- Email: [email protected]
Effect of nickel additive on the grain growth of tungsten wire
Journal of the Less-Common
Metals, 153 (1989)
EFFECT OF NICKEL ADDITIVE TUNGSTEN WIRE IN-HYUNG
MOON,
Department (Received
SUNG-TAG
275
ON THE GRAIN GROWTH OF
OH and YOUNG-LIP
of Materials Engineering,
January
275 - 283
KIM
Han Yang University, Seoul 133-791
(Korea)
6,1989)
Summary
The dependence of grain growth rate in tungsten on the amount of nickel additive was investigated to provide information on the role of nickel in tungsten-activated sintering. Undoped tungsten wires 250 pm in diameter, with various initial grain sizes, were sintered at 1400 “C with or without the nickel activator, and the change of grain size was determined by metallographic analysis. The addition of nickel to tungsten wire accelerated the grain growth as expected. However, the grain growth kinetics during tungsten-activated sintering seems to be different from that in the liquid-phase sintering of the W-Ni system; the rate constant for grain growth increased with increasing thickness of the nickel interlayers.
1. Introduction
Many mechanisms have been proposed to explain the phenomenon of nickel-activated sintering of tungsten powder compacts. Most of these mechanisms are based on the possible role of nickel as a carrier phase for tungsten atom movement; the nickel-rich phase, regardless of where it is located in the contact area between the tungsten powder particles, in the tungsten grain boundary or on the tungsten surface, provides a high-diffusivity path for tungsten self-diffusion [ 1,2]. If the nickel-rich phase present in the tungsten grain boundary can act as a highdiffusivity path for the tungsten atom, a diffusion-controlled process such as tungsten grain growth should be influenced by the presence of this nickel-rich phase, and the rate of growth would be dependent on the thickness of the nickel-rich phase interlayer at the tungsten grain boundary. Nickel added to tungsten filament was found to be segregated mostly at the tungsten grain boundary and to promote grain growth in the tungsten wire [ 3, 41. The thickness of such a segregated nickel interlayer given in the literature [3, 5,6] is diverse, ranging from a mono-atomic layer up to 10 pm. Geguzin and Kibets [6] gave the kinetics of the grain growth of a nickeldoped tungsten powder compact by the following relationship between 0022-5088/89/$3.50
0 Elsevier Sequoia/Printed
in The Netherlands
276
mean grain size R and heat treatment time t: R 3 - Ro3 a t, where R, is the initial grain size; this suggests that the grain growth mechanism might be controlled by tungsten transport through the nickel interlayer. The above relationship between R and t is also valid for describing the grain growth of W-Ni by Ostwald ripening during liquid-phase sintering [ 71. However, Kang and Y oon [ 81 reported that the time-dependent growth exponent decreased from three as the nickel content decreased in the grain growth of the W-Ni system during liquid-phase sintering. This suggests that the kinetics of tungsten grain growth might be dependent on the amount of nickel added to tungsten. The objective of our study is to reconfirm the role of nickel in tungsten grain growth and to examine the relation between the kinetics of tungsten grain growth and the amount of nickel for a possible explanation of materials transport in activated sintering of nickel-doped tungsten.
