J Mater Sci (2012) 47:5766–5773 DOI 10.1007/s10853-012-6469-0
Pressure-less spark plasma sintering effect on non-conventional necking process during the initial stage of sintering of copper and alumina Yann Aman • Vincent Garnier • Elisabeth Djurado
Received: 18 January 2012 / Accepted: 2 April 2012 / Published online: 24 April 2012 Ó Springer Science+Business Media, LLC 2012
Abstract In the present study, we focus on the characterization of the necking mechanisms during the early stages of pressure-less spark plasma sintering (PL-SPS) compared to conventional sintering (CS) of two different types of powdered materials (Cu and a-Al2O3). SEM observations of the evolution of particle morphology and necks from the as-received powders to sintered ones show the nature of the neck between particles which were either in contact or not. For alumina, no particular necking process (melt or viscous bridge) was observed regardless of the sintering conditions (PL-SPS and CS), even for a very high heating rate 455 °C/min. For copper, this neck morphology is unequivocally not typical of conventional ones, thus, suggesting mass transport by an ejection mechanism. This particular morphology was seen occasionally. In comparison, the conventionally sintered Cu particles presented a smoother surface, with conventional curved necks suggesting the contribution of surface diffusion mechanisms. Based on partial pressure calculations, a direct thermal effect might not explain the observed non-conventional neck for copper. On the other hand, local field enhancement effect and local favourable thermal breakdown voltage conditions are described and discussed in order to support the experimental results.
Y. Aman V. Garnier (&) Laboratoire MATEIS UMR CNRS 5510, INSA Lyon, Universite´ de Lyon, 7 avenue Jean Capelle, 69621 Villeurbanne, France e-mail:
[email protected] Y. Aman E. Djurado Laboratoire LEPMI UMR CNRS 5631, Grenoble INP, Universite´ Joseph Fourier, Domaine Universitaire 1130 rue de la piscine, 38402 Saint Martin d’He`res, France
123
Introduction In the recent decades, the emergence of spark plasma sintering (SPS) has been seen as one of the most effective sintering technologies for powdered materials including metals, ceramics, glass, biomaterials and even polymers [1–5]. Since being first patented by Inoue half a century ago [6], SPS has developed based on the assumption that using discharge or plasma induced by pulsed electric currents coupled with the application of uniaxial mechanical pressure could help in the low temperature and short-term sintering of advanced nanostructured materials. However, despite the tremendous technological advances in electric current-activated/assisted sintering [6], a large gap still exists between the numerous experimental studies of the process parameters and the fundamental understanding of the underlying mechanisms of SPS. This difficulty in understanding the salient mechanisms could be due to the numerous physical couplings (electrical–thermal–mechanical) [7, 8] encountered in this process. Among the nonelucidated hypothetical mechanisms, such as plasma (or micro-discharge) cleaning of particle surfaces before sintering activation [9], Joule heating [10], electromigration [11], local melting and evaporation [12], or thermal diffusion by Soret effect [13], the existence of plasma has been cited routinely without providing evidence or justification. Recently, Hulbert et al. [14] raised doubts about the presence of plasma or glow discharges from in situ atomic emission spectroscopy and ultrafast voltage experiments. Nevertheless, their conclusions did not provide arguments opposing the existence of a particular ‘SPS effect’, which could explain the true advantages offered by the SPS process. On the other hand, in a recent comparative study between SPS and hot pressing (HP) of a-Al2O3, Langer et al. [15] concluded that densification proceeded similarly
J Mater Sci (2012) 47:5766–5773
in both cases by grain-boundary diffusion mechanism; these authors attributed the beneficial SPS effect to a higher number of necks during the initial stage of SPS due to temperature overshoot. Kun et al. [16] have also shown that SPS enhanced neck growth and accelerated the atom diffusion compared to HP. Although it is generally accepted that the consolidation mechanisms during sintering start at the neck formation and neck growth stage, i.e. the initial stage of sintering, it is noteworthy that the characterization of early formation of necks during SPS has not been well explored so far [17]. An initial experimental study of neck formation during SPS was proposed by Song et al. [18], who based on their theoretical analysis of temperature distribution found that the boiling point could be reached at contacting surfaces of copper particles, resulting in neck growth at low temperatures by a ‘self-adjusting mechanism’. However, while using a low (but not zero) mechanical pressure in their SPS experiments, these researchers only observed some dimples on the compacted spherical copper particles, and finer particles located at the contacting zones, which they attributed to neck formation by local melting and rapid solidification. Recently, Demirskyi et al. [19] have shown that neck growth rate is 100 times faster by SPS than by conventional sintering (CS). Grain-boundary diffusion and power law creep were responsible for neck growth during spark-plasma sintering. However, our study focuses on the neck growth and not on the initial neck formation. In addition, a mechanical pressure is applied. Knowing that metals or ceramic oxides may undergo plastic deformation/ particle surface softening under compressive pressure at relatively low temperatures [20], the application of mechanical pressure should be avoided if one wants to shed light on a probable ‘SPS effect’ on neck formation during sintering activation. Thus, in the present study, a simple modification of the die and punch assembly is proposed, and attention is focused on the characterization of the necking mechanisms at early stages of pressure-less SPS (PL-SPS) in comparison to CS of two different types of powdered materials (Cu, a-Al2O3).
