`
`Thin Solid Films, 171 (1989) 5-31
`
`STRESS-RELATED EFFECTS IN THIN FILMS
`
`JOHN A. THORNTON
`
`Coordinated Science Laboratory and Department ofMetaI1urgy, University of Illinois. Urbana, IL 61801
`( U.S.A.)
`D. W. HOFFMAN
`
`Research Stafl. Ford Motor Co., Dearbarn, MI 48121-2053 ( U.S.A.)
`
`Nore: This summary article first appeared in the Proceedings of the 22nd Annual
`Technical Conference of the Society of Vacuum Coaters in 1979. It is reproduced
`here, with permission, for both its technical and historical significance. In it are
`found a number of the insights and instructive sketches characteristic of John
`Thornton, which in this case were known only to those few having access to the
`proceedings volume.
`
`Virtually all vacuum-deposited coatings are in a state of stress. The total stress
`is composed of a thermal stress and an intrinsic stress. The thermal stress is due to
`the difference in the thermal expansion coeflicients of the coating and the substrate
`materials. The intrinsic stress is due to the accumulating effect of the crystallo-
`graphic flaws that are built into the coating during deposition. In soft, low melting
`point materials such as aluminum, bulk diffusion tends to relax the internal stresses
`and to prevent their accumulation. However, these diffusion processes can cause
`flaws such as holes and hillocks to form. In hard, higher melting point materials such
`as chromium deposited at low temperatures (e.g. the case of decorative coatings on
`plastic substrates), intrinsic stresses accumulate and tend to dominate over thermal
`stresses. Stress cracking and buckling are commonly observed. The fundamental
`nature of the internal stresses that are found in both evaporated and sputtered
`coatings is reviewed in this paper from the point of View of decorative coating
`applications. Recent sputtering studies are described which indicate that apparatus
`geometry is particularly important in determining the state of stress that forms in
`deposits.
`
`1.
`
`INTRODUCTION
`
`Virtually all metallic and inorganic compound films are in a state of stress. This
`behavior is independent of the method of deposition and applies, for example, to
`evaporated films‘. The total stress is composed of a thermal stress and a so-called
`intrinsic stress. The thermal stress is due to the difference in the thermal expansion
`coefficients of the coating and substrate materials. The intrinsic stress is due to the
`accumulating effect of the crystallographic flaws that are built into the coating
`"“”“g"""°““°“' —
`
`0040-6090/89
`
`Elsevier Sequoia/Printed in The Netherlands
`
`GILLETTE 1017
`
`GILLETTE 1017
`
`
`
`6
`
`J. A. THORNTON. D. W. HOFFMAN
`
`The internal stresses in a thin film can give rise to seemingly unrelated behavior
`that can seriously influence the film’s performance. The parameter T_;’Tm, where T
`(K) is the substrate temperature and Tm is the coating material melting point. is
`particularly important in cataloging stress-related behavior for differing materials.
`In soft, low melting point materials such as aluminum, typical deposition conditions
`involve a relatively high T/‘Tm. Under these conditions. bulk difiusion. which
`becomes increasingly important with increasing T,-‘Tm. relaxes the intrinsic stresses
`and prevents their accumulation. However. the thermal stresses resulting from the
`temperature changes which occur at the conclusion of deposition or in subsequent
`annealing cycles can drive these diffusion processes in such a way that material
`transport occurs. and holes and hillocks are produced in the films. In hard, higher
`melting point materials such as chromium, typical deposition conditions involve a
`relatively low T/Tm. Under these conditions intrinsic stresses can accumulate and
`dominate over thermal stresses.
`
`The film~to—substrate bond must be capable of withstanding the force produced
`by the integrated stress throughout the film. In the intrinsic stress case this force
`increases with the film thickness and can be much larger than the forces provided, for
`example, by a typical tape adhesion test. Thus stress cracking, buckling, and poor
`adhesion are commonly observed when the film thickness exceeds a critical value
`which may be as low as several hundred angstroms.
`The purpose of this paper is to review some of the considerations that are
`important in the occurrence of stresses in metal coatings. The discussion is generally
`from the point of view of decorative coatings. Thus particular emphasis is given to
`the stress-related effects that occur when brittle, high temperature coatings such as
`chromium are deposited on low temperature substrates such as acrylonitrile
`butadiene styrene (ABS). Recent sputtering studies will be described which indicate
`that the apparatus geometry is particularly important in determining the state of
`stress that forms in such coatings.