2. Grain growth in W-Ni Normal grain growth kinetics has been described R(t) = kt””
by (1)
where R is average grain size, t is time, k is a temperature-dependent constant, and n is the growth exponent, which has a value of two to four for pure materials [9]. If the grain growth is assumed to be driven only by surface energy and it is assumed that the grain growth rate is proportional to the average curvature of the grain boundaries, II has a value of two for the single-component system. If R, is the mean grain size at t = 0, eqn. (1) takes the following form: R2-Ro2=kt
(2)
The kinetics of grain growth in a composite structure with an interlayer phase between the grains, such as in the W-Ni system, is different from that for the pure element system. It is usually controlled either by a process of grain boundary reaction or by diffusional mass transfer through the interlayer phase. However, the grain growth law for a composite system can be also expressed as follows: R”-R,“=kt
(3)
If the kinetics of grain growth is controlled by interface reaction, such as solution or precipitation at a grain boundary, the exponent n is equal to two [ 61, resulting in the formula given as eqn. (2). If the kinetics is controlled by a diffusional process in the interlayer phase, the value of n is three [6] ; R3-Ro3=kt form
(4)
The grain growth laws given in eqns. (2) and (4) are identical to the which is valid for the grain growth of solid particles in liquid-phase
277
sintering according to the Lifshitz-Slyozov-Wagner theory [‘7]. Such a similarity can be understood by assuming that the interlayer in the grain boundary plays the same role as the liquid matrix phase in liquid-phase sintering . If it is assumed that the nickel added to tungsten mainly forms a nickel-rich interlayer phase at the tungsten grain boundaries except for a small portion of nickel which is in solution in the tungsten grains, the thickness of the nickel interlayer is related to tungsten grain size in the following form [ 61: 2 CNi X=_----_ 3 CNi* where X is the thickness of the nickel interlayer, cNi is the mean volume concentration of nickel atoms, and CNi* is the equilibrium volume concentration of nickel atoms in the nickel-rich interlayer. As given in eqn. (5), X is directly proportional to R. Therefore the thickness of the nickel-rich interlayer will be increased by the grain growth of tungsten in the presence of a fixed amount of nickel in the W-Ni system. If the grain growth kinetics is controlled by diffusional mass transfer in this interlayer, the grain growth rate will then depend also on the thickness of this interlayer.
3. Experimental
details
The tungsten wire used for the experiment was an undoped wire 250 pm in diameter with a minimum purity of 99.98%, supplied by Alfa Products of Karlsruhe. The wire was first annealed in a hydrogen atmosphere at 850 “C for 1 h, to reduce the surface oxide and to soften the wire, and then was wound tightly on an alumina spool. The initial tungsten grain size was controlled by varying the preheating time at 1400 “C in a hydrogen atmosphere. The average grain diameter of recrystallized tungsten wire was 7.8 and 21.1 I.tm after heat treatments of 40 and 300 min respectively. Nickel (0.45 wt.%) was added to the tungsten wire spool by a salt solution and a reducing treatment [lo]. These spools, with various initial tungsten grain size, and pure tungsten specimens with an equivalent grain size were sintered at 1400 “C in a hydrogen atmosphere for varying times. The grain size was measured by Jeffries’ procedure 1111. The time dependence of the mean grain size during isothermal sintering was determined by measurement on specimens withdrawn from the furnace at predetermined times. Further microstructural development during sintering was observed by scanning electron microscopy. Specifications of the specimens used for this experiment are given in Table 1.
278 TABLE
1
Specifications
Specimen
of specimens
no.
la 2 3 4
Initial W grain size (w-a)
Amount of Ni additive
Not availableb Not available 7.8 21.2
None; 0.45 0.45 0.45
(wt.%) pure W
‘Specimen 1 was pure tungsten wire, used as a reference specimen. bSpecimens 1 and 2 were subjected only to the annealing treatment at 850 “C for 1 h. Therefore they had the as-drawn structure at the beginning of sintering; their grain size was very fine.
4. Experimental
results and discussion
Figure 1 shows photomicrographs of the tungsten grains after various times of sintering at 1400 “C for the tungsten wire specimens with nickel addition (specimen 2) and without nickel addition (specimen 1). As shown in this figure, the average grain size of specimen 2 is much larger than that of specimen 1, as expected. Figure 2 shows the growth of the tungsten grains during sintering and the initial state in specimens 3 and 4. The initial size of tungsten grains was 7.8 and 21.1 pm respectively, and 0.45 wt.% nickel was added to them. Nickel added to tungsten accelerated the tungsten grain growth regardless of the initial size of the tungsten grain at the moment of nickel addition. Figure 3 shows the relationship between the measured mean grain size and sintering time for the various specimens. As shown in this figure, the grain growth rate of the nickel-added tungsten wires was always higher than that of the pure tungsten wire at any moment during the sintering times investigated here. If this relation is plotted on a logarithmic scale, approximate linearity b&ween log R and log t is obtained, as shown in Fig. 4. The slope of the line for specimen 1, pure tungsten wire, is 0.49, whereas that for specimen 2 is about 0.63; specimens 3 and 4 show relatively lower slopes (0.44 and 0.41 respectively) than specimens 1 and 2. The growth exponent of 0.49 obtained from specimen 1 is comparable with eqn. (2), suggesting that the grain growth kinetics in the pure tungsten wire was controlled by a grain boundary reaction as described previously. However, the growth exponent of 0.63 for specimen 2, to which nickel was added without any recrystallization treatment, is an unexpected one. This value is much higher than the exponent 0.5 measured by Kwon et al. [12] during the sintering of nickel-doped tungsten powder compact. However, their data were obtained from a porous tungsten compact (porosity less than 10%). Geguzin and Klinchuk [13] measured a growth exponent of l/3 in
279
specimen
specimen
1
2
60 rnin.
