Interfacial reactions in AgZn thin film couples

Thin Solid Films, 186 (1990) 8 7 4 8

87

INTERFACIAL REACTIONS IN Ag-Zn THIN FILM COUPLES A. K. BANDYOPADHYAY AND S. K. SEN Department of Materials Science, Indian Association for the Cultivation of Science, Jadavpur, Calcutta 700032 (India) SUCHITRA SEN Central Glass and Ceramic Research Institute, Calcutta 700032 (India) (Received January 31, 1989; revised June 19, 1989; accepted September 5, 1989)

Interfacial reactions in Ag-Zn thin film couples have been investigated by measuring the contact resistance and composite electrical resistance with time and temperature in order to understand the kinetic behaviour of the system. The resistivity measurements have been supplemented by X-ray diffraction (XRD), scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The activation energy of diffusion has been found to be 6.1 x 10-20 j from electrical resistivity measurements where the model assumes rapid grain boundary diffusion followed by a defect-assisted path for diffusion into the grain and 8.3 × 10- 20 j from contact resistance measurements where the model is based on grain boundary diffusion. XRD indicates the growth of 13'-Ag-Zn phase even at room temperature which changes to the 13phase above 250 °C. SEM confirms the diffusion of silver through zinc grain boundaries, especially at the interface boundary. TEM indicates the growth of grain size with annealing and confirms the presence of the 13'phase.

1. INTRODUCTION Extensive interactions between thin films occur at temperatures lower than the corresponding reactions of well-annealed bulk specimens. This is because of the fact that, in a thin film, the interface constitutes a major fraction of the total volume compared with the bulk specimen, thereby giving prominence to the processes related to the surface. Sometimes the diffusion may proceed at a temperature as low as room temperature, whereas diffusion at such a temperature might not be pronounced in the bulk specimen. In our previous study 1 the diffusion of Ag/Sn bimetallic films was shown to proceed at room temperature with the rapid growth of an intermetallic phase. However, tin belongs to the so-called fast diffuser system where the bulk diffusivity of a monovalent metal such as silver is very high, almost of the order of the grain boundary diffusivity. In this paper we have chosen the system Ag/Zn where lattice diffusivity is many times slower. It was previously observed by Simic and Marinkovic 2 that zinc reacts with silver at room temperature. Until now no detailed work on this system has been reported. Thus the purpose of the present investigation is to identify and monitor the

0040-6090/90/$3.50

© ElsevierSequoia/Printedin The Netherlands

88

A. K. BANDYOPADHYAY, S. K. SEN, S. SEN

interfacial reactions in this system. In order to obtain both quantitative and qualitative data about the diffusion kinetics, phase identification, microstructural changes and diffusion mechanism, electrical resistivity measurements have been supplemented by X-ray diffraction (XRD), transmission electron microscopy (TEM) and scanning electron microscopy (SEM) studies. 2. EXPERIMENTAL DETAILS

The interfacial reaction is monitored through the change in electrical resistance measurement with time. Two types of sample configurations (Fig. 1) have been used. One type is for composite resistance measurement (Fig. l(a)) and the other is for contact resistance measurement (Fig. l(b)). The experiments in both cases were performed in high vacuum in situ. The necessary experimental set-up for this purpose was published previously 1. The change in resistance was automatically recorded in a microvolt amplifier recorder supplemented by a microvolt potentiometer for occasional checking of the recorded data.

4i

g

•---.~v

)/

(a)

! (b) /~V Fig. 1. Schematic arrangements for (a) composite resistance measurement and (b) contact resistance measurement: I, current leads; V, potential leads.

3. RESULTS 3.1. The composite resistance

For composite resistivity measurements, silver was first deposited on a glass substrate and annealed at 100 °C and then cooled to room temperature. Zinc was subsequently deposited. The fractional change in composite resistance with time has been formulated by Hall et al. 3 as AR 4~ Co R - 7lz1/2 (Dt) 1/2 .