`
`2. GENERAL ASPECTS OF THERMAL ANT) INTRINSIC STRESSES
`
`2.1. Thermal .s‘Ires.s'e.s'
`
`from its
`is different
`When a coated substrate is at a temperature that
`temperature during deposition, a thermal stress will be present as a result of the
`differences in the film and substrate thermal expansion coeificients. For the thin
`films used in decorative coatings, the film thicknesses are generally less than 10"
`times the substrate thickness. Under these conditions, plastic flow in the substrate
`can generally be neglected and the thermal stress induced in the film by the film
`substrate bonding is given in a one-dimensional approximation (neglecting the
`Poisson effect) by“
`
`9'm 2 Eriar '“ Ktlirs ‘“ Ta)
`
`(I)
`
`where E; is Young’s modulus, oz, and as are the average coefficients of thermal
`expansion for the film and substrate, 7} is the substrate temperature during
`deposition, and Ta is the temperature during measurement. A positive value of Cr”,
`corresponds to a tensile stress.
`
`
`
`STRESS~RELATED EFFECTS IN THIN FILMS
`
`7
`
`Table I gives elastic constants and thermal expansion coefficients for several
`coating metals. Table II gives thermal expansion coefficients for typical substrate
`materials encountered in decorative coating. There seems to be no evidence that the
`thermal expansion coefficient’ and the modulus of elasticity3 for a film are very
`different from the bulk values. Thus the values given in the table may be used for
`making engineering estimates.
`
`TABLE I
`PROPERHES or COATING MATERIALS
`
`Material
`
`Melting
`temperature
`cc)
`
`T/Tm 3
`
`Young '5
`modulus E,
`<GN m”)
`
`Yield
`strength
`(GNm"l
`
`Thermal expansion
`coefiicient
`cc-1)
`
`A1
`Sb
`Cr
`Cu
`Au
`Fe
`Pb
`Ni
`W
`
`660
`630
`1875
`1083
`1063
`1534
`325
`1453
`3410
`
`0.34
`0.35
`0.15
`0.23
`0.23
`0.17
`0.53
`0.18
`0.08
`
`63
`80
`260
`120
`81
`204
`16
`208
`351
`
`0.11
`0.011
`0.16
`0.32
`0.21
`0.25
`0.009
`0.32
`1.8
`
`2.39 x 10‘5
`0.9 x 10‘5
`0.62 X 10-5
`1.65 x 10”5
`1.52 x 10‘5
`l.l7>< 1O‘5
`2.93 x 10"
`1.33 x 10”
`0.43 x 10”
`
`3 Based on T= 40 “C (typical deposition temperature on plastic substrate).
`
`TA BLE I1
`PROPERTIES or SUBSTRATE MATERIALS
`
`Material
`
`ABS
`Polypropylene
`Nylon (mineral filled)
`Polyester (glass filled)
`Glass (soda—lime)
`
`"’ Softening point.
`
`Heat distortion
`temperature (“C)
`
`Thermal expansion
`caeflicient (“C " I }
`
`95
`85
`175
`250
`700 ‘*
`
`9.5 X 10’ 5
`8.5 x 10’ 5
`7 x l0‘5
`3 x 10” 5
`0.92 x 10' 5
`
`Consider the case of a thin chromium film (ocf=0.62 x 10” 5 “C ' 1)deposited on
`an ABS substrate (as = 9.5 x 10‘ 5 °C -1) at a temperature of 40 “C (104 °F) and
`heated to 82 °C (180 °F) during a typical thermal cycle test for chromium—coated
`plastic automotive parts. A tensile stress of about 0.97 GN m “2 (140 000 lbs in '2),
`which is larger than the yield strength of the chromium (0.16 GN In '2) is predicted.
`Next consider the case of an aluminum coating (on, = 2.4 x 105 °C’ 1) deposited
`on a silicon substrate (as = 7.6 x 10 ” 5 °C ' ‘) at a temperature of 150 °C (302 °‘F) and
`then annealed at 400 “C (752 °F) as part of an electronic processing application. A
`compressive stress of about 0.26 GN m‘ 2 (37 000 lbs in’ 2), which is larger than the
`yield strength of aluminum (0.11 GN in‘ 2), is predicted.
