Вы находитесь на странице: 1из 12

Wear 233–235 Ž1999.

1–12
www.elsevier.comrlocaterwear

ELSI conference: invited lecture


Liquid impact: theory, experiment, applications
J.E. Field )
UniÕersity of Cambridge, Department of Physics, CaÕendish Laboratory, Cambridge CB3 0HE, UK

Abstract

Liquid impact comes into a wide range of technological problems. The theory has developed historically from two areas: those
interested in ‘‘water entry’’ and the action of waves on ships and structures; and those studying liquid drops or jets impinging on
surfaces. Erosion problems caused by liquid drops impacting surfaces occur with steam-turbines and with aircraft or missiles travelling at
high speed through rain. Cavitation damage is also closely related to liquid impact erosion. More beneficial uses of liquid drops and jets
are for cleaning surfaces or cutting materials. The reason a liquidrsolid impact can cause so much damage is that in the initial stages of
impact the contact periphery expands supersonically Ži.e., faster than stress waves in either liquid or solid.. There is, therefore, a period
Žthe duration depending on the impact velocity and geometry in the contact region. when compressible loading results and pressures are
not released by flow. This paper describes the background theory, the various areas of application, techniques for producing controlled
impacts in the laboratory for the velocity range from a few meters per second to several thousands of meters per second, the use of
high-speed photography for studying such impacts, and the damage processes in the solid. An area of current interest is producing IR
transmitting ‘‘window’’ materials for aircraft which have sufficient strength to resist erosion damage by rain, sand, ice and bird-strike. IR
materials such as zinc sulphide and germanium are relatively weak. However, it is now possible to produce free-standing ‘‘windows’’ and
domes made from chemically vapour-deposited diamond ŽCVDD.. Recent advances in this area are presented. q 1999 Elsevier Science
S.A. All rights reserved.

Keywords: Liquid impact; Chemically vapour-deposited diamond; Jet

1. Introduction mechanics of liquid impact contains important material


from the viewpoint of applications to material damage.
An understanding of the dynamics of liquids undergo- Brunton and Rochester’s article w6x in the same volume
ing collisions with either solids or other liquids is needed covers similar material and is highly recommended for a
in a number of technological situations. Some examples general overview of the liquid erosion problem. A compre-
are splashes w1x, impact of structures on liquid surfaces hensive bibliography on rain erosion with annotated refer-
w2,3x and impact of liquid drops or jets on erodable sur- ences on work performed up to 1972 has been compiled by
faces. Historically, the erosion of steam turbine blades Sims and Trevett w7x. Other general sources are the book
stimulated interested in the subject, but there is now by Springer w8x, the series of international conferences on
considerable research on the rain erosion of aircraft and ‘‘Erosion by Liquid and Solid Impact’’ w9–13x, the jet-cut-
missiles, particularly the ‘‘window’’ materials. Cavitation ting conferences sponsored by the British Hydrodynamics
damage is also closely linked to liquid impact. Research Association at two yearly intervals from 1972
Several other reviews of liquid erosion problems have w14x, and the proceedings of the ASTM conference on
appeared in recent years, though the emphasis has been on erosion w15x.
the materials science aspects of the subject. The reader can Another earlier review by Lesser and Field w16x covered
obtain a general orientation from the volume on erosion the theory of liquid impact in some detail; only key points
edited by Preece w4x. The article by Adler w5x on the are emphasised here. The present review concentrates on
the equipment developed for the laboratory simulation of
liquid impact and its application to infra-red transmitting
)
Tel.: q44-1223-337318; fax: q44-1223-350266; E-mail: materials research, particularly the development of chemi-
jef1003@phy.cam.ac.uk cally vapour-deposited diamond ŽCVDD..

0043-1648r99r$ - see front matter q 1999 Elsevier Science S.A. All rights reserved.
PII: S 0 0 4 3 - 1 6 4 8 Ž 9 9 . 0 0 1 8 9 - 1
2 J.E. Field r Wear 233–235 (1999) 1–12

2. Liquid impact theory Pressure release commences after a time of


3rV
Liquid impact consists of two main stages; initially, the Ž 4.
liquid behaves in a compressible manner generating the ts 12
.
2C
so-called ‘‘ water-hammer’’ pressures. These high pres- The release waves reach the central axis and terminate
sures are responsible for most of the damage resulting the high-pressure stage after a time
from liquid impact and are maintained while the edge of
the contact area between the impacting liquid and the solid 3rV
X
Ž 5.
moves supersonically with respect to the shock speed in t s 12
.
2C
the w16–19x. Fig. 1 illustrates the situation a short time
after impact. Once incompressible stream line flow is established, the
As pointed out by Lesser w19x, it is possible to use pressure on the central axis falls to the much lower
Huygen’s construction to find the shape of the shock Bernoulli stagnation pressure
envelope. The pressure on the central axis is given by rV 2
Pc s Vr 1 C1 r 2 C2 , Pi s .
Ž 6.
r C qr C Ž 1. 2
1 1 2 2 In calculating the pressures for high velocity liquid
where V is the impact velocity and r 1 , r 2 and C1 , C2 are impact, it is essential to use the appropriate shock velocity.
the densities and the shock velocities of the liquid and the This is related to the acoustic velocity, C0 Žf 1500 m sy1
solid, respectively. For impact on a rigid target, the pres- for water. by
sure is: C1 s C0 q kV Ž 7.
P s r 1 C1V , Ž 2. where k is a constant which has a value close to 2 for
frequently referred to as the ‘‘water-hammer’’ pressure water in the velocity range for V up to 1000 m sy1 w18x.
w20x. The simulation of drop impact by jet impact relies on
The pressures at the contact periphery are somewhat reproducing the local liquidrsolid geometry for the all-im-
higher and reach f 3r 1C1V at the instant the shock enve- portant compressible stage of collision. The incompressible
lope overtakes the contact periphery and starts to move up flow stages with the jet and drop will be different but at
the free surface of the drop w19x. Due to the very short this stage the pressures are much lower. For example, the
duration of this pulse Ža few nanoseconds., this is usually ratio of Pc rPi for velocities, V s 50, 100, and 500 m sy1
ignored. are 64, 34 and 10, respectively.
The high pressures are generated over a radius of Photographic evidence for the initial shock structures
contact given by and the onset of jetting have been obtained for impact with
drops w21,22x and liquid wedges w23x. Pressure measure-
R s rVrC1 , Ž 3. ments have been made by Rochester and Brunton w24x.
where r is the radius of curvature of the drop Žliquid mass.
in the region of contact.
3. Experimental

3.1. Multiple impact deÕices

A variety of techniques have been devised for studies of


liquid–solid impact. Historically, the first was the wheel-
and-jet method, which has been used in various forms
since the 1920s. In a typical configuration, specimens are
attached to the edge of a disk so that they protrude
radially. As the disk rotates, the specimens impact one or
more cylindrical jets of liquid. The wheel and jets are
usually in a chamber that is pumped to as low a pressure
as possible to reduce aerodynamic distortion of the jets.
Anyone who has used such an apparatus will know that
Fig. 1. Initial stage of impact between a water drop and a solid target with specimen alignment is critical for obtaining reproducible
the contact edge moving faster than the shock velocity in the liquid Ži.e., data. The reasons for this are discussed by Lesser and
supersonically.. The shock envelope is made up of many wavelets which
can be found from a Huygens-type construction. The liquid behind the
Field w16x. If the specimen is perfectly aligned with the
envelope is compressed and the target beneath this area subjected to high face of the specimen parallel to the axis of the jet, initial
contact is along a line OO Žsee Fig. 2a and b.. There is
X
pressure.
J.E. Field r Wear 233–235 (1999) 1–12 3

main difficulties with this technique are Ža. decelerating


the specimen after impact without adding further damage;
and Žb. distortion of the drop by gravity and the air blast
ahead of the projectile. It is clear from the theory of liquid
impact that if a drop is flattened it will have higher radii of
curvature in the contact region and will behave effectively
as a larger drop and thus be more damaging. Adler’s and
Fig. 2. Impacts between specimen and jet in a wheel-and-jet apparatus. James’ tests w37x are conducted in a helium atmosphere at
Ža. Plan view; Žb. side view for a well-aligned specimen; Žc. an angled an ambient pressure of 0.1 Torr to minimise distortion.
specimen. The low pressure necessitates using a drop made of a
solution of water and ethylene glycol, with 80% water by
volume, to reduce evaporation. The other workers have
then compressible behaviour with no jetting until the con- devised baffle systems to reduce the aerodynamic effects.
tact area has a width 2 x. For a brittle material, tensile Rickerby w39x was able to suspend drops of up to 6-mm
failure occurs only outside this central region. This unfrac- diameter using a water–glycerine solution with a few
tured central strip corresponds to the circular region sur- percent of glycerine. Adler w41x has analysed various
rounded by circumferential cracks for spherical-drop im- sources of published data on single-drop impact patterns
pact. However, if the specimen is misaligned through an on PMMA. For nominally similar-sized drops, the dimen-
angle b Žsee Fig. 2c., the contact is not instantaneous sions of the damage patterns were noticeably smaller for
X
along OO but moves upwards with a velocity that is the low-pressure experiments or for those in which special
supersonic for sufficiently small b and subsonic for large care had been taken to reduce drop distortion. Adler’s
b. In the first of these cases, high-edge pressures will be latest review of his techniques is given in Ref. w42x.
generated at the contact point but there will be no flow, An alternative approach devised by Bowden and Brun-
while for large b a high-velocity jet will move upward. ton w43,44x is to keep the specimen stationary and to
The three cases of perfect alignment, small b, and large b project a jet of liquid against. A projectile is fired into a
are very different. A key point is that for the velocity used stainless steel chamber containing a small quantity of
in this test the angle b that determines whether or not the water sealed in by a neoprene disk. The projectile and
X
contact point moves supersonically along OO is a few neoprene move forward as a piston and extrude the liquid
degrees at most. Hancox and Brunton w25x give examples through a narrow orifice. The ratio of jet to projectile
of damaged polymethylmethacrylate ŽPMMA. specimens velocity is typically 3 to 5 times. This method has been
eroded in this test, one for perfect alignment and one used extensively for studies of the mechanisms of erosion
inclined with b s 98. The first specimen shows cracks and w17,38,44–46x. Although a basically simple and inexpen-
flow damage outside an undamaged central strip. The sive technique, the design and machining of the nozzles
angled specimen exhibits arc-like cracks and erosion cut- are critical if coherent, smooth-fronted and reproducible
tings across the central axis. Very different erosion rates jets are to be produced. Fig. 3 is a single-shot photograph
were also recorded. of a jet from a 0.8-mm nozzle. The umbrella of spray
A development of the wheel-and-jet approach for comes mainly from the water in the parallel section of the
higher-velocity testing are whirling-arm devices. Here, the orifice. It is made up of droplets of micron size that do not
specimens are attached to a long arm rotated in an atmo- contribute to the damage. The jet-front velocity increases
sphere of water drops. Velocities of up to Mach 3 in air over a distance of ; 10 mm. At distances of 2 to 3 times
have been achieved w26,27x. A variety of linear test facili- this, the onset of Taylor instability begins to disrupt the
ties exist: specimens have been launched ballistically from front surface. A stand-off distance of 10 mm is recom-
a gas gun through an atmosphere of drops w28x; specimens mended in all experiments. The acceleration after leaving
and complete components have been attached to rocket- the orifice results from the liquid from the main part of the
driven sledges and propelled on rails through a simulated chamber taking a finite time to move through the slower-
rain field w28–32x; droplets have been accelerated to Mach moving liquid from the orifice section. The essential point
0.83 in a wind tunnel and impacted against targets; and is that there is a central core of coherent liquid which at a stand-
specimens have been attached to an aircraft and flown at off distance of 10 mm has a smooth, slightly curved front
high speed through rain w33x. profile. This has allowed the simulation of drop impact
with these jets. The ‘‘equivalence’’ curves relating drop and
3.2. Single-impact techniques jet impact are given in Ref. w47x for both normal and
angled impacts.
The attraction of producing controlled single impacts is A number of schemes have been proposed for creating
that it allows detailed study of both liquid and solid. One high-speed jets of liquid for cleaning, cutting, and mining
approach is to fire a specimen mounted at the front of a applications. For the most part, these methods utilise
projectile against a suspended or falling drop w34–40x. The steady-state jets: however, transient or non-steady methods
4 J.E. Field r Wear 233–235 (1999) 1–12

An alternative approach for projecting a liquid mass at


velocities up to ; 5 km sy1 has been described by
Sullivan and Hockridge w32x. A water-gelatine drop with
97% water is carried in a sabot fired by a helium gun. The
sabot falls away, leaving the drop to impact the specimen.

3.3. Multiple impact jet apparatus (MIJA)

The apparatus used to simulate rain erosion in our


studies is the MIJA Žsee Fig. 4.. For a full description, see
Seward et al. w53–55x. This apparatus uses a two-stage
pressure reservoir to accelerate a nylon piston onto a
titanium shaft positioned in the rear of a water-filled
nozzle. The shaft’s rapid insertion into the nozzle forces a
high velocity water jet from the exit orifice onto the
sample which is sitting on a computer-controlled x–y
stage. This arrangement allows the entire impacting pro-
cess to be automated via an interface with a personal
computer. The jet velocity is measured by a series of light
beams between optical fibres and continually monitored by
the computer. MIJA can produce a jet every 5 s with
velocities in the range 30–600 m sy1 with a spread in
velocities of less than 1%. Any chosen damage array is
achieved by having the sample on the computer-controlled
stage which has a positional accuracy of 10 mm.
The damage caused by these water jets was evaluated
and compared to that resulting from impacts with spherical
water drops by Hand and Field w47x. These data showed
that the diameter of water drop which gave the same
Fig. 3. Single-shot photograph of a jet from a 0.8-mm nozzle. The central damage pattern as a particular diameter water jet depended
core is coherent liquid. The umbrella spray is made up of micron-sized on the impact velocity. This is because the front of the jet
droplets, which do not contribute to the damage. Note the smooth,
is not a true hemisphere, but slightly flattened, which
slightly curved, front profile. The horizontal extent of the picture is 4.4
mm ŽField et al., 1979b..
means that at low velocities the jet gives the damage
observed from a large water drop and as the velocity
increases the equivalent drop diameter decreases Žsee Fig.
appear more favourable for attaining velocities that are 5. .
supersonic with respect to the liquid. The transient method
w48–50x typically involves firing a packet of liquid through
a converging nozzle. In this process, the front region of the
liquid is rapidly compressed and accelerated. The jet veloc-
ity is at its maximum as it cross the orifice plane and can
be as high as 10 to 12 times the projectile velocity w51x or,
in absolute terms, up to ; 5 km sy1 . This is higher than
with the Brunton nozzle, but in the rain erosion simulation
Ždiscussed above. the coherence and front profile of the
liquid are of primary importance, whereas in the mining
application, high velocity and coherence are the key fac-
tors. However, there are problems with projectile liquid
jets at very high velocities, For example, if the jet velocity
exceeds the speed of sound in the liquid, then decompres-
sion occurs after the liquid leaves the nozzle. This causes
disruption of the jet. At a later stage, deceleration by air
drag and Taylor instability cause further break-up. Field
and Lesser w52x have discussed these aspects of the me-
chanics of high-speed liquid jets. Fig. 4. The MIJA used for simulating high-velocity raindrop impact.
J.E. Field r Wear 233–235 (1999) 1–12 5

impacts Ži.e., DTV 300 impacts. is therefore very close to


the ADTV of the material. The ADTV decreases as the
impacting drop or jet diameter increases. This is related to
the number and size of flaws which can be sampled and is
discussed later. The values quoted in the following work
are all for a 0.8-mm diameter jet. A more detailed descrip-
tion of the threshold velocity evaluation procedure can be
found in Refs. w54,55x.
The damage pattern observed under liquid impact for
brittle materials is typically a series of discrete circumfer-
ential fractures around the undamaged central loaded zone.
The fractures are caused by the Rayleigh surface wave
Fig. 5. The equivalent drop diameters for the 0.8-mm MIJA jet used in emanating from the impact w17x. The pressure pulses pro-
the study.
duced by liquid impact are intense because of the com-
pressible behaviour of the liquid in the first stages of
impact Žsee Section 2.. An example of damage produced in
The material’s rain erosion resistance is characterised a brittle material by liquid impact is given in Fig. 7. In
some material, repeated impact produces local failure on or
by determining its absolute damage threshold velocity
ŽADTV.. This is the velocity below which, for a given near the impact axis. Bowden and Brunton w44x found this
water drop size, the sample will never experience any with PMMA. In this case, the damage was located beneath
the surface at a depth of about half the radius of the
damage regardless of the number of impacts to which it is
contact region, R. This is where the Hertz theory for
exposed. Because of the high accuracy of the MIJA jet
elastic contact would predict the maximum in shear stresses
velocity and positioning, this parameter can be simply
Žsee, for example, Ref. w56x., and Bowden and Brunton
obtained from a single sample. The sample Žtypically a
w44x suggested this as an explanation. However, for such a
25-mm diameter disc. has up to 15 sites positioned over its
surface, each one allocated an impact velocity and suffi- stress wave dominated loading, this is unlikely to be the
ciently separated from adjacent sites so that there is no full explanation. Recent experiments by Bourne et al. w57x
interaction. Each site is initially impacted once at that show that the subsurface axial cracks in PMMA form
velocity and inspected for damage under a microscope at when the release waves from the contact periphery interact
giving a net tension.
100 = magnification. The lowest velocity at which dam-
Interestingly, the release waves will also travel in the
age is observed after a single impact is recorded as DTV
Žone impact. Žthe damage threshold velocity after one liquid giving cavitation when they cross. This has been
impact. and the sample returned to MIJA so that each site suggested theoretically by Field et al. w23x and there is
can be impacted again. This process is repeated until a full experimental evidence w58,59x. The nuclei for such cavities
curve of DTV against number of impacts is obtained. A could be the air trapped at the interface during impact.
threshold curve for zinc sulphide is given in Fig. 6 and
shows that the curve tends to flatten out after, typically, 50
impacts. The intercept on the velocity axis after 300

Fig. 6. A typical zinc sulphide damage threshold curve obtained from Fig. 7. A typical zinc sulphide damage pattern caused by a 300 m sy1
MIJA using a 0.8-mm diameter nozzle. The undamaged central region impact from a 0.8-mm MIJA nozzle. The undamaged central region has a
has a diameter of ca. 1 mm. diameter of ca. 1 mm.
6 J.E. Field r Wear 233–235 (1999) 1–12

When the cavities themselves collapse they could give


peaks of pressure which damage the surface.
A third mechanism for damage at or near the central
axis is by the action of compressive or shear loading which
generates tensile failure at the boundaries between grains
depending on their orientation and anisotropy. Once a pit
develops, hydraulic loading by trapped liquid can develop
damage as shown by Field w60x.
Central damage is likely to be less important than the
circumferential cracking in the rain erosion situation since
it depends on multiple impacts on the same site. A situa-
tion which can be realised in the laboratory using MIJA
but not by other techniques and only after very long
exposures in the practical application. Fig. 9. The damage threshold curve for zinc sulphide and zinc sulphide
In most of the experiments reported here, the samples coated with various thicknesses of BP.
were acoustically backed by arranging for the product of
the density, r, and the shock velocity, C, to be the same
for both the target and the backing material. The surfaces tions on the onset of damage in diamond has been dis-
of the two materials need to be smooth and flat for this. A cussed elsewhere w64x.
thin layer of grease allows compression waves to pass over
the interface but prevents reflected tensile waves re-enter-
ing the sample. The sample then behaves effectively semi- 4. Currently used substrate materials
infinitely.
In the practical situation, the component is unbacked, The threshold velocity curves for a range of infrared
and it is possible that rear surface stress wave reflections transparent materials have been obtained and the ADTVs
could affect the threshold velocity and any impact damage. plotted as a function of material properties. The best fit is
In soda-lime glass, this effect was detectable for impacts obtained by plotting ADTV against the log of the static
with a 0.8-mm jet diameter at velocities ) 250 m sy1 fracture toughness K Ic Žsee Fig. 8.. In this graph, 0.8-mm
w61,62x. This is well above the velocity threshold for liquid jet ADTV data has been converted into threshold velocities
impact damage in soda-lime glass which has a low acous- for a 2-mm diameter spherical drop using equations dis-
tic attenuation coefficient. cussed in Refs. w47,65x so as to eliminate the variation in
Diamond is difficult to acoustically match because of equivalent drop diameter with jet velocity from the correla-
the difficulty of finding materials of similar very high tion.
acoustic impedance, though gold is a possibility w63x. It is immediately clear that the threshold velocities are
Diamond has a very high stress wave velocity and this considerably lower than the maximum velocities at which
means that with samples of less than about 1 mm thickness the components will be flown and this emphasises the
the pulse from the rear surface, for an impact of f 600 m severity of the problem even with materials such as dia-
sy1 returns to the impact area while the compressive mond and sapphire. It is also clear that currently there are
loading phase is still operating. This implies that free- materials available for the 3–5 mm applications Žsapphire
standing CVDD should have a higher threshold velocity
for greater thicknesses. The effect of stress wave interac-

Fig. 8. The ADTVs for 2 mm diameter water drops as a function of Fig. 10. The damage threshold curve for germanium and germanium
fracture toughness. coated with various thicknesses of BP.
J.E. Field r Wear 233–235 (1999) 1–12 7

Fig. 11. Example of rear surface damage seen on CVDD with the damage predominantly intergranular. The grain size is 100 mm. The horizontal extent of
the picture is ca. 100 mm.

and spinel. that can survive impact velocities considerably be used as a free-standing dome material, it is also being
higher than those that transmit in the 8–12 mm waveband investigated as a protective coating.
Žzinc sulphide and germanium.. It is, therefore, the latter
group that has been the focus of efforts to improve the
erosion performance of infrared materials. These efforts 5. Coatings
concentrated on modifying the existing material, coating it
Cambridge research on coated IR samples, including
with a protective layer andror finding a more erosion
resistant replacement. The latter efforts have centred on the CVD diamond, is detailed in the paper by Coad et al. w65x.
use of CVDD techniques to grow a complete diamond Coatings are added either to improve the optical properties
dome or window. Diamond is optically suitable, has excel- Žfor example, an anti-reflection coating. or in an attempt to
lent mechanical properties and is also attractive because it increase mechanical performance by increasing the mate-
is highly resistant to the thermal shock associated with rial’s ADTV. Few coatings have proved successful in
rapid acceleration to high velocities Žanother serious prob- improving the ADTV significantly above that of the sub-
lem that the missile designer must overcome, see for strate alone. This may be because the preparation of the
instance Ref. w66x.. While the ultimate aim is that diamond substrate for coating reduces its strength, or because the
coating adds tensile residual stresses to the substrate. The
coatings are usually deposited at high temperatures and
any thermal expansion mismatch between it and the sub-
strate will result in thermal stresses. In addition, there are
intrinsic stresses in the coating which also increase with
deposition temperature and rate and coating thickness. In
the systems of interest, these stresses are usually compres-
sive in the coating and tensile in the substrate. These

Fig. 12. DTV Žrear surface damage. vs. sample thickness for the growth Fig. 13. Bulk waves and surface waves associated with the enlarged ring
and nucleation sides of CVDD. cracks seen on the impacted surface of CVDD.
8 J.E. Field r Wear 233–235 (1999) 1–12

with only a small error introduced, but for an opaque


material like germanium, this cannot be done. Some care
must, therefore, be taken when determining the signifi-
cance of the higher ADTV of the 13-mm BP on germa-
nium compared to that of a similar thickness of BP on zinc
sulphide. However, after a sufficiently large number of
impacts, any damage in the substrate should propagate
through the coating and become visible and it is believed
Fig. 14. Two possible first order reinforcements which cause the enlarged
ring crack in CVDD; Ža. dilatational, C1 , to dilatational and, Žb. mode
that the 300 impacts used in the DTV experiments are
change at the rear surface, dilatational to transverse, C2 , and vice versa sufficient for this to be the case. The difference in ADTVs
reinforcing with the Rayleigh wave, C R . between BP coated germanium and BP coated zinc sul-
phide is therefore a real effect. However, the increased
DTV at low numbers of impacts, with increased coating
stresses will increase the coating’s tendency to debond thickness, on the germanium graph ŽFig. 10., is probably
from the substrate and will effectively weaken the sub- due to the increase in the number of impacts required for
strate, in some cases to the point of spontaneous fracture of the damage, which has initiated in the substrate, to propa-
the sample as it cools down to room temperature. gate through to the coating surface where it is first ob-
One of the most successful systems to date is boron served. The assumption that damage initiates first in the
phosphide ŽBP.. The threshold curves for a range of BP substrate is verified by inspection of the damaged impact
coating thicknesses on zinc sulphide substrates together sites on BP coated germanium which have a hexagonal
with that of the uncoated material are given in Fig. 9. The symmetry indicative of the cleavage planes in the crys-
coating has provided considerable protection to the sub- talline substrate.
strate and there appears to be a benefit afforded by increas- BP has also been shown to improve the solid particle
ing the coating thickness from 11 up to 17 mm. Fig. 10 erosion performance of zinc sulphide and germanium w67–
shows a similar series of graphs, but this time with germa- 69x.
nium as the substrate material. The inclusion of the 5-mm
thick coating indicates an increase in ADTV with coating
thickness Žbetween 0 and 12 mm.. 6. Results for free-standing diamond
It should be noted that since the evaluation of damage
with MIJA is purely an optical one, there is a possibility CVDD has been tested on MIJA for 5 years and during
that an opaque coating will give artificially high DTV that time enormous advances have been made with the
values since it conceals the damage in the substrate from quality, the thickness and, recently, the shape of the mate-
the observer. In the case of the optically translucent zinc rial tested, with full hemispherical domes provided for
sulphide, substrate damage can be inspected from the rear evaluation w70x.

Fig. 15. Example of the enlarged ring crack seen on CVDD, diameter of ring is 5.1 mm. The small bubbles are within the epoxy resin mounting and not in
the CVDD.
J.E. Field r Wear 233–235 (1999) 1–12 9

CVDD is tested using the small sample technique over


a number of impact sites to evaluate the homogeneity of
the material and to allow enough area on the sample to test
both sides. Since diamond has a very high acoustic
impedance, it is very difficult to match acoustically except
with a layer of gold which is impractical in IR applica-
tions. This has resulted in the samples being impacted
unbacked, which is closer to the situation in practical
applications.
Due to a combination of the high shock wave velocity
Ž18 mm msy1 ., the low shock wave attenuation and the
geometry of the samples, a number of damage mechanisms
are seen frequently in CVDD.

6.1. Rear surface damage

This damage initiates on the rear surface of the sample Fig. 17. Transition between the first damage mode seen in CVDD discs
on the axis of impact. The damage takes the form of a when impacted on the growth surface.
crazing which extends outwards from a central initiation
point. This damage develops due to a mismatch between
the acoustic impedance of diamond, Z D Ž63.4 = 10 6 kg
y1 the stronger of the two sides. The grains tend to increase in
my2 s ., and the impedance of the mounting material,
often an epoxy resin, Z E Ž2.7 = 10 6 kg my2 sy1 . or, with size away from the nucleation surface and so the growth
free-standing samples, air. Due to this impedance mis- side is usually weaker and the difference between the
match and the fact that the transition in impedances for the nucleation and growth side strengths diverges with increas-
compressive wave into the backing is from a high value to ing specimen thickness.
a lower one, a large percentage of the compressional wave The second observation is the rate at which the im-
is reflected at the interface back into the diamond as a provement in the DTV Žrear surface. occurs with thick-
tensile wave. This tension on the rear face causes the ness. This is much higher for the nucleation side than for
fractures which are predominantly seen along the grain the growth side. The DTV Žrear surface. increases with
boundaries, see Fig. 11. thickness for both sides due to geometric attenuation of the
The threshold velocity for the rear surface ŽDTV Žrear compressive wave across the thickness of the sample. This
surface.. has been studied for both the growth and the geometric attenuation is the only factor for the nucleation
nucleation side for a range of CVDD thicknesses Ž500 to side as the grain size is unaffected by the sample thickness.
2000 mm., see Fig. 12. There are two main results which However, for the growth side, there is an additional con-
can be drawn. Firstly, the DTV Žrear surface. for the sideration which is the effective weakening of the surface
nucleation side is always greater than for the growth side. due to the larger grains for thicker samples.
This is due to the strength of CVDD being grain size
dependent and increasing as the grain size increases. The
nucleation side invariably has smaller grains and thus is

Fig. 16. Diameter of enlarged ring crack seen on the impacted surface of Fig. 18. Example of edge failure seen when the impact site is too close to
CVDD vs. sample thickness. sample edge. The horizontal extent of the picture is 15.7 mm.
10 J.E. Field r Wear 233–235 (1999) 1–12

the transition at which the first damage occurs is at about


1100 " 50 mm. There is no such transition seen when the
sample is impacted on the nucleation side.

