tabulated sublimation energy. (2) Remove n electrons from the
neutral atom, one by one, leaving them in remote field-free space, well separated
from each other and from the resulting ion: this takes work Hn equal to the sum of
the first n ionization energies of the atom. (3) Take the electrons, one-by-one, to the
emitter Fermi level, via paths that all pass through the position x somewhat
outside the surface, and then move the ion to position x: in this combined process,
an amount of energy n is retrieved, where is the local work function of the part
of the specimen surface that faces the external ion. It follows that
X
n
Q0n 0 Hn n 0 Is n, B:4
s1
where Is is the sth ionization energy. This quantity Q0n is not really an activation
energy as usually defined, hence it is sometimes called the thermodynamic term
and denoted by Kn0.
In the second step, move the ion closer to the surface, to a new position x within
the range of attractive ion-surface forces but outside the range of short-range
repulsive forces, and assume (somewhat arbitrarily and not necessarily correctly)
that the strongest attractive force is due to image forces. This step introduces an
image-PE contribution n2e2/160x, where 0 is the electric constant. In the third
step, apply an electric field of mean strength F: this introduces an electrostatic PE
contribution neFx. The final (approximate) expression for Wn(x, F) is
Now ask what is the maximum field FnM that (for a given value of n) can be
predicted from this equation. Applying the condition dFE/dxe0 0 finds the max-
imum field as
FnM 40 =n3 e3 Q0n 2 0:6944616 V nm1 eV2 Q0n 2 =n3 : B:7
Equation (B.7) has been called Mullers formula, because a formula of this type
(albeit incorrect in detail) was first given by Muller [8]. (In fact, both Gomer and
Swanson [13] and Brandon [17] contributed to giving the correct Muller-type
formula for an n+ ion.) Correspondingly, FnM can be called the Muller escape
field for an n+ ion, and xnM the Muller escape distance for an n+ ion.
Evaluation of values for xnM suggests that, for many species, the distance xnM is
noticeably different from estimates (made in other ways) of the distance xb of the
base of the bonding well from the specimens electrical surface. (In particular, when
xnM < xb, the calculation cannot be physically realistic, because strong repulsive
forces are being ignored.) Physically, one expects the position of the base of the
bonding well to be the main factor that influences where escape occurs. Hence, it is
better in principle to set xe0 xb, and rearrange (B.6) into the form
FnE xb Q0n n2 e2 =160 xb =nexb : B:9
This formula, introduced by Tsong [18], was used by Muller and Tsong for their
systematic 1969 calculations [19], in which they set xb equal to the Pauling atomic
radius [20] for the element of interest.
Equation (B.9) can also be used to define a correction factor n(xb) by
FnE xb n xb FnM : B:10
It has been shown [21] that n(x) is a slowly varying function of distance; explicitly,
calculations for 55 elements (Forbes, unpublished work) have shown that for many
elements n(xb) is in the range: 0.90 n(xb) 1. Consequently, for a charge-
exchange-type model, the numerical value of FnM can be used as a satisfactory
estimate for FnE xb . Values of FnM were first systematically calculated by Tsong
[22] and are tabulated in many AP textbooks, usually for n 1, 2, 3.
The value of the parameter FnE xb depends on n. Brandons criterion [17] is to
take the lowest of these values as the theoretical estimate of zero-barrier evapora-
tion field FE, and the corresponding value of n as the theoretically predicted escape
charge state. Elementary FEV theory (as found in APT textbooks) normally uses the
Muller-Brandon approach, in which ZBEF estimates are obtained by applying
Brandons criterion to the Muller escape fields as determined above.
Note that the derivation of Mullers formula given above relates primarily to the
energetics of the escape process, does not involve the assumption that escape takes
place via the classical image-hump mechanism, but does show that Mullers
formula is applicable (as an approximation) to charge-exchange-type escape mech-
anisms [14, 21].
Mullers formula was originally derived mathematically in a different way, by
assuming that escape takes place via the classical image-hump mechanism, and was
originally called the image-hump formula. The continued use of this name is
unfortunate, because it can give the spurious (logically false) impression that the
success of Mullers formula justifies the use of the image-hump (IH) model in the
262 Appendix B: Field Evaporation
Fig. B.1 The melting temperature of various elements versus their theoretical and experimentally
determined evaporation fields. This chart allows a zeroth-order estimate of the evaporation field
for an element or compound if one is unavailable from the Muller-Brandon approach
context of field sensitivity, whereas the facts of the matter are that the IH-model
field-sensitivity predictions have failed badly (quantitatively) on almost every
occasion in the last 70 years on which they have been seriously tested. There can
also be a spurious (logically false) conclusion that because the image-hump model
fails badly, the so-called image-hump formula is not useful. In order to avoid both
these types of false impression/conclusion, the name Mullers formula is to be
preferred. However, it does need to be emphasized that the Muller-Brandon
approach estimates the ZBEF and that it is an approximation. For many materials,
Mullers formula appears to be satisfactory as a simple first approximation, but we
do not currently know how good it is. Significant work remains to develop a theory
of FEV escape that is predictive and applicable for both metals and nonmetals in
both laser and voltage pulse modes.
If the user desires an estimate of an evaporation field for an element or com-
pound, a zeroth-order estimate may be obtained by plotting the melting temperature
of various elements versus their theoretical and experimentally determined evapo-
ration field as shown in Fig. B.1. Although there is significant scatter in the
phenomenological power law fit (R2 ~ 0.7) shown on the plot, this method may
provide at least a starting point for the evaporation field if such a value is
unavailable. It is clear that for compounds such an estimate will be more accurate
for miscible alloys as opposed to ordered phases.
Appendix B: Field Evaporation 263
From the above discussion it is clear that the rate of field evaporation can be
enhanced simply by increasing the applied field on the specimen. Field evaporation
may also be significantly enhanced by irradiation of the specimen by a beam of
photons or electrons. It is exactly this effect that provides the operating mechanism
for the pulsed-laser atom probe [28]. Tsong [5] proposed that the enhancement in
field evaporation could be caused by two effects. The first, and simplest, effect is
emitter heating by the particle beam. The inelastic interaction of the particles with
the emitter produces a temperature rise in the specimen, and hence an increase in
the FEV rate constant in accordance with (B.1). With a reliable model for Q(F)
(or reliable empirical data for the FEV operating flux in use), the change in the field
evaporation characteristics could be calculated.
264 Appendix B: Field Evaporation
References
10. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion Micros-
copy. Oxford University Press, Oxford (1996)
11. Forbes, R.G., Biswas, R.K., Chibane, K.: Field evaporation theory: a re-analysis of published
field sensitivity data. Surf. Sci. 114, 498514 (1982)
12. Gomer, R.: Field desorption. J. Chem. Phys. 31, 341345 (1959)
13. Gomer, R., Swanson, L.W.: Theory of field desorption. J. Chem. Phys. 38, 16131629 (1963)
14. Forbes, R.G.: Field evaporation theory: a review of basic ideas. Appl. Surf. Sci. 87, 111
(1995)
15. Wang, L.C., Kreuzer, H.J.: Kinetic theory of field evaporation of metals. Surf. Sci. 237,
337346 (1990)
16. Sanchez, C.G., Lozovoi, A. Y., Alavi, A. Field evaporation from first principles. Mol. Phys.
102, 10451055 (2004)