`Thus we see that in many applications temperature differences between the
`
`
`
`8
`
`J. A. TH()RNTOt‘\’. D. W’. HOFFMAN
`
`conditions of deposition and application. or associated with post—deposition
`annealing processes. can result in thermal stresses that exceed the yield strength of
`the film and are capable of fracturing even relatively strong coating—to—substrate
`bonds.
`
`2.2.
`
`Iiizrirzsic .s'tr‘e.\‘sc.s'
`
`The intrinsic stress can be defined as that component of the total measured
`stress that cannot be attributed to thermal stress‘. It is due to the accumulating effect
`of the crystallographic fiaws that are built into the coating during deposition. The
`intrinsic stresses are similar to the internal stresses that are formed in a bulk material
`
`during cold working. In the cold work case, the stresses result from the strain which
`is associated with the various lattice defects created by the deformation, However,
`the density of defects that are frozen into a film during deposition can be two orders
`of magnitude higher than that produced by the severest cold~worl<. treatment of a
`bulk material3. Intrinsic stresses that are in the range 0.1-3 GN m " 3 {l5000v
`450 000 lbfin" 5)., and therefore comparable with the yield strength of most metals,
`can develop in thin films deposited at low T/Tm. The intrinsic stresses are strongly
`dependent on the conditions ofdeposition.
`
`2.3. Re<*<)vcry andrec‘r_r.s'ta.//1':aI{on
`The energy which is stored in a film or cold—w0rl<ed metal by the stresses acts as
`a thermodynamic driving force which tends to relax the stresses (by vacancy.
`interstitial, and dislocation movement) ifthe temperature is such that T/Tm is in the
`range 0.l—0.3. The phenomenon is known as recovery”. At higher T_:‘Tm (0.3~—0.5).
`the stresses are relaxed by the recrystallization ofthe strained grains into new strain-
`free gtains5"’. Thus at substrate temperatures such that T/Tm is greater than 0.2.
`recovery and recrystallization relax the intrinsic stresses and reduce their accumul-
`ation in a growing coating.
`
`2.4. Influemre Q/'dep0.s‘itI'0n trwzdiriuns
`The microstructure and thus the intrinsic stresses in vacuum—deposited
`coatings are sensitive to the deposition conditions. Figure I shows a schematic
`representation of the microstructure dependence of sputtered coatings on T/7",,, and
`on the pressure of the argon working gasi. The growth of vacuum-deposited
`coatings involves atoms (1) arriving in a distribution that depends through self-
`shadowing on the coating atom arrival directions, and on the roughness of the
`coating surface, and then (2) diffusing over the surface until they become trapped in
`low energy lattice sites and are incorporated into the growing coating. Finally. the
`deposited atoms may readjust their positions within the coating lattice by recovery
`and recrystallization.
`.
`The T/Tm dependence in Fig. 1 results because the surface diffusion processes.
`like the bulk diffusion processes that characterize recovery and recrystallization (see
`Fig. 3(c)). are dependent on T/Tm. The pressure dependence is believed to result
`because collisions between the sputtered atoms and the argon atoms at elevated
`pressures cause the coating atoms to arrive at the substrate in randomized directions
`that promote shadowing.
`
`
`
`STRESS-RELATED EFFECTS IN THIN FTLMS
`
`9
`
`
`
`(lb
`sussnmr
`0-5
`TEMPERATURf ¥T"Tml
`
`ARDON
`PRESSURE
`«mom
`
`Fig. 1. Schematic representation of the influence of substrate temperature and argon pressure on the
`microstructure of metal coatings deposited using cylindrical magnetron sputtering sources. T (K) is the
`substrate temperature and 7;, (K) is the melting point of the coating material. (See ref. 7.)
`
`Coatings in the zone T (transition) region have a dense fibrous structure with a
`smooth, highly reflective surface. They form on smooth substrates at low T/Tm when
`the coating flux arrives in a direction that is largely normal to the substrate surface,
`so that shadowing effects are minimized. Coatings with the zone T structure are
`desired for decorative applications. Large intrinsic stresses can form in the zone T
`region.
`Coatings in the zone 1 region are characterized by a structure consisting of
`tapered crystals separated by open, voided boundaries. The structure results from
`shadowing because high points on the growing surface receive more coating flux
`than valleys do, particularly when a significant oblique component is present in the
`arriving coating flux. This structure is promoted by an elevated working gas
`pressure or a substrate surface that is rough or is arranged at a large angle relative to
`the coating flux. The extreme zone 1 structure is too porous to support stresses and
`has a rough, poorly reflecting surface.
`The zone 2 structure consists of columnar grains separated by distinct,
`intercrystalline boundaries. Recovery limits the intrinsic stresses in zone 2.
`Zone 3 is defined as that range of conditions where bulk diffusion has a
`dominant influence on the final structure of the coating. Recrystallization occurs if
`sufiicient strain is built
`into the coating during deposition. Recovery and/or
`recrystallization limits the intrinsic stresses in this region.
`
`2.5. T/Tm influence on internal stress
`Figure 2 shows an idealized representation of the total stress generated in a thin
`film as a function of T/Tm (ref. 2). The substrate temperature T is assumed to be
`greater than the final measurement temperature (room temperature), and the film
`thermal expansion coefficient is assumed to be greater than that of the substrate, so
`that a tensile thermal stress is generated (see eqn. (1)) when the substrate is cooled.
`The intrinsic stress is also assumed to be tensile (generally the case in evaporated
`films). At low T/Tm, the intrinsic stress dominates over the thermal stress. When
`
`
`
`10
`
`J. A. TXIORNTOIN} D. W’. HOFFMAN
`
`
`
`
`
`THERMAL
`STRESS
`
`$EPOSITlO\ TEMP£R/¥l|.lRE 'T’Tm?
`Fig. 2. Schematic representation olthermal and intrinsic stress contributions.
`
`T/‘Tm exceeds about ().25—0.3 (zone 2), the recovery processes are operative so that
`the intrinsic component of the stress is reduced. At higher T)-"Tm, the thermal stress
`dominates. A linear increase in thermal stress with T/Tm is shown. This would be
`approximately the case for an increasing substrate temperature T and a given
`coating material Tm.
`When depositing decorative coating on plastics. the substrate temperature
`must generally be limited to less than 100 " C (see Table 11). Thus, for such
`applications. T is fixed and the T/Tm dependence is determined primarily by the
`coating material through the melting point Tm and this becomes a parameter for
`discriminating between coating materials. Figure 3 shows the T,/T,,, dependence of
`the hardness (Fig. 3(a)). thermal expansion coefficient (Fig. 3(b)). activation energy
`for self—dilTusion (Fig. 3(c)), and electroplated stress (Fig. 3(d)) for various metals of
`engineering interest; it
`is based arbitrarily on a substrate temperature (40 “C
`(104 “F)) that is typical for electroplating and for vacuum depositing thin coatings on
`plastic substrates. The fundamental nature of T,/Tm as an indexing parameter is
`apparent. The materials can be divided into two classes: (1) the low melting point
`materials (T/Tm is greater than 0.25) that are soft, have a high thermal expansion
`coefiicient. a low activation energy for self-diffusion, and a low intrinsic stress. and
`(2) the higher melting point materials (T;"T,,, is less than 0.25) that are harder, have a
`lower thermal expansion coefficient. a higher activation energy for self-diffusion, and
`a high intrinsic stress.
`Thus the low T,/Tm side of Fig. 2 corresponds to coating metals such as
`chromium, molybdenum and tungsten, while the high T,/Tm side corresponds to
`metals such as aluminum, antimony and lead. The total stress may have a minimum
`at some intermediate T,/Tm as a result of the combined effects of the intrinsic and
`thermal stresses. It should be noted that the thermal stress contribution may
`increase more rapidly than linearly with increasing T/Tm when the T/Tm increase is
`due to a decreasing Tm because of the temperature dependence of the thermal
`expansion coefiicient (Fig. 3(b)). It should also be noted that while the intrinsic stress
`forms during deposition. the thermal stress does not form until the substrate is
`cooled. If the coating material melting point is sufficiently low that T/Tm is greater
`
`HIGH PIIELTING
`POINT (Tm)
`MATERIALS
`
`LOW MELTING
`POINT lTmI
`MATFRIALS
`
`‘
`
`TOTAL
`lI\ITER!\lAL
`STRESS
`
`SIREES
`
`INTRINSIC OR
`GROWTH STRESS
`
`
`
`STRESS-RELATED EFFECTS IN THIN FILMS
`
`l l
`
`HARDNESS
`
`THERMAL EXPANSION
`COEFHCIENT
`7 = 60°C
`
`mmontss
`
`(at
`
`%
`; 7
`3
`E 6
`5 5
`;
`g 4
`3 3
`2
`
`I 0
`
`0
`
`0,1
`
`0.6
`0.5
`0.3 0.4
`0.2
`TEMPERATURE RATIOKT/Tm;
`
`0.7
`
`ACTIVATION ENERGY
`FOR SELF DIFFUSION
`'[.q)°C
`
`0
`
`o
`
`0.1
`
`0.5 0.0
`0.4
`0.2 0.3
`TEMPERATURE RATl0 |T[Tm|
`
`0.7
`
`0
`
`0.1
`
`0.2 0.3 0.0 0.5 0.6
`TEMPERATURE RATlO(TiTmI
`
`0.1
`
`(d)
`(c)
`Fig. 3. Dependence of various material parameters on T/Tm. T(K) is the temperature and Tm (K) is the
`melting point. (Electroplated stress data from ref. 8.)