6.3. Edge failure

If the impact site on a flat CVDD disc is too close to


the edge of the sample then an extensive crack can be
generated, see Fig. 18. This is due to the reinforcement of
Fig. 19. Damage mechanism which results in edge failure in CVDD the outward bound Rayleigh wave and the bulk reflected
samples, see Table 1. wave from the side of the disc, see Fig. 19. As the waves
are both generated at impact, the ratio of the distances
travelled gives a ratio of their velocities. Three sites im-
pacted on different samples developed this damage and the
6.2. Enlarged ring crack measurements and ratios of the distances travelled are
given in Table 1. As can be seen, there is a very good
This damage mechanism occurs on the impacted surface agreement between the ratio of the distances travelled by
when the bulk waves, reflected from the rear surface, the two waves and the ratio of the Rayleigh wave velocity,
interact with the slower Rayleigh surface wave. The inter- C R , and the dilatational wave velocity, C1.
action can be either with a returning dilational wave or a
slower transverse wave, see Figs. 13 and 14. The result is a 6.4. Circumferential cracking
section of increased tension on the surface which results in
a large ring crack, see Fig. 15. This damage occurs at a Like all brittle materials, CVDD has a DTV Ž300
radius greater than is expected if the damage were to have impacts. for damage caused by the Rayleigh wave alone
been solely introduced by the Rayleigh wave. Upon analy- Žno reinforcement effects.. For thin samples, the described
sis, the radius of the ring crack is found to be the result of damage modes are seen before any evidence of circumfer-
the interaction between the Rayleigh wave and a returning ential cracking is initiated. Impacting at higher velocities,
transverse bulk wave w17,64x. As this damage relies upon a as is the case for the thicker samples, causes short circum-
returning bulk wave, there is a strong relationship with the ferential cracks at a radius similar to that for a more
sample thickness and radius of the enlarged ring crack, see conventional brittle material Žtypically 500 mm for a 0.8-
Fig. 16. For all thicknesses of sample tested, there is good mm jet.. This damage has been seen on a few samples and
agreement with the theoretical radius and the experimen- only when the growth surface is the impacted face. The
tally obtained value w65x. DTV Ž300 impacts. for the growth side for samples ) 1100
For thin samples ŽF 1000 mm. impacted on the growth mm thick is 450 " 25 m sy1 , is evaluated using the small
side Žas is the possible situation for a CVDD dome during sample technique. There appears to be no variation in the
service., the first damage mechanism to be witnessed DTV Ž300 impacts. for the growth side with sample
Žwhen tested using the small sample technique. is rear thickness and, therefore, surface flaw size. This is in
surface damage with the enlarged ring crack initiating at a agreement with a computer model which has been devel-
higher velocity. However, this order is reversed for thicker oped for impact on brittle materials.
samples Ž) 1000 mm., where the first damage to be
witnessed is the enlarged ring crack on the growth side 6.5. CVDD domes
followed by rear surface damage on the nucleation side.
Fig. 17 shows the threshold velocities for these two dam- Two complete diamond domes were supplied for evalu-
age modes vs. sample thickness. As can be clearly seen, ation; 750 mm thick and 1300 mm thick at the apex and an

Table 1
Measurements for three impact sites, showing agreement with the ratio of velocities that caused damage
Sample Sample Distance b Žmm., Distance Theoretical arŽ a q 2 b .
number thickness travelled by see Fig. 14 travelled by the value of
Žmm. Rayleigh wave, dilatational CR rC1 for
a Žmm. wave, a q 2 b Žmm. diamond
1 750 " 10 5.0 " 0.1 1.7 " 0.1 8.4 " 0.2 0.594 0.60 " 0.03
2 860 " 10 3.3 " 0.1 1.1 " 0.1 5.5 " 0.2 0.594 0.60 " 0.03
3 580 " 10 3.8 " 0.1 1.3 " 0.1 6.6 " 0.2 0.594 0.59 " 0.03
J.E. Field r Wear 233–235 (1999) 1–12 11

Recent research on IR ‘‘ windows’’ and ‘‘domes’’ is


described. Producing a successful coating is difficult for a
whole variety of reasons Žsee Section 5.. In fact, there are
cases when the coated system is weaker than the substrate
alone. Of the coatings studied, BP has useful properties for
improving the rain and solid particle erosion of zinc
sulphide and germanium ‘‘windows’’.
The paper ends with a discussion of data obtained on
CVDD specimens, ‘‘windows’’ and domes. This technol-
ogy is advancing in an exciting way. However, even when
the diamond rain erosion resistance has been optimised it
is unlikely that the ADTV will reach the maximum veloci-
ties at which it will be carried on an aircraft and will not
even approach the velocities of the fastest missiles. The
erosion of infrared transparent materials will remain a
Fig. 20. Example of full hemispherical CVDD dome tested in the problem, and an active research area, for some time.
Cavendish Laboratory.

inner radius of 35 mm, see Fig. 20. The thinner dome was Acknowledgements
impacted on both sides and the DTV Žrear surface. gave
very good agreement with the data obtained for the flat The author thanks the many former students and post-
discs. The thicker dome has only been impacted on the docs who have contributed to the Cambridge research,
growth side with rear surface damage observed on the particularly Drs. E.J. Coad, D.A. Gorham, R.J. Hand,
nucleation side. The DTV Žrear surface. was lower than C.S.J. Pickles, C.R. Seward, D. Townsend, and Professors
had been seen for the flat disc samples. This was thought M.J. Mathewson and S. van der Zwaag. Collaborative
to be due to the poor surface finish on the inside of the research with Professor M.B. Lesser Žnow at KTH Stock-
dome after removal from the substrate. holm. contributed significantly. The research has been
Comparison between CVDD and DTV data on natural carried out with the support of the Defence Evaluation
and high-temperature, high-pressure ŽHTHP. synthetic dia- Research Agency and De Beers. The author would like to
mond is given by Coad et al. w65x. Both natural and HTHP thank Dr. J.A. Savage of DERA Malvern for his encour-
can have DTVs of almost 600 m sy1 for impact with 2 agement, and R. Marrah for technical support with SIJA
mm drops Žsee Fig. 8.. CVDD has improved significantly and MIJA.
over the last few years, but still has a DTV about 200 m
sy1 less than good-quality natural diamond.
References
7. Discussion
w1x A.M. Worthington, A Study of Splashes, Longmans, New York,
1908; republished Macmillan, New York, 1963.
There has not been space in this review to do more than w2x H. Wagner, Z. Angew. Math. Mech. 12 Ž1932. 193–215.
mention and reference the various technologies which w3x R. Skalak, D. Feit, J. Eng. Ind. Trans. ASME 88B Ž1966. 325–331.
involve liquid impact. These range from the ‘‘problem’’ w4x C.M. Preece ŽEd.., Treatise on Materials Science and Technology,
areas such as erosion, to the practically advantageous such Vol. 16, Erosion, Academic, New York, 1979, 450 pp.
w5x W.F. Adler, in: C.M. Preece ŽEd.., Erosion, Academic, New York,
as jet cleaning or cutting. The theory of liquid impact in
1979, pp. 127–184.
Section 2 does show, however, why liquid impact is not an w6x J.H. Brunton, M.C. Rochester, in: C.M. Preece ŽEd.., Erosion,
innocuous event associated with low pressures but, be- Academic, New York, 1979, pp. 185–248.
cause of compressibility, can give very high transient w7x M.G. Sims, P.W. Trevett, Bibliography of Rain Erosion, RAE
pressures. Library Bibliogr. No. 327, MOD, London, 1972, 115 pp.
w8x G.S. Springer, Erosion by Liquid Impact, Wiley, New York, 1976,
The various techniques of producing liquid impact are
264 pp.
described. Cambridge research has concentrated on devel- w9x A.A. Fyall, R.B. King ŽEds.., 1st, 2nd, 3rd, 4th Int. Conf. Rain
oping laboratory techniques which keep the target station- Erosion Assoc. Phenom., RAE Farnborough, England, 1965, 1967,
ary and move the liquid Žbasically, this is simpler, more 1970, 1974.
easy to control and less expensive.. It involves producing w10x J.E. Field ŽEd.., 5th Int. Conf. Erosion Liquid and Solid Impact,
coherent, smooth-fronted jets which can simulate drop Cavendish Laboratory, Cambridge, England, 1979.
w11x J.E. Field, N.S. Corney ŽEds.., 6th Int. Conf. Erosion Liquid and
impact. MIJA is an equipment which can achieve these Solid Impact, Cavendish Laboratory, Cambridge, England, 1983.
objectives. It can be used to determine ADTVs for materi- w12x J.E. Field, J.P. Dear ŽEds.., 7th Int. Conf. Erosion Liquid and Solid
als. Impact, Cavendish Laboratory, Cambridge, England, 1987.
12 J.E. Field r Wear 233–235 (1999) 1–12

w13x I.M. Hutchings, J.A. Little ŽEds.., 8th Int. Conf. Erosion Liquid and by Solid and Liquid Impact, Cavendish Laboratory, Cambridge,
Solid Impact, Wear, 1995, pp. 186–187. England, 1979, Paper 13, 11 pp.
w14x British Hydraulic Research Association, Proc. Int. Symp. Jet Cutting w39x D.G. Rickerby, PhD Thesis, University of Cambridge, 1977.
Technol., BHRA, Cranfield, England, 1972 onwards at two-yearly w40x P.W. Blair, PhD Thesis, University of Cambridge, 1981.
intervals. w41x W.F. Adler, in: J.E. Field ŽEd.., 5th Int. Conf. Erosion by Solid and
w15x W.F. Adler ŽEd.., Erosion: Prevention and Useful Applications, STP Liquid Impact, Cavendish Laboratory, Cambridge, England, 1979,
664, Am. Soc. Test Mater., Philadelphia, PA, 1979, 643 pp. Paper 9, 10 pp.
w16x M.B. Lesser, J.E. Field, Ann. Rev. Fluid Mech. 15 Ž1983. Ž1983. w42x W.F. Adler, 9th Int. Conf. Erosion by Liquid and Solid Impact,
97–122. Wear, 1998, in press.
w17x F.P. Bowden, J.E. Field, Proc. R. Soc. London, Ser. A 232 Ž1964. w43x F.P. Bowden, J.H. Brunton, Nature 181 Ž1958. 873–875.
331–352. w44x F.P. Bowden, J.H. Brunton, Proc. R. Soc. London, Ser. A 263
w18x F.J. Heymann, J. Appl. Phys. 40 Ž1969. Ž1969. 5113–5122. Ž1961. 433–450.
w19x M.B. Lesser, Proc. R. Soc. London, Ser. A 377 Ž1981. 289–308. w45x D.A. Gorham, J.E. Field, J. Phys. D 9 Ž1976. 1529–1541.
w20x S.S. Cook, Proc. R. Soc. London, Ser. A 119 Ž1928. 481–488. w46x D.A. Gorham, M.J. Matthewson, J.E. Field, in: W.F. Adler ŽEd..,
w21x J.P. Dear, J.E. Field, J. Appl. Phys. 63 Ž1988. 1015–1021. Erosion: Prevention and Useful Applications, STP 664, Am. Soc.
w22x J.E. Field, J.P. Dear, J.-E. Ö gren, J. Appl. Phys. 65 Ž1989. 533–540. Test Mater., Philadelphia, PA, 1979, pp. 320–342.
w23x J.E. Field, M.B. Lesser, J.P. Dear, Proc. R. Soc. London, Ser. A 401 w47x R.J. Hand, J.E. Field, Eng. Fract. Mech. 37 Ž1990. 293–311.
Ž1985. 225–249. w48x B.V. Vortsekhovksy, Br. Pat. 1,109,286.
w24x M.C. Rochester, J.H. Brunton, in: J.E. Field ŽEd.., Proc. 5th Int. w49x W.C. Cooley, in: G.B. Clark ŽEd.., 12th Int. Symp. Rock Mech.,
Conf. on Erosion by Liquid and Solid Impact, Cavendish Labora- AIME, New York, 1971, pp. 653–665.
tory, Cambridge, 1979, pp. 6.1–6.7. w50x B. Edney, Proc. Int. Symp. Jet Cutting Technol., Chicago, IL,
w25x N.L. Hancox, J.H. Brunton, Philos. Trans. R. Soc. London, Ser. A BHRA, Cranfield, England, Paper B2.
260 Ž1966. 121–139. w51x I.L. Rhyming, Z. Angew. Math. Phys. 24 Ž1973. 149–164.
w26x N.W. Wahl, in: A.A. Fyall, R.B. King ŽEds.., 3rd Int. Conf. Rain w52x J.E. Field, M.B. Lesser, Proc. R. Soc. London, Ser. A 357 Ž1977.
Erosion Assoc. Phenom., RAE Farnborough, England, 1970, pp. 143–162.
13–38. w53x C.R. Seward, PhD Thesis, University of Cambridge, 1992.
w27x W. Kunz, in: A.A. Fyall, R.B. King ŽEds.., 4th Int. Conf. Rain w54x C.R. Seward, C.S.J. Pickles, R. Marrah, J.E. Field, Proc. SPIE 1760
Erosion Assoc. Phenom., RAE Farnborough, England, 1974, pp. Ž1992. 280–290.
187–198. w55x C.R. Seward, E.J. Coad, C.S.J. Pickles, J.E. Field, Proc. SPIE 2286
w28x R.B. Mortenson, in: A.A. Fyall, R.B. King ŽEds.., 4th Int. Conf. Ž1994. 285–300.
Rain Erosion Assoc. Phenom., RAE Farnborough, England, 1974, w56x K.L. Johnson, Contact Mechanics, Cambridge Univ. Press, Cam-
pp. 155–185. bridge, 1985.
w29x R.B. King, in: A.A. Fyall, R.B. King ŽEds.., 1st Int. Conf. Rain w57x N.K. Bourne, T. Obara, J.E. Field, Philos. Trans. R. Soc. London,
Erosion Assoc. Phenom., RAE Farnborough, England, 1965, pp. Ser. A 355 Ž1997. 607–632.
49–57. w58x J.H. Brunton, J.J. Camus, Proc. 9th Int. Congr. High-Speed Photog-
w30x W.G. Reinecke, in: A.A. Fyall, R.B. King ŽEds.., 4th Int. Conf. Rain raphy, Society of Motion Picture and Television Engineers, Denver,
Erosion Assoc. Phenom., RAE Farnborough, England, 1974, pp. CO, 1970.
209–233. w59x N.K. Bourne, T. Obara, J.E. Field, Proc. R. Soc. London, Ser. A 452
w31x G.F. Schmitt, W.G. Reinecke, G.D. Waldman, STP 567, Am. Soc. Ž1996. 1497–1502.
Test. Mater., Philadelphia, PA, 1974, pp. 219–238. w60x J.E. Field, in: A.A. Fyall, R.B. King ŽEds.., 2nd Int. Conf. Rain
w32x R.J. Sullivan, R.R. Hockridge, in: A.A. Fyall, R.B. King ŽEds.., 4th Erosion Assoc. Phenom., RAE Farnborough, England, 1967, pp.
Int. Conf. Rain Erosion Assoc. Phenom., RAE Farnborough, Eng- 593–603.
land, 1974, pp. 109–154. w61x S. van der Zwaag, J.E. Field, Philos. Mag. A 53 Ž1983. 101–111.
w33x A.A. Eyall, R.B. King, G.R. Higgs, J.H. Fawcett, H.C.J. Williams, w62x S. van der Zwaag, J.E. Field, Eng. Fract. Mech. 17 Ž1983. 376–379.
RAE Farnborough Tech. Rep. 75080, 1975, pp. 1–84. w63x J.E. Field ŽEd.., The Properties of Diamond, Academic Press,
w34x D.C. Jenkins, J.D. Booker, in: E.G. Richardson ŽEd.., Aerodynamic London, 1979.
Capture Particles, Pergamon, Oxford, 1960, pp. 97–103. w64x C.R. Seward, E.J. Coad, J.E. Field, J. Hard Mater. 5 Ž1994. 49–62.
w35x A.A. Fyall, in: A.A. Fyall, R.B. King ŽEds.., 2nd Int. Conf. Rain w65x E.J. Coad, C.S.J. Pickles, C.R. Seward, G.H. Jilbert, J.E. Field,
Erosion Assoc. Phenom., RAE Farnborough, England, 1967, pp. Proc. R. Soc. London, Ser. A 454 Ž1998. 213–238.
563–586. w66x C.A. Klein, Proc. SPIE 1760 Ž1992. 338–357.
w36x A.A. Fyall, P.A. Smith, US Avionics Lab., Symp. Electromagn. w67x E.M. Waddell, D.R. Gibson, Proc. SPIE 2286 Ž1994. 364–375.
Windows, Vol. 1, 1968, pp. 55–77. w68x E.M. Waddell, D.R. Gibson, Proc. SPIE 2286 Ž1994. 376–385.
w37x W.F. Adler, T.W. James, ONR Tech. Rep. for Contract No. w69x G.H. Jilbert, J.E. Field, Wear 21 Ž1998. 15–23.
N0001476C-0744, 1978, 144 pp. w70x E.J. Coad, C.S.J. Pickles, G.H. Jilbert, J.E. Field, Diamond Relat.
w38x J.E. Field, D.A. Gorham, J.T. Hagan, M.J. Matthewson, M.V. Mater. 5 Ž1996. 640–643.
Swain, S. van der Zwaag, in: J.E. Field ŽEd.., 5th Int. Conf. Erosion

Вам также может понравиться