Experimental procedure For the purpose of this study, Cu spherical powders (99 % purity, -325 mesh, d50 * 10 lm, #13990, Alfa Aesar France, France), and a-Al2O3 nanopowders ([99.9 % purity, d50 * 170 nm, #BMA 15, Baı¨kowski Chemicals, France) were used as starting materials. Powdered and sintered samples were characterized by scanning electron microscopy (FEG-SEM Ultra 55, Carl Zeiss AG, Germany). Micrograph analysis and particle size distribution of
5767
Fig. 1 Schematic illustration of the experimental PL-SPS device
starting powders were performed using the public licence software Image J (rsbweb.nih.gov/ij). Sintering experiments were carried out using an FCT SPS apparatus (HP D 25/1-FCT Gmbh, Germany). For PLSPS, the powdered samples were loaded in a graphite die and punch assembly (20 mm inner diameter) lined with graphite foil (0.35-mm-thick). The die/punch assembly has been modified so that the mechanical pressure is only applied on the die and not on the powdered samples (Fig. 1). Owing to this modification, the electrical contact between die, punch and sample is maintained as long as there is no shrinkage of the powder(s). The advantage of this setup is to remove the effect of mechanical pressure. However, this experimental setup does not allow the sample densification shrinkage to be followed during heating. Hence, only the early stage of neck formation can be studied at low temperatures. To do this, the sintering temperature was regulated with a K-type thermocouple (1-mm-diameter) directly introduced through the graphite mould in the powdered sample. The pulse pattern consisted of two pulses lasting 10 ms followed by one pause period of 5 ms, until the desired sintering temperature is reached. Then the power was shut off to allow the sample to cool without any holding time. A vacuum (\1 mbar) was maintained throughout the PL-SPS cycle. Test-error experiments were necessary to find the sintering temperatures beyond which shrinkage occurred. Hence, two PL-SPS sintering conditions (heating rate/sintering temperature) were adopted for each powder: Cu (#1: 465 °C/min/565 °C; #2: 180 °C/min/ 782 °C). Al2O3 (#1: 455 °C/min/632 °C; #2: 175 °C/min/ 747 °C). For all PL-SPS experiments, the maximum current intensity in the graphite mould assembly was IRMS * 2.55 kA for an applied voltage of URMS * 5.1 V during the heating stage. High heating rate were selected to favour local temperature gradients at interparticle contacts shown to enhance grain growth [21, 22]. In order to compare the neck morphologies with those encountered during CS, the same powders were sintered in the same graphite mould assembly, without any mechanical pressure, and using the following conditions in a conventional electric
123
5768
J Mater Sci (2012) 47:5766–5773
Fig. 2 SEM observations of sintering necks in Cu: a raw powders, b and c PL-SPS at 565 °C, d PL-SPS at 782 °C, e CS at 600 °C and f CS at 800 °C
furnace: Cu (#1: 10 °C/min/600 °C—5 min under vacuum; #2: 10 °C/min/800 °C—5 min under vacuum). Al2O3 (#1: 10 °C/min/600 °C— 1 h; #2: 10 °C/min/800 °C— 1 h under vacuum).
