- Email: [email protected]
Spark plasma sintering behaviour of copper powders having different particle sizes and oxygen contents
Powder Technology 291 (2016) 170–177
Contents lists available at ScienceDirect
Powder Technology journal homepage: www.elsevier.com/locate/powtec
Spark plasma sintering behaviour of copper powders having different particle sizes and oxygen contents C. Menapace a,⁎, G. Cipolloni a, M. Hebda b, G. Ischia a a b
Department of Industrial Engineering, University of Trento, via Sommarive 9, 38123 Trento, Italy Institute of Materials Engineering, Cracow University of Technology, ul.Warszawska 24, 31-155 Kraków, Poland
a r t i c l e
i n f o
Article history: Received 20 July 2015 Received in revised form 9 December 2015 Accepted 19 December 2015 Available online 21 December 2015 Keywords: Spark plasma sintering Copper Oxygen content
a b s t r a c t Two different copper powders with different initial oxygen contents (H = high, L = low) were milled and spark plasma sintered to obtain nanosized bulk copper. The characteristics of the milled powders were analysed together with their behaviour in sintering. Spark plasma sintering was carried out at 950 °C for 1 min, applying a pressure of 30 MPa at 700 °C. The consistent difference observed between the two powders is related to the much lower particle size of powder H and its consequent higher surface oxidation. It was observed that even if the finer particle size should lead to a faster and better densification, the much higher oxide amount predominates, slowing down the sinterability of the powder. Moreover, during the sintering process, a higher electrical resistance was measured for powder H due to its higher surface (particle–particle contact area) and oxide content. The degassing process occurring on sintering, caused by the decomposition of the stearic acid used as a process control agent in ball milling, is also markedly influenced by the powder particle size, showing an anticipated decomposition in sample H due to the higher reactivity of this powder. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Spark plasma sintering (SPS) is a powder consolidation process based on the simultaneous application of a low-voltage, high-energy pulse current and uniaxial pressure [1,2]. The importance of this sintering method has been demonstrated by the large number of papers published during the past decade. SPS is used to sinter metallic powders, as well as ceramics, intermetallics and composites [3–6]. One of the main benefits of this technology is its localized heat input and short sintering time, which reduces the grain growth linked to high temperatures. Therefore, SPS is one of the most suitable sintering processes for materials that are very sensitive to temperature like nanostructured powders, being able to consolidate the compact to near full density with limited grain growth. Different parameters influence the sintering behaviour of a powder; two of the most important are the particle size and the oxygen content. Regarding the particle size, it is well known that the thermodynamic driving force for sintering is proportional to the specific surface area of the powder. Particle size also influences the number and the extension of the contact areas and, in turn, the local pressure and the electrical resistance, which are of great importance in SPS due to its peculiar heating mechanism [7,8]. The second parameter is the powder oxygen content, which refers to the amount of oxides present on the powder particle surface. In ⁎ Corresponding author. E-mail address: [email protected] (C. Menapace).
http://dx.doi.org/10.1016/j.powtec.2015.12.020 0032-5910/© 2015 Elsevier B.V. All rights reserved.
sintering, the presence of oxides on the particle surface is detrimental because it slows down all of the mechanisms of mass transport responsible for neck formation. On the other hand, if the metallic powder is milled before sintering, the presence of oxides on the particles is beneficial because they exert a cutting action, promoting the fragmentation of the powder [9]. In the present paper, two different Cu powders with different initial oxygen levels were milled for 100 h to obtain a nanometric grain size and then consolidated in SPS to retain their nanosized structure. Many researchers have focused their attention on the spark plasma sintering behaviour of ultra-fine and nanosized copper [10–16], paying attention mainly to the evolution of the grain size and the microstructural characteristics during sintering. Less attention has been dedicated to the study of the influence of the powder oxygen content on the milling and sintering processes. Because commercially pure copper powders contain a certain amount of oxygen and a drawback associated with mechanical milling is the further unavoidable oxygen contamination by the milling media, atmosphere and process control agent (PCA), it is of great importance to evaluate the effect of the oxygen content on the behaviour of this powder. 2. Materials and experimental procedure Two different Cu powder grades with different initial oxygen contents (powders H = high oxygen amount = 0.5 wt.% and L = low oxygen amount = 0.1 wt.%) were milled for 100 h and sintered in SPS to obtain nanostructured copper compacts.
C. Menapace et al. / Powder Technology 291 (2016) 170–177
The two powders have a similar particle size and morphology, as shown in Fig. 1. The particle size, apparent density and oxygen content are summarized in Table 1. Milling was carried out in a Fritsch Pulverisette 5 planetary ball mill (Idar-Oberstein, Germany) with a vial and balls of hardened steel at a rotation speed of 400 rpm and a ball to powder ratio (BPR) of 10. 0.5 wt.% of stearic acid was used as a PCA (process control agent). To avoid oxidation, the vial was set under vacuum before milling. The milled powders were examined under light optical microscopy (LOM), scanning electron microscopy (SEM), and transmission electron microscopy (TEM) and through X-ray diffraction. For the microstructural analysis, a conventional grinding and polishing procedure was used. The grain size and lattice strain of the milled powder and sintered discs were determined through X-ray diffraction (XRD) with a Cu Kα (λ = 0.154056 nm) source and an image plate detector over the 2θ range from 30° to 120° in reflection geometry. The experimental spectra were elaborated using materials analysis using diffraction (MAUD). The dislocation density was calculated through the formula [17] ρ1=2 ¼
pffiffiffiffiffiffiffiffiffiffiffi 2 3ε1=2 Db
ð1Þ
171
Table 1 Characteristics of the two initial powders.
Particle size % cumulative
Apparent density (g/cm3) O2
N75 μm N45 μm and b75 μm b45 μm
H
L
8.7 57.6 33.7 2.32 0.5
5 52.8 42.2 3.23 0.1
700 °C. The heating rate was 100 °C/min up to 900 °C and 50 °C/min up to the sintering temperature. A holding time of 1 min at the maximum temperature was applied. During the whole SPS sintering cycle, the voltage, current, displacement of the upper punch and temperature were recorded. The density of the sintered discs was measured by the water displacement method, and the oxygen and nitrogen contents were measured by a LECO TC400 machine. 3. Results and discussion 3.1. Powders
where ε = microstrain, D = mean grain size and b = Burgers vector. Thermogravimetric analysis (TGA) of the milled powders was carried out to study their degassing behaviour. TGA was carried out under an argon atmosphere, heating the samples up to 1060 °C at a heating rate of 20 °C/min; these analyses were combined with quadrupole mass spectrometry (QMS) using an STA409CD apparatus (Netzsch) to identify the gases emitted during heating. The powders were sintered in a DR.SINTER® SPS1050 (Sumitomo Coal & Mining, now SPS Syntex, Inc.) apparatus with graphite punches and dies. The geometry of the samples was a disc with a height of 5 mm and a diameter of 20 mm. SPS was performed at a nominal temperature (measured with a thermocouple inserted into a blind hole in the die wall) of 950 °C, with a uniaxial pressure of 30 MPa applied at
Fig. 1. Powders H (a) and L (b) before milling.