(1)

INTERFACIAL REACTIONS IN

A g - Z n COUPLES

89

where ~ is the increase in resistivity for unit change in concentration, 2 the grain size, t the time of diffusion, D the diffusion coefficient and Co the saturation value of the diffusion along the grain boundaries. The thicknesses of silver and zinc films were approximately 200 nm and 300 nm respectively. The temperature of the substrate was found to increase by 8 °C just after deposition, and it cooled down to room temperature slowly after about 30 min. Naturally the resistance of the sample was not steady for few minutes. However, when necessary corrections for the temperature coefficient of resistivity were allowed for, the initial resistance was found to be 3.166 t2, which increases slowly, much more slowly than was found for the Ag/Sn system. The normalized change AR/R in resistance against time at room temperature has been plotted on logarithmic graph paper with those found at higher temperatures (Fig. 2). The sample was next heated to 200 °C for more than 3 h and the change in resistance with time recorded. Subsequently the sample was heated to 300 °C for approximately 4 h. The experiment at 102 °C was performed using another sample heated for 5 h. In fact after 1 h of annealing the specimens exhibited a saturation effect (Fig. 2). The initial portion of the curves has a slope of 1/2, showing that the reaction is diffusion controlled. The mean temperature coefficient of resistance ~ = (1/R)dR/dT of the samples has been found to be about 1.4 x 10- 3 °C- 1 within the temperature range between room temperature and 300 °C. The samples when annealed also exhibit a saturation effect. The saturation value of AR/R has been taken to be approximately 1.0. With ~C o = 1 and 2, the average grain size, taken as 200nm in eqn. (1), the diffusion co-efficients at room temperature, 102,200 and 300 °C have been plotted in Fig. 3. From an r.m.s, fitting of the graph, the activation energy of diffusion was found to be 6.1 × 10- 2o j.

I0

IOO MINUTES IN THE FURNACE

IO~)O

tO( :)O

Fig. 2. Fractional change in resistance of Ag/Zn composite film with aging time at various temperatures. The initial slope of the curve is parallel to the curve having a t ~/2 dependence.

90

A. K. BANDYOPADHYAY, S. K. SEN, S. SEN

-SG

sgo

200

T(°C)

,o~

//

~x

o -45

_¢

-4(:

2.0

2'.5 3'.0 3.5 iO00/T r K-') Fig. 3. •ariati•ninthedi•usi•nc•ef•cient•nD(T)•as•btainedfr•mthep••tspresentedinFig.2•withthe reciprocal absolute temperature liT.

3.2. The contact resistance For contact resistance measurement specimens were prepared by depositing 100 nm of silver and subsequently 100 nm of zinc without breaking the vacuum in a special geometrical configuration as shown in Fig. l(b). The contact resistance is defined as the resistance across two adjacent legs of the couple when a potential is applied across the other two adjacent legs. The specimens were annealed at different temperatures and different times while their electrical contact resistances were monitored. The contact resistance AR c is given by ARc = Rc(t,T)- Rc(0,T) = K(T) t"

(2)

where K (T) denotes a temperature-dependent and time-independent parameter and n denotes a time exponent. For diffusion-controlled processes n is 0.5. A simple model developed by Bauer and Jordan a associates K(T) with the diffusion coefficients D for thin film couples. It has been assumed that rapid diffusion occurs at the early stages of diffusion from the interface towards the free surface along a periodic array of isolated and independent parallel grain boundaries characterized by a thickness 6 and separation 2 (the grain size). The rate constant K(T) can be expressed as 4

Ap6 K(T) = g ~ - O 1/2

(3)

where Ap is the difference between the average electrical resistivity Po of the elemental components and the maximum resistivity of the resultant alloy or

INTERFACIAL REACTIONS IN

Ag-Zn

91

COUPLES

0.!

0.0~

o - . R.T A ..- ~,0°=

0.0(

o-- 200°©

0.0~

00;

0

(a)

i

i

l

2

I

tVz (h,.V,)3

I

4

0.20r

0.16

i..0-~

0.12

o

o am°c 6 327°C n 36900

0

(b)

0.45

0.9

1.35

tvt (hr.',t)

Fig. 4. Change A R c ( t , T ) in contact resistance as a function of the square root t 1/2 of time at various temperatures.