-
120
-
300
-
Fig. 1. Microstructure tungsten wire) sintered
-of specimen 1 (pure at 1400 “C.
tungsten
wire) and specimen
2 (nickel-added
the grain growth of a nickel-doped tungsten sintered compact, suggesting that the grain growth kinetics might be controlled by a diffusions process of tungsten in the nickel interlayer. Furthermore, it is remarkable that Kang and Yoon [ 81 measured a growth exponent of 0.4 for the 99%W-l%Ni system, whereas they obtained an exponent of l/3 for specimens with higher nickel content, such as 3%Ni97~W, in the liquid-phase sintering of W-Ni at 1540 “C. This finding suggests that the grain growth kinetics might be controlled by a diffusional process in the liquid matrix phase in the presence of an ample nickel addition, but another kinetic mechanism, possibly controlled by a grain boundary reaction, must begin to operate when the amount of the interlayer liquid phase is reduced. The slope values, 0.44 and 0.41, obtained from the curves of specimens 3 and 4 respectively, suggest that the grain growth kinetics may not be related to the diffusion-controlled process in the nickel interlayer. However, the curves given in Fig. 4 are not totally linear; they show a slight downward deflection at a common point in time. Therefore, we can divide the curve
Fig. 2. Microstructure of tungsten state in specimens 3 and 4.
grain
grown
during
sintering
at 1400
“C and
110 100
-
70
;
60
2 ;:
50
z ;
40
$3
E 30 3 z
20 10
01 0
“I’,
~‘~“‘~’ I20
240
360
SINTERINC
Fig. 3. Dependence
of mean
480 TIME
600
720
(min.)
grain size on sintering
time for the various
specimens.
initial
281
01’ IO
30 60 >‘+,I 120 SLNTERLNG TIME (min.)
Fig. 4. Logarithmic
plot of relationship
(a) Fig. 5. Scanning electron micrograph heat treatment for 500 min.
4x0
700
between
mean
grain size and sintering
time.
(b) (a) and its nickel
mapping
(b) for specimen
4 after
into two parts, as indicated by the broken lines in Fig. 4; the first part of the curve has a somewhat higher slope than the last part. After such a modification a slope of 0.55 is obtained for specimen 2, and 0.38 and 0.33 for specimens 3 and 4 for the Iater stage of sintering indicated by the broken lines in Fig. 4. The last two values of the slope approach l/3. This means that the continued grain growth beyond a mean grain diameter above 50 pm might be controlled by the diffusion of tungsten in the nickel interlayer. If all of the nickel added to the tungsten is assumed to be present at the tungsten grain boundary, the thickness of the nickel interlayer h can be estimated from eqn. (5) by putting C,i = 0.01, CNiP= 0.7 and R = 25 pm, and a value of about 0.24 pm is obtained. This value is an overestimate, because part of the nickel forms aggregate pools at the grain boundary junctions [14], and another part will be soluted in the tungsten grains [15]. Figure 5 shows scanning electron micrograph (a) and its nickel mapping (b)
282
for specimen 4 after heat treatment for 500 min. As shown in this figure, a nickel-rich interlayer with measurable thickness is present at the tungsten grain boundary. The change of the growth exponent from values close to l/2 to about l/3 during grain growth could be related to the increase of a nickel-rich phase in the tungsten grain boundary as a result of tungsten grain growth. As specimens 3 and 4 show similar slope values, we can assume that the kinetics of grain growth is controlled by the same mechanism for both types of specimen. Inserting into eqn. (3) the measured value of this slope (which is equivalent to the reciprocal) and the initial grain sizes of specimens 3 and 4 (R = 7.8 ym and R = 21.1 pm), we obtain an approximately linear relation- R,2*36) and time t, as shown in Fig. 6. The tempership between (R2.36 ature-dependent rate constant of grain growth k can be estimated from the slope of the above curves. Specimen 4 had a higher slope than specimen 3. This means that the larger the initial size of the tungsten grains, the higher the rate constant of the grain growth. These results are contrary to previous reports [7,13]. 120
100
t
SINTERING
TIME
(min.)