The milled H and L powders are shown in Fig. 2. A large difference in particle size between the two powders is clearly evident, as also confirmed by the particle size distribution reported in Fig. 3. In the same figure, median values of particle size are reported; they are different by one order of magnitude (7 vs. 78 μm). It is well known that the powder particle morphology and size change with the milling time. The milling process can be indeed divided into two different stages: the flaking stage and the fragmentation stage [9,18]. In the first stage, highly deformed flakes are formed due to collision with the balls. Upon increasing the milling time, the size of the flakes increases through cold welding, as they become thicker and highly strain-hardened so that they can be fragmented during the further collisions (second stage). After 100 h of milling, the fragmentation process of powder H is in an advanced
Fig. 2. 100 h milled Cu powders: powders H (a) and L (b).
172
C. Menapace et al. / Powder Technology 291 (2016) 170–177 Table 3 Mean crystallite size, microstrain and dislocation density of the milled H and L powders.
H L
Mean crystallite size (nm)
Lattice strain (%)
Dislocation density (m−2)
28 22
0.004 0.22
2.97E+16 2.89E+17
a nanocrystalline material and thus decreasing its density [21]. The theoretical density d of a nanocrystalline material can be calculated using the lattice parameter according to Eq. (2) [20] V¼
Fig. 3. Cumulative distribution of particle size (mean diameter) of milled powders.
stage with respect to powder L, leading to much finer powder particles. This is due to the larger amount of oxides in the first powder that assists in the fragmentation process. The oxygen and carbon contents of the milled powders are reported in Table 2. The oxygen uptake by powder H is three times higher than that of powder L because H is much finer. XRD analysis revealed the presence of 21 vol.% Cu2O in powder H, while in the L powder, no oxides were detected, which means that the oxygen forms a supersaturated solid solution. The carbon content is only due to the stearic acid added as PCA and is therefore almost the same for the two powders. The complete crystallographic data derived from the X-ray analysis are summarized in Table 3. Both powders have a very fine crystallite size, while the microstrain and, as a consequence, dislocation density, are lower for powder H. The strain softening of powder H is due to the exit of oxygen from the copper lattice [19] and the corresponding formation of Cu2O, as shown by the XRD spectra in Fig. 4. 3.2. Spark plasma sintering The microstructures obtained by SPS, the milled powders H and L, are reported in Fig. 5. They are really fine, and no grain boundaries are resolved under LOM. The X-ray data reported in Table 4 confirm the nanostructure of these compacts. The sintered sample H has a larger grain size than sample L, together with a lower dislocation density, as a consequence of the higher grain size and lower dislocation density of the corresponding powder. The density and porosity of the SPSed samples is reported in Table 5. The relative density is quite low, even if no pores were observed under LOM. The porosity is completely closed. It is well known that nanostructured materials generally have lower densities than the corresponding microstructured ones because of the presence of a large numbers of defects that increase the molar volume [20]. The dislocations in plastically deformed crystals can be divided into “geometrically necessary” and “statistically stored”, both of them increasing the volume of
Table 2 Oxygen and carbon contents of powders H and L milled for 100 h.
H L
Oxygen wt.%
Carbon wt.%
3.56 1.14
0.40 0.38
VCE M N¼ Z d
ð2Þ
where VCE is the cell volume (for cubic crystals = a3 where a is the lattice parameter, calculated from XRD analyses: a = 3.6231 Å for sample H and a = 3.6257 Å for sample L), Z is the number of atoms per unit cell (4), N is Avogadro's number (6022 ∗ 1023) and M is the molar weight (63.546 g/mol). The theoretical density is 8875 g/cm3 and 8856 g/cm3 for materials H and L, respectively, in agreement with the higher density of defects in the latter. Considering the theoretical density of a microstructured annealed wrought copper, 8.96 g/cm3, the densities of the nanostructured coppers H and L are 99.05% and 98.83% of the theoretical value, as a result of the presence of defects introduced throughout the nanostructure. Comparing these values with the experimentally measured density data in Table 5, the amount of pores can be calculated to be 18.75% and 12.23% for H and L, respectively. Porosity was detected only by a TEM analysis. As shown in the TEM bright field pictures of Fig. 6a and b (referring to sample H), there is a consistent amount of nanopores, located mainly at the grain boundaries. Also clearly evident in Fig. 6b is the pinning effect exerted by these pores, as the grain size is much lower in the area, indicated by the arrow on the picture, where pores are visible at the grain boundaries. Generally, these pores are very fine (b100 nm), but some larger ones (N100 nm) were also observed. A similar effect was observed by Zhang et al. [11]. The microhardness and hardness reported in Table 6 are in agreement with the density data because they are influenced by the presence of nanopores. The difference between the two materials is related to their different grain sizes (L is much finer) and different amounts of nanopores (porosity of L is much lower). The grain size and nanoporosity concur to define the microhardness and hardness. The nanopores in the sintered specimens may originate from both the internal voids of the powder agglomerates and by the entrapment of gases during sintering [15,16,22]. Intra-agglomerate pores may only be eliminated if the agglomerates are fragmented during sintering; the other pores are mostly formed during sintering if the products of degassing are entrapped by the densifying material when the process is not properly conducted (i.e., the gas emission is prevented by the application of pressure) [16]. To study this second type of porosity and with the purpose of understanding the behaviour during the sintering of the two different copper powders, the SPS parameters were recorded during the whole sintering cycle. The first two parameters are the upper punch displacement and the chamber pressure, which are shown in Fig. 7 as a function of temperature. The displacement vs. temperature curve shows the same trend for the two materials: it increases slowly at the beginning of the heating and then shows a rapid increase when pressure is applied (700 °C) until it reaches a plateau, indicating that even if the temperature is further increased, no more reduction in the sample height can be obtained, meaning that the density does not increase further. The majority of the densification occurs between 700 and 950 °C because it is sustained by the application of pressure. The chamber pressure curves are characterized by the presence of some peaks representative of the degassing phenomena occurring
C. Menapace et al. / Powder Technology 291 (2016) 170–177
173
Fig. 4. XRD spectra of powders H and L.
during heating. These curves, presented as dashed lines in Fig. 7, are completely different between the two materials. For powder H, there are three peaks at 187, 297, 496 °C, while powder L displays only one peak at 420 °C. In the SPS chamber pressure curve, peaks were detected only up to 700 °C when pressure is applied during the sintering process, as it prevents any further evolution of gas. To interpret these peaks, TGA combined with QMS was carried out (Fig. 8). All of the peaks detected the on SPS curves and TGA/mass spectroscopy are described in Table 7 and indicated by a number in Figs. 7 and 8. All of the peaks observed are due to the evolution of the gases (H2, H2O, CO, CO2) produced by the decomposition of the residues of stearic acid present in the powder after milling. Stearic acid has an autoignition temperature of 395 °C and decomposes in the presence of oxygen according to the following reactions: C18 H36 O2ðsÞ þ 27O2ðgÞ →18CO2ðgÞ þ 18H2 OðgÞ
ð3Þ
C18 H36 O2ðsÞ þ 18O2ðgÞ →18COðgÞ þ 18H2 OðgÞ
ð4Þ
In the absence of oxygen, as in the case of the TGA atmosphere (argon), H2 is formed instead of H2O. In powder H, peak 1 at 187 °C is observed in SPS but not in TGA and is probably due to humidity absorbed by the powder [23]. The heat treatment in TGA carried out after the application of vacuum in the TG chamber followed by a flux of argon reduces the formation of humidity, thus eliminating this peak. Peaks 2 and 3 are detected by both the SPS
pressure chamber and TGA/mass spectroscopy and are due to the evolution of H2O, H2 and CO2. Peak 4, on the other hand, is not observed in SPS because the application of pressure at 700 °C prevents degassing at a higher temperature. In powder L, peak 5 is due to the evolution of H2, and peak 6 is analogous to peak 4 of powder H, due to the formation of CO2 being prevented by the application of pressure at lower temperature. Comparing the two powders, it can be observed that in powder L, there is no evolution of CO2 up to 915 °C, while in powder H, peaks 2 (at 280–300 °C) and 3 (at 475–500 °C) are both due to the production of CO2. This is in agreement with the drop in the carbon content of this powder when it is sintered: in sample H, it decreases from 0401 to 0346 wt.%, as reported in Table 8, while it remains almost unvaried in powder L. Peaks 2 and 3 being present in powder H but missing in powder L are explained by considering the different particle sizes of the two powders. Powder H is so fine that an anticipated decomposition of stearic acid is observed, as this lubricant is much finer as well and therefore much more reactive [23]. The voltage applied and the current flowing through the samples were recorded during the whole sintering cycle, as well. These curves provide further information regarding the sintering behaviour of powders H and L. The current vs. voltage curves are shown in Fig. 9a and b. This curve shows a change in slope (indicated in Fig. 9a and b by an arrow) at a voltage of 2.11 V for powder H and 2.24 V for powder L. This point indicates the breakdown of the surface oxide and the beginning of
174
C. Menapace et al. / Powder Technology 291 (2016) 170–177
Fig. 5. Microstructures of SPSed discs of H (a) and L (b) powders.
the current flow through the copper particles. Up to this point, the current also flows through the oxide layer, other than through the die walls. Cu2O is in fact a p-type semiconductor with a conductivity that is very sensitive to temperature [24]. The conductivity of the oxide surface layer depends on its thickness, and the different current values observed up to the breakdown point into the compact H and L are in line with the different oxidation states of the two powders. The current measured in sample H is indeed less than half that in L, confirming a much higher surface oxidation in the first powder. This is in agreement with the X-ray analysis, which measured 21 vol.% of cuprous oxide in powder H, while no oxides were detected in powder L. Moreover, from the graphs of Fig. 9a and b, it is evident that the current increase in the L specimen is more pronounced than in H, indicating the lower resistance of the former. The temperature measured on samples H and L is reported as a function of the power supplied (V ∗ I) during the sintering cycle in Fig. 10a and b.
Table 4 Mean crystallite size, microstrain and dislocation density of the SPSed compacts. Sample
Mean crystallite size (nm)
Lattice strain (%)
Dislocation density (m−2)
H L
82 57
0.056 0.27
3.90E+16 1.24E+17
Table 5 Density and porosity of the SPSed discs.
Fig. 6. Examples of nanopores observed in SPSed discs.
A higher temperature (reached by the Joule effect) is observed in sample H relative to L, which is a confirmation of the higher resistance of the former. It should be considered that the electrical resistance of the powder compact changes during the sintering cycle due to neck growth, increasing the area (A) through which the current passes. The initial contact area was calculated using the contact mechanics, which states that between two spheres of the same radius r in contact under a certain force F, the contact area (A) is A ¼ ð3FR=4EÞ1=3
ð5Þ
where F is calculated from the initial force applied in the SPS divided by the number of particles, and R is defined as R ¼ r=2
ð6Þ
where r is the radius of the spherical particles in contact. For r median values were taken which are 3.5 μm for powder H and 39 μm for L as reported in Fig. 3. The number of particles was calculated knowing the diameter of the sintered disc and supposing to have a cubic arrangement of spherical particles. E ¼ 2 1−ν2 =E
ð7Þ
where ν and E are the Poisson's ratio end the elastic modulus of copper, respectively.
Table 6 Microhardness and hardness of the SPSed discs.
sample
Relative density (%)
Total porosity (%)
Sample
Microhardness, HV0.1
Hardness, HV5
H L
80.3 86.6
19.7 13.4
H L
43 99
40 97
C. Menapace et al. / Powder Technology 291 (2016) 170–177
175
Fig. 8. TGA and mass spectroscopy of powders H (a) and L (b).
Fig. 7. Displacement and chamber pressure vs. temperature during the sintering cycles of powders H (a) and L (b).
It is assumed that the increase of the neck area follows the displacement vs. temperature trend, starting from the initial condition of a rearrangement of spherical particles in contact under the application of the initial force and the final condition given by the section of the sample multiplied by the relative density. The increase in A with temperature is reported in Fig. 11a and b for samples H and L, respectively. Knowing the neck area, it is possible to express the current density i as the ratio between the current I (plotted in Fig. 9) and the area A (plotted in Fig. 11):
i ¼ I=A
ð8Þ
The resistance of compact H is higher than that of L because a) it has a higher contact surface between the powder particles, b) it is more oxidized, at least up to the oxide breakdown (indicated by the arrows in Fig. 12), and c) it has an intrinsic higher porosity from the beginning due to its tendency to form agglomerates (intra-agglomerate pores), as this powder is very fine.
Furthermore, the trend of curve H is completely different with respect to that of sample L. It shows an initial rapid increase, which is due to the high oxidation grade of this powder, but it can also be related to the initial powder particle rearrangement that is also confirmed by the rapid increase of the displacement curve (Fig. 7a). The
Table 7 Data derived from SPS and TGA/mass spectroscopy curves.
Then, the resistance R of the compact during the sintering cycle can be calculated according to Ohm's law as
Sample
Peak
Peak temperature of chamber pressure (°C)
R ¼ V=i
H
1 2 3 4 5 6
187 297 496
ð9Þ
R is plotted as a function of temperature in Fig. 12 for the two copper compacts.
L
420
Peak temperature of mass spectroscopy (°C) and gas detected
Temperature of mass decrease (°C)
280 (H2O and CO2) 475 (H2 and CO2) 820 (CO and CO2) 416 (H2) 915 (CO2)
290 408 790 413 920
176
C. Menapace et al. / Powder Technology 291 (2016) 170–177
Table 8 Comparison between the oxygen and carbon contents of the powder and the SPSed samples. O, wt.%
H L
C, wt.%
Powder
Sintered part
Powder
Sintered part
3.561 1.140
0.609 0.589
0.401 0.376
0.346 0.380
rearrangement indeed leads to a consistent increase in the contact points between the powder particles and thus to a rapid increase in the contact surface, as shown in Fig. 11a. A very rapid increase is observed from room temperature up to approximately 100 °C, while the oxide breakdown is observed at approximately 45 °C in powder H. Increasing the contact surface, the current density i decreases and the resistance R increases. The large difference between the real contact surfaces of systems H and L during the sintering cycle accounts for the great difference in the resistance up to the temperature of the final densification (which starts with the pressure application at 700 °C). From that point, the difference in R between the two materials is much less (in the range between 12 and 19%) and related only to their different porosities. In fact, if it is assumed that the electrical resistance of a sintered part (Rs) is related to the resistance of the corresponding
Fig. 10. Temperature vs. power supplied during the sintering cycles of powders H (a) and L (b).
wrought material (R0) and its porosity (ε) through the following formula [25] Rs ¼ R0
1 1 2ε
ð10Þ
A difference of 18% is obtained between the resistances of samples H and L, inserting in this formula the porosity data experimentally measured on the sintered discs (18.7% for H and 12.2% for L). This value is in very good agreement with the R data of Fig. 12, which indicates a difference of 19% at the last point of the sintering cycle. 4. Conclusions
Fig. 9. Current vs. voltage during the sintering cycles of powders H (a) and L (b).
Two copper powders having different initial oxygen contents (H and L) were ball milled and spark plasma sintered to produce bulk nanostructured copper. The different initial oxide amounts influence the milling process, leading to a much finer particle size in the powder having a higher initial oxidation state (powder H), which however shows a worse sintering densification. The SPS parameters (current and electrical resistance) during the sintering cycle, as well as all of the degassing phenomena occurring on heating, are markedly influenced by the different particle size and oxidation state of the two powders. At the end, powder L, in spite of its higher particle size, shows the best densification because of its lower oxidation state and lower grain size.
C. Menapace et al. / Powder Technology 291 (2016) 170–177
177
References
Fig. 11. Neck area during the sintering cycles of powders H (a) and L (b).
Fig. 12. Resistance of compacts H and L during the sintering cycle.
[1] V. Mamedov, Spark plasma sintering as advanced PM sintering method, Powder Metall. 45 (4) (2002) 322–328. [2] Z.A. Munir, U. Anselmi-Tamburini, M. Ohyanagi, The effect of electric field and pressure on the synthesis and consolidation of materials: a review of the spark plasma method, J. Mater. Sci. 41 (2006) 763–777. [3] R. Orrù, J. Woolman, G. Cao, Z.A. Munir, Synthesis of dense nanometric MoSi2 through mechanical and field activation, J. Mater. Res. 16 (5) (2001) 1439–1448. [4] F. Bernard, F. Charlot, E. Gaffet, Z.A. Munir, One-step synthesis and consolidation of nanophase iron aluminide, J. Am. Ceram. Soc. 84 (5) (2001) 910–914. [5] M. Ohyanagi, T. Yamamoto, H. Kitaura, Y. Kodera, T. Ishii, Z.A. Munir, Consolidation of nanostructured SiC with disorder–order transformation, Scr. Mater. 50 (2004) 111–114. [6] E.M. Heian, S.K. Khalsa, J.W. Lee, Z.A. Munir, T. Yamamoto, M. Ohyanagi, Synthesis of dense, high-defect-concentration B4C through mechanical activation and fieldassisted combustion, J. Am. Ceram. Soc. 87 (5) (2004) 779–783. [7] S. Diouf, A. Molinari, Densification mechanisms in spark plasma sintering: effect of particle size and pressure, Powder Technol. 221 (2012) 220–227. [8] S. Diouf, C. Menapace, A. Molinari, Study of effect of particle size on densification of copper during spark plasma sintering, Powder Metall. 55 (3) (2012) 228–234. [9] C. Suryanarayana, Mechanical alloying and milling, Prog. Mater. Sci. 46 (1–2) (2001) 1–184. [10] Z.H. Zhang, F.C. Wang, L. Wang, L. Shukui, S. Osamu, Sintering mechanism of largescale ultrafine-grained copper prepared by SPS method, Mater. Lett. 62 (2008) 3987–3990. [11] Z.H. Zhang, F.C. Wang, L. Wang, Ultrafine grained copper prepared by spark plasma sintering process, Mater. Sci. Eng. A476 (2008) 201–205. [12] X. Song, X. Liu, J. Zhang, Neck formation and self-adjusting mechanism of neck growth of conducting powders in spark plasma sintering, J. Am. Ceram. Soc. 89 (2006) 494–500. [13] Z.H. Zhang, F.C. Wang, S.K. Lee, Y. Liu, J.W. Cheng, Y. Liang, Microstructure characteristic, mechanical properties and sintering mechanism of nanocrystalline copper obtained by SPS process, Mater. Sci. Eng. A523 (2009) 134–138. [14] O. Yanagisawa, H. Kuramoto, K. Matsugi, M. Komatsu, Observation of particle behavior in copper powder compact during pulsed electric discharge, Mater. Sci. Eng. A350 (2003) 184–189. [15] A.S. Sharma, K. Biswas, B. Basu, D. Chakravarty, Spark plasma sintering of nanocrystalline Cu and Cu-10 Wt Pct Pb alloy, Metall. Mater. Trans. A 42 (7) (2011) 2072–2084. [16] H. Wen, Y. Zhao, Z. Zhang, O. Ertorer, S. Dong, E.J. Lavernia, The influence of oxygen and nitrogen contamination on the densification behavior of cryomilled copper powders during spark plasma sintering, J. Mater. Sci. 46 (2011) 3006–3012. [17] Y.H. Zhao, H.W. Sheng, K. Lu, Microstructure evolution and thermal properties in nanocrystalline Fe during mechanical attrition, Acta Mater. 49 (2001) 365–375. [18] J.Y. Huang, Y.K. Wu, H.Q. Ye, Ball milling of ductile metals, Mater. Sci. Eng. A199 (1995) 165–172. [19] Y.V.R.K. Prasad, K.P. Rao, Influence of oxygen on rate-controlling mechanisms in hot deformation of polycrystalline copper: oxygen-free versus electrolytic grades, Mater. Lett. 58 (2004) 2061–2066. [20] T.D. Shen, J. Zhang, Y. Zhao, What is the theoretical density of a nanocrystalline material? Acta Mater. 56 (14) (2008) 3663–3671. [21] M.F. Ashby, The deformation of plastically non-homogeneous materials, Philos. Mag. 21 (1970) 399–424. [22] S. Diouf, C. Menapace, M. D'Incau, A. Molinari, G. Ischia, Spark plasma sintering of cryomilled copper powder, Powder Metall. 56 (5) (2013) 420–426. [23] G. Cipolloni, M. Pellizzari, A. Molinari, M. Hebda, M. Zadra, Contamination during the high-energy milling of atomized copper powder and its effects on spark plasma sintering, Powder Technol. 275 (2015) 51–59. [24] P.G. Slade (Ed.), Electrical Contacts: Principles and Applications, Marcel Dekker Inc., NY 1999, p. 40. [25] G.F. Bocchini, The influence of porosity on the characteristics of sintered materials, Int. J. Powder Metall. 22 (3) (1986) 185–202.
Contents lists available at ScienceDirect
Powder Technology journal homepage: www.elsevier.com/locate/powtec
Spark plasma sintering behaviour of copper powders having different particle sizes and oxygen contents C. Menapace a,⁎, G. Cipolloni a, M. Hebda b, G. Ischia a a b
Department of Industrial Engineering, University of Trento, via Sommarive 9, 38123 Trento, Italy Institute of Materials Engineering, Cracow University of Technology, ul.Warszawska 24, 31-155 Kraków, Poland
a r t i c l e
i n f o
Article history: Received 20 July 2015 Received in revised form 9 December 2015 Accepted 19 December 2015 Available online 21 December 2015 Keywords: Spark plasma sintering Copper Oxygen content
a b s t r a c t Two different copper powders with different initial oxygen contents (H = high, L = low) were milled and spark plasma sintered to obtain nanosized bulk copper. The characteristics of the milled powders were analysed together with their behaviour in sintering. Spark plasma sintering was carried out at 950 °C for 1 min, applying a pressure of 30 MPa at 700 °C. The consistent difference observed between the two powders is related to the much lower particle size of powder H and its consequent higher surface oxidation. It was observed that even if the finer particle size should lead to a faster and better densification, the much higher oxide amount predominates, slowing down the sinterability of the powder. Moreover, during the sintering process, a higher electrical resistance was measured for powder H due to its higher surface (particle–particle contact area) and oxide content. The degassing process occurring on sintering, caused by the decomposition of the stearic acid used as a process control agent in ball milling, is also markedly influenced by the powder particle size, showing an anticipated decomposition in sample H due to the higher reactivity of this powder. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Spark plasma sintering (SPS) is a powder consolidation process based on the simultaneous application of a low-voltage, high-energy pulse current and uniaxial pressure [1,2]. The importance of this sintering method has been demonstrated by the large number of papers published during the past decade. SPS is used to sinter metallic powders, as well as ceramics, intermetallics and composites [3–6]. One of the main benefits of this technology is its localized heat input and short sintering time, which reduces the grain growth linked to high temperatures. Therefore, SPS is one of the most suitable sintering processes for materials that are very sensitive to temperature like nanostructured powders, being able to consolidate the compact to near full density with limited grain growth. Different parameters influence the sintering behaviour of a powder; two of the most important are the particle size and the oxygen content. Regarding the particle size, it is well known that the thermodynamic driving force for sintering is proportional to the specific surface area of the powder. Particle size also influences the number and the extension of the contact areas and, in turn, the local pressure and the electrical resistance, which are of great importance in SPS due to its peculiar heating mechanism [7,8]. The second parameter is the powder oxygen content, which refers to the amount of oxides present on the powder particle surface. In ⁎ Corresponding author. E-mail address: [email protected] (C. Menapace).
http://dx.doi.org/10.1016/j.powtec.2015.12.020 0032-5910/© 2015 Elsevier B.V. All rights reserved.
sintering, the presence of oxides on the particle surface is detrimental because it slows down all of the mechanisms of mass transport responsible for neck formation. On the other hand, if the metallic powder is milled before sintering, the presence of oxides on the particles is beneficial because they exert a cutting action, promoting the fragmentation of the powder [9]. In the present paper, two different Cu powders with different initial oxygen levels were milled for 100 h to obtain a nanometric grain size and then consolidated in SPS to retain their nanosized structure. Many researchers have focused their attention on the spark plasma sintering behaviour of ultra-fine and nanosized copper [10–16], paying attention mainly to the evolution of the grain size and the microstructural characteristics during sintering. Less attention has been dedicated to the study of the influence of the powder oxygen content on the milling and sintering processes. Because commercially pure copper powders contain a certain amount of oxygen and a drawback associated with mechanical milling is the further unavoidable oxygen contamination by the milling media, atmosphere and process control agent (PCA), it is of great importance to evaluate the effect of the oxygen content on the behaviour of this powder. 2. Materials and experimental procedure Two different Cu powder grades with different initial oxygen contents (powders H = high oxygen amount = 0.5 wt.% and L = low oxygen amount = 0.1 wt.%) were milled for 100 h and sintered in SPS to obtain nanostructured copper compacts.
C. Menapace et al. / Powder Technology 291 (2016) 170–177
The two powders have a similar particle size and morphology, as shown in Fig. 1. The particle size, apparent density and oxygen content are summarized in Table 1. Milling was carried out in a Fritsch Pulverisette 5 planetary ball mill (Idar-Oberstein, Germany) with a vial and balls of hardened steel at a rotation speed of 400 rpm and a ball to powder ratio (BPR) of 10. 0.5 wt.% of stearic acid was used as a PCA (process control agent). To avoid oxidation, the vial was set under vacuum before milling. The milled powders were examined under light optical microscopy (LOM), scanning electron microscopy (SEM), and transmission electron microscopy (TEM) and through X-ray diffraction. For the microstructural analysis, a conventional grinding and polishing procedure was used. The grain size and lattice strain of the milled powder and sintered discs were determined through X-ray diffraction (XRD) with a Cu Kα (λ = 0.154056 nm) source and an image plate detector over the 2θ range from 30° to 120° in reflection geometry. The experimental spectra were elaborated using materials analysis using diffraction (MAUD). The dislocation density was calculated through the formula [17] ρ1=2 ¼
pffiffiffiffiffiffiffiffiffiffiffi 2 3ε1=2 Db
ð1Þ
171
Table 1 Characteristics of the two initial powders.
Particle size % cumulative
Apparent density (g/cm3) O2
N75 μm N45 μm and b75 μm b45 μm
H
L
8.7 57.6 33.7 2.32 0.5
5 52.8 42.2 3.23 0.1
700 °C. The heating rate was 100 °C/min up to 900 °C and 50 °C/min up to the sintering temperature. A holding time of 1 min at the maximum temperature was applied. During the whole SPS sintering cycle, the voltage, current, displacement of the upper punch and temperature were recorded. The density of the sintered discs was measured by the water displacement method, and the oxygen and nitrogen contents were measured by a LECO TC400 machine. 3. Results and discussion 3.1. Powders
where ε = microstrain, D = mean grain size and b = Burgers vector. Thermogravimetric analysis (TGA) of the milled powders was carried out to study their degassing behaviour. TGA was carried out under an argon atmosphere, heating the samples up to 1060 °C at a heating rate of 20 °C/min; these analyses were combined with quadrupole mass spectrometry (QMS) using an STA409CD apparatus (Netzsch) to identify the gases emitted during heating. The powders were sintered in a DR.SINTER® SPS1050 (Sumitomo Coal & Mining, now SPS Syntex, Inc.) apparatus with graphite punches and dies. The geometry of the samples was a disc with a height of 5 mm and a diameter of 20 mm. SPS was performed at a nominal temperature (measured with a thermocouple inserted into a blind hole in the die wall) of 950 °C, with a uniaxial pressure of 30 MPa applied at
Fig. 1. Powders H (a) and L (b) before milling.
The milled H and L powders are shown in Fig. 2. A large difference in particle size between the two powders is clearly evident, as also confirmed by the particle size distribution reported in Fig. 3. In the same figure, median values of particle size are reported; they are different by one order of magnitude (7 vs. 78 μm). It is well known that the powder particle morphology and size change with the milling time. The milling process can be indeed divided into two different stages: the flaking stage and the fragmentation stage [9,18]. In the first stage, highly deformed flakes are formed due to collision with the balls. Upon increasing the milling time, the size of the flakes increases through cold welding, as they become thicker and highly strain-hardened so that they can be fragmented during the further collisions (second stage). After 100 h of milling, the fragmentation process of powder H is in an advanced
Fig. 2. 100 h milled Cu powders: powders H (a) and L (b).
172
C. Menapace et al. / Powder Technology 291 (2016) 170–177 Table 3 Mean crystallite size, microstrain and dislocation density of the milled H and L powders.
H L
Mean crystallite size (nm)
Lattice strain (%)
Dislocation density (m−2)
28 22
0.004 0.22
2.97E+16 2.89E+17
a nanocrystalline material and thus decreasing its density [21]. The theoretical density d of a nanocrystalline material can be calculated using the lattice parameter according to Eq. (2) [20] V¼
Fig. 3. Cumulative distribution of particle size (mean diameter) of milled powders.
stage with respect to powder L, leading to much finer powder particles. This is due to the larger amount of oxides in the first powder that assists in the fragmentation process. The oxygen and carbon contents of the milled powders are reported in Table 2. The oxygen uptake by powder H is three times higher than that of powder L because H is much finer. XRD analysis revealed the presence of 21 vol.% Cu2O in powder H, while in the L powder, no oxides were detected, which means that the oxygen forms a supersaturated solid solution. The carbon content is only due to the stearic acid added as PCA and is therefore almost the same for the two powders. The complete crystallographic data derived from the X-ray analysis are summarized in Table 3. Both powders have a very fine crystallite size, while the microstrain and, as a consequence, dislocation density, are lower for powder H. The strain softening of powder H is due to the exit of oxygen from the copper lattice [19] and the corresponding formation of Cu2O, as shown by the XRD spectra in Fig. 4. 3.2. Spark plasma sintering The microstructures obtained by SPS, the milled powders H and L, are reported in Fig. 5. They are really fine, and no grain boundaries are resolved under LOM. The X-ray data reported in Table 4 confirm the nanostructure of these compacts. The sintered sample H has a larger grain size than sample L, together with a lower dislocation density, as a consequence of the higher grain size and lower dislocation density of the corresponding powder. The density and porosity of the SPSed samples is reported in Table 5. The relative density is quite low, even if no pores were observed under LOM. The porosity is completely closed. It is well known that nanostructured materials generally have lower densities than the corresponding microstructured ones because of the presence of a large numbers of defects that increase the molar volume [20]. The dislocations in plastically deformed crystals can be divided into “geometrically necessary” and “statistically stored”, both of them increasing the volume of
Table 2 Oxygen and carbon contents of powders H and L milled for 100 h.
H L
Oxygen wt.%
Carbon wt.%
3.56 1.14
0.40 0.38
VCE M N¼ Z d
ð2Þ
where VCE is the cell volume (for cubic crystals = a3 where a is the lattice parameter, calculated from XRD analyses: a = 3.6231 Å for sample H and a = 3.6257 Å for sample L), Z is the number of atoms per unit cell (4), N is Avogadro's number (6022 ∗ 1023) and M is the molar weight (63.546 g/mol). The theoretical density is 8875 g/cm3 and 8856 g/cm3 for materials H and L, respectively, in agreement with the higher density of defects in the latter. Considering the theoretical density of a microstructured annealed wrought copper, 8.96 g/cm3, the densities of the nanostructured coppers H and L are 99.05% and 98.83% of the theoretical value, as a result of the presence of defects introduced throughout the nanostructure. Comparing these values with the experimentally measured density data in Table 5, the amount of pores can be calculated to be 18.75% and 12.23% for H and L, respectively. Porosity was detected only by a TEM analysis. As shown in the TEM bright field pictures of Fig. 6a and b (referring to sample H), there is a consistent amount of nanopores, located mainly at the grain boundaries. Also clearly evident in Fig. 6b is the pinning effect exerted by these pores, as the grain size is much lower in the area, indicated by the arrow on the picture, where pores are visible at the grain boundaries. Generally, these pores are very fine (b100 nm), but some larger ones (N100 nm) were also observed. A similar effect was observed by Zhang et al. [11]. The microhardness and hardness reported in Table 6 are in agreement with the density data because they are influenced by the presence of nanopores. The difference between the two materials is related to their different grain sizes (L is much finer) and different amounts of nanopores (porosity of L is much lower). The grain size and nanoporosity concur to define the microhardness and hardness. The nanopores in the sintered specimens may originate from both the internal voids of the powder agglomerates and by the entrapment of gases during sintering [15,16,22]. Intra-agglomerate pores may only be eliminated if the agglomerates are fragmented during sintering; the other pores are mostly formed during sintering if the products of degassing are entrapped by the densifying material when the process is not properly conducted (i.e., the gas emission is prevented by the application of pressure) [16]. To study this second type of porosity and with the purpose of understanding the behaviour during the sintering of the two different copper powders, the SPS parameters were recorded during the whole sintering cycle. The first two parameters are the upper punch displacement and the chamber pressure, which are shown in Fig. 7 as a function of temperature. The displacement vs. temperature curve shows the same trend for the two materials: it increases slowly at the beginning of the heating and then shows a rapid increase when pressure is applied (700 °C) until it reaches a plateau, indicating that even if the temperature is further increased, no more reduction in the sample height can be obtained, meaning that the density does not increase further. The majority of the densification occurs between 700 and 950 °C because it is sustained by the application of pressure. The chamber pressure curves are characterized by the presence of some peaks representative of the degassing phenomena occurring
C. Menapace et al. / Powder Technology 291 (2016) 170–177
173
Fig. 4. XRD spectra of powders H and L.
during heating. These curves, presented as dashed lines in Fig. 7, are completely different between the two materials. For powder H, there are three peaks at 187, 297, 496 °C, while powder L displays only one peak at 420 °C. In the SPS chamber pressure curve, peaks were detected only up to 700 °C when pressure is applied during the sintering process, as it prevents any further evolution of gas. To interpret these peaks, TGA combined with QMS was carried out (Fig. 8). All of the peaks detected the on SPS curves and TGA/mass spectroscopy are described in Table 7 and indicated by a number in Figs. 7 and 8. All of the peaks observed are due to the evolution of the gases (H2, H2O, CO, CO2) produced by the decomposition of the residues of stearic acid present in the powder after milling. Stearic acid has an autoignition temperature of 395 °C and decomposes in the presence of oxygen according to the following reactions: C18 H36 O2ðsÞ þ 27O2ðgÞ →18CO2ðgÞ þ 18H2 OðgÞ
ð3Þ
C18 H36 O2ðsÞ þ 18O2ðgÞ →18COðgÞ þ 18H2 OðgÞ
ð4Þ
In the absence of oxygen, as in the case of the TGA atmosphere (argon), H2 is formed instead of H2O. In powder H, peak 1 at 187 °C is observed in SPS but not in TGA and is probably due to humidity absorbed by the powder [23]. The heat treatment in TGA carried out after the application of vacuum in the TG chamber followed by a flux of argon reduces the formation of humidity, thus eliminating this peak. Peaks 2 and 3 are detected by both the SPS
pressure chamber and TGA/mass spectroscopy and are due to the evolution of H2O, H2 and CO2. Peak 4, on the other hand, is not observed in SPS because the application of pressure at 700 °C prevents degassing at a higher temperature. In powder L, peak 5 is due to the evolution of H2, and peak 6 is analogous to peak 4 of powder H, due to the formation of CO2 being prevented by the application of pressure at lower temperature. Comparing the two powders, it can be observed that in powder L, there is no evolution of CO2 up to 915 °C, while in powder H, peaks 2 (at 280–300 °C) and 3 (at 475–500 °C) are both due to the production of CO2. This is in agreement with the drop in the carbon content of this powder when it is sintered: in sample H, it decreases from 0401 to 0346 wt.%, as reported in Table 8, while it remains almost unvaried in powder L. Peaks 2 and 3 being present in powder H but missing in powder L are explained by considering the different particle sizes of the two powders. Powder H is so fine that an anticipated decomposition of stearic acid is observed, as this lubricant is much finer as well and therefore much more reactive [23]. The voltage applied and the current flowing through the samples were recorded during the whole sintering cycle, as well. These curves provide further information regarding the sintering behaviour of powders H and L. The current vs. voltage curves are shown in Fig. 9a and b. This curve shows a change in slope (indicated in Fig. 9a and b by an arrow) at a voltage of 2.11 V for powder H and 2.24 V for powder L. This point indicates the breakdown of the surface oxide and the beginning of
174
C. Menapace et al. / Powder Technology 291 (2016) 170–177
Fig. 5. Microstructures of SPSed discs of H (a) and L (b) powders.
the current flow through the copper particles. Up to this point, the current also flows through the oxide layer, other than through the die walls. Cu2O is in fact a p-type semiconductor with a conductivity that is very sensitive to temperature [24]. The conductivity of the oxide surface layer depends on its thickness, and the different current values observed up to the breakdown point into the compact H and L are in line with the different oxidation states of the two powders. The current measured in sample H is indeed less than half that in L, confirming a much higher surface oxidation in the first powder. This is in agreement with the X-ray analysis, which measured 21 vol.% of cuprous oxide in powder H, while no oxides were detected in powder L. Moreover, from the graphs of Fig. 9a and b, it is evident that the current increase in the L specimen is more pronounced than in H, indicating the lower resistance of the former. The temperature measured on samples H and L is reported as a function of the power supplied (V ∗ I) during the sintering cycle in Fig. 10a and b.
Table 4 Mean crystallite size, microstrain and dislocation density of the SPSed compacts. Sample
Mean crystallite size (nm)
Lattice strain (%)
Dislocation density (m−2)
H L
82 57
0.056 0.27
3.90E+16 1.24E+17
Table 5 Density and porosity of the SPSed discs.
Fig. 6. Examples of nanopores observed in SPSed discs.
A higher temperature (reached by the Joule effect) is observed in sample H relative to L, which is a confirmation of the higher resistance of the former. It should be considered that the electrical resistance of the powder compact changes during the sintering cycle due to neck growth, increasing the area (A) through which the current passes. The initial contact area was calculated using the contact mechanics, which states that between two spheres of the same radius r in contact under a certain force F, the contact area (A) is A ¼ ð3FR=4EÞ1=3
ð5Þ
where F is calculated from the initial force applied in the SPS divided by the number of particles, and R is defined as R ¼ r=2
ð6Þ
where r is the radius of the spherical particles in contact. For r median values were taken which are 3.5 μm for powder H and 39 μm for L as reported in Fig. 3. The number of particles was calculated knowing the diameter of the sintered disc and supposing to have a cubic arrangement of spherical particles. E ¼ 2 1−ν2 =E
ð7Þ
where ν and E are the Poisson's ratio end the elastic modulus of copper, respectively.
Table 6 Microhardness and hardness of the SPSed discs.
sample
Relative density (%)
Total porosity (%)
Sample
Microhardness, HV0.1
Hardness, HV5
H L
80.3 86.6
19.7 13.4
H L
43 99
40 97
C. Menapace et al. / Powder Technology 291 (2016) 170–177
175
Fig. 8. TGA and mass spectroscopy of powders H (a) and L (b).
Fig. 7. Displacement and chamber pressure vs. temperature during the sintering cycles of powders H (a) and L (b).
It is assumed that the increase of the neck area follows the displacement vs. temperature trend, starting from the initial condition of a rearrangement of spherical particles in contact under the application of the initial force and the final condition given by the section of the sample multiplied by the relative density. The increase in A with temperature is reported in Fig. 11a and b for samples H and L, respectively. Knowing the neck area, it is possible to express the current density i as the ratio between the current I (plotted in Fig. 9) and the area A (plotted in Fig. 11):
i ¼ I=A
ð8Þ
The resistance of compact H is higher than that of L because a) it has a higher contact surface between the powder particles, b) it is more oxidized, at least up to the oxide breakdown (indicated by the arrows in Fig. 12), and c) it has an intrinsic higher porosity from the beginning due to its tendency to form agglomerates (intra-agglomerate pores), as this powder is very fine.
Furthermore, the trend of curve H is completely different with respect to that of sample L. It shows an initial rapid increase, which is due to the high oxidation grade of this powder, but it can also be related to the initial powder particle rearrangement that is also confirmed by the rapid increase of the displacement curve (Fig. 7a). The
Table 7 Data derived from SPS and TGA/mass spectroscopy curves.
Then, the resistance R of the compact during the sintering cycle can be calculated according to Ohm's law as
Sample
Peak
Peak temperature of chamber pressure (°C)
R ¼ V=i
H
1 2 3 4 5 6
187 297 496
ð9Þ
R is plotted as a function of temperature in Fig. 12 for the two copper compacts.
L
420
Peak temperature of mass spectroscopy (°C) and gas detected
Temperature of mass decrease (°C)
280 (H2O and CO2) 475 (H2 and CO2) 820 (CO and CO2) 416 (H2) 915 (CO2)
290 408 790 413 920
176
C. Menapace et al. / Powder Technology 291 (2016) 170–177
Table 8 Comparison between the oxygen and carbon contents of the powder and the SPSed samples. O, wt.%
H L
C, wt.%
Powder
Sintered part
Powder
Sintered part
3.561 1.140
0.609 0.589
0.401 0.376
0.346 0.380
rearrangement indeed leads to a consistent increase in the contact points between the powder particles and thus to a rapid increase in the contact surface, as shown in Fig. 11a. A very rapid increase is observed from room temperature up to approximately 100 °C, while the oxide breakdown is observed at approximately 45 °C in powder H. Increasing the contact surface, the current density i decreases and the resistance R increases. The large difference between the real contact surfaces of systems H and L during the sintering cycle accounts for the great difference in the resistance up to the temperature of the final densification (which starts with the pressure application at 700 °C). From that point, the difference in R between the two materials is much less (in the range between 12 and 19%) and related only to their different porosities. In fact, if it is assumed that the electrical resistance of a sintered part (Rs) is related to the resistance of the corresponding
Fig. 10. Temperature vs. power supplied during the sintering cycles of powders H (a) and L (b).
wrought material (R0) and its porosity (ε) through the following formula [25] Rs ¼ R0
1 1 2ε
ð10Þ
A difference of 18% is obtained between the resistances of samples H and L, inserting in this formula the porosity data experimentally measured on the sintered discs (18.7% for H and 12.2% for L). This value is in very good agreement with the R data of Fig. 12, which indicates a difference of 19% at the last point of the sintering cycle. 4. Conclusions
Fig. 9. Current vs. voltage during the sintering cycles of powders H (a) and L (b).
Two copper powders having different initial oxygen contents (H and L) were ball milled and spark plasma sintered to produce bulk nanostructured copper. The different initial oxide amounts influence the milling process, leading to a much finer particle size in the powder having a higher initial oxidation state (powder H), which however shows a worse sintering densification. The SPS parameters (current and electrical resistance) during the sintering cycle, as well as all of the degassing phenomena occurring on heating, are markedly influenced by the different particle size and oxidation state of the two powders. At the end, powder L, in spite of its higher particle size, shows the best densification because of its lower oxidation state and lower grain size.
C. Menapace et al. / Powder Technology 291 (2016) 170–177
177
References
Fig. 11. Neck area during the sintering cycles of powders H (a) and L (b).
Fig. 12. Resistance of compacts H and L during the sintering cycle.
[1] V. Mamedov, Spark plasma sintering as advanced PM sintering method, Powder Metall. 45 (4) (2002) 322–328. [2] Z.A. Munir, U. Anselmi-Tamburini, M. Ohyanagi, The effect of electric field and pressure on the synthesis and consolidation of materials: a review of the spark plasma method, J. Mater. Sci. 41 (2006) 763–777. [3] R. Orrù, J. Woolman, G. Cao, Z.A. Munir, Synthesis of dense nanometric MoSi2 through mechanical and field activation, J. Mater. Res. 16 (5) (2001) 1439–1448. [4] F. Bernard, F. Charlot, E. Gaffet, Z.A. Munir, One-step synthesis and consolidation of nanophase iron aluminide, J. Am. Ceram. Soc. 84 (5) (2001) 910–914. [5] M. Ohyanagi, T. Yamamoto, H. Kitaura, Y. Kodera, T. Ishii, Z.A. Munir, Consolidation of nanostructured SiC with disorder–order transformation, Scr. Mater. 50 (2004) 111–114. [6] E.M. Heian, S.K. Khalsa, J.W. Lee, Z.A. Munir, T. Yamamoto, M. Ohyanagi, Synthesis of dense, high-defect-concentration B4C through mechanical activation and fieldassisted combustion, J. Am. Ceram. Soc. 87 (5) (2004) 779–783. [7] S. Diouf, A. Molinari, Densification mechanisms in spark plasma sintering: effect of particle size and pressure, Powder Technol. 221 (2012) 220–227. [8] S. Diouf, C. Menapace, A. Molinari, Study of effect of particle size on densification of copper during spark plasma sintering, Powder Metall. 55 (3) (2012) 228–234. [9] C. Suryanarayana, Mechanical alloying and milling, Prog. Mater. Sci. 46 (1–2) (2001) 1–184. [10] Z.H. Zhang, F.C. Wang, L. Wang, L. Shukui, S. Osamu, Sintering mechanism of largescale ultrafine-grained copper prepared by SPS method, Mater. Lett. 62 (2008) 3987–3990. [11] Z.H. Zhang, F.C. Wang, L. Wang, Ultrafine grained copper prepared by spark plasma sintering process, Mater. Sci. Eng. A476 (2008) 201–205. [12] X. Song, X. Liu, J. Zhang, Neck formation and self-adjusting mechanism of neck growth of conducting powders in spark plasma sintering, J. Am. Ceram. Soc. 89 (2006) 494–500. [13] Z.H. Zhang, F.C. Wang, S.K. Lee, Y. Liu, J.W. Cheng, Y. Liang, Microstructure characteristic, mechanical properties and sintering mechanism of nanocrystalline copper obtained by SPS process, Mater. Sci. Eng. A523 (2009) 134–138. [14] O. Yanagisawa, H. Kuramoto, K. Matsugi, M. Komatsu, Observation of particle behavior in copper powder compact during pulsed electric discharge, Mater. Sci. Eng. A350 (2003) 184–189. [15] A.S. Sharma, K. Biswas, B. Basu, D. Chakravarty, Spark plasma sintering of nanocrystalline Cu and Cu-10 Wt Pct Pb alloy, Metall. Mater. Trans. A 42 (7) (2011) 2072–2084. [16] H. Wen, Y. Zhao, Z. Zhang, O. Ertorer, S. Dong, E.J. Lavernia, The influence of oxygen and nitrogen contamination on the densification behavior of cryomilled copper powders during spark plasma sintering, J. Mater. Sci. 46 (2011) 3006–3012. [17] Y.H. Zhao, H.W. Sheng, K. Lu, Microstructure evolution and thermal properties in nanocrystalline Fe during mechanical attrition, Acta Mater. 49 (2001) 365–375. [18] J.Y. Huang, Y.K. Wu, H.Q. Ye, Ball milling of ductile metals, Mater. Sci. Eng. A199 (1995) 165–172. [19] Y.V.R.K. Prasad, K.P. Rao, Influence of oxygen on rate-controlling mechanisms in hot deformation of polycrystalline copper: oxygen-free versus electrolytic grades, Mater. Lett. 58 (2004) 2061–2066. [20] T.D. Shen, J. Zhang, Y. Zhao, What is the theoretical density of a nanocrystalline material? Acta Mater. 56 (14) (2008) 3663–3671. [21] M.F. Ashby, The deformation of plastically non-homogeneous materials, Philos. Mag. 21 (1970) 399–424. [22] S. Diouf, C. Menapace, M. D'Incau, A. Molinari, G. Ischia, Spark plasma sintering of cryomilled copper powder, Powder Metall. 56 (5) (2013) 420–426. [23] G. Cipolloni, M. Pellizzari, A. Molinari, M. Hebda, M. Zadra, Contamination during the high-energy milling of atomized copper powder and its effects on spark plasma sintering, Powder Technol. 275 (2015) 51–59. [24] P.G. Slade (Ed.), Electrical Contacts: Principles and Applications, Marcel Dekker Inc., NY 1999, p. 40. [25] G.F. Bocchini, The influence of porosity on the characteristics of sintered materials, Int. J. Powder Metall. 22 (3) (1986) 185–202.