`
`than 0.2-0.3 at room temperature, then the thermal stresses will also be reduced by
`recovery.
`
`2.6. Stress distribution
`
`The details of the stress distribution between a coating and substrate can be
`very complex; however, some useful generalizations can be made. Figure 4 shows an
`idealized representation of the normal stresses across the cross-section of a
`homogeneous film and free~standing substrate at a point far from an edge. The stress
`distribution is determined by requirements that the sum of the normal forces and
`bending moments over the total cross-section be zero’. The first requirement means
`that stresses of an opposite sign must form in the substrate to balance those in the
`
`llglmmzt
`0. 6
`
`0
`
`0.1
`
`0.6
`0.5
`0.4
`0.2 0.3
`TEMPERATURE RAlI0lTIl’m!
`
`0.7
`
`(b)
`1.0
`
`INTERNAL STRESS IN
`ELECTROPLATED COATINGS
`T*40°C
`
`0.8
`
`.7
`E
`E
`a 0.4
`if
`m
`
`02
`
`
`
`
`
`I7
`
`J. A. TI-!ORl\'TON. D. W’. HOFFMAN
`
`TENSION |%‘ITERNAI_ STRESS IN FII_.‘:‘.
`
` INTERFACE
`
`CONIPRESSIVE STRESS
`IN SUBSTRATE
`
`
`--——-« NELITRAL PLA\§iE
`
`
`SUBSTRATE
`
`
`
`
`TENSION AI OPPOSITE EDGE OF SUBSTRATE
`
`Fig. 4. Schematic representation of stress distribution across coating and substrate cross-section.
`
`film. The stresses thus reverse sign at the interface. The second requirement means
`that the substrate must bend to balance the bending moment which is produced by
`the stressed film on one ofits surfaces. Thus the stresses in the substrate must change
`
`sign as shown in the figure, so that the substrate stresses on the surface opposite the
`coating have the same sign as those in the coating. In most decorative coating cases.
`the-film thickness is much less than that of the substrate. so that the stress in the
`
`substrate is only a very small fraction of that in the film. Under these conditions the
`substrate bending is negligible. Accordingly no variation in film stress, due to
`substrate bending, is shown in Fig. 4.
`The intrinsic stress must become zero at a free edge of the coating, as shown in
`Fig. 5. Typically, the intrinsic stresses are relatively constant throughout the coating
`thickness, as shown in Fig. 4. Under this condition the interface shear stress is
`proportional to the coating thickness and to the gradient in the average intrinsic
`stress along the coating”. Thus, the interface shear stress is concentrated at the film
`edge, as indicated in Fig. 5. A de—adherence stress, which is normal to the interface
`and dependent on the gradient of the shear stress, is also predicted to develop”.
`Consequently, the interface bond must withstand a shear force per unit width of the
`
`MAGNITUDE OF INTRINSIC
`gmggg INHLM
`
`INTRINSIC STRESS
`IN FILM
`
`COATING
`
`
`
`\
`
`i:*fi‘:J$§;:if€€\\‘
`
`
`NIAGNITUDE OE SHEAR
`STRESS AT INTERFACE
`
`Fig. 5. Schematic representation ofinterfaee shear stress distribution.
`
`
`
`STRESS-RELATED EFFECTS IN THIN FILMS
`
`coating which is proportional to the coating thickness (integrated stress) and an
`associated normal force. Therefore, for a given intrinsic stress level and coating—to-
`substrate bonding, there is a maximum coating thickness that can be tolerated
`before loss of adhesion results.
`The concentrated shear forces at the film edges distort the substrate and film in
`such a way that the strain is continuous across the film—substrate interface at points
`away from the edgel. If the film is scratched, cracked, or if poor adhesion develops at
`some point on the surface, then a redistribution of these shear stresses must occur so
`that the substrate and coating strains are again matched at the interface. Thus a
`scratch can be the precursor to a catastrophic failure of coating adhesion. This
`points out the usefulness of forming scribed scratch patterns prior to performing the
`tape—pu1l type of adhesion tests.
`to
`The considerations described above, although discussed with respect
`intrinsic stresses, also apply to thermal stresses. In fact, a thermal stress analogy is
`particularly useful in illustrating the edge nature of the bonding stresses as discussed
`in the preceding paragraph“). Consider two plates with different thermal expansion
`coefiicients that are fastened together as shown in Fig. 6(a) with bolts that do not
`deflect. Now let the plates be heated. If the bolts were removed, the differences in
`thermal expansion would force the holes out of alignment as shown in Fig. 6(b).
`However, if the end holes are forced into alignment and the end bolts inserted, then
`all the other holes will be in line as shown in Fig. 6(c), z'.e. the strains will match in the
`two plates. Thus the other bolts may be inserted, but they will carry no load. The
`load is carried totally by the end bolts. Now if the end bolts deflect, the load will be
`shared by the inner bolts. If the non-deflecting end bolts are cut, then the load shifts
`to the next bolt in etc. If the upper plate is severed, the load must also appear on the
`bolts on each side of the cut, as shown in Fig. 6(d).
`
`PLATES WI H DIFFERENT
`THERMAL EXPANSION
`
`COEFFICIENIS
`
`LC-\'\»’ TEv\\PERATL?RE
`
`HIGH rrrtiPERAruRE
`
`LOAD on sons
`
`.
`
`was .TE:\iPtRAwR£
`
`Novutrtzcrerc
`, BOLTS
`’
`
`I
`
`f
`
`LOAD ONBOHS
`\
`
`/PLATESEVERID
`i I Q
`
`HIGH VIMPERAILARE
`
`(3)
`
`(b)
`
`(c)
`
`(d)
`
`Fig. 6. Behavior of bolted joints under thermal loads (see ref. 10).
`
`
`
`l4
`
`J. A. THORNTON. D. W’. HOFFMAN
`
`3. EVA PORATED COATINGS
`
`3.]. Law melting point materials
`For materials such as aluminum and lead. the T/Tm values at room temperature
`(0.3 for aluminum and 0.5 for lead) are large enough (greater than 0.2-0.3) to permit
`recovery. Thus even those thermal stresses which are formed in bringing the
`substrates to room temperature relax. Accordingly, what may be observed is not the
`mechanical consequences of the stress, such as adhesion failures, but the conse-
`quences of the stress-driven diffusion. An example is the formation of overgrowth
`known as whiskers or hillocks on thin film surfaces. (One-dimensional growths are
`called whiskers;
`the others are generally called hillocks.) l-lillocks have been
`observed to grow extensively in aluminum, gold. and lead films under conditions of
`high compressive stress’ ‘. The hillock size is often many times the thickness of the
`film. Under some circumstances hillock formation can be reversible. ie. they grow
`when the coating is under compressive stress and shrink when the coating is in
`tension. The exact mechanism for hillocl-1 growth is still under investigation. One
`type of model suggests that compressive stresses are relaxed by the diffusion of
`material from the bulk to the surface of the coating1 2‘ 1 3. Fortunately. the substrate
`temperatures in decorative coating applications are generally too small to cause
`hilloclc and hole formation.
`
`3.2. High melting point materials
`In higher melting point materials. the intrinsic stresses become increasingly
`important. A common manifestation is poor adhesion. Many measurements of the
`internal stresses in evaporated coatings have been reported. Many of the data
`appear inconsistent3. This is probably due to difficulties in controlling the
`deposition parameters. The substrate temperature, residual gas level. deposition
`rate and coating thickness have been found to be important? The substrate
`temperature is particularly important because of its role in determining the relative
`levels of the intrinsic and thermal stresses (see Fig. 2)“. Excellent reviews of the
`internal stresses in evaporated coatings have been provided by HoiTman2"‘.
`Despite the diversity of results, some useful generalizations can be made
`regarding the intrinsic stresses in evaporated metal coatings3'3.
`(1) The intrinsic stress does not seem to be strongly dependent on the substrate
`material.
`
`(2) The incremental stress is relatively independent ofthe film thickness.
`(3) The intrinsic stress in evaporated metal films is almost exclusively tensile,
`with values that can be of the same order of magnitude (0.1 1 GN rn ’ 3) as the yield
`strength of the coating material.
`Compressive stresses in evaporated metal films are occasionally observed and
`generally attributed to impurities”. Evidence of a compressive contribution of
`0.4 GN m “ 2 for every atomic per cent of oxygen has been reported for evaporated
`nickel films”. The primary source of the oxygen appeared to be the evaporation
`filament.
`
`Various models have been suggested to explain the intrinsic tensile stresses
`which develop in evaporated metal films. All the models are similar. in that they
`
`
`
`STRESS~RELATED EFFECTS IN THIN FILMS
`
`I5
`
`invoke some mechanism for the constrained shrinkage during the film growth
`that is required to produce the tensile stress. One class of model assumes that
`arriving atoms are initially incorporated into the growing coating as a disordered
`surface layer because they become buried before they can form a completely ordered
`structure as indicated in Fig. 7(a). Subsequent rearrangement and shrinkage of the
`disordered material produces the tensile stress” 18. A second class invokes the use
`of grain boundaries” ‘9. One such model considers that the stress is built up by the
`volume contraction which results when two grains coalesce to form a single grain
`and, therefore, eliminate a grain boundary”. The shrinkage is equal to the difference
`between the thickness of the eliminated grain boundary and the interatomic spacing.
`A second model argues that a tensile stress is a natural consequence of a pure grain
`boundary (i.e. no trapped impurity species) because the combination of the free
`surface and grain boundary energies per unit area yields an asymmetric force field
`which produces a net tensile strain’ 5. Thus, for example, in a columnar coating, such
`as that shown in Fig. 7(b), the grain boundary area and the intrinsic tensile stress
`increase linearly with the coating thickness.
`
`COATWG ‘W
`
`DISORDERED
`SURFACE LAYER
`
`
`ckvsmunt -
`MATERIAL
`
`3
`
`ANNEAUNG DURING
`,
`comumsmou mooucrs
`CRYSTALLIZATION SHRINKAGE
`AND TENSILE srresss
`
`R
`
`
`
`ta)
`
`STRATE
`5
`
`\\ \3{\\\\\\~*
`‘COATING FLUX
`
`TENSILE STRESS
`
`GRAI N BOUNDARY
`
`
`
`SUBSTRATE
`k\\\\\\\\\\\\\
`(b)
`Fig. 7. Schematic representation of models for intrinsic stress formation: (a) recrystallization model”;
`(b) grain boundary model‘ 5.
`
`4. ELECTRODEPOSITED COATINGS
`
`The stresses in electroplated coatings are of interest because of their similarity
`to evaporated coatings. Electroplating bath temperatures are typically in the
`range 30—100°C. Thus the deposition conditions for high melting point materials
`involve sufficiently low T/Tm values that little recovery can occur. Consequently,
`the structural disorder that forms during deposition is frozen into the coatings,
`producing tensile intrinsic-type stresses similar to those seen in evaporated deposits.
`The higher the melting point, the lower T/Tm and the degree of structural recovery,
`and the higher the stress. This dependence is shown in Fig. 3(d).
`
`
`
`16
`
`J. A. THORNTON. l). w. HOFFMAN
`
`The tensile stress levels in electroplated deposits‘‘ are typically two to three
`times less than those in corresponding evaporated coatings. The reason is
`undoubtedly the improved structural order which one finds in electrodeposited
`coatings as compared with coatings formed by vacuum deposition at
`low
`temperatures. Models for stresses in electroplated coatings are similar to those in
`evaporated coatings‘.
`
`S. SPUTTERED COATINGS
`
`5 .1 . Deposition parameters
`In sputtering the following deposition parameters are expected to influence the
`coating stress: (1) substrate temperature; (2) working gas species; (3) working gas
`pressure; (4) deposition rate; (5) angle of incidence (orientation of deposition surface
`relative to direction of coating flux); (6) apparatus geometry; (7) distance between
`the substrate and the source2°‘“. The substrate temperature assumes a secondary
`importance in the following discussion because of our primary interest in decorative
`coatings deposited on plastic substrates. The apparatus geometry proves to be a
`particularly important consideration. Therefore, the discussion is organized with
`respect to this parameter.