92

A. K. B A N D Y O P A D H Y A Y , S. K. SEN, S. SEN

compound, d the film thickness and g an empirical parameter equal to about 0.2. Ap, 6 etc. are uncertain factors in the above expression. However, if the interdiffusion proceeds for a long time i.e. when the diffusion length l ~> d, AR~ approaches a constant value given by 5

ARc = g Ap/d

(4)

If the experimental value of ARc is used, the value of Ap is found to be about 5p0. The initial room temperature contact resistance R c was found to be 65.9 mr2 for the couple consisting of 100 nm silver and 100 nm zinc. The change in resistance AR~ and the corresponding K(T) at room temperature have been found to be lower than those for the Ag/Sn system. K(T) is found to be 5.6 x 10 5 f2 s 1/2. The sample was subsequently heated to 80 and 200 °C and the evolution of the contact resistance ARc(t,T ) was measured. The results have been plotted as AR~ against t ~'2 in Fig. 4(a). Similar graphs of AR~ against t 1/z (Fig. 4(b)) were obtained when the experiment was repeated with another sample at 281, 327 and 369°C. The corresponding K(T) values are shown in Table I. A parabolic growth law has been observed. The results also show distinctly another parabolic region at a later stage of the diffusion. Similar behaviours have been observed by Shearer et al. s with the AI/Cu system. The first parabolic region might be due to short-circuit interdiffusion in both silver and zinc and the second might be due to homogenization of grain interiors. An approximate value of the diffusion coefficient D can be found from the time t* of termination of the first parabolic region which is given by t* ~ dia/D

(5)

where D refers to the grain boundary diffusion. The diffusion coefficients calculated from the above formula are 4.2 x 10- 16 m 2 S i and 2.4 x 10 ~5 m 2 s 1 at 80 °C and 200 ~C respectively. From the constants of the initial slopes of Fig. 4 K(T) values have been determined at different temperatures and they have been plotted as function of the reciprocal of the absolute temperature (Fig. 5). The activation energy is found to be 8.3 x l0 2oj. TABLE 1 T H E RATE C O N S T A N T S

K(T) A T

7'( Ct K(TJ(~s

DIFFERENT TEMPERATURES

27 12)

5.6x10

5

4.6x10

80

200

281

~

2.4x10 -3

7.4x10 ~

327 9.6x10

"~

369 1.4×10

2

3.3. X-ray diffraction Two types of films, one with silver at the bottom and zinc at the surface and another with zinc at the bottom and silver at the surface, on glass substrates were subjected to XRD studies. The diffraction patterns of unannealed films and films annealed at 300~'C for 2 h have been chart recorded with a Phillips X-ray diffractometer (model PW 1050.51) using monochromatic LiF-filtered CuKct radiation. The X-ray diffractograms are shown in Figs. 6 and 7. In the first case, the unannealed sample exhibits mostly zinc lines with lY-Ag-Zn oriented in the 202 and 300 directions. After annealing, the (300) intensity of 13'-Ag-Zn is seen to decrease

INTERFACIAL REACTIONS IN

Ag-Zn

93

COUPLES

T ('C) -JO

,

,

i

=

"~.5

2.0

' 2,5

l 3.0

3.5

moo& (K-') Fig. 5. Variation in the parameter In K(T), as obtained from the slopes of the plots presented in Fig. 4, with the reciprocal absolute temperature 1/T.

o qZ

o

~

,-,

o

(a)

ta

(a)

t.

.'Nc

;. I

i

i

,

,

I

45

43

41

39

37

35

N

37 .55 45 43 41 39 20 Oeorees) (b) 2 0 degrees (b) Fig. 6. X-ray diffractograms of an Ag/Zn film having the configuration of silver at the bottom and zinc at the top: (a) unannealed; (b) annealed at 300 °C for 2 h. Fig. 7. X-ray diffractograms of an Ag/Zn film having the configuration of zinc at the bottom and silver at the top: (a) unannealed; (b) annealed at 300 °C for 2 h.

94

A. K. BANDYOPADHYAY, S. K. SEN, S. SEN

whereas the 13-Ag-Zn alloy (110) intensity is found to grow. In the second case, i.e. when silver is at the t o p a n d zinc at the b o t t o m , silver lines are p r o m i n e n t in the u n a n n e a l e d sample; these diminish in intensity with annealing a n d 13'-Ag-Zn is seen to exist for b o t h the a n n e a l e d a n d the u n a n n e a l e d samples.

3.4. Scanning electron microscopy and transmission electron microscopy A convenient w a y of d e t e r m i n i n g the diffusion m e c h a n i s m is t h r o u g h the use of an offset film c o n f i g u r a t i o n 6 in which, d u r i n g the fabrication of bimetallic couples, the second metal film is d e p o s i t e d slightly offset with respect to the first. In this technique three s e p a r a t e regions o f pure silver, pure zinc a n d A g / Z n o v e r l a p have been p r e p a r e d L The u n a n n e a l e d s a m p l e does not exhibit any m a r k e d diffusion at r o o m t e m p e r a t u r e (Fig. 8(a)). The samples which were a n n e a l e d at 300 °C for 2 h

(a)

(b)

(c)

(d)

te) Fig. 8. Scanning electron micrographs of Ag/Zn samples showing the boundary between Ag/Zn overlap and zinc under different conditions: (a) unannealed specimen; (b) annealed at 300 °C for 2 h (zinc at the bottom, silver at the top); (c) same as (b) but at a higher magnification; (d) annealed at 300 °C for 2 h (silver at the bottom, zinc at the top); (e) same as (d) but in the backscanered mode.

INTERFACIAL REACTIONS IN

A g - Z n COUPLES

95

have two configurations, one set with zinc at the bottom and silver at surface and another set having the opposite configuration. In the first case, it is seen that the boundary between the interface and zinc becomes blurred. A short distance from the interface on the zinc side white pimples are seen whereas voids are observed at the interface (Figs. 8(b) and 8(c)). In the second case, when silver is at the bottom and zinc is at the top, some sort of island structure (Fig. 8(d)) is seen to appear along the interface boundary. When the atomic contrast mechanism is used in the backscattered mode, these pimples and particles are found to be either silver or an Ag-Zn alloy (Fig. 8(e)). This confirms the mass transport of silver through zinc. The T E M measurements have been performed using a 200 kV Phillips (JEOL) microscope. The diffraction pattern of the unannealed sample has rings with more or less the same intensities and the corresponding microstructure exhibits particles of a

(a)

(b)

(c)

(d)

'

i

5

% (e) Fig. 9. (a) SAD pattern from the A g - Z n film in the unannealed condition. (b) Microstructure of the Ag-Zn film in the unannealed condition. (c) SAD pattern from the A g - Z n film annealed at 250 °C for 2 h. (d) Bright field micrograph of the Ag-Zn film annealed at 250 °C for 2 h. (e) Dark field micrograph of the A g - Z n film annealed at 250 °C for 2 h.

96

A. K. BANDYOPADHYAY, S. K. SEN, S. SEN

very small grain size (Figs. 9(a) and 9(b)). The sample annealed at 250 °C for 2 h displays a strong (1 ! 1) line of silver and (101) lines of zinc with spotted weak lines of [Y-Ag-Zn alloy (Fig. 9(c)). The bright field micrograph (Fig. 9(d)) exhibits some large particles about 50-70 nm in size surrounded by a void region. The small spots in the void region are 10-15 nm in size. Dark field imaging revealed the existence of three phases, i.e. silver, zinc and Ag-Zn (Fig. 9(e)). 4. DISCUSSION 4.1. Measurement of the evolution of contact and composite resistance The measurements of the compositional depth profile with conventional technique such as Auger electron spectroscopy, secondary ion mass spectrometry and Rutherford backscattering are undoubtedly accurate for calculating interdiffusion but it is equally helpful to follow the degradation process continuously through electrical resistivity measurements. The two methods employed in this work to follow the degradation process are based on two different geometrical configurations of the diffusing elements. The configuration for composite resistance measurements is based on a model 3 where diffusion proceeds instantaneously along the grain boundaries to a saturation value and then laterally from the boundary (type B kinetics). The diffusion coefficient is of the order of 10-18mZs -1 at 300°C (Fig. 3) whereas the activation energy is 6.1 × 10-z°J. The bulk value of the activation energy of silver in zinc is about 2.9 × 10- ~9 j and that of zinc in silver is 1.7 × 10-19 j. The activation energy is low compared with the bulk value. According to Hall et al. 3, this might be due to the fact that the diffusion in the thin film is defect assisted. The defect density in thin films is high and the activation energy of the diffusion coefficient might simply be a result of the movement of frozen-in defects where no energy is needed for the creation of these defects. The activation energy in thin films is lower in such a case. For contact resistance measurements the model is based on type C kinetics where the diffusion predominantly takes place along the grain boundaries, especially during the initial period. Consequently the diffusion coefficients derived with this model are found to be about six orders of magnitude higher than those found using the composite configuration. The activation energy is also found to be slightly higher. Similar results have been found for the diffusion of palladium in gold by Hall et al. 3 The approximate values of the diffusion coefficients obtained using eqn. (5), however, are found to be smaller. 4.2. X-ray diffraction, scanning electron microscopy and transm&sion electron microscopy According to the phase diagram a stable hexagonal [3' phase is seen to form at room temperature which is transformed to the b.c.c. [3 phase from 260 °C, more or less at the same composition. Recently, Matsuo and Torri 7 have also reported similar transitions. Therefore our XRD study confirms the above results. For samples where silver is on top the [3'-[3 transformation could not be detected, probably because of the masking of silver lines on the weak [3line. From the SEM photographs it is clear that there are silver-depleted regions at