17. Brandon, D.G.: The structure of field-evaporated surfaces. Surf. Sci. 3, 118 (1964)
18. Tsong, T.T.: On the mechanism of field evaporation. Surf. Sci. 10, 102117 (1968)
19. Muller, E. W., Tsong, T.T.: Field Ion Microscopy: Principles and Applications. Elsevier,
New York, NY, 1969
20. Pauling, L.: The Nature of the Chemical Bond. Cornell University Press, Ithaca, NY, 1960
21. Forbes, R.G.: A new formula for predicting low temperature evaporation field. Appl. Phys.
Lett. 40, 277279 (1982)
22. Tsong, T.T.: Field ion image formation. Surf. Sci. 70, 211233 (1978)
23. Muller, E.W., Krishnaswamy, S.V.: High ion charges in field-evaporating 5d transition
metals. Phys. Rev. Lett. 37, 10111014 (1976)
24. Ernst, N: Experimental investigation on field evaporation of singly and doubly charged
rhodium. Surf. Sci. 87 469482 (1979)
25. Haydock, R., Kingham, D.R.: Post-ionization of field-evaporated ions. Phys. Rev. Lett. 44,
15201523 (1980)
26. Kingham, D.R.: The post-ionization of field evaporated ions: a theoretical explanation of
multiple charge states. Surf. Sci. 116, 273301 (1982)
27. Gault, B, Moody, M.P., Cairney, J.M., Ringer S.P.: Atom Probe Microscopy. Springer,
New York, NY, 2012
28. Kellogg, G.L., Tsong, T.T.: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys.
51(2), 11841194 (1980)
29. Bunton, J.H., Olson, J.D., Lenz, D., R., Kelly, T.F.: Advances in pulsed-laser atom probe:
instrument and specimen design for optimum performance. Microsc. Microanal. 13, 418427
(2007)
30. Vella, A., Mazumder, B., Da Costa, G., Deconihout, B.: Field evaporation mechanism of bulk
oxides under ultra fast laser illumination. J. Appl. Phys. 110, 044321 (2011)
Appendix C: Reconstruction Geometry
Overview
In this appendix several geometric exercises are used to illustrate the definitions
essential for spatial reconstructions by using and building on approaches previously
established in the literature [14]. We start by defining common terms that are
relied upon later, and then calculations are developed for the lateral projection and
the volume increment. We finish by showing a calculation of the shank angle from
the voltage curve. In all of these calculations the apex is assumed to be spherical
and situated on a conical shank. We try to follow the convention that distances on
the scale of the whole instrument will be in upper case (flight path LDet for example)
while local properties of the tip will be in lower case (rTip for example).
Image Compression
The projection model is illustrated in Fig. C.1 and assumes that a single projection
point exists at a distance mrTip beyond the center of the spherical apex, where
1 + m is known as the image compression factor. Assuming that an ion is
launched normal to the sphere surface (with angle Launch), the effect of is to
reduce the projection angle to Det and with it the magnification.
The reconstruction process is the task of transforming a detector position into a
launch position on the tip, rLaunch. By considering the detector/flight path triangle,
we know the compressed angle:
LDet RDet
cos Det q and sin Det q : C:1
L2Det R2Det L2Det R2Det
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 267
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
268 Appendix C: Reconstruction Geometry
s2 2mr Tip s cos Det m2 r 2Tip cos 2 Det r 2Tip m2 r 2Tip m2 r 2Tip cos 2 Det
Appendix C: Reconstruction Geometry 269
and we find
p
s r Tip m cos Det 1 m2 sin 2 Det :
and we write
s r Tip C:5
r Launch r Tip sin Det : C:6
sin Det
tan Launch : C:8
cos Det m
To evaluate the small-angle limit of this expression we will substitute back in for
the definition of to write
p
m sin Det cos Det sin Det 1 m2 sin 2 Det
tan Launch p :
m sin 2 Det cos Det 1 m2 sin 2 Det
We make the approximations sin (Det) Det and cos (Det) 1. Noting that
we have a term proportional to unity in the denominator but must keep Det to first
order in the numerator we get
m 1Det
tan Launch :
1
With a small-angle limit on the (larger) launch angle this reduces to the useful
approximation
Lateral Magnification
In developing our expression for the compressed trajectory we found (C.6). This
equation is very important in itself as it relates the position on the detector to the
radial launch position on the tip. So we will go back to (C.1)(C.3) in order to
evaluate the ratio RDet/rLaunch.
sRDet
r Launch q
L2Det R2Det
and then
This is the expression that applies to large field of view (FOV) local electrode
atom probes. In order to obtain the classical small FOV expression for lateral
magnification we take the small-angle limits for s and cos(Det) to get
RDet r Tip
r Launch
LDet
RDet LDet
: C:10
r Launch r Tip
The radius of the apex is a key component in understanding all aspects of the spatial
reconstruction. It is important to both the radial magnification as well as the depth
increment per atom. As shown in Fig. C.2, we start with the simple case where the
sphere of the apex and the cone of the shank meet with tangential continuity.
Appendix C: Reconstruction Geometry 271
Fig. C.2 Tip geometry for variation of tip radius as a function of analysis depth and tangential
continuity
We begin by writing down the expected variation of tip radius with analysis depth
for a fixed shank half-angle, . Note that the angle describes how the radius of the
spherical cap varies with the depth of the spherecone intersection:
dr Cap
tan : C:11
dzShank
We are trying to evaluate how the apex radius varies with analysis depth, zA:
dr Tip
K : C:12
dzA
We want the derivative of zA, so we get ready by finding the derivative of zShank
first. Again from the central triangle we have
r Cap
cos
r Tip
272 Appendix C: Reconstruction Geometry
Fig. C.3 Tip geometry for variation of tip radius as a function of analysis depth without tangential
continuity
dzShank cos
:
dr Tip tan
dzA cos
sin 1
dr Tip tan
sin
K : C:14
1 sin
In this more general specimen geometry shown in Fig. C.3, we relax the require-
ment that the apex meets smoothly with the shank. We hold that the tip radius rTip is
just a fixed multiple of the shank radius rCap:
r Tip
f S= : C:15
C r Cap
Appendix C: Reconstruction Geometry 273
r Cap 1
cos : C:16
r Tip f S=
C
Writing zA zShank zCap and zCap rTip zS and considering the central
triangle to see that zS rTip sin () we can write
We calculate derivatives as before. This time our expression for the derivative of
zShank is
dzShank cos
dr Tip tan
dzA cos
sin 1:
dr Tip tan
dr Tip f S= tan
K , f C
r C:17
dzA 1 f S= tan tan f 2S 1
C =C
In this section we calculate the size of the analysis volume incorporating a fixed
shank angle and image compression. Looking at Fig. C.4, one can see that the
acquired analysis volume has three geometric components:
A spherical cap on top defined by the initial tip radius r0Tip and the point of
0
intersection of the sphere with the analysis cone, Z0
0 0
The main analysis cone section from Z0 to Z1
A spherical cap, not acquired, on the bottom defined by the final tip radius r1Tip
0
and the point of intersection of the sphere with the analysis cone, Z1
274 Appendix C: Reconstruction Geometry
Rd
qAn
ZZ0 ZZ1 ZAC
Rd
Ld
Z0
Z '0
r0An
Z1 Z '1 r0Tip
r1An
qF
mr0Tip q 0
z
r1Tip
qF
mr1Tip
q 1z
zR 3
z3 2R Z3
V Cap R2 z R2 Z :
3 zZ 3 3
Appendix C: Reconstruction Geometry 275
For both the top and bottom cap sections, the relevant radius is the current tip
radius rTip, and the Z of interest is zS as defined by (C.7):
so each of the two cap sections can have their volumes written as
!