Fig. 6. Plot of (R2*36 - R,2.36) as a function of sintering time t.
If it is accepted that the growth kinetics is controlled by diffusion through the nickel interlayer, the rate constant k is expected to decrease with increasing tungsten grain size, because the thickness of the nickel interlayer increases, as given in eqn. (5). This relation was experimentally observed in the grain growth of the W-Ni powder compact during liquid-phase sintering [8]. However, the result in Fig. 6 is in contradiction to this. The results of the present experiment suggest strongly that the grain growth kinetics of the nickel-added tungsten wire may not be related to the conventionally accepted diffusional process, i.e., the diffusion of tungsten atoms across the nickel interlayer as in the case of the liquid-phase sintering of the W-Ni system.
283
The increase of rate constant with increasing thickness of the nickel interlayer might suggest a positive role of the nickel interlayer for tungsten mass transfer in grain growth. If we can accept the coalescence mechanisms as the possible mechanism for grain growth, as in liquid-phase sintering [7], the higher h value for the thicker nickel interlayer might be explained in part as follows: as the nickel-rich interlayer thickens, it can provide a larger cross-sectional area for tungsten atom transfer from the tungsten contact area to the neck outside. 5. Conclusion Nickel added to tungsten wire accelerated the growth of the tungsten grains, as expected. Based on the measured growth exponent, the grain growth kinetics in nickel-added tungsten wire seems to be controlled by grain boundary reaction in the initial stage of grain growth, but as the grain size continues to increase, the value of the exponent slowly approaches a value characteristic of a diffusioncontrolled process. During the sintering phase that follows, the grain growth kinetics of tungsten in the presence of a small amount of nickel seems to be different from that of W-Ni during liquid-phase sintering, in that the rate constant for grain growth increases with increasing thickness of the nickel interlayer. This is contrary to the relation found in the liquid-phase sintering of W-Ni. Acknowledgments We gratefully acknowledge the financial support of the Korea Science and Engineering Foundation and Deutsche Forschungsgemeinschaft. We also thank Prof. Dr. G. Petzow, Max-Planck Institut fiir Metallforschung, Stuttgart, F.R.G. , for his interest in this work. References 1 R. M. German, Sci. Sintering, 15 (1983) 27. 2 G. H. Gessinger and H. F. Fischmeister, J. Less-Common Met., 27 (1972) 129. 3 T. G. Nieh, Scripfa Metall., 18 (1984) 1279. 4 S. Friedmann and J. Brett, Trans. AIM& 242 (1968) 2121. 5 J. Hofmann, S. Hofmann and L. Tillmann, 2. Metallk., 65 (1974) 721. 6 Ya Y. Geguzin and V. I. Kibets, Fiz. Met. Metalloved., 36 (1973) 1043. 7 R. M. German, Liquid Phase Sintering, Plenum Press, New York, 1985, p. 133. 8 T. K. Kang and D. N. Yoon, Metall. Trans., 9A, (1978) 433.
9 M. P. Anderson, D. J. Stolovitz, G. S. Grest and P. S. Sahni, Acta Metall., 32 (1984) 783. 10 I. H. Moon, Int. J. Powder Metall. Powder Technol., 11 (1975) 27. 11 ANSI/ASTM, E 112-80, Annual Book of ASTM Standards, 1980, Part II, American Society for Testing and Materials, Philadelphia, PA, p. 186. Y. S. Kwon, J. S. Lee and I. H. Moon, Mod. Dev. Powder Metall., 14 (1980) 147. Ya Y. Geguzin and Yu. I. Klinchuk, Fiz. Met. Metalloved., 37 (1974) 1099. I. H. Moon and Y. S. Kwon, Scripta Metall., 13 (1979) 33. 15 A. Gabriel, H. L. Lukas, C. H. Allibert and I. Ansara, 2. Mefallkd., 76 (1985) 589.