INTERFACIAL REACTIONS IN

Ag-Zn

COUPLES

97

the boundary of the zinc and Ag/Zn overlap regions. The formation of pimples on the zinc side indicates grain boundary diffusion of silver through zinc grain boundaries. The backscattered mode indicates evidence of mass transport of silver through zinc along the interface. The diffraction pattern of the annealed specimen is spotty, indicating thereby the increase in grain size with annealing. This is confirmed when we look at the micrograph. The grain size of the unannealed specimen is much smaller than that of the annealed specimen. The micrograph of the annealed specimen reveals large grains of zinc, some of which have been alloyed (13' phase), and fine-grained silver particles between large grains. The dark field micrograph of the same annealed specimen indicates the presence of alloyed grains. 5. CONCLUSION

The basic process occurring during interdiffusion in Ag/Zn thin film couples can now be visualized as follows. When silver and zinc are brought into contact with each other, the diffusion of silver into zinc grain boundaries occurs but the rate is much slower than that seen in Ag/Sn couples. The Ag-Zn (13') phase is nucleated. With annealing, the grain size of the zinc film increases and above 250 °C some of the 13'phase is changed to the 13phase. In our previous study of the diffusion of Ag/Sn couples it was not possible to detect the diffusion above room temperature resistometrically. This is perhaps because the solubility of silver in tin is very small (0.06~o) whereas the solubility of silver in zinc is much higher (about 3~o).The bulk diffusion coefficient of silver in tin is several times higher than that of silver in zinc, as a result of which the formation of a uniform solid solution in the interior of tin grains is possible even at room temperature. An intermetallic compound of 7-Ag3Sn at the interface might also be responsible for hindering further diffusion at higher temperature. However, in the Ag/Zn couples the solubility is higher and most of the diffusion takes place along the grain boundary. Although an intermetallic alloy of silver and zinc (lY-Ag-Zn) is found to be formed initially, this changes to the fl form as the diffusion proceeds at higher temperatures. From the electrical resistivity measurements it can be seen that the kinetics of the growth of intermetallic phases is diffusion controlled and apparently the activation energy of phase formation is not affected by the structural change. As the grain size increases with temperature the grain boundary paths become fewer. Since the temperature is not high enough for lattice diffusion, defectenhanced diffusion might occur. However, when the solubility limit is reached, further diffusion does not take place. Therefore a saturation effect is also observed in Ag/Zn samples at a higher temperature. ACKNOWLEDGMENTS

Thanks are due to Professor S. P. Sen Gupta for his assistance in the XRD work and keen interest in this research. One of us (S. Sen) is thankful to the Director, Central Glass and Ceramic Research Institute, for his kind permission to publish the work.

98

A . K . BANDYOPADHYAY, S. K. SEN, S. SEN

REFERENCES 1 2 3 4 5 6 7

S.K. Sen, A. Ghorai, A. K. Bandyopadhyay and S. Sen, Thin Solid Films, 155 (1987) 243 253. V. Simic and Z. Marinkovic, Thin Solid Films, 61 (1979) 149-160. P.M. Hall, J. M. Morabito and J. M. Poate, Thin Solid Films, 33 (1976) 107 134. C.L. BauerandA. G. Jordan, Phys. StatusSolidiA, 47(1978) 321. M.P. Shearer, S.K. SenandC. L. Bauer, Phys. StatusSolidiA, 69(1982) 139. S. Nakahara, R. J. McCoy, L. Buene and J. M. Vandenberg, Thin Solid Films, 84 (1981) 185 196. Y. Matsuo and Y. Torri, J. Jpn. Inst. Metall., 51 (1987) 31 36.