2 cos Det m3
V Cap r 3Tip cos Det m : C:19
3 3
f cos Det m
2 f3
V Cap r 3Tip f :
3 3
It is important to note that An is not the specimen shank angle but rather
describes how the FOV evolves through the analysis depth. From Fig. C.4, we
see that
r 0An
tan An
ZZ 0
where
0
ZZ0 Z AC Z0 :
Recalling (C.6)
r 0Tip
ZAC
K , f
Using (C.7),
0
z0 r 0Tip r 0Tip cos Det m
So
r 0Tip
ZZ 0 r 0Tip 1 cos Det m
K , f
Then
We simplify this to
sin Det K , f
tan An :
1 cos Det K , f
g sin Det
1 cos Det K , f
gK , f
tan An
r 0Tip
ZZ 0
K , f
r 1Tip
ZZ 1
K , f
Appendix C: Reconstruction Geometry 277
Simplifying
g2 1 3 0 3
V Spec 3f 2 f 3 r Tip r Tip : C:20
3 K , f
Volume Increment
dr Tip
K , f :
dzTip
dV Spec
g2 K , f 3f 2 f 3 r 2Tip
dzTip
278 Appendix C: Reconstruction Geometry
zTip i :
g2 K , f 3f 2 f 3 r 2Tip
If we assume that the shank angle is zero, then K,f is zero. In the limit of
small-angle we obtain a per-atom z increment of
i L2Det
zTip :
r 2Tip R2Det 2
Starting with our formula for the acquisition volume (C.20) we note that all shank
angle dependence is in the K,f terms, so we consolidate the remaining terms into a
placeholder, , and write
3
g2 3
V Spec r 1Tip r 0Tip
3 K , f
where
g2 cos Det 3f 2 f 3 :
g2 SVR
K , f C:21
3 SVR
where
3 3
r 1Tip r 0Tip
SVR :
V
The volume VSpec is typically accumulated from the assigned atomic volumes,
and the initial and final tip radii are evaluated with R v/KF. So we will
typically use
3
3
vTip v0Tip
1
SVR P
FK 3
Appendix C: Reconstruction Geometry 279
g2 SVR
tan q C:22
f S=C 3 SVR g2 SVR g2 SVR f 2S=C 1
g2 SVR
sin : C:23
3 SVR g2 SVR
References
1. Blavette, D., Sarrau, J.M., Bostel, A., Gallot, J.: Direction et distance danalyse a la sonde
atomique. Rev. Phys. Appl. 17, 435440 (1982)
2. Bas, P., Bostel, A., Deconihout, B., Blavette, D.: A general protocol for the reconstruction of
3D atom probe data. Appl. Surf. Sci. 87/88, 298304 (1995)
3. Walck, S.D., Buyuklimanli, T., Hren, J.J.: Extended depth profiling with the IAP. J. Phys. 47-
C2(3), 451458 (1986)
4. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers (2000)
Appendix D: Mass Spectral Performance
As a time of flight (TOF) mass spectrometer, the mass resolving power (MRP) of an
atom probe is one of the critical parameters that describe its performance. On the
surface, this parameter seems straightforward to define and measure. However, as
we shall see below, there are many subtleties that can influence the determination
and reporting of MRP [1].
1 1 L2
nev mv2 m 2 D:1
2 2 t
where n is the charge state, e is the elementary unit of charge, V is the potential
difference, m is the ion mass, v is the ion speed, L is the ion flight path length, and
t is the ion TOF; then the resulting equation may be solved for TOF as
m L2
t2 : D:2
n 2ev
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 281
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
282 Appendix D: Mass Spectral Performance
For a given ion type, m, n, and e are constants and differentiating (D.2) gives
t L V
t 2 : D:3
2 L V
If a single ion type is measured multiple times with a perfect timer, then the
timing variation that develops is due to voltage (V/V ) or flight length (L/L)
variations. This component of timing precision will be referred to as the physical
timing uncertainty, tP:
L 1 V
tP t t : D:4
L 2 V
Timer electronics achieve a finite precision which will be denoted as tT. This
imprecision is independent of the physical timing imprecision. Furthermore, there
is timing uncertainty associated with the temporal duration of field evaporation
during a pulse. This term is called time-of-departure spread (TODS) and is denoted
as tD. This timing imprecision is also independent of both the timer precision and
the physical timing precision. These three terms may be added in quadrature as
follows:
t2 tP 2 tT 2 tD 2 : D:5
Finally, MRP is related to the time resolving power through (D.2) by noting that
m,
t
2 n m D:7
t
n
or
r
m 1 t 1 m 1
MRP L: D:8
m 2 t 2t n 2eV
p
From (D.2) it is apparent that flight time, and hence MRP, will vary with mn,
p
V 1 , and L. Thus if MRP is to be compared from one experiment to the next, it
must be reported at standard values for m/n, V, and L, or the values must be stated
explicitly.
Appendix D: Mass Spectral Performance 283
Sources of V/V
Normalized energy deficits, E/E, which can come from a variety of sources, are
indistinguishable from voltage fluctuations in the accelerating voltage, V/V, and
so all such terms will be grouped in the V/V term. When doing this, it is essential
to remember that a V operating on an ion of n +2 produces a E that is twice
that of an n +1 ion. When a large energy spread exists, flight time differences for
a given ion type get worse with longer total flight time, and longer flight paths do
not lead to improved timing resolution. In laser pulsing, energy deficits may be
small if they are primarily due to thermal energy spreads (0.025 eV at 300 K). Thus
the TOF spread, t, is unchanged during ion flight, and greater MRP can be
achieved by making the flight time longer. This effect was demonstrated dramat-
ically by Liu et al. [2] where they achieved MRP of 4,000 by making very long
flight times in an 8-m-long laser pulsed atom probe.
In voltage pulsing, energy spreads develop because the accelerating voltage is
pulsed by V/V 20 % to induce synchronized field evaporation events. The field
evaporation event actually occurs over a narrower range of voltage which induces
energy spreads of E/E V/V 12 % for n +1 ions. This is a large value
which limits MRP in voltage pulsing to about 200 regardless of flight path length
unless some additional measure is taken (e.g., local electrode or reflectron) to
counter this effect. In LEAPs, the energy spread in field pulsing is reduced com-
pared to a remote electrode geometry because the distance over which any ion is
accelerated is reduced (distance between tip and aperture of local electrode) and the
ion enters the field-free region of the flight path more quickly.
Sources of L/L
The flight length of an ion, L, is generally stated for the distance to the center of the
detector (or the shortest distance to the detector in the case of a reflectron-based
system). In APT, the detector records hit position coordinates on the detector which
makes it possible to correct for geometric differences in flight length with angle
from the central axis such that all flight times are corrected to correspond to L. In the
LEAP, a flight length correction known as the bowl correction was developed by
Oltman [3] and is discussed in detail in Chap. 4.
Sources of tT/t
The imprecision of the clocks used to time ion flights is a hard limit on the timing
precision. However, there are other factors that may limit precision further. The
imprecision introduced by the electronics that pick an arrival time from pulses that
284 Appendix D: Mass Spectral Performance
have traversed the delay line anode (DLA) is likely to be greater than the clock
imprecision and is usually the largest limitation imposed by the hardware. Further-
more, this limitation can vary across the face of any DLA. MCPs generally have a
fast response and low time spread relative to the timer precision, so they should not
degrade the precision overall.
Sources of tD/t
Physically, the temporal spread of the evaporation window is the sum in quadrature
of the timing imprecisions. In field pulsing, voltage pulses of a few nanoseconds are
used, but the evaporation events are restricted to the topmost portion of the pulse
and the TODS is correspondingly small for good electrical conductors. The energy
spread terms dominate in field pulsing.
For thermal pulsing, energy variations can be small, and it is the TODS that
usually dominates. The challenge for making small TODS is cooling the specimen
quickly after heating. Certainly, heating the smallest possible volume helps accom-
plish this. It is for this crucial reason that laser-capable LEAPs have always been
designed to focus to the smallest possible spot size. The shortest TODS reported to
date was 114 ps on an aluminum specimen (good thermal diffusivity) with 355 nm
laser light on a LEAP 4000X (<3 m diameter, 4 spot) [1]. Note that this TODS is
much greater than the 12 ps pulse width of the laser which is a confirmation that the
cooling time dominates the TODS. Other materials, such as steels, can have more
shallow skin depths (depth over which energy is absorbed) than aluminum which
can also aid in heating small volumes and result in a reduction in TODS.
Fig. D.1 The role that discrete discrimination plays in image quality in APT. In (a) the apparent
oxygen signal in the silicon substrate (right side of image) is due to overlap with the large tail from
the silicon peaks. In (b) with greater MRP comes improved discrete discrimination, so there is no
misidentified oxygen signal in the silicon substrate to the right. Although the peaks may be
decomposed collectively to obtain accurate bulk composition, (b) shows the benefit of discrete
discrimination for imaging as very few oxygen atoms are actually present in the substrate silicon
strong silicon signals (located at m/n 14, 14.5, and 15 Da) in the mass spectrum.
Even though a clear oxygen peak is observed at 16 Da, a majority of the counts in
this mass range are actually silicon ions. Since there is no way to discriminate
individual ions that have different identities but the same m/n, these signals cannot
be discriminated discretely in an image.
Collective discrimination is defined as the ability to discern, as a fraction of
some group, the identity of each subgroup. Collective discrimination would be
accomplished, for example, by decomposition of a single unresolved peak or
overlapping peaks into their constituents using deconvolution or isotopic distribu-
tion information (see Chap. 6). Collective discrimination is important for determi-
nation of bulk composition within a particular volume.
In Fig. D.2, this distinction is illustrated with regard to resolving two peaks.