Results Using the above experimental PL-SPS conditions, all the sintered samples remained powdered, without any
123
macroscopic shrinkage of the green samples. The SEM observations of the evolution of particle morphology and necks from the as-received powders to sintered ones are shown in Figs. 2 and 3. In the case of copper (Fig. 2), the starting powder presented spherical and rough micronic particles with no capillary necks even between the closest particles (Fig. 2a). After PL-SPS at 565 °C, very few consolidated Cu particles could be distinguished. Particle sphericity and surface roughness were unchanged (Fig. 2b), but more noteworthy is the nature of the neck
J Mater Sci (2012) 47:5766–5773
5769
Fig. 3 SEM observations of sintering necks in a-Al2O3: a raw powder, b PL-SPS at 632 °C, c PL-SPS at 747 °C, d CS at 600 °C under vacuum, e CS at 600 °C under air and f CS at 800 under vacuum
between spherical particles which were very close but not in contact. This neck morphology is unequivocally not typical of that encountered during CS which presents local curvatures due to local mass transfer via surface diffusion or evaporation–condensation mechanisms [23]. Furthermore, angular lines rather than smooth curves were observed at the interface of connected particles (Fig. 2b), suggesting mass transport by an ejection mechanism. When the spherical Cu particles were in contact, no neck could be observed during low temperature PL-SPS (Fig. 2c). This particular morphology was seen occasionally in the
micrographs, although no statistical evaluation was performed in the present study. In comparison, the Cu particles conventionally sintered (CS) at 600 °C presented a smoother surface, with conventional curved necks (Fig. 2e). The latter suggests contributions from surface diffusion mechanisms and particle surface in the necking process during CS, contrary to low temperature PL-SPS of Cu. A high PL-SPS temperature (782 °C) resulted in more consolidated smoother particles, with a significant increase in size of the highly angular neck morphology coexisting with other conventional necks (Fig. 2d). The increase in
123
5770
J Mater Sci (2012) 47:5766–5773
neck growth rate under high current density during the initial SPS stage of Cu was also reported by Frei et al. [24], who attributed it to enhanced evaporation and surface diffusion mechanisms due to the contribution of electromigration forces. However, in the present study, this increase in neck size was also observed for the CS of Cu samples, with the presence of terrace-step patterns on particle surface (Fig. 2f), suggesting surface diffusion mechanisms [25]. In addition, it is also important to note that at high temperatures, all the observed necks (PL-SPS and CS) have a non-densifying character, since they did not bring the sintered particles closer, and no macroscopic shrinkage of the green sample was observed (Fig. 2d, f). Concerning the ultrafine alumina powder (Fig. 3), no noticeable change was observed between the starting powder and the PL-SPS at 632 °C (Fig. 3a, b) although the green sample was subjected to possible macroscopic temperature gradients during the high heating rate (455 °C/ min). No particular necking process (melt or viscous bridge) was observed regardless of the sintering conditions (PL-SPS and CS). This contradicts the ambiguous ‘liquidstate bridging’ and ‘whisker-like’ necks reported by Kumeda et al. [26] in alumina balls sintered at 1500 °C. However, increasing the sintering temperature in PL-SPS and CS resulted in more spherical and agglomerated particles (Fig. 3c–f), probably due to surface diffusion mechanisms, without any macroscopic shrinkage of the green samples because the neck formation associated with the first stage of sintering does not lead to shrinkage. The latter is consistent with the low activation energy (236 kJ/ mol) for the non-densifying necking process during the initial stage of SPS of ultrafine a-Al2O3, as reported by Stanciu et al. [27], which was ascribed to enhanced surface diffusion under an electric field.
Discussion Thermal effect Our results indicate that the necking process during the early sintering stage of SPS is strongly dependent on the nature, size, and surface characteristics of the powder. However, from the observation of a non-conventional neck in the case of electrically conductive Cu powder, it would be reasonable to think that this could be attributed to a temperature effect. According to Alcock et al. [28], the vapour pressure PCu of Cu(g) over the temperature range [298–357 K] can be expressed as: logðP=atmÞ ¼ 9:123 ÿ 17748 T ÿ1 ÿ 0:7317 log T;
ð1Þ
where T is the absolute temperature (K). Hence, at an SPS temperature of 565 °C, PCu & 6.4 9 10-15 atm, and at
123
782 °C, PCu & 1.22 9 10-10 atm, which is negligible in comparison to the vapour pressure PCu at the melting point (*5910-7 atm at 1084 °C) and boiling point of copper (*1.03 atm at 2561 °C), calculated from Geiger et al. [29]. Hence, the particular neck observed in the case of PLSPS of Cu cannot be ascribed to vaporization induced by the sintering temperature. In the case of a-Al2O3, the situation is more complex because it does not vaporize as alumina gas, but decomposes into a mixture of Al, O2, AlO, Al2O, and AlO2, and the proportion of each component changes as a function of temperature. Brewer and Searcy [30] have shown that AlO(g) was the principal aluminium species when Al2O3(s) is volatilized in inert gas or vacuum, according to the reaction (2): 2Al2 O3ðsÞ ! 4AlOðgÞ + O2ðgÞ