`
`5.2. Cylindrical post nzagnenvrrs
`A cylindrical post magnetron sputtering source25‘2" is shown in Fig. 8. The
`stresses were determined by deposition coatings of known thickness on thin glass
`wafers (about 0.15 mm thick) and interferometrically determining the deflection
`which the stressed coatings induced in the wafers”. The wafer deflection is
`proportional to the integrated stress within the coating (force per unit width). as
`CYL I NDR l CAL MAGNETRON
`
`SUSSTRAHS
`
`SPUTTFRING SOURCE
`
`
`
`CONFINED PLASMA sum
`“WORM MAGNETMLLY
`
`T‘ VACUUM CHAMBER
`v
`v
`~«
`V
`~»
`«
`FIRCULATIN’ Exé ttrrttzom miaretv
`
`Fig. 8. Cylindrical post magnetron sputtering source.
`
`
`
`STRESS—RELATED EFFECTS IN THIN FILMS
`
`l7
`
`discussed previously. By depositing a series of coatings of various thicknesses under
`common deposition conditions, it is possible to determine the thickness dependence
`of the integrated stress and thus the incremental internal stress.
`Cylindrical post magnetrons can be operated over a wide pressure range
`extending from less than 0.13 Pa (1mTorr) to over 13 Pa ( 100 mTorr). Figure 9
`shows the measured internal stress as a function of the working pressure for
`chromium coatings deposited at normal incidence at about 1 nm s ‘ 1 (600 A min “ 1)
`using argon and krypton as working gases and both cast (99.99‘;g) and electroplated
`chromium sputtering targets. Coatings deposited at sufficiently low pressures are
`seen to be in a state of high compression. The magnitude is considerably greater than
`reported for chromium coatings deposited with a d.c. triode type of sputtering
`source”. Coatings deposited above a critical pressure of about 0.27 Pa (2 mTorr)
`develop a tensile stress comparable with that reported for evaporated coatings”. At
`higher pressures, typical of the operating range for conventional diode sputtering
`sources, the tensile stresses are reduced and similar in magnitude to those reported
`for these sources”. At very high deposition pressures, the stresses become negligible.
`The data in Fig. 9 indicate that the working gas species and sputtering target
`purity are of secondary importance compared with the gas pressure. A small
`influence has been seen due to deposition rate. High rates promote larger
`compressive stresses and increase the critical pressure. Additional studies of the
`Im:'orr>
`
`1
`
`IC
`
`100
`
`C‘.'Ll\lDP.lCAL '°OST ’\\»‘«CNETF<‘OhS
`FJDRMAL ANGLE OF WCIDENCE
`UEVOSITION RATE = L0 nm"S
`COATING YHICKNESS = 9U~25O on
`
`DC TRIOUE
`SPlIT.YEREU
`C‘-IR-0:‘.\|‘U\l
`
`:K’)1r
`
`-130
`
`no
`‘
`
`. ELECTRUPIATFD CATHGDI mr
`0 EASI CATHODE rW.°° ‘
`iv
`
`0,
`
`LG
`:=ar>\l RE w9a\
`
`ill
`
`
`
`
`
`WCREMENIALSTRESS4CNIm2>
`
`Fig. 9. Internal stress in sputtered chromium coatings as a function of working gas pressure. (Data from
`refs. 20 and 24.)
`
`
`
`18
`
`J. A. THORNTON. D. w. HOFFMAN
`
`influence of deposition rate, working gas species and radial substrate location are
`currently underway.
`The general stress—pressure dependence shown in Fig. 9 has been observed for
`over 15 elemental metals and alloys2°‘“2“. Typical integrated stress data are shown in
`Fig. 10. The magnitude of the maximum compressive stress formed under typical
`deposition conditions varies with T/Tm as shown in Fig. 11. The similarity with
`Fig. 3(d) should be noted. The critical pressure which separates the compression and
`tension deposition conditions is seen in Fig. 10 to increase with the atomic mass of
`the coating material“ 23. The dependence is shown in Fig. 12.
`
`CYLINDRICAL POST MAGNETRONS
`NORMAL ANGLE OF INCIDENCE
`ARGON WORKING GAS
`Q
`COAHNG THICKNESS ~2CO0A
`
`DEPOSITION RATE ~6oo£i/min.
`
`“/
`NICKEL
`(MASS-59)7/
`. _ . _ _ . . _ _ _ _ . . _ _ _.