For collective discrimination, it should be possible to separate two peaks by
deconvolution when there are enough counts to perform a statistically significant
separation. For peaks of about equal height, as shown in Fig. D.2a, discrimination is
possible when they are separated by their FWHM.
For discrete discrimination, the requirements on MRP are much greater. Ideally,
every atom displayed would be unambiguously identified. For discussion purposes,
the minimum requirement is five times that for collective discrimination (5
FWHM), as shown in Fig. D.2b. When peaks are not of equal magnitude or width,
or when the number of counts is low, then these peak separations may need to be
even greater.
A practical example is shown in Fig. D.3. Here the 28Si+ peak and the 14N2+ peak
are just separated enough for good collective discrimination. A composition deter-
mined from this volume would be expected to have good accuracy; however, if
286 Appendix D: Mass Spectral Performance
Fig. D.2 Examples of peak overlaps for peaks of equal magnitude. The MRP in (a) is adequate for
collective peak discrimination through deconvolution. The peaks in (b) are much more separated,
relative to their width making discrete discrimination possible
Fig. D.3 Collective discrimination of the 14N2+ and 28Si+ peaks. With sufficient counts, numerical
decomposition is straightforward. The two shaded boxes are placed such that they meet at the
minimum between the peaks to suggest the usual range delimiters that might be used in plotting
these counts in an image
Appendix D: Mass Spectral Performance 287
ranges were set as shown by the two shaded areas, then many of the counts in each
range would be misidentified. Discrete display of the hits in this spectrum using
these ranges would lead to a more ambiguous result. The conclusion of this
discussion is that the required MRP depends strongly on whether composition
determination or image display is the analytical goal.
Standard MRP
References
1. Oltman, E., Kelly, T.F., Prosa, T.J., Lawrence, D., Larson, D.J.: Measuring contributions to
mass resolving power in atom probe tomography. Microsc. Microanal. 17(S2), 754755 (2011)
2. Liu, J., Wu, C.-w., Tsong, T.T.: Measurement of the atomic site specific binding energy of
surface atoms of metals and alloys. Surf. Sci. 246(13), 157162 (1991)
3. Oberdorfer, C., Eich, S.M., Schmitz, G.: A full-scale simulation approach for atom probe
tomography. Ultramicroscopy 128, 5567 (2013), doi: http://dx.doi.org/10.1016/j.ultramic.
2013.01.005
Appendix E: Additional Considerations
for LEAP Operation
Before routine work on a LEAP can be pursued, there are certain instrument
calibrations and preparations that must be made. This appendix summarizes the
essential considerations that should be made before routine operation of the LEAP.
Calibration Files
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 289
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
290 Appendix E: Additional Considerations for LEAP Operation
voltage changes. Fortunately, this only impacts the live mass spectrum quality.
Each individual analysis is properly corrected as a step in data processing (Chap. 5),
so the quality of the reconstructed data will not be impaired by a choice of a
calibration file that does not match the experimental conditions well.
Local electrodes (LEs) are critical for good LEAP performance (for a detailed
description of the LE see Chap. 3). Each LE is factory tested to ensure good
ultrahigh-vacuum (UHV) compatibility, correct shape, and high resistance to elec-
tron emission through the voltage operation range of the LEAP. The nature of APT
is to analyze specimens that fail with some significant probability. With each failure
there is opportunity to damage the LE leading to deceased performance (i.e., partial
occlusion of the open aperture itself or electron emission from the LE that results in
a very poor signal-to-noise ratio). A challenge of tool management is to store, test,
and clean LEs in a way that ensures maximum performance and lifetime. New
electrodes from CAMECA have been qualified for electrical performance to 15 kV
with less than 0.2 % data rate (DR) using the procedure outlined in the Local
Electrode Flat-Test Procedure technical bulletin.
Storage
LEs are designed to operate in both voltage pulsed and laser pulsed modes. The
plastic insulator (polyether ether ketone or PEEK) that separates the voltage pulsed
nickel apex from the grounded puck body absorbs moisture when exposed to a
non-UHV environment. Consequently, LEs should be stored in the LEAP when
possible and for best results should reside in the instrument for 37 days before
being used to analyze specimens. A dust-free, dry-pumped vacuum storage unit is
the best alternative to the LEAP, followed by a dust-free desiccator.
Heating
conditioning period. Figure E.1 shows the mass spectra from a silicon microtip
analyzed with an LE that has been put into service 4 h after storage outside of
vacuum (unbaked) and another LE that has been baked for 4 h and put into service.
The microtip is made of antimony-doped silicon and, after analyzing through the
surface, should not contain oxygen or hydrogen. The relative number of detected
events associated with contamination (peaks labeled H2O, SiO, and SiH2O) is
reduced from 3.5 to 0.35 % of the total spectrum for the baked LEs.
Generally, electrodes stored in the LEAP do not experience changes in electrical
performance unless put into service. For example, an LE that tests to 10 kV (shows
electron emission) will continue to test to 10 kV even after a month of storage
within the LEAP buffer chamber. Though time in UHV does not affect electrical
performance it does impact the presence of water-bearing and hydrogen-bearing
contamination emitted during acquisition. For any unbaked LEs that experience
contact with the ambient environment, it is recommended that they be placed in
LEAP UHV for a minimum of 37 days before being put into service. Long-term
trends for a group of baked and unbaked LEs are shown in Fig. E.2. A bake of local
electrodes in the load lock for 4 h at 150
C appears equivalent to conditioning LEs
under vacuum without baking for 4 days (note that the unbaked LE results are
shifted by 96 h to highlight the trend). The additional downward curvature of the
baked LE contamination trends immediately after baking for the H2O and SiO
peaks suggests that time in vacuum alone may not be sufficient to achieve the best
cleanliness for LEs.
292 Appendix E: Additional Considerations for LEAP Operation
Testing
Electron emission from the local electrode, referred to as electrode turn-on, is the
primary failure mode that limits LE lifetime. Sharp debris or damage to the LE
surface produces a geometry that enables FowlerNordheim electron emission [3].
It is believed that this emission current terminates somewhere on the specimen,
enabling ion emission that at some critical voltage overwhelms the normal imaging
operation of the atom probe (see Fig. E.3). Depending on the type of debris or
damage, either these emission regions on the LE can be mitigated through plasma
treatments or they may require refurbishment of the LE assembly with a new
aperture cone.
Testing for electron emission is illustrated in Fig. E.4. The LE is positioned over
a flat test sample (aligned in this case near a microtip for reference) in distance
similar to normal microtip analysis (~150 m above the test flat). This distance is
based on the measured separation between the LE and its reflection (see the
CAMECA Local Electrode Flat-Test Procedure technical bulletin for instrument-
specific details). LEs are considered safe to use with a specimen up to the voltage
for which the flat-test begins to exhibit significant emission (0.2 % DR or measured
electron current). Minimum electrical turn-on performance guidelines suggest that
the maximum operational voltage for analysis of a wire specimen is 1.52.0 times
the flat-test value and for a microtip specimen is 1.21.4 times the flat-test value due
to the higher electric field imposed upon the electrode surface when running
microtips due to the microtip geometry.
Appendix E: Additional Considerations for LEAP Operation 293
Fig. E.3 Schematic illustrating possible sources of observed LE electron emission. (1)
FowlerNordheim electron emission from LE. (2) Ionization of gas and/or substrate atoms when
electron beam travels to high-voltage surface and strikes substrate surface. (3) Ionized material
follows electric field to terminate on LE or passes through to the detector. (4) High-ion-emission
region (hot spot) observed at detector (as well as broader emission region) indicating significant
LE emission
Plasma Treatment
Plasma treatment has been found to improve the electron emission characteristics of
LEs in general. The precise nanoscale modifications to the LE are not well
understood, but the benefits are clear. During the LE plasma treatment process,
the plasma likely removes carbonaceous contamination leading to a cleaner surface.
For recovering damaged or contaminated LEs, plasma treatment can improve flat-
test turn-on voltages by 510 kV in some cases.
294 Appendix E: Additional Considerations for LEAP Operation
Once electrodes leave vacuum storage at the factory, they are prone to pick up
small particle contamination of various sorts. Should these particles be of the right
size, shape, and location, they can degrade the electrical performance of the
electrode. This degradation in flat-test performance can almost always be undone
(~95 % of cases) with proper conditioning of the electrode. In order to remove
problematic contamination, the following procedures should be followed for new
electrodes introduced from an ambient environment into the UHV environment of
the LEAP:
1. Immediately before entry into the load lock, electrodes should be physically
treated with a flow of clean, dry, ~15 psig pressure gas (canned dry air or N2) to
remove any loose particle contamination on the electrode surface. Spray gas
directly at the front opening of the electrode (position a few centimeters away
and spray for 12 second bursts).
2. If available, then plasma treat the electrode for 12 h with the integrated plasma
treatment system (check with CAMECA for guidelines regarding gas type,
power, and pressure of operation for the plasma).
3. Perform a flat-test procedure as described in the technical bulletin by turning
up voltage until a DR of 0.2 % is achieved or upon reaching 11 kV. Sometimes
an electrical discharge may occur, requiring the operator to perform the test
again. These electrical discharges often remove stubborn particle contamina-
tion that has not been removed via the high-pressure gas and plasma treat-
ments and should not necessarily be avoided for new electrodes. These
electrical discharges often happen after introduction of new electrodes into
the LEAP from storage in air and do not negatively affect the performance of
new electrodes.
4. Should electrodes fail to yield satisfactory flat-test results, then repeat the gas
flow and plasma treatment an additional time. If plasma treatment is unavailable,
then remove and repeat physical treatment with gas and re-test.
5. Any new electrode failing to meet minimal electrical performance characteris-
tics upon entry and after testing/conditioning described above should be imme-
diately returned for replacement/requalification (contact your CAMECA service
representative for details).
6. For the purposes of acquiring the highest quality data possible, electrodes should
be placed into the LEAP as soon as received from CAMECA and tested/
conditioned for subsequent use. Should space considerations require that some
electrodes need to be stored external to the LEAP, keep them covered and in a
dry, oil-free, dust-free environment and test/condition them upon entry into
the LEAP.
Appendix E: Additional Considerations for LEAP Operation 295
Acquisition Control
Fig. E.5 The proportional control algorithm in action showing (top) target and actual detection
rate and (bottom) voltage response for the same 10 million detected events. The arrow shows
excursions in the detected rate that occur during a typical laser Drift Comp scan. The large +
shows the occurrence of an IEC event and the resulting fast drop in voltage which resulted in an
equally fast reduction in observed detection rate
Pulses Per Voltage Update specifies that after every n pulses the firmware
calculates the current DR (based on the filter depth as determined above),
calculates an error term based on the difference between the current and target
DRs, and calculates a voltage-update step size equal to the product of the error
term and the proportional coefficient setting. This parameter defines n and hence
determines the frequency of voltage adjustments.
Proportional Coefficient (%/%) is a multiplier for converting the detection-rate
error term to a voltage adjustment value.
Max upward voltage step (V ) is the maximum upward-step size per voltage-
update interval. The final voltage step for an increasing voltage step is clipped
against this setting.
Max downward voltage step (V ) is the maximum downward-step size per
voltage-update interval. The final voltage step for a decreasing voltage step is
clipped against this setting.
Appendix E: Additional Considerations for LEAP Operation 297
f 200 kHz
V Up Volts=s f V Max Up =N Pulses Per Update ! 20 Volts=s
f 200 kHz
V Down Volts=s f V Max Down =N Pulses Per Update ! 20 Volts=s
Advanced users can edit and modify the EvapControl.INI file to include addi-
tional algorithms or modify the default algorithms. In order to reset the
EvapControl.INI file to its default settings, delete or rename the EvapControl.INI
file and then close and reopen the LEAP Control Center software, and a new default
file is generated.
The behavior of each default algorithm is summarized in Table E.1 where
the number of pulses per update is also converted to time per update for a
200 kHz pulse frequency. Typical rate determinations are averaged over a 10,000
or 50,000 pulse window. Voltages are updated every 5,000, 10,000, or 25,000
pulses depending on the specific algorithm. The proportional coefficient ranges
from 0.05 to 10, while the maximum voltage change per step is between 0.5 and 5 V
going upward or 0.5 and 1,000 V going downward.
When operating the LEAP, the user selects target DR and initial voltage
(or engages the algorithm during the experiment), and the algorithm manipulates
the voltage from there. Two of the most obvious effects of the algorithm choice are
the minimum DR allowed by the software and the maximum rate of voltage
increase (which will often affect the time it takes to go from near-zero ion emission
to the target DR). For many materials that require low DRs to survive, only the Low
Detection Rate and Interface Low Rate algorithms are appropriate, although these
algorithms have reasonably large maximum upward voltage steps.
In terms of optimizing analysis yield or data quality, the current algorithms are
blind. They merely try to hold the evaporation process relatively constant.
Multiphase materials or layered structures can present widely varying local evap-
oration effects. Alternative control algorithms, dependent on acquisition informa-
tion other than the DR, might allow for improved yield and/or data quality.
298
Over time, the LEAP acquisition control algorithm has evolved to include a fast
response feature termed interface evaporation control (IEC) that has demonstrated
yield benefit for some sample types. The goal is to detect abnormal detection
statistics (a possible signal preceding specimen failure) and minimize the chances
for failure by quickly lowering the evaporation rate. It is difficult to rapidly change
the specimen voltage due to the relatively high capacitance of the specimen stage
and HV power supply, so schemes were developed to rapidly stop the applied
pulse to the tip. The original implementation in software calculated an average DR
over a pulse window (defined by the user). When the average rate exceeded the
target rate by some factor, defined by the user, the software would stop firing of
the voltage pulser (or deflect the laser off the tip) and lower the voltage by some
amount (also defined by the user). Although the response of this algorithm is
relatively quick compared to the control algorithm, a firmware implementation
is significantly faster. Note that not all LEAP tools have the firmware implemen-
tation option.
The modern implementation of the IEC in LCC 4.2 uses the firmware to monitor
ions emitted during consecutive pulses and responds immediately (voltage pulse
generation is interrupted or laser pulse picking terminates, and the standing voltage
is lowered prior to the next pulse). The user selects the sensitivity of the response
(low, medium, or high), so when an extreme number of consecutive instrument
pulses result in a detected ion (36 out of 40 pulses, 18 out of 20 pulses, or 9 out of
10 pulses, respectively) the instrument immediately lowers the voltage by an
amount defined by the user. For typical target DRs (1 % or 1 ion for every
100 pulses) it is statistically unlikely for this sequence of detected events to occur
randomly and indicates initiation of rapid tip change (the odds are less than 1067,
1033, and 1017 for low, medium and high IEC, respectively). Quickly stopping
this behavior should improve the chances for specimen survival for a wide variety
of sample types. The occurrence of an IEC event is illustrated in Fig. E.5 by the
large plus sign positioned immediately following the lowering of voltage which
results in a correlated drop in DR. IEC settings are sample type specific, and a too
sensitive setting will result in a lower average detection rate and a degraded signal-
to-noise ratio (SNR).
Keeping DR constant by manipulating only the voltage may not provide for the best
yield or data quality. For example, background counts resulting from the field
evaporation of adsorbed vacuum contaminants should be a function of DR, apex
surface area, and pulse rate. Consequently, background levels might be improved
by operating under constant evaporation flux (ion detection rate per unit area) rather
300 Appendix E: Additional Considerations for LEAP Operation
than DR (ion detection per time) [6]. For material systems that rarely fail during
analysis, constant ion flux likely provides for a better SNR, more consistent
evaporation physics, and possibly improved spatial reconstructions. In order to
estimate the ion flux from the surface, an estimate of the specimen surface area is
necessary. LCC allows the operator to enter an estimated evaporation field that it
then uses with the specimen voltage to estimate surface area (see Chap. 5).
There is some danger or added complexity in implementing a constant flux
control for inhomogeneous materials, especially for multilayer systems. First,
relative to the constant DR algorithm, the electric field will be higher in order to
increase the DR as emission area increases. This increases the chance of specimen
failure as the analysis proceeds. Second, flux estimates are complicated for mate-
rials with differing evaporation fields present in the field of view simultaneously.