ð2Þ
According to their experimental results, the vapour pressure PAlO for Al2O3 was between 1.03 9 10-5 and 1.66 9 10-5 atm at the melting point of Al2O3 (2072 °C). This partial pressure decreases rapidly, as shown by Lou, Mitchell and Heuer [31], displaying a maximum equilibrium pressure line between Al2O3(s) and gaseous sub-oxides of aluminium at high temperatures. At 1627 °C, the maximum PAlO is *5910-9 atm. This result suggests that PAlO will therefore be much smaller at 632 and 747 °C, since it is expected that weight loss due to evaporation in a hightemperature experiment is significant for vapour pressures of evaporating species exceeding 1.02 9 10-8 atm [31]. Therefore, the low temperatures used in the PL-SPS experiments do not allow evaporation–condensation to occur during the initial sintering stage of Al2O3. One should also consider the effect of possible vapour pressure PAl for aluminium that may condense to form liquid aluminium. Levi et al. [32] have shown that liquid aluminium may favour Al2O3 epitaxial growth at low temperatures (below 1100 °C) and Sang Ho Oh et al. [33] observed that such growth effectively occurs for sapphire nanowires by interfacial diffusion of oxygen through the ordered liquid aluminium atoms. However, this phenomenon may be very limited due to the presence of a significant oxygen partial pressure inside the alumina sample that directly oxidizes aluminium vapour to gaseous Al2O. The thermal effect mechanisms described in this section may not give a significant contribution to the formation of the specific neck morphology observed which consequently cannot be ascribed purely to a thermal effect. Electromigration effect For conductive material like copper, the current on mass transport had an effect [11]. The electron density passing through the sample creates an electron wind (electromigration)
J Mater Sci (2012) 47:5766–5773
which gives an added force to the specie diffusivity. This enhanced copper diffusivity may occur in our SPS conditions. However, matter ejection could not be associated with an electromigration mechanism. On the other hand, the electron wind goes from the cathode to the anode, and therefore may act to increase the mass transport in only one direction. This anisotropic phenomenon may thus give an asymmetric neck as observed in Fig. 2d, but could not be associated with particular necks like those observed in Fig. 2b. Field enhancement effect and breakdown voltage In the case of two nearly touching conducting perfect spheres, Lekner [34] has established theoretically the field enhancement factor f for a given ratio of sphere separation s to radius r of the spheres, so that ÿ1 Eav p2 r 1 r f ¼ ln þ ln 2 þ c ; ð3Þ E0 3 s 2 s where E0 is the external electric field, Eav the average field in the gap between the spheres, along the line of centres, and c & 0.5772 is the Euler’s constant. Applying this equation to the case of spherical copper as observed on Fig. 2b with the assumption that copper is an excellent conducting material and for a given ratio r/s = 25 (r = 5 lm, s = 200 nm), the field enhancement factor is &29. Thus, since the applied external electric field during SPS operation slightly exceeds 300 V/m (i.e. 6 V/2 cm), it can be deduced that in the present case of the conducting nearly touching copper spheres, the local electric field could reach 8400 V/m. However, larger electric field would normally be necessary to create the gaseous breakdown voltage and cause the gas to partially ionize and begin conducting: creating a spark/plasma. The breakdown voltage was first studied by Paschen [35] using parallel plates in a gas as a function of pressure and gap distance. Paschen’s law states that the voltage required
5771
to spark a specific gas is proportional to the ‘pressure times gap distance’: V¼
aðpdÞ lnðpdÞ þ b
where V is the breakdown voltage, a and b are constants depending on the gas composition, p is the pressure and d the gap distance between the plates. Paschen’s law predicts a minimum breakdown voltage, as shown in Fig. 4. However, experimental results by Dhariwal et al. [36] and Slade et al. [37], as well as breakdown simulation models proposed by Tirumala et al. [38] and Radmilovi0 cRadjenovi0 c et al. [39] are also shown in Fig. 4. No large difference in breakdown voltage can be seen for gap spacing larger than 6 lm. Successive ionization of gas molecules occurs and induces the gas breakdown (Townsend mechanism). On the other hand, substantial deviations occur between experimental/simulated data and Paschen’s law for gaps smaller than 6 lm. In this case, the number of gaseous molecules/ions is too small in the innerelectrode space for the Townsend mechanism to occur. The electron production necessary to induce breakdown is partially due to the electron field emission (i.e. emission of electrons induced by external electromagnetic fields). To reduce the breakdown voltage, the field emission could be enhanced when positive gaseous ions approach the cathode and reduce the potential barrier at the cathode, thus making