12 13 14
Metals, 153 (1989)
EFFECT OF NICKEL ADDITIVE TUNGSTEN WIRE IN-HYUNG
MOON,
Department (Received
SUNG-TAG
275
ON THE GRAIN GROWTH OF
OH and YOUNG-LIP
of Materials Engineering,
January
275 - 283
KIM
Han Yang University, Seoul 133-791
(Korea)
6,1989)
Summary
The dependence of grain growth rate in tungsten on the amount of nickel additive was investigated to provide information on the role of nickel in tungsten-activated sintering. Undoped tungsten wires 250 pm in diameter, with various initial grain sizes, were sintered at 1400 “C with or without the nickel activator, and the change of grain size was determined by metallographic analysis. The addition of nickel to tungsten wire accelerated the grain growth as expected. However, the grain growth kinetics during tungsten-activated sintering seems to be different from that in the liquid-phase sintering of the W-Ni system; the rate constant for grain growth increased with increasing thickness of the nickel interlayers.
1. Introduction
Many mechanisms have been proposed to explain the phenomenon of nickel-activated sintering of tungsten powder compacts. Most of these mechanisms are based on the possible role of nickel as a carrier phase for tungsten atom movement; the nickel-rich phase, regardless of where it is located in the contact area between the tungsten powder particles, in the tungsten grain boundary or on the tungsten surface, provides a high-diffusivity path for tungsten self-diffusion [ 1,2]. If the nickel-rich phase present in the tungsten grain boundary can act as a highdiffusivity path for the tungsten atom, a diffusion-controlled process such as tungsten grain growth should be influenced by the presence of this nickel-rich phase, and the rate of growth would be dependent on the thickness of the nickel-rich phase interlayer at the tungsten grain boundary. Nickel added to tungsten filament was found to be segregated mostly at the tungsten grain boundary and to promote grain growth in the tungsten wire [ 3, 41. The thickness of such a segregated nickel interlayer given in the literature [3, 5,6] is diverse, ranging from a mono-atomic layer up to 10 pm. Geguzin and Kibets [6] gave the kinetics of the grain growth of a nickeldoped tungsten powder compact by the following relationship between 0022-5088/89/$3.50
0 Elsevier Sequoia/Printed
in The Netherlands
276
mean grain size R and heat treatment time t: R 3 - Ro3 a t, where R, is the initial grain size; this suggests that the grain growth mechanism might be controlled by tungsten transport through the nickel interlayer. The above relationship between R and t is also valid for describing the grain growth of W-Ni by Ostwald ripening during liquid-phase sintering [ 71. However, Kang and Y oon [ 81 reported that the time-dependent growth exponent decreased from three as the nickel content decreased in the grain growth of the W-Ni system during liquid-phase sintering. This suggests that the kinetics of tungsten grain growth might be dependent on the amount of nickel added to tungsten. The objective of our study is to reconfirm the role of nickel in tungsten grain growth and to examine the relation between the kinetics of tungsten grain growth and the amount of nickel for a possible explanation of materials transport in activated sintering of nickel-doped tungsten.
2. Grain growth in W-Ni Normal grain growth kinetics has been described R(t) = kt””
by (1)
where R is average grain size, t is time, k is a temperature-dependent constant, and n is the growth exponent, which has a value of two to four for pure materials [9]. If the grain growth is assumed to be driven only by surface energy and it is assumed that the grain growth rate is proportional to the average curvature of the grain boundaries, II has a value of two for the single-component system. If R, is the mean grain size at t = 0, eqn. (1) takes the following form: R2-Ro2=kt
(2)
The kinetics of grain growth in a composite structure with an interlayer phase between the grains, such as in the W-Ni system, is different from that for the pure element system. It is usually controlled either by a process of grain boundary reaction or by diffusional mass transfer through the interlayer phase. However, the grain growth law for a composite system can be also expressed as follows: R”-R,“=kt
(3)
If the kinetics of grain growth is controlled by interface reaction, such as solution or precipitation at a grain boundary, the exponent n is equal to two [ 61, resulting in the formula given as eqn. (2). If the kinetics is controlled by a diffusional process in the interlayer phase, the value of n is three [6] ; R3-Ro3=kt form
(4)
The grain growth laws given in eqns. (2) and (4) are identical to the which is valid for the grain growth of solid particles in liquid-phase
277
sintering according to the Lifshitz-Slyozov-Wagner theory [‘7]. Such a similarity can be understood by assuming that the interlayer in the grain boundary plays the same role as the liquid matrix phase in liquid-phase sintering . If it is assumed that the nickel added to tungsten mainly forms a nickel-rich interlayer phase at the tungsten grain boundaries except for a small portion of nickel which is in solution in the tungsten grains, the thickness of the nickel interlayer is related to tungsten grain size in the following form [ 61: 2 CNi X=_----_ 3 CNi* where X is the thickness of the nickel interlayer, cNi is the mean volume concentration of nickel atoms, and CNi* is the equilibrium volume concentration of nickel atoms in the nickel-rich interlayer. As given in eqn. (5), X is directly proportional to R. Therefore the thickness of the nickel-rich interlayer will be increased by the grain growth of tungsten in the presence of a fixed amount of nickel in the W-Ni system. If the grain growth kinetics is controlled by diffusional mass transfer in this interlayer, the grain growth rate will then depend also on the thickness of this interlayer.
3. Experimental
details
The tungsten wire used for the experiment was an undoped wire 250 pm in diameter with a minimum purity of 99.98%, supplied by Alfa Products of Karlsruhe. The wire was first annealed in a hydrogen atmosphere at 850 “C for 1 h, to reduce the surface oxide and to soften the wire, and then was wound tightly on an alumina spool. The initial tungsten grain size was controlled by varying the preheating time at 1400 “C in a hydrogen atmosphere. The average grain diameter of recrystallized tungsten wire was 7.8 and 21.1 I.tm after heat treatments of 40 and 300 min respectively. Nickel (0.45 wt.%) was added to the tungsten wire spool by a salt solution and a reducing treatment [lo]. These spools, with various initial tungsten grain size, and pure tungsten specimens with an equivalent grain size were sintered at 1400 “C in a hydrogen atmosphere for varying times. The grain size was measured by Jeffries’ procedure 1111. The time dependence of the mean grain size during isothermal sintering was determined by measurement on specimens withdrawn from the furnace at predetermined times. Further microstructural development during sintering was observed by scanning electron microscopy. Specifications of the specimens used for this experiment are given in Table 1.
278 TABLE
1
Specifications
Specimen
of specimens
no.
la 2 3 4
Initial W grain size (w-a)
Amount of Ni additive
Not availableb Not available 7.8 21.2
None; 0.45 0.45 0.45
(wt.%) pure W
‘Specimen 1 was pure tungsten wire, used as a reference specimen. bSpecimens 1 and 2 were subjected only to the annealing treatment at 850 “C for 1 h. Therefore they had the as-drawn structure at the beginning of sintering; their grain size was very fine.
4. Experimental
results and discussion
Figure 1 shows photomicrographs of the tungsten grains after various times of sintering at 1400 “C for the tungsten wire specimens with nickel addition (specimen 2) and without nickel addition (specimen 1). As shown in this figure, the average grain size of specimen 2 is much larger than that of specimen 1, as expected. Figure 2 shows the growth of the tungsten grains during sintering and the initial state in specimens 3 and 4. The initial size of tungsten grains was 7.8 and 21.1 pm respectively, and 0.45 wt.% nickel was added to them. Nickel added to tungsten accelerated the tungsten grain growth regardless of the initial size of the tungsten grain at the moment of nickel addition. Figure 3 shows the relationship between the measured mean grain size and sintering time for the various specimens. As shown in this figure, the grain growth rate of the nickel-added tungsten wires was always higher than that of the pure tungsten wire at any moment during the sintering times investigated here. If this relation is plotted on a logarithmic scale, approximate linearity b&ween log R and log t is obtained, as shown in Fig. 4. The slope of the line for specimen 1, pure tungsten wire, is 0.49, whereas that for specimen 2 is about 0.63; specimens 3 and 4 show relatively lower slopes (0.44 and 0.41 respectively) than specimens 1 and 2. The growth exponent of 0.49 obtained from specimen 1 is comparable with eqn. (2), suggesting that the grain growth kinetics in the pure tungsten wire was controlled by a grain boundary reaction as described previously. However, the growth exponent of 0.63 for specimen 2, to which nickel was added without any recrystallization treatment, is an unexpected one. This value is much higher than the exponent 0.5 measured by Kwon et al. [12] during the sintering of nickel-doped tungsten powder compact. However, their data were obtained from a porous tungsten compact (porosity less than 10%). Geguzin and Klinchuk [13] measured a growth exponent of l/3 in
279
specimen
specimen
1
2
60 rnin.
-
120
-
300
-
Fig. 1. Microstructure tungsten wire) sintered
-of specimen 1 (pure at 1400 “C.
tungsten
wire) and specimen
2 (nickel-added
the grain growth of a nickel-doped tungsten sintered compact, suggesting that the grain growth kinetics might be controlled by a diffusions process of tungsten in the nickel interlayer. Furthermore, it is remarkable that Kang and Yoon [ 81 measured a growth exponent of 0.4 for the 99%W-l%Ni system, whereas they obtained an exponent of l/3 for specimens with higher nickel content, such as 3%Ni97~W, in the liquid-phase sintering of W-Ni at 1540 “C. This finding suggests that the grain growth kinetics might be controlled by a diffusional process in the liquid matrix phase in the presence of an ample nickel addition, but another kinetic mechanism, possibly controlled by a grain boundary reaction, must begin to operate when the amount of the interlayer liquid phase is reduced. The slope values, 0.44 and 0.41, obtained from the curves of specimens 3 and 4 respectively, suggest that the grain growth kinetics may not be related to the diffusion-controlled process in the nickel interlayer. However, the curves given in Fig. 4 are not totally linear; they show a slight downward deflection at a common point in time. Therefore, we can divide the curve
Fig. 2. Microstructure of tungsten state in specimens 3 and 4.
grain
grown
during
sintering
at 1400
“C and
110 100
-
70
;
60
2 ;:
50
z ;
40
$3
E 30 3 z
20 10
01 0
“I’,
~‘~“‘~’ I20
240
360
SINTERINC
Fig. 3. Dependence
of mean
480 TIME
600
720
(min.)
grain size on sintering
time for the various
specimens.
initial
281
01’ IO
30 60 >‘+,I 120 SLNTERLNG TIME (min.)
Fig. 4. Logarithmic
plot of relationship
(a) Fig. 5. Scanning electron micrograph heat treatment for 500 min.
4x0
700
between
mean
grain size and sintering
time.
(b) (a) and its nickel
mapping
(b) for specimen
4 after
into two parts, as indicated by the broken lines in Fig. 4; the first part of the curve has a somewhat higher slope than the last part. After such a modification a slope of 0.55 is obtained for specimen 2, and 0.38 and 0.33 for specimens 3 and 4 for the Iater stage of sintering indicated by the broken lines in Fig. 4. The last two values of the slope approach l/3. This means that the continued grain growth beyond a mean grain diameter above 50 pm might be controlled by the diffusion of tungsten in the nickel interlayer. If all of the nickel added to the tungsten is assumed to be present at the tungsten grain boundary, the thickness of the nickel interlayer h can be estimated from eqn. (5) by putting C,i = 0.01, CNiP= 0.7 and R = 25 pm, and a value of about 0.24 pm is obtained. This value is an overestimate, because part of the nickel forms aggregate pools at the grain boundary junctions [14], and another part will be soluted in the tungsten grains [15]. Figure 5 shows scanning electron micrograph (a) and its nickel mapping (b)
282
for specimen 4 after heat treatment for 500 min. As shown in this figure, a nickel-rich interlayer with measurable thickness is present at the tungsten grain boundary. The change of the growth exponent from values close to l/2 to about l/3 during grain growth could be related to the increase of a nickel-rich phase in the tungsten grain boundary as a result of tungsten grain growth. As specimens 3 and 4 show similar slope values, we can assume that the kinetics of grain growth is controlled by the same mechanism for both types of specimen. Inserting into eqn. (3) the measured value of this slope (which is equivalent to the reciprocal) and the initial grain sizes of specimens 3 and 4 (R = 7.8 ym and R = 21.1 pm), we obtain an approximately linear relation- R,2*36) and time t, as shown in Fig. 6. The tempership between (R2.36 ature-dependent rate constant of grain growth k can be estimated from the slope of the above curves. Specimen 4 had a higher slope than specimen 3. This means that the larger the initial size of the tungsten grains, the higher the rate constant of the grain growth. These results are contrary to previous reports [7,13]. 120
100
t
SINTERING
TIME
(min.)
Fig. 6. Plot of (R2*36 - R,2.36) as a function of sintering time t.
If it is accepted that the growth kinetics is controlled by diffusion through the nickel interlayer, the rate constant k is expected to decrease with increasing tungsten grain size, because the thickness of the nickel interlayer increases, as given in eqn. (5). This relation was experimentally observed in the grain growth of the W-Ni powder compact during liquid-phase sintering [8]. However, the result in Fig. 6 is in contradiction to this. The results of the present experiment suggest strongly that the grain growth kinetics of the nickel-added tungsten wire may not be related to the conventionally accepted diffusional process, i.e., the diffusion of tungsten atoms across the nickel interlayer as in the case of the liquid-phase sintering of the W-Ni system.
283
The increase of rate constant with increasing thickness of the nickel interlayer might suggest a positive role of the nickel interlayer for tungsten mass transfer in grain growth. If we can accept the coalescence mechanisms as the possible mechanism for grain growth, as in liquid-phase sintering [7], the higher h value for the thicker nickel interlayer might be explained in part as follows: as the nickel-rich interlayer thickens, it can provide a larger cross-sectional area for tungsten atom transfer from the tungsten contact area to the neck outside. 5. Conclusion Nickel added to tungsten wire accelerated the growth of the tungsten grains, as expected. Based on the measured growth exponent, the grain growth kinetics in nickel-added tungsten wire seems to be controlled by grain boundary reaction in the initial stage of grain growth, but as the grain size continues to increase, the value of the exponent slowly approaches a value characteristic of a diffusioncontrolled process. During the sintering phase that follows, the grain growth kinetics of tungsten in the presence of a small amount of nickel seems to be different from that of W-Ni during liquid-phase sintering, in that the rate constant for grain growth increases with increasing thickness of the nickel interlayer. This is contrary to the relation found in the liquid-phase sintering of W-Ni. Acknowledgments We gratefully acknowledge the financial support of the Korea Science and Engineering Foundation and Deutsche Forschungsgemeinschaft. We also thank Prof. Dr. G. Petzow, Max-Planck Institut fiir Metallforschung, Stuttgart, F.R.G. , for his interest in this work. References 1 R. M. German, Sci. Sintering, 15 (1983) 27. 2 G. H. Gessinger and H. F. Fischmeister, J. Less-Common Met., 27 (1972) 129. 3 T. G. Nieh, Scripfa Metall., 18 (1984) 1279. 4 S. Friedmann and J. Brett, Trans. AIM& 242 (1968) 2121. 5 J. Hofmann, S. Hofmann and L. Tillmann, 2. Metallk., 65 (1974) 721. 6 Ya Y. Geguzin and V. I. Kibets, Fiz. Met. Metalloved., 36 (1973) 1043. 7 R. M. German, Liquid Phase Sintering, Plenum Press, New York, 1985, p. 133. 8 T. K. Kang and D. N. Yoon, Metall. Trans., 9A, (1978) 433.
9 M. P. Anderson, D. J. Stolovitz, G. S. Grest and P. S. Sahni, Acta Metall., 32 (1984) 783. 10 I. H. Moon, Int. J. Powder Metall. Powder Technol., 11 (1975) 27. 11 ANSI/ASTM, E 112-80, Annual Book of ASTM Standards, 1980, Part II, American Society for Testing and Materials, Philadelphia, PA, p. 186. Y. S. Kwon, J. S. Lee and I. H. Moon, Mod. Dev. Powder Metall., 14 (1980) 147. Ya Y. Geguzin and Yu. I. Klinchuk, Fiz. Met. Metalloved., 37 (1974) 1099. I. H. Moon and Y. S. Kwon, Scripta Metall., 13 (1979) 33. 15 A. Gabriel, H. L. Lukas, C. H. Allibert and I. Ansara, 2. Mefallkd., 76 (1985) 589.
12 13 14