Underestimating flux would again result in higher DRs, higher applied field, and
higher likelihood of specimen failure. Finally, for multilayer systems, proceeding
from a layer of low-field material to high-field material for example, the increasing
specimen voltage necessary for evaporation of the high-field material may be
incorrectly interpreted by a constant flux algorithm as an increase in emission
area. The algorithmic demand for voltage (field) increase will be twofold: Constant
flux requires the DR to be increased because of a perceived increase in emission
area (increase in applied voltage), and the increasing field required by the material
requires higher field (increase in applied voltage) to maintain even a steady
DR. The combination of these two demands for increased field (voltage) increases
likelihood of fracture, so any implementation of this type of algorithm must be
considered carefully.
For pulsed laser acquisition, constant apex field/temperature is desirable for high
data quality based on reasoning similar to that for constant flux. This type of control
requires an ability to monitor the apex temperature and an algorithm to interpret this
variable in order to modify pulse energy during acquisition. Because the evapora-
tion field required to reach a particular rate is a function of temperature, charge-
state ratios (CSRs) can provide an indirect indication of apex temperature [7]. The
sensitivity of any given CSR to the actual temperature (at the time of evaporation)
depends on the details of that particular ion species (rate of CSR change per rate of
field change) and its response as a function of temperature (rate of field change as a
function of temperature). A variety of theoretical and experimental behaviors have
been observed [8, 9]. Implementation of such an algorithm might be straightfor-
ward because the DR could be controlled by voltage and the laser pulse energy
could be controlled by CSR, so both controls are separate and decoupled; however,
some of the same dangers recognized for constant ion flux exist for constant CSR
mode applied to inhomogeneous systems. The particular ion species used to control
laser pulse energy needs to be present everywhere in the specimen; otherwise,
Appendix E: Additional Considerations for LEAP Operation 301
The LEAP is designed to deliver maximum LPE density to the tip apex by utilizing a
diffraction-limited laser spot (see Chap. 3) [10]. This approach avoids heating
unnecessary regions of the tip and optimizes heat transfer from the apex to the
bulk, thereby improving mass resolving power [10, 11]. This approach however is
not without design challenges. The small spot size requires accurate targeting of the
laser to the apexas the specimen moves, the laser must follow, and when the laser
moves or drifts it must be promptly repositioned back to the specimen apex. This
coordinated alignment is susceptible to various instrumental instabilities which
might include thermal expansion or contraction of various elements comprising
the laser path, positioning optics, specimen, or specimen stage and can cause the
laser position and/or apex position to drift on the order of a micron to a few microns
during a multi-hour acquisition. Sub-micrometer repositioning corrections are
required to keep the laser optimally aligned to the tip. A set of laser scanning
algorithms are available and designed for periodic tracking of the laser spot to the
tip apex as determined by the maximum observed DR.
Three scanning types are defined and can be modified by the user: Scout,
Standard, and Drift Comp scans (see Table E.2). The Scout scan is intended to
sweep large areas around a specimen to allow initial alignment of the laser spot to
the specimen apex. For the LCC 4.2, a 12 12 m2 scanning area with 0.4 m
resolution is a practical limit to the scan size. The number of points in any scan is
limited, and scanning with resolutions coarser than 0.4 m risks missing detection
of the maximum entirely. The Standard scan is intended to more accurately place
the laser at the maximum detection position by utilizing a smaller scanning area
which allows for finer scan resolution. Here the scan area is intended to include the
entire area over which emission from the specimen can be observed and is a
function of the actual tip dimensions coupled with the laser spot size. The overall
dimensions of this scan can vary as a function of specimen type, DR, and LPE, and
Table E.2 Default scans defined for LCC 4.2 (see Appendix H of the LEAP User Manual for
instrument-specific details on peak finding, thresholds, dwell time, and other laser scan
information)
Laser position scans (m) Scout Standard Drift Comp
Tip scan edge size 12 1.2 0.25
Tip scan step size 0.4 0.12 0.025
Focus scan depth 20 10 3
Focus scan step size 4 2 1
302 Appendix E: Additional Considerations for LEAP Operation
so some adjustment to the default settings may be warranted by the user. The Drift
Comp scan is intended for use during stable acquisition. In this mode only small
corrections in laser position are necessary, and it is not desirable to allow the DR to
be significantly reduced (any rate reduction likely causes a relative increase in
signal background); hence, the drift scan covers only a small scan area. Similar to
the Standard scan, the area over which the data rate is observed to change notice-
ably is specimen dependent, and so some tailoring of the scan range to specimen
type may be warranted. A simple guideline is for the Drift Comp scan edges
(corners) to dip to not more than ~50 % of the target DR and for the number of
scan steps to fall between 25 and 100 (5 510 10 array steps). The control
algorithm adjusts the dwell time for each step to attempt to ensure adequate
counting statistics, so at low DRs, scan times can become significant.
Besides coordination of the lateral laser position with the specimen apex, the
small focus depth of the laser requires coordination of the laser focus with
the specimen apex as well. The elongated depth of focus reduces the relative
sensitivity to motion along this direction, and so the correction is less significant.
For LEAP 3000X systems, a focus correction at the beginning of an analysis may
not need significant correction for many hours. LEAP 4000X systems have a lesser
depth of focus requiring more frequent focus corrections (~hourly).
Also shown in Table E.2 are the default settings for the LCC 4.2 Scout, Standard,
and Drift Comp focus scans. The Scout focus scan is intended to cover a wide range
of focus depth using the Scout aerial scan and might only be needed after a local
electrode change when the required position of the laser focus changes signifi-
cantly. The Standard focus scan uses the Standard aerial scan and is intended to
cover a more narrow range of focus depth. The Standard focus scan is suggested as
an initial scan at the beginning of an analysis that has significant and stable ion
emission. The Drift focus scan uses the Drift aerial scan and is intended to be quick
and provide only small adjustments to the laser focus position. This is the scan that
is used during automated acquisition. Note that not all materials run well under
default automated focus scan conditions.
Two notable features exist to aid the performance of the tip tracking algorithm:
Mass Filter Ranges and Extended Dwell Time are both defined on the LCC
Advanced Controls tab within the Laser Atom Probe tab. When scanning over a
specimen, a variety of emitted ion types can be observed. The goal of proper tip
tracking is for the laser to be positioned in the place that results in maximum in-time
specimen ion emission. Depending on the specimen structure, there may be places
on the specimen that will emit ions at greater intensities but are dominated by
out-of-synch or adsorbed ion species (the platinum weld on a lift-out specimen for
example). To avoid this, the user can define up to three ranges to use for calculating
DR during a laser scan. Default ranges exclude hydrogen and water ion peaks. The
Extended Dwell Time check box allows the user to triple the amount of time
devoted to individual laser steps. This may be necessary when scan statistics at
low DRs are insufficient to result in acceptable tip scanning. All of the user-selected
features on the Advanced Control tab can be saved, enabling customized scan sizes
for various specimen classes.
Appendix E: Additional Considerations for LEAP Operation 303
Fig. E.6 IVAS Wizard plot of laser position and ambient temperature for a 24-h microtip run
(top). The laser scanning frequency and size were insufficient to compensate for an external
temperature variation that occurred during the middle of the analysis (indicated by the accompa-
nying rise in voltagebottom plot). Higher scan frequency or larger area may be warranted.
Hourly focus scans reveal changes in focus of up to 1.5 m, and these are correlated to large lateral
laser shifts as well. For the 4000X, 1 m focus changes are significant. More frequent focus scans
may be warranted in this example
References
1. Lunger, M.J., Vergason, G.E.: Reducing pump-down time with dry air venting. Soc of
Vacuum Coaters 43rd Annual Technical Conference Proceedings, p 432 (2000)
2. Dylla, H.F., Manos, D.M., Lamarche, P.H.: Correlation of outgassing of stainless-steel and
aluminum with various surface treatments. J. Vac. Sci. Tech. 11(5), 26232636 (1993),
DOI:10.1116/1.578617
3. Fowler, R.H., Nordhiem, L.: Electron emission in intense electric fields. Paper presented at
the Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathe-
matical and Physical Character
4. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York, NY (2000)
5. Li, Y., Ang, K.H., Chong, G.C.Y.: Patents, software and hardware for PID control - an
overview and analysis of the current art. IEEE Contr. Syst. Mag. 26(1), 4254 (2006)
6. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer Series
in Materials Science, vol. 160. Springer (2012)
7. Kingham, D.R.: The post-ionization of field evaporated ions: a theoretical explanation of
multiple charge states. Surf. Sci. 116, 273301 (1982)
8. Tsong, T.T.: Field ion image formation. Surf. Sci. 70, 211233 (1978)
9. Southworth, H.N., Ralph, B.: J. Microsc., 90 (1969)
10. Bunton, J.H., Olson, J.D., Lenz, D.R., Larson, D.J., Kelly, T.F.: Optimized laser thermal
pulsing of atom probe tomography: LEAP 4000X. Microsc. Microanal. 16(Suppl 2), 1011
(2010)
11. Bunton, J.H., Olson, J.D., Lenz, D.R., Kelly, T.F.: Advances in pulsed-laser atom probe:
instrument and specimen design for optimum performance. Microsc. Microanal. 13, 418427
(2007)
Glossary
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 305
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
306 Glossary
Background Recorded event in the mass spectrum which do not originate from
the specimen or are not found within a peak (not correlated to a voltage or a laser
pulse).
Backside Specimen preparation orientation such that the analysis proceeds from
below the original surface and toward the exposed surface (opposite of top-side
or top-down analysis).
Bakeout The process of heating the LEAP instrument to approximately 150
C.
A bakeout is performed after the LEAP vacuum chambers have been exposed to
air and assists in achieving high or ultrahigh vacuum.
Base pressure The standard operating pressure within a particular vacuum cham-
ber in the LEAP instrument.
BCC Body-centered cubic.
Bgd See background.
Bias voltage The DC voltage applied to the specimen during data acquisition.
Bowl correction A time-of-flight correction to retain the relationship that all ions
of the same mass and charge record the same time flight for all locations on the
detector.
Buffer chamber A vacuum chamber between the load-lock and analysis cham-
bers. This is the second of two chambers in which a carousel may be located and
the chamber from which specimen and local electrode pucks are stored for use.
Calibration specimen A standard specimen (often aluminum or doped silicon)
that produces consistent data quality when the LEAP system is operating
properly such that it can be used to monitor system performance.
CAMECA ROOT A software application developed within the framework of
ROOT, a Lesser General Public License object-oriented framework for data
analysis developed by the European organization CERN.
CAMECA SAS A company based in Gennevilliers, France, that produces scien-
tific instrumentation including SIMS, EPMA, LEXES, and APT. CAMECA was
originally an acronym for Compagnie des Applications Mecaniques et
Electroniques au Cinema et a lAtomistique but has been out of the film business
since the late 1960s. It has several subsidiaries, including CAMECA Instru-
ments, Inc., Madison, WI-based headquarters for CAMECA in North America.
Capping Deposition of sacrificial material on top of a sample to protect it during
specimen preparation.
Carousel A carousel is a platter onto which is loaded either local electrode pucks
or specimen pucks. The carousel resides in the load-lock or storage chamber.
Carousel heater An optional load-lock upgrade that enables baking of local
electrodes and/or specimens in the load-lock chamber to rapidly degas them.
CERN Conseil Europeen Pour La Recherche Nucleaire, now referred to as the
European Laboratory for Particle Physics.
Client PC A desktop computer loaded with the LCC software through which a
LEAP system operator controls LEAP functions such as data acquisition, spec-
imen-to-local electrode alignment, and ISdb functions.
Glossary 307
Co-evaporation When multiple ions evaporate from the specimen during a single
pulse.
Cold finger The part of the cryostat at which refrigeration is concentrated. The
cold finger is mechanically isolated from the specimen stage to reduce specimen
vibration during data acquisition. It is thermally connected to the specimen stage
by thermally conductive helium gas.
Complex ion An ion that contains more than one atom, also known as a molecular
ion.
Constant CSR A data collection mode where the laser power control algorithm
calculates and maintains a constant CSR for a species in the specimen.
Constant flux A data collection mode where the specimen voltage control algo-
rithm calculates and maintains a constant ion/nm2/s detection rate.
Cross Section Refers to preparing a specimen such that the analysis proceeds
parallel to the original exposed surface (also known as parallel orientation).
Cryogenic temperature controller This device, located in the Instrument rack,
controls the temperature of the cold finger that cools the specimen. The cold
finger has a constant cooling power; the cryogenic temperature controller adds
heat to the cold path with a resistive coil to regulate temperature.
CSR Charge-state ratio.
Da Daltonthe SI atomic mass unit (AMU).
Data acquisition See atom probe analysis.
DAVis Data acquisition and visualization system. The LEAP software component
that provides 2D, digital specimen analysis data to the workstation. This soft-
ware also provides run-time mass spectra and monitors specimen voltage, ion
detection rates, etc.
Delay line detector A PSD designed to provide high-speed resolution of the time
of flight and XY-coordinate encoding of a single or multiple ions. The DLD
consists of serpentine DLAs and timing circuits. A charge cloud produced by the
microchannel plates impinges on the DLD creating an electromagnetic pulse that
travels each direction to the ends of the delay line and out to the ATDC timing
circuits.
Delocalization A smoothing function allowing discrete data produced by the
LEAP to create the 3D concentration grid used to create surfaces in IVAS.
Deprocessing A specimen preparation step where unwanted regions are removed
from near the region of interest. Often used when preparing microelectronic
samples.
Desiccant A drying agent typically used in dry-box applications to prevent mois-
ture from accumulating on local electrodes and specimens in storage. Only dust-
free desiccants are recommended for local electrode storage.
Desorption Field evaporation of gas ions adsorbed onto the specimen surface.
Detector A device within the LEAP instrument that includes the MCP and DLD
and which detects ions released during data acquisition.
Detector space A coordinate system based on the curved specimen surface
projected onto the flat detector and is displayed in live time in DAVis. Detector
308 Glossary
repelled from the tip. The curvature of the surface near the tip causes a natural
magnificationions are repelled in a direction roughly perpendicular to the
surface, producing a point projection effect. A DLD or a phosphor screen is
used to visualize the ions.
FIM See field ion microscopy.
Flash See specimen fracture.
Flat-test The procedure to test the quality of a local electrode.
Flat-top microtip array A silicon coupon with an array of blunt-topped specimen
posts. These coupons provide an efficient way to do precise and accurate FIB
lift-out specimen preparation techniques.
Flight path The space inside the atom probe that ions pass through when they
leave the specimen and travel to the detector.
Focused ion beam A specimen preparation and milling technique typically using
gallium as the focused ion medium to micro-machine samples into specimens.
FOV Field of view.
Fracture See specimen fracture.
FW0.01M Full width 1/100th maxa measure of a spectral peak width or mass
resolving power.
FW0.1M Full width 1/10th maxa measure of a spectral peak width or mass
resolving power.
FWHM Full width half maxa measure of a spectral peak width or mass resolv-
ing power.
HAADF High-angle annular dark fieldan imaging technique and detector type
used in TEM.
Hit event A hit event is a per-pulse detection of an ion on any of the detector
channels in the LEAP microscope.
Horizontal transfer rod The manually operated, magnetically coupled, horizon-
tal rod that moves a specimen or a local electrode puck from the carousel in the
buffer chamber to the analysis chamber.
HR The LEAP model including the large-angle reflectron flight path.
HVAC Heating, ventilation, and air conditioning.
IEC See interface evaporation control.
Image correction factor A variable in a large field-of-view reconstruction to
correct for the fact that ions off of the center of the specimen apex do not travel
in straight lines toward the detector, but rather ions leaving from locations closer
to the specimen shank are increasingly compressed toward the center of the
specimen apex.
Imago The proper noun for, and an acceptable truncation of, Imago Scientific
Instruments Corporation. Acquired by AMETEK, Inc. Imago is now part of
CAMECA Instruments, Inc.
Instrument flange The front flange on the LEAP instrument Analysis chamber,
upon which is mounted the specimen stage, detector, pulser electronics, etc.
310 Glossary
LEAP microscope The part of the LEAP system that is mounted on a chassis,
contains vacuum chambers and other equipment and instruments, and from
which specimen data originates. A CAMECA product designed to analyze
specimens at the atomic scale.
LEAP system schematic The LEAP system schematic is a digital, pictorial, user
interface in the LCC software representing the functions and features of the
LEAP microscope, from which parts of the microscope are controlled.
LEAP system server See LSS.
LEAPnet A LEAP ethernet network that allows the workstation to communicate
with the instrument-rack PC, which, in turn, communicates with other instru-
ments in both the rack and on the LEAP instrument.