it easier for electrons to tunnel due to the electric field [40]. This mechanism called ‘Ion-enhanced field emission’ corresponds to the so-called modified Paschen’s curve [38]. The field enhancement factor f for two nearly touching copper particles and the small gap between the copper particles are consistent with an ion-enhanced field emission mechanism. Furthermore, our SPS experiments were conducted up to 782 °C where the vapour pressure of copper may also contribute to this mechanism, and play the role of charge carrier supporting the current in the space charge region. The large current intensity (up to IRMS * 2.55 kA) may also directly contribute to the electron production necessary to induce the breakdown. Finally, temperature also plays a crucial role because heat induces thermal electron emission. The electrons gain their energy from heat to be liberated from the Fermi level, Richardson shows that the electrons produced from heated metal depend exponentially on the temperature [41], i.e. the emission current per unit area J is related to the temperature according to the Richardson–Dushman equation [42]: J ¼ AT 2 expðÿeU=kTÞ
Fig. 4 Breakdown voltage (V) versus gap spacing (lm) for copper
ð4Þ
ð5Þ
where A is a constant (1.2 9 106 A m-2 K-2 for copper), e is the electronic charge, / is the work function (4.41 eV
123
5772
for copper), k Boltzmann’s constant and T is the absolute temperature. The current density reaches 1.12 nA m-2 at 782 °C helping us increase the local electron production in the space charge region and induce the breakdown. The specific SPS conditions (temperature, pulsing current and electromagnetic field) are favourable for creating local voltage breakdown inducing local spark/plasma as shown in Fig. 2b, whereas an electrical discharge increases the gap distance because of erosion of the two electrodes when an arc strikes them. The local plasma generated between the two copper particles brings enough matter and energy (temperature) to bond the particles by an ejection mechanism. Other studies also exist to confirm this behaviour. Similar modified surfaces such as ours (craters and frozenin liquid) have been observed by AFM (atomic force microscopy) after electrical sparks and arc plasma discharges [43]. With currents as low as 1.3 A/cm2, Cordier and co-workers [44] have observed at yttria-stabilized zirconia granule surfaces protrusion formation which may establish contacts with other granules. Recently, Demirskyi et al. [45] also considered a possible discharge effect at low heating temperatures and high reversed heating gradient, as an explanation of unexpected microstructures obtained by SPS. An in situ experiment [46], inside a TEM, has shown that neck formation between Ni particles could occur due to the applied electrical current and electrical field but in the absence of an external heat source. However, this result was observed for already contacting particles. Another study [47] shows evidence of evaporation during SPS leading to enhanced neck growth due to the effect of the pulsed current. All these results are favourable to also support the possible existence of a local spark/plasma.
Conclusion To conclude, using two different types of powdered materials (Cu, a-Al2O3), we have characterized the necking mechanisms at early stages of PL-SPS and compared them to the mechanisms of CS. The die/punch assembly of the SPS apparatus has been modified so that the mechanical pressure is only applied on the die and no longer on the powdered samples. This setup is thus able to avoid the mechanical pressure effect during sintering. No particular necking process or enhanced diffusion mechanism was observed in the case of alumina nanopowders. PL-SPS experiments, in comparison with reported results in literature suggest neck formation by a surface diffusion mechanism at low SPS sintering temperature. On the other hand, the present study highlighted unambiguously a non-conventional necking mechanism in the case of the conductive copper powder. Based on
123
J Mater Sci (2012) 47:5766–5773
thermodynamic calculations, it was shown that a single thermal effect could not explain the non-conventional neck. The possible local field enhancement has been calculated, and the required parameters controlling the breakdown voltage are evaluated. The non-conventional neck therefore is formed due to favourable local conditions produced by SPS (temperature, pulsing current and electromagnetic fields), which create local thermal breakdown inducing local sparks/plasma. Acknowledgements The authors gratefully acknowledge the support received from ‘Re´gion Rhoˆne Alpes—MACODEV’ for this study (M. Yann Aman thesis), Pr. G. Fantozzi and G. Bonnefont from the SPS consortium at INSA Lyon.
References 1. Zhang ZH, Wang FC, Wang L, Li S-K (2008) Mater Sci Eng A 476:201 2. Shen Z, Johnsson M, Zhao Z, Nygren M (2002) J Am Ceram Soc 85:1921 3. Riello P, Bucella S, Zamengo L, Anselmi-Tamburini U, Francini R, Pietrantoni S, Munir ZA (2006) J Eur Ceram Soc 26:3301 4. Gu YW, Loh NH, Khor KA, Tor SB, Cheang P (2002) Biomaterials 23:37 5. Omori M (2000) Mater Sci Eng A 287:183 6. Grasso S, Sakka Y, Maizza G (2009) Sci Technol Adv Mater 10:053001 7. Mondalek P, Silva L, Durand L, Belleta M (2010) AIP Conf Proc 1252:697 8. Vanmeensel K, