IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 66, NO.
8, AUGUST 2019
The Resistivity Size Effect in Epitaxial Nb(001)
and Nb(011) Layers
Erik Milosevic, Sit Kerdsongpanya, Mary E. McGahay, Baiwei Wang, and Daniel Gall
Abstract — Epitaxial Nb(011) and Nb(001) layers are sput-
ter deposited onto a -plane and r -plane sapphire substrates,
respectively, and their resistivity ρ measured in situ, ex situ,
and at 77 K as a function of layer thickness d = 4–400 nm.
The resistivity increase with decreasing d is independent
of layer orientation and is described with the model by
Fuchs and Sondheimer (FS), providing a value for the bulk
electron mean free path λ = 20 ± 2 nm at room temperature.
Exposure to air causes a 1.5-nm-thick surface oxide and
an increase in ρ by up to 74%, suggesting a decrease
in the surface scattering specularity from p1 = 0.9 ± 0.1
at the Nb-vacuum interface to completely diffuse scattering
(p1 = 0) at the oxidized Nb surface. Alternatively, this
increase in resistance can be attributed to roughening
during surface oxidation while retaining completely diffuse
scattering, yielding a lower bound for the room-temperature
λ of 9.0 ± 0.4 nm. The product of the bulk resistivity ρo times
λ is temperature-independent and, depending on either
choosing the roughness or the specularity interpretation,
ρo λ = 14 × 10−16 or 30 × 10−16 m2 , respectively. These
values are 3.9 and 8.5 times larger than ρo λ from a previous
theoretical prediction, indicating a dramatic break down of
the classical FS model for Nb and indicating that the resis-
tivity size effect in Nb is considerably larger than predicted
earlier. They are also larger than for W, Ru, and Co, making
Nb not promising for high-conductivity narrow interconnect
lines.
Index Terms — Alternative metals, back end of
line (BEOL), interconnects, niobium, middle of line (MOL),
mean free path, resistivity scaling, surface scattering.
I. I NTRODUCTION
T HE electrical resistivity of metallic conductors increases
with decreasing dimensions [1], [2]. This size effect is
a major limiting factor in the performance of current and
next-generation interconnects [3], [4] and is due to electron
scattering at surfaces [5]–[11], grain boundaries [12]–[18], and
Manuscript received April 8, 2019; revised May 29, 2019 and
June 12, 2019; accepted June 19, 2019. Date of publication July 8,
2019; date of current version July 23, 2019. This work was supported
in part by the Semiconductor Research Corporation (SRC) under
Grant 2881, in part by the NY State Empire State Development’s
Division of Science, Technology and Innovation (NYSTAR) through
the Focus Center-NY–RPI under Contract C150117, and in part
by NSF under Grant 1740271 and Grant 1712752. The review
of this paper was arranged by Editor R. Wang. (Corresponding author:
Daniel Gall.)
The authors are with the Department of Materials Science and Engi-
neering, Rensselaer Polytechnic Institute, Troy, NY 12180 USA (e-mail:
galld@rpi.edu).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TED.2019.2924312
0018-9383 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Georgia Institute of Technology. Downloaded on February 02,2021 at 20:11:11 UTC from IEEE Xplore. Restrictions apply.
3473
is exacerbated by surface roughness [19]–[23]. The resistance
contribution due to surface scattering is commonly quantified
using the semiclassical model by Fuchs and Sondheimer (FS)
[24], [25]. The FS model predicts a resistivity increase that
is approximately proportional to ρo λ/d, where ρo is the bulk
resistivity, λ is the bulk mean free path, and d is the distance
between scattering interfaces, which is the thickness of a film
or the width of an interconnect line. Therefore, in the limit
of narrow wires (d → 0), metals with small ρo λ products
are expected to exhibit the largest conductivities [26], [27]
and are thus promising as interconnect materials for narrow
high-conductivity lines. Recent investigations have quantified
the resistivity size effect for various elemental metals including
W [28]–[30], Ru [31]–[33], and Co [33], [34] by determining
λ and ρo λ from measured ρ versus d curves. First-principles
simulations predict a relatively low product ρo λ = 3.56 ×
10−16 m2 for Nb [35]. This value is 40% smaller than for
Cu, and is also smaller than the predictions for W, Ru, and
Co [26], [35]. Thus, Nb is a promising candidate for narrow
interconnect lines, which motivates this study on measuring
the resistivity size effect in Nb.
In this paper, we report on the resistivity size effect in
epitaxial Nb(011) and Nb(001) films. We determine the elec-
tron mean free path and corresponding ρo λ product from
the thickness-dependence of the electrical resistivity measured
both in situ and ex situ at room temperature and at 77 K.
Epitaxial layers are used because the absence of grain bound-
aries removes the confounding effects from grain boundary
scattering, allowing a direct quantification of the resistivity
due to surface scattering. The measured ρ is independent of
film orientation, which is distinctly different from the reported
anisotropy in the resistivity size effect in W [28], [30]. The
ex situ measured resistivity is larger than if measured in
situ prior to air exposure, suggesting either a transition from
specular ( p1 = 0.9) to diffuse ( p1 = 0) electron scattering
at the Nb surface or significant surface roughening during
surface oxidation, yielding values of ρo λ = 30 × 10−16 or
14×10−16 m2 , which are both much larger than the previous
theoretical prediction.
II. P ROCEDURE
Nb(011) and Nb(001) layers were deposited by dc mag-
netron sputtering on Al2 O3 (112̄0) and Al2 O3 (1̄012) sub-
strates, respectively, in a three-chamber ultrahigh vacuum
system with a base pressure of 10−9 Torr [36], [37].
3474 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 66, NO. 8, AUGUST 2019

All substrates were cleaned in consecutive ultrasonic baths of

trichloroethylene, acetone, isopropanol, and deionized water.
Then, they were mounted onto a Mo holder with colloidal
silver paint and introduced into the vacuum system via a load
lock. Prior to deposition, all substrates were degassed at the
deposition temperature of 500 ◦ C for 1 h. Deposition was done
in 3 mTorr 99.999% pure Ar with a fixed power of 60 or
180 W to a 99.95% pure Nb target facing the substrate at
a distance of 30 cm, yielding a deposition rate of 0.23 or
0.7 nm/s. For both substrate orientations, deposition time was
varied to generate nominal Nb film thicknesses ranging from
4 to 400 nm. After deposition, the films were left inside
the vacuum chamber to cool down to room temperature for
12 h, followed by in situ transfer to an analysis chamber for
electrical transport measurements with an in situ linear four-
point-probe which was operated with a current of 1–100 mA. Fig. 1. Representative XRR results from a 5.0-nm-thick Nb(011)
Measurements at 77 K were taken with an ex situ four- layer (blue) and from a 35.4-nm-thick Nb(001) layer (red). Solid lines:
point-probe by submerging each sample in liquid nitrogen measured intensity. Dotted lines: curve fitting.
immediately after removal from the load-lock that was vented TABLE I
with dry N2 . Ex situ room-temperature resistivity measure- T HICKNESS AND R ESISTIVITY OF E PITAXIAL N B (011) AND N B (001)
L AYERS B EFORE (In Situ ) AND A FTER (Ex Situ ) A IR E XPOSURE
ments were taken with the same four-point-probe after sample
warm-up using a continuous flow of dry N2 gas to inhibit ice
formation.
X-ray diffraction (XRD) and X-ray reflectivity (XRR)
analyses were performed using a PANalytical X’pert PRO
MPD system with a Cu source and a parabolic mirror
yielding a parallel beam with a <0.055◦ divergence, and a
0.27◦ parallel plate collimator in front of a scintillation point
detector. XRR data was fit using the PANalytical X’Pert
Reflectivity software. For this purpose, the densities of Al2 O3
and Nb are fixed to the known values of 3.99 and 8.57 g/cm3,
respectively. The best fit to the XRR data was achieved with
a surface oxide with a density of 4.6 g/cm3, equal to the
known bulk Nb2 O5 density.

III. R ESULTS AND D ISCUSSION

All Nb layers, in this paper, are either epitaxial
Nb(011)/Al2O3 (112̄0) or Nb(001)/Al2O3 (1̄012) layers.
The epitaxy is confirmed with XRD methods [6], [32], [38],
[39] using a combination of θ –2θ scans, rocking curves and ϕ-
scans of the Nb 011 reflections. The Nb layer microstructure
and epitaxial relationships are similar to what has been
extensively reported for this material system [40], [41].
For example, both the Nb(011) and Nb(001) layers exhibit
an in-plane orientation such that Nb(1̄11)||Al2O3 (0001)
[42]–[48]. To accommodate this relationship in the Nb(001)
films, the Nb(001) planes are misoriented by 3◦ with respect
to the (1̄012) planes of the sapphire [47]. The measured
rocking-curve widths are 0.18◦ and 0.14◦ for the Nb
011 and 002 reflections of the 298-nm-thick Nb(011) and the
380-nm-thick Nb(001), respectively.
Fig. 1 shows representative XRR results from two samples:
A Nb(011) layer with a nominal thickness of dNb = 5 nm
(blue) and a dNb = 35-nm-thick Nb(001) layer (red). The
measured intensity is plotted as solid lines in a logarithmic
scale as a function of the scattering angle 2θ , while the
results from curve fitting are shown as dotted lines, offset by
a factor of 102 above their respective experimental data for
clarity. The curve from the Nb(011) film is best described by
a 5.0 ± 0.2 nm thick Nb layer covered by a 1.5 ± 0.4 nm
thick Nb2 O5 surface oxide, the known stable oxide of Nb at
room temperature [49], [50]. We determine the as-deposited
in situ film thickness dNb = 5.6 nm from the sum of the
measured 5.0 nm and the equivalent thickness of 0.6 nm of
metallic Nb contained within the measured 1.5-nm Nb2 O5
layer. The curve fitting also provides a value for the root-
mean-square (rms) roughness ω of the Nb2 O5 surface of 0.4
± 0.2 nm, as well as 0.4 ± 0.3 nm and 0.2 ± 0.1 nm for the
Nb2 O5 –Nb and Nb–Al2 O3 interfaces, respectively. Similarly,
the curve from the Nb(001) film with nominal dNb = 35 nm
is best described by a 35.4 ± 0.6 nm thick Nb layer with a
1.5 ± 0.5 nm Nb2 O5 surface oxide which corresponds to an
in situ thickness of dNb = 36.0 nm. Measured rms roughness
values are ω = 0.8 ± 0.3 nm, 0.9 ± 0.4 nm, and 0.2 ± 0.1 nm
for the Nb2 O5 surface, and the Nb2 O5 –Nb and Nb–Al2 O3
interfaces, respectively. The XRR analyses were completed
for all samples with dNb < 100 nm yielding the results that
are tabulated in Table I.

Authorized licensed use limited to: Georgia Institute of Technology. Downloaded on February 02,2021 at 20:11:11 UTC from IEEE Xplore. Restrictions apply.
MILOSEVIC et al.: RESISTIVITY SIZE EFFECT IN EPITAXIAL Nb(001) AND Nb(011) LAYERS
Fig. 2. Resistivity ρ of epitaxial Nb(011) (blue) and Nb(001) (red) layers
versus thickness d, measured in situ (solid circles) and ex situ (gray
circles) in vacuum and air at 295 K, and immersed in liquid N2 at 77 K
(inset, triangles). The lines are from data fitting using the FS model.
The XRR analyses indicate that the thickness of the Nb2 O5
surface oxide is similar for all samples regardless of Nb
thickness and orientation, with average oxide thickness values
of 1.8 ± 0.5 nm and 1.6 ± 0.1 nm for the Nb(011) and
Nb(001) films, respectively, where the stated error bar cor-
responds to the standard deviation. The layer thickness for the
thickest samples, dNb(011) = 298 nm and dNb(001) = 381 nm,
are determined from the deposition rate and time, since XRR
fringes cannot be resolved for dNb > 100 nm.
Fig. 2 shows a plot of the in situ and ex situ measured room-
temperature resistivity ρ of Nb(011) and Nb(001) layers as
a function of thickness dNb . The inset is the corresponding
plot at 77 K. All data sets show an increase in ρ with
decreasing dNb, which is attributed to electron surface scatter-
ing that becomes increasingly dominant with decreasing dNb.
The measured ρ for any given dNb is, within experimental
uncertainty, independent of layer orientation. This is distinctly
different from reports on W(011) and W(001) layers where
the anisotropy leads to differences in ρ of up to a factor of
two, which can be directly and quantitatively attributed to the
anisotropy in the Fermi surface [28]. In contrast, we measure
no apparent anisotropy for the resistivity size effect in Nb,
which is not a priori evident from the Nb Fermi surface but
may be attributed to Nb having one electron less per atom than
W and therefore, exhibiting a more symmetric closed second-
zone hole surface centered at  [51]. Thus, in the following
discussion, we no longer distinguish between Nb(011) and
Nb(001) layers, since they effectively have the same resistivity.
The in situ measured resistivity of the thickest layers is
14.9 ± 1.2 μcm and 15.3 ± 0.9 μcm for dNb = 298
and 380 nm, respectively. This is in good agreement with the
reported Nb bulk resistivity of 15.2 μcm [52]. The resistivity
remains approximately constant as dNb is decreased to dNb =
97.9 and 100.2 nm, with values of ρ = 14.9 ± 0.6μcm and
14.6 ± 0.4 μcm, respectively, indicating that the resistiv-
ity contribution from electron-surface scattering is negligible
for dNb > 100 nm. However, as dNb is further decreased,
Authorized licensed use limited to: Georgia Institute of Technology. Downloaded on February 02,2021 at 20:11:11 UTC from IEEE Xplore. Restrictions apply.
3475
the in situ resistivity continuously increases to reach 31.3 ±
1.2 μcm and 33.2 ± 1.8 μcm at dNb = 3.8 and 4.2 nm.
These results are also summarized in Table I. The solid line
shown in Fig. 2 is obtained by fitting the in situ resistivity
data using the exact form of the semiclassical framework
of FS [24], [25]. A well-known challenge [7], [30], [32]
of this approach is that it does not allow for independent
determination of the surface scattering specularity p and the
bulk mean free path λ. More specifically, for any arbitrary
choice of p within the physical constraint 0 ≤ p ≤ 1, a value
for λ can be found with λmin ≤ λ ≤ ∞ such that the
model predicts a ρ versus d curve that matches the measured
data, where λmin is a lower bound for the mean free path
consistent with a specific ρ versus d data set. Correspondingly,
we determine λmin = 9.0 ± 0.4 nm from the in situ data by
fitting with the FS model using completely diffuse electron
scattering at both the upper and lower film surfaces ( p1 =
p2 = 0) [7] and fixing ρo to the reported bulk Nb resistivity
of 15.2 μcm. Independent fitting of the Nb(011) and Nb(001)
data sets yields λmin = 8.9 ± 0.4 nm and 9.0 ± 0.4 nm.
These values are the same within experimental uncertainty,
confirming the absence of anisotropy in the resistivity size
effect of Nb.
Comparing the in situ and ex situ data shown in Fig. 2
indicates no detectable effect of air exposure for dNb > 30 nm.
However, dNb < 30 nm yields an ex situ resistivity that is
considerably higher than the values measured in situ, as also
summarized in Table I. A similar increase in ρ upon air expo-
sure has previously been reported for epitaxial Cu(001) [5],
[6], [27], [53] and Co(0001) [34] layers and has been attributed
to the surface oxidation of Cu and Co which 1) causes
atomic-scale roughening of the conductive metal surface which
promotes diffusive electron scattering and 2) consumes a finite
metal thickness which is incorporated into the developing
oxide layer, effectively reducing the conductive cross-sectional
area. We note that our XRR data allows to directly account
for the latter effect, since the measured thicknesses of both
the Nb and the surface oxide provide, in turn, values for both
the ex situ Nb thickness excluding the oxide and the in situ
as-deposited Nb thickness. The former thickness is smaller
for all samples, as some of the conductive Nb is lost to the
developing surface oxide as also shown in Table I and reflected
in the ex situ data points in Fig. 2 which are shifted slightly
to the left with respect to the in situ data points. Therefore,
we attribute the remaining difference in the ex situ and in situ
resistivity to a decrease in surface scattering specularity upon
air exposure. To quantitatively study this effect, we first use
the FS model to describe the ex situ data, which yields the
dashed line in Fig. 2 and a λmin = 20 ± 2 nm. This value for
λmin corresponds to the mean free path that best describes the
ex situ data for completely diffuse scattering at the Nb–Nb2O5
interface. In a second step, we fit the in situ data using a fixed
λ = 20 nm and as a free fitting parameter the specularity p1
of the Nb-vacuum interface. This yields p1 = 0.9 ± 0.1 and
a curve that is indistinguishable from the solid black line in
Fig. 2. Therefore, the Nb resistivity data at 295 K is best
described by a constant mean free path of λ = 20 ± 2 nm and
a surface scattering specularity that transitions from nearly
3476
specular ( p1 = 0.9 ± 0.1) to completely diffuse ( p1 = 0)
upon air exposure. The corresponding product ρo λ is 30 ×
10−16 m2 and is 8.5 times larger than the first principles
prediction of 3.56 × 10−16 m2 from [35].
The inset of Fig. 2 shows the resistivity ρ of the Nb(011)
and Nb(001) films measured at 77 K as a function of thick-
ness d. Similar to above, our first step in analyzing the data
is curve fitting with the FS model using p1 = p2 = 0 and
ρo = 2.5 μcm, which is the reported bulk resistivity for Nb
at 77 K [54], [55]. This yields the dotted line in the inset and
a value for the lower bound for the mean free path at 77 K of
λmin = 70 ± 10 nm. The corresponding lower bound for the
product ρo λ is 17×10−16 m2 , which is 42% smaller than the
ρo λ at 295 K. Classical or semiclassical bulk electron transport
descriptions predict a temperature-independent ρo λ product
[12]. Thus, the lower ρo λ at 77 K suggests 1) a thickness-
dependent electron-phonon coupling [56], [57]; 2) a wave
vector dependence in the electron-phonon scattering cross
sections [28]; 3) a breakdown of the FS model in the limit of
small thickness and/or low temperature [29], [58], [59]; and/or
4) a finite specularity at the film-liquid nitrogen interface
[32]. To illustrate this last possible explanation, we fit the
low-temperature resistivity using a temperature-independent
ρo λ = 30 × 10−16 m2 from the room-temperature ex situ
measurements but allow the Nb film surface specularity p1
at 77 K to vary as a free fitting parameter. This results in
p1 = 0.8 and a curve that describes well the low-temperature
data and is nearly identical to the curve shown in the inset.
Thus, all three data sets (in situ, ex situ, 77 K) can be
simultaneously described with the FS model using a single ρo λ
product of 30×10−16 m2 and a surface scattering specularity
that decreases from p1 = 0.9 for the Nb-vacuum interface to
p1 = 0.8 for the Nb-liquid N2 interface to p1 = 0 for the Nb
surface that has been exposed to air. We note that the dotted
line in the inset does not perfectly describe the data. More
specifically, the measured resistivity increases steeper at low
d than predicted from the FS model. This is consistent with
some other studies on W [29] and Cu [5] that indicate an
experimental ρ versus d dependence that exhibits a stronger
curvature than predicted by FS, suggesting the breakdown of
FS at small d, which may be attributed to the transition from
a 3-D to a 2-D transport description [59] but may also be
associated with the increasing importance of surface roughness
at decreasing d [22].
Now, we consider an alternative interpretation for the
three resistivity data sets, using a direct surface roughening
argument instead of a decreasing surface specularity during
air exposure. For this purpose, we envision an as-deposited
relatively smooth Nb surface which roughens upon air
exposure, leading to the measured increase in ρ, while the
surface scattering is assumed to remain completely diffuse for
all samples such that the mean free path corresponds to the
lower bound of λ = 9.0 ± 0.4 nm determined from the in situ
data. The Nb2 O5 –Nb interface roughness develops due to the
oxidation during air exposure and is determined from the XRR
measurements. The resistivity increase due to this roughness
is described using the step edge model in [22], which predicts
a resistivity contribution that is proportional to the rms
Authorized licensed use limited to: Georgia Institute of Technology. Downloaded on February 02,2021 at 20:11:11 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 66, NO. 8, AUGUST 2019
TABLE II
P RODUCT OF THE B ULK R ESISTIVITY T IMES THE B ULK M EAN F REE
PATH ρo λ F ROM F IRST-P RINCIPLE P REDICTIONS [26], [35] AND
M EASURED FOR C U [7], W [28], C O [33], [34],
R U [32], [33], AND N B ( THIS PAPER )
roughness ω divided by the lateral correlation length ξ of
the surface morphology. ξ of the buried Nb2 O5 –Nb interface
cannot be determined from our XRR data. Thus, instead,
we determine ξ that is required such that the increase in ρ
upon air exposure can be attributed to the measured roughness
and find ξ = 1.0, 1.0, 1.2 nm for the Nb(011) layers with
thickness dNb = 3.3, 5.0, 10.7 nm and ξ = 0.9, 1.1, 1.5 nm
for the Nb(001) layers with thickness dNb = 3.6, 4.9, 10.2 nm.
These values are obtained with the step edge model using a
Nb ballistic conductance go = 3/(4ρo λ) = 5.5×1014 −1 m−2
estimated from the free electron approximation. Their average
is ξ = 1.0 ± 0.2 nm, which is similar to the thickness of
the Nb2 O5 surface oxide of 1.6–1.8 nm and can be attributed
to surface oxide nucleation followed by isotropic growth
leading to oxide surface grains with 1–2 nm diameters.
The resistivity at 77 K can similarly be explained using
the roughness model. For this purpose, we use a constant
lower-bound ρo λ product of 14 × 10−16 m2 determined
from the in situ room-temperature measurements and find
an average ω/ξ = 0.06 ± 0.05 at 77 K. This value is
nearly zero, suggesting a negligible surface oxidation during
the brief air exposure between sample removal from the
deposition system and sample submerging in liquid nitrogen.
Applying the same analysis (backward) with a constant
ρo λ = 14 × 10−16 m2 to the room temperature in situ
and ex situ data sets yields ω/ξ = 0.04 ± 0.02 and 0.38
± 0.05, respectively. Thus, in summary, all three data sets
can be explained with completely diffuse surface scattering
and a room temperature λ = 9.0 ± 0.4 nm corresponding
to a temperature-independent ρo λ = 14 × 10−16 m2 , and
a transition from a smooth Nb surface both in situ and in
liquid nitrogen to an ex situ oxidized surface with a rough
Nb2 O5 –Nb interface with 1.0 ± 0.2 nm wide mounds.
We note that the two “competing” interpretations of
the measured resistivity data using changes in scattering
specularity and surface roughness are not mutually exclusive
to one another. That is, the explanations represent two
extremes and a combination of a smaller loss of specularity
with a smaller increase in scattering from surface roughness
can just as adequately describe the measured resistivities.
This yields a range of possible ρo λ products from 14 × 10−16
to 30 × 10−16 m2 . This range for ρo λ is significantly larger
than the first principles prediction of 3.56 × 10−16 m2
from [35] as well as the reported ρo λ products for Cu
(6.7 × 10−16 m2 ) [7], W(110) (10.1 × 10−16 m2 ) [28],
MILOSEVIC et al.: RESISTIVITY SIZE EFFECT IN EPITAXIAL Nb(001) AND Nb(011) LAYERS
Co(0001) (12.2 × 10−16 m2 ) [33], [34], and Ru(0001)
(5.06 × 10−16 m2 ) [32], [33], as also summarized in
Table II. Thus, our measured ρ versus d data indicates that
the resistivity size effect in Nb is much stronger than the
first-principles calculations suggest. It is also larger than the
size effect of all competing interconnect metals listed above,
such that we conclude that Nb is not a promising metal for
high-conductivity narrow interconnect lines.
IV. C ONCLUSION
Epitaxial Nb(011) and Nb(001) layers are grown on a-plane
and r -plane sapphire substrates. The XRR analyses indicate
a 1.5-nm-thick native oxide layer that forms as a result of
air exposure. The resistivity measured in situ, ex situ, and at
77 K can be described by (1) a ρo λ product of 30×10−16 m2
which corresponds to a room temperature λ = 20 ± 2 nm, and
electron surface scattering that is completely diffuse ( p = 0)
at the ex situ oxidized Nb surface, but is partially specular
at the pristine in situ Nb-vacuum ( p = 0.9 ± 0.1) and the
Nb-liquid nitrogen ( p = 0.8) interfaces, (2) a ρo λ product of
14 × 10−16 m2 with a corresponding room temperature
λ = 9.0 ± 0.4 nm, completely diffuse surface scattering
at the Nb surface, and a roughness that has a negligible
resistivity contribution in situ and in liquid nitrogen, but
increases significantly during air exposure to cause a resistivity
increase, or (3) a combination of (1) and (2). The resulting
ρo λ ranges from 14 × 10−16 to 30 × 10−16 m2 which is
3.9 to 8.5 times larger than previously predicted from first
principles. Correspondingly, the resistivity size effect in Nb is
considerably larger than in other metals that are considered
for narrow interconnect lines, rendering Nb not promising for
interconnect applications.
R EFERENCES
[1] S. M. Rossnagel and T. S. Kuan, “Alteration of Cu conductivity in the
size effect regime,” J. Vac. Sci. Technol. B, Microelectron. Nanometer
Struct. Process., Meas., Phenomena, vol. 22, no. 1, p. 240, Jan. 2004.
doi: 10.1116/1.1642639.
[2] J. J. Thomson, “On the theory of electric conduction through thin
metallic films,” Proc. Cambridge Philos. Soc. , vol. 11, pp. 120–123,
1901.
[3] J. Kelly et al., “Experimental study of nanoscale Co damascene
BEOL interconnect structures,” in Proc. IEEE Int. Interconnect Technol.
Conf. /Adv. Metallization Conf. (IITC/AMC), May 2016, pp. 40–42.
doi: 10.1109/IITC-AMC.2016.7507673.
[4] J. S. Chawla et al., “Resistance and electromigration performance of
6 nm wires,” in Proc. IEEE Int. Interconnect Technol. Conf. /Adv. Met-
allization Conf. (IITC/AMC), May. 2016, pp. 63–65. doi: 10.1109/IITC-
AMC.2016.7507682.
[5] P. Y. Zheng, R. P. Deng, and D. Gall, “Ni doping on Cu sur-
faces: Reduced copper resistivity,” Appl. Phys. Lett., vol. 105, no. 13,
Sep. 2014, Art. no. 131603. doi: 10.1063/1.4897009.
[6] P. Zheng, T. Zhou, and D. Gall, “Electron channeling in TiO2 coated
Cu layers,” Semicond. Sci. Technol., vol. 31, no. 5, Mar. 2016,
Art. no. 055005. doi: 10.1088/0268-1242/31/5/055005.
[7] J. S. Chawla, F. Gstrein, K. P. O’Brien, J. S. Clarke, and D. Gall,
“Electron scattering at surfaces and grain boundaries in Cu thin films
and wires,” Phys. Rev. B, Condens. Matter, vol. 84, no. 23, Dec. 2011,
Art. no. 235423. doi: 10.1103/PhysRevB.84.235423.
[8] J. S. Chawla and D. Gall, “Specular electron scattering at single-crystal
Cu(001) surfaces,” Appl. Phys. Lett., vol. 94, no. 25, Jun. 2009,
Art. no. 252101. doi: 10.1063/1.3157271.
Authorized licensed use limited to: Georgia Institute of Technology. Downloaded on February 02,2021 at 20:11:11 UTC from IEEE Xplore. Restrictions apply.
3477
[9] E. Milosevic and D. Gall, “Copper interconnects: Surface state
engineering to facilitate specular electron scattering,” IEEE Trans.
Electron Devices, vol. 66, no. 6, pp. 2692–2698, Jun. 2019.
doi: 10.1109/TED.2019.2910500.
[10] F. Zahid, Y. Ke, D. Gall, and H. Guo, “Resistivity of thin Cu films coated
with Ta, Ti, Ru, Al, and Pd barrier layers from first principles,” Phys.
Rev. B, Condens. Matter, vol. 81, no. 4, Jan. 2010, Art. no. 045406.
doi: 10.1103/PhysRevB.81.045406.
[11] J. M. Purswani and D. Gall, “Electron scattering at single crystal Cu
surfaces,” Thin Solid Films, vol. 516, nos. 2–4, pp. 465–469, Dec. 2007.
doi: 10.1016/j.tsf.2007.07.146.
[12] A. F. Mayadas and M. Shatzkes, “Electrical-resistivity model for poly-
crystalline films: The case of arbitrary reflection at external surfaces,”
Phys. Rev. B, Condens. Matter, vol. 1, no. 4, pp. 1382–1389, Feb. 1970.
doi: 10.1103/PhysRevB.1.1382.
[13] M. César, D. Gall, and H. Guo, “Reducing grain-boundary resistivity of
copper nanowires by doping,” Phys. Rev. Appl., vol. 5, no. 5, May 2016,
Art. no. 054018. doi: 10.1103/PhysRevApplied.5.054018.
[14] M. César, D. Gall, and H. Guo, “Calculated resistances of single grain
boundaries in copper,” Phys. Rev. Appl., vol. 2, no. 4, Oct. 2014,
Art. no. 044007. doi: 10.1103/PhysRevApplied.2.044007.
[15] D. Choi, “The electron scattering at grain boundaries in tung-
sten films,” Microelectron. Eng., vol. 122, pp. 5–8, Jun. 2014.
doi: 10.1016/j.mee.2014.03.012.
[16] T.-H. Kim et al., “Large discrete resistance jump at grain boundary in
copper nanowire,” Nano Lett., vol. 10, no. 8, pp. 3096–3100, Jul. 2010.
doi: 10.1021/nl101734h.
[17] J. M. Rickman and K. Barmak, “Simulation of electrical conduc-
tion in thin polycrystalline metallic films: Impact of microstruc-
ture,” J. Appl. Phys., vol. 114, no. 13, Oct. 2013, Art. no. 133703.
doi: 10.1063/1.4823985.
[18] N. A. Lanzillo et al., “Defect and grain boundary scattering in tungsten:
A combined theoretical and experimental study,” J. Appl. Phys., vol. 123,
no. 15, Apr. 2018, Art. no. 154303. doi: 10.1063/1.5027093.
[19] Y. Namba, “Resistivity and temperature coefficient of thin metal films
with rough surface,” Jpn. J. Appl. Phys., vol. 9, no. 11, pp. 1326–1329,
Nov. 1970. doi: 10.1143/JJAP.9.1326.
[20] Y. Ke, F. Zahid, V. Timoshevskii, K. Xia, D. Gall, and H. Guo,
“Resistivity of thin Cu films with surface roughness,” Phys. Rev. B,
Condens. Matter, vol. 79, no. 15, Apr. 2009, Art. no. 155406.
doi: 10.1103/PhysRevB.79.155406.
[21] P. Y. Zheng, T. Zhou, B. J. Engler, J. S. Chawla, R. Hull, and D. Gall,
“Surface roughness dependence of the electrical resistivity of W(001)
layers,” J. Appl. Phys., vol. 122, no. 9, Sep. 2017, Art. no. 095304.
doi: 10.1063/1.4994001.
[22] T. Zhou, P. Zheng, S. C. Pandey, R. Sundararaman, and D. Gall, “The
electrical resistivity of rough thin films: A model based on electron
reflection at discrete step edges,” J. Appl. Phys., vol. 123, no. 15,
Apr. 2018, Art. no. 155107. doi: 10.1063/1.5020577.
[23] V. Timoshevskii, Y. Ke, H. Guo, and D. Gall, “The influence of surface
roughness on electrical conductance of thin Cu films: An ab initio
study,” J. Appl. Phys., vol. 103, no. 11, Jun. 2008, Art. no. 113705.
doi: 10.1063/1.2937188.
[24] K. Fuchs, “The conductivity of thin metallic films according to the
electron theory of metals,” Math. Proc. Cambridge Philos. Soc., vol. 34,
no. 1, pp. 100–108, Jan. 1938. doi: 10.1017/S0305004100019952.
[25] E. H. Sondheimer, “The mean free path of electrons in
metals,” Adv. Phys., vol. 1, no. 1, pp. 1–42, Jan. 1952.
doi: 10.1080/00018735200101151.
[26] D. Gall, “Electron mean free path in elemental metals,” J. Appl. Phys.,
vol. 119, no. 8, Feb. 2016, Art. no. 085101. doi: 10.1063/1.4942216.
[27] J. S. Chawla and D. Gall, “Epitaxial Ag(001) grown on MgO(001)
and TiN(001): Twinning, surface morphology, and electron surface
scattering,” J. Appl. Phys., vol. 111, no. 4, Feb. 2012, Art. no. 043708.
doi: 10.1063/1.3684976.
[28] P. Zheng and D. Gall, “The anisotropic size effect of the electrical
resistivity of metal thin films: Tungsten,” J. Appl. Phys., vol. 122, no. 13,
Oct. 2017, Art. no. 135301. doi: 10.1063/1.5004118.
[29] D. Choi, X. Liu, P. K. Schelling, K. R. Coffey, and K. Barmak, “Failure
of semiclassical models to describe resistivity of nanometric, polycrys-
talline tungsten films,” J. Appl. Phys., vol. 115, no. 10, Mar. 2014,
Art. no. 104308. doi: 10.1063/1.4868093.
[30] D. Choi et al., “Electron mean free path of tungsten and the electrical
resistivity of epitaxial (110) tungsten films,” Phys. Rev. B, Condens.
Matter, vol. 86, no. 4, Jul. 2012, Art. no. 045432. doi: 10.1103/Phys-
RevB.86.045432.
3478
[31] S. Dutta et al., “Thickness dependence of the resistivity of
platinum-group metal thin films,” J. Appl. Phys., vol. 122, no. 2,
Jul. 2017, Art. no. 025107. doi: 10.1063/1.4992089.
[32] E. Milosevic, S. Kerdsongpanya, A. Zangiabadi, K. Barmak,
K. R. Coffey, and D. Gall, “Resistivity size effect in epitaxial Ru(0001)
layers,” J. Appl. Phys., vol. 124, no. 16, Oct. 2018, Art. no. 165105.
doi: 10.1063/1.5046430.
[33] E. Milosevic, S. Kerdsongpanya, and D. Gall, “The resistivity size
effect in epitaxial Ru(0001) and Co(0001) layers,” in Proc. IEEE
Nanotechnol. Symp. (ANTS), Nov. 2018, pp. 1–5. doi: 10.1109/NAN-
OTECH.2018.8653560.
[34] E. Milosevic, S. Kerdsongpanya, M. McGahay, A. Zangiabadi, and
K. Barmak, “ELectron-surface scattering in epitaxial Co(0001) films,”
J. Appl. Phys., vol. 125, no. 24, Jun. 2019, Art. no. 245105. doi:
10.1063/1.5086458.
[35] D. Gall, “Metals for low-resistivity interconnects,” in Proc. IEEE
Int. Interconnect Technol. Conf. (IITC), Jun. 2018, pp. 157–159.
doi: 10.1109/IITC.2018.8456810.
[36] B. Wang, S. Kerdsongpanya, M. E. McGahay, E. Milosevic, P. Patsalas,
and D. Gall, “Growth and properties of epitaxial Ti1−x Mgx N(001)
layers,” J. Vac. Sci. Technol. A, Vac. Surf. Films, vol. 36, no. 6,
Nov. 2018, Art. no. 061501. doi: 10.1116/1.5049957.
[37] J. S. Chawla, X. Y. Zhang, and D. Gall, “Effective electron mean free
path in TiN(001),” J. Appl. Phys., vol. 113, no. 6, 2013, Art. no. 063704.
doi: 10.1063/1.4790136.
[38] R. Deng, B. D. Ozsdolay, P. Y. Zheng, S. V. Khare, and D. Gall, “Optical
and transport measurement and first-principles determination of the ScN
band gap,” Phys. Rev. B, Condens. Matter, vol. 91, no. 4, Jan. 2015,
Art. no. 045104. doi: 10.1103/PhysRevB.91.045104.
[39] P. Zheng, B. D. Ozsdolay, and D. Gall, “Epitaxial growth of tungsten
layers on MgO(001),” J. Vac. Sci. Technol. A, Vac. Surf. Films, vol. 33,
no. 6, Nov. 2015, Art. no. 061505. doi: 10.1116/1.4928409.
[40] J. L. Du et al., “What determines the interfacial configuration of
Nb/Al2 O3 and Nb/MgO interface,” Sci. Rep., vol. 6, Oct. 2016,
Art. no. 33931. doi: 10.1038/srep33931.
[41] A. R. Wildes, J. Mayer, and K. Theis-Bröhl, “The growth and structure
of epitaxial niobium on sapphire,” Thin Solid Films, vol. 401, nos. 1–2,
pp. 7–34, 2001. doi: 10.1016/S0040-6090(01)01631-5.
[42] J. Mayer, W. Mader, D. Knauss, F. Ernst, and M. Rühle, “Struc-
tures of Nb/Al2 O3 interfaces produced by different experimental
routes,” MRS Online Proc. Library Arch., vol. 183, p. 55, Jan. 1990.
doi: 10.1557/PROC-183-55.
[43] G. Gutekunst, J. Mayer, and M. Rühle, “Atomic structure of epitaxial
Nb-Al2 O3 interfaces I. Coherent regions,” Philos. Mag. A, vol. 75, no. 5,
pp. 1329–1355, 1997. doi: 10.1080/01418619708209859.
[44] G. Gutekunst, J. Mayer, V. Vitek, and M. Rühle, “Atomic structure of
epitaxial Nb-Al2 O3 interfaces II. Misfit dislocations,” Philos. Mag. A,
vol. 75, no. 5, pp. 1357–1382, 1997. doi: 10.1080/01418619708209860.
[45] E. J. Grier, M. L. Jenkins, A. K. Petford-Long, R. C. C. Ward, and
M. R. Wells, “Misfit dislocations of epitaxial (110) niobium
− ) sapphire interfaces grown by molecular beam epitaxy,” Thin
(1120
Solid Films, vol. 358, no. 1, pp. 94–98, Jan. 2000. doi: 10.1016/S0040-
6090(99)00675-6.
Authorized licensed use limited to: Georgia Institute of Technology. Downloaded on February 02,2021 at 20:11:11 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 66, NO. 8, AUGUST 2019
[46] S. M. Durbin, J. E. Cunningham, and C. P. Flynn, “Growth of
single-crystal metal superlattices in chosen orientations,” J. Phys. F,
Metal Phys., vol. 12, no. 6, pp. L75–L78, Jun. 1982. doi: 10.1088/0305-
4608/12/6/001.
[47] S. Di Nunzio, K. Theis-Bröhl, and H. Zabel, “Epitaxial growth of
Co/Cu(001) superlattices on sapphire substrates,” Thin Solid Films,
vol. 279, nos. 1–2, pp. 180–188, Jun. 1996. doi: 10.1016/0040-
6090(95)08197-6.
[48] D. X. Li, P. Pirouz, A. H. Heuer, S. Yadavalli, and C. P. Flynn,
“The characterization of Nb-Al2 O3 and Nb-MgO interfaces in MBE
grown Nb-MgO-Nb-Al2 O3 multilayers,” Acta Metallurgica et Mater.,
vol. 40, pp. S237–S247, 1992. doi: 10.1016/0956-7151(92)90282-J.
[49] C. E. Sittig, M. Textor, N. D. Spencer, M. D. Wieland,
and P. H. Vallotton, “Surface characterization of implant materials
c.p. Ti, Ti-6Al-7Nb and Ti-6Al-4V with different pretreatments,”
J. Mater. Sci. Mater. Med., vol. 10, no. 1, pp. 35–46, 1999.
doi: 10.1023/A:1008840026907.
[50] I. Lindau and W. E. Spicer, “Oxidation of Nb as studied by
the uv-photoemission technique,” J. Appl. Phys., vol. 45, no. 9,
pp. 3720–3725, 1974. doi: 10.1063/1.1663849.
[51] M. H. Halloran, J. H. Condon, J. E. Graebner, J. E. Kunzier, and
F. S. L. Hsu, “Experimental study of the fermi surfaces of niobium and
tantalum,” Phys. Rev. B, Condens. Matter, vol. 1, no. 2, pp. 366–372,
Jan. 1970. doi: 10.1103/PhysRevB.1.366.
[52] W. M. Haynes, “Electrical resistivity of pure metals,” in CRC
Handbook of Chemistry and Physics. Boca Raton, FL, USA: CRC Press,
2014.
[53] J. S. Chawla, F. Zahid, H. Guo, and D. Gall, “Effect of O2 adsorption
on electron scattering at Cu(001) surfaces,” Appl. Phys. Lett., vol. 97,
no. 13, Sep. 2010, Art. no. 132106. doi: 10.1063/1.3489357.
[54] J. E. Jensen, W. A. Tuttle, R. B. Stewart, H. Brechna, and A. G. Prodell,
“Electrical resistivity of some metallic elements and commercial alloys,”
Brookhaven Natl. Lab. Sel. Cryog. Data Noteb., Upton, NY, USA,
Tech. Rep., 1980. [Online]. Available: https://www.bnl.gov/magnets/staff/
gupta/cryogenic-data-handbook/Section10.pdf
[55] L. A. Hall, “Survey of electrical resistivity measurements on 16 pure
metals in the temperature range 0 to 273 K,” Cryogenics Division, Inst.
Basic Standards, Nat. Bureau Standards, Boulder, CO, USA, NBS Tech.
Note 365, 1968, p. 20, vol. 365.
[56] S. Kim, H. Suhl, and I. K. Schuller, “Surface phonon scattering in the
electrical resistivity on Co/Ni superlattices,” Phys. Rev. Lett., vol. 78,
no. 2, pp. 322–325, Jan. 1997. doi: 10.1103/PhysRevLett.78.322.
[57] Y. P. Timalsina et al., “Evidence of enhanced electron-phonon coupling
in ultrathin epitaxial copper films,” Appl. Phys. Lett., vol. 103, no. 19,
Nov. 2013, Art. no. 191602. doi: 10.1063/1.4829643.
[58] Y. P. Timalsina et al., “Effects of nanoscale surface roughness on
the resistivity of ultrathin epitaxial copper films,” Nanotechnology,
vol. 26, no. 7, Jan. 2015, Art. no. 075704. doi: 10.1088/0957-4484/26/
7/075704.
[59] T. Zhou and D. Gall, “Resistivity scaling due to electron surface
scattering in thin metal layers,” Phys. Rev. B, Condens. Matter, vol. 97,
no. 16, Apr. 2018, Art. no. 165406. doi: 10.1103/PhysRevB.97.165406.