Load-lock chamber This chamber is the chamber into which you place carousels
and is the first step in loading specimens and local electrodes into the instrument.
Local electrode The cone-shaped piece of metal mounted on a puck and posi-
tioned directly opposite the specimen in the analysis chamber. A brief, high-
voltage pulse, supplied by the high-voltage pulser, which is applied to the local
electrode during a voltage pulsed atom probe experiment. The local electrode
improves the mass resolution, field of view, and acquisition speed of atom probe
analysis and simplifies specimen preparation for atom probe analysis.
Local electrode puck The local electrode puck provides a mounting surface for
the local electrode within the LEAP instrument. See Local Electrode.
LPE Laser pulse energy.
LPS Laser positioning server. The LEAP software component that controls laser
position.
LSS LEAP system server. The LEAP software component that allows all LEAP
software and components to communicate and interact.
m/n Mass-to-charge-state ratio, also expressed as m/c, m/e, m/q, m/z, and u/e.
Mass-to-charge-state ratio The mass-to-charge-state ratio of an ion is the ions
mass (in mass units) divided by its charge state (in charge units). The mass-to-
charge-state ratio is on the x-axis of a mass spectrum and is indicated in SI units
of Daltons (Da).
MCP See microchannel plate.
Metadata Metadata is information a user enters into a database and that can be
used for sorting and searching the database. Adding specimens to the ISdb often
requires adding to the database metadata about the specimen.
Microchannel plate Part of the atom probe detector system. The MCP is a
transducer/amplifier that amplifies a single, charged species by approximately
1 106. Gain is controlled by the voltage difference between the front and back
plate.
Microchannel plate amplifier See microchannel plate.
Micro-positioning stage controller An electronic component in the instrument
rack that controls the micro-positioning motors in the specimen stage. These
motors move the specimen relative to the fixed position of the local electrode.
312 Glossary
Roughing pump Roughing pumps are mechanical pumps for initial pumping of
the load-lock and buffer chambers. Roughing pumps are required before apply-
ing the turbo pumps, the latter capable of achieving ultrahigh vacuum (UHV).
RRAW The raw data file: RRAW files contain the raw detector data and the order
of ion collection from the LEAP as well as the history of various parameters
measured or controlled throughout the acquisition.
RRNG ASCII format files that describe how mass ranges are associated with
particular ion types.
Sample A sample is a representative part or a single item from a larger whole or
group. For example, a sample is obtained from a wafer, and a specimen is
prepared from a sample.
Sampling Taking a random subset of the collected information to increase calcu-
lation or display speeds. Typically, reconstruction calculations and charts in the
IVAS wizard are limited to 500,000 ions or fewer.
SDM Spatial distribution map.
Shank angle A measure of specimen taper.
SIMS Secondary ion mass spectrometry.
Site specific Specimen preparation that targets a specific ROI.
SNR Signal-to-noise ratio.
Specimen The needle-shaped object of interest from which the LEAP microscope
attains an image.
Specimen analysis See atom probe analysis.
Specimen database A LEAP software component where all specimens, local
electrodes, and run conditions are recorded. See ISdb.
Specimen DB See ISdb.
Specimen failure See specimen fracture.
Specimen fracture A specimen fracture occurs when a specimen mechanically
fails. Many common terms have been invented to describe failure including
flash, rupture, failure, or zot.
Specimen puck See puck.
Specimen stage That part of the LEAP hardware, located inside the analysis
chamber and part of the Instrument flange assembly, to which specimen and
local electrode pucks are mounted.
Specimen stage controller See micro-positioning stage controller.
SRM Standard reference material.
Stage controller See micro-positioning stage controller.
Storage chamber See buffer chamber.
STR STR files contain the raw detector data and the order of ion collection from
the LEAP as well as the history of various parameters measured or controlled
throughout the acquisition. They are used in the latest LEAP systems to create
RHIT files.
Stub A tube of metal, usually copper or nickel, which can accept wire geometry
specimens and can be secured in a specimen puck. Also known as crimps or
tubes.
314 Glossary
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 315
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
316 Index
E I
Efficiency. See Detection efficiency IEC. See Interface Evaporation Control (IEC)
Electron emission, 3, 5, 64, 66, 106 Image
Electropointer, 29 compression (ICF), 120, 123, 134, 136,
Electropolishing, 19, 2529, 49, 201 137, 140, 141, 146148, 154, 156, 157,
Energy 267269
spread, 11, 16, 18, 59, 60 hump model, 139, 259262
Erosion parameter, 181, 182 Imaging atom probe (IAP), 7, 8
Evaporation Implant, 16, 3335, 39, 118,145, 149154, 156,
field, 15, 3436, 38, 40, 43, 8284, 90, 91, 157, 168, 235
96, 103, 122124, 133136, 138, 139, Initial radius, 125, 148
146, 148151, 155, 197, 233 Interface, 34, 40, 41, 43, 44, 81, 8385, 127,
in-time, 16, 66, 80, 164 139141, 148, 152155, 175, 176, 178,
out-of-time, 80, 86, 89, 90, 103, 164, 167 179, 202, 203, 208, 218, 223, 224, 227,
232, 234236
Interface Evaporation Control (IEC), 93, 101,
F 102, 299
Faceting, 221 Intergranular attack (IGA), 204207
Feature size, 175177 In-time. See Evaporation, in-time
FEEM. See Field electron emission microscope Iso-concentration, 1, 152154, 173, 175179,
(FEEM) 195, 196, 207, 220, 222, 226, 229, 233
Ferritin, 232233 Isosurface, 174, 178, 179
Ferroelectric, 216217
Field
emission, 3, 96, 225 K
enhancement, 10, 60, 63, 64 k-Factor, 122, 124, 127, 133, 136, 146, 148
evaporation, 15, 55, 64, 79, 134136,
258264
ionization, 5, 5557, 73 L
of view (FOV), 1, 58, 112, 218 Laser positioning control (drift, focus, scan),
Field electron emission microscope (FEEM), 75, 98105, 301303
35 Launch angle, 267270
Field ion microscope (FIM), 46, 69, 72, 257 LE. See Local Electrode (LE)
File formats, 110, 249256 LED. See Light Emitting Diode (LED)
FIM. See Field ion microscope (FIM) Level,
Fin-shaped field-effect transistor (FinFET), critical, 167, 170, 171
3739, 235238 detection, 166, 170, 171
Flat test, 98, 290294 quantification, 168, 170, 171
Flight, 6, 7, 16, 60, 67, 113 Lift-out, 16, 17, 19, 32, 33, 3638, 4143,
path, 7, 18, 56, 5861, 66, 88, 110, 45, 4749, 65, 158, 201, 213, 224,
113115, 133, 146, 147 225, 231
Fracture, 64, 75, 81, 83, 92, 98, 104 Light emitting diode (LED), 223226
Friends-of-friends, 180 Loadlock, 7072, 74
Local electrode (LE)
care, 6971, 106, 290294
G general, 2, 3, 1016, 49, 55, 5866, 91,
Gaussian, 59, 165, 166, 174176, 178 92, 96104, 106, 112, 122, 123,
Geologic material, 210217 148, 157
Grid, 142, 144, 154, 163, 173177, 186 Low energy cleanup, 3840, 43, 150
Index 317
P S
Parallel. See Analysis direction Sampling, 119, 128
Parallel-orientation analysis deprocessing, Scan size. See Laser positioning control (drift,
40, 84 focus, scan)
Passivation, 208 Scanning. See Laser positioning control (drift,
Patterson function, 186 focus, scan)
Peak identification, 117, 118 Scout scan. See Laser positioning control (drift,
Piezoelectric, 73, 216217 focus, scan)
Pile-up, 171, 173 SDM, neighborhood, 130, 188, 189, 194
Plasma, 292294 Segregation, 179, 204, 227, 234, 236
Poisson, 166 Shank angle, 27, 46, 6264, 90, 121124, 128,
Poschenrieder lens, 7 137, 140, 141, 146155, 278279
Position-sensitive atom probe (PoSAP), 3, Shape (apex, hemispherical, non-spherical, non-
810 hemispherical, geometry), 10, 11, 29, 30,
Position sensitive detector, 8, 56, 60, 67, 39, 47, 6266, 70, 106, 124, 127, 133,
68, 132 135, 138, 139, 142, 149, 150, 154, 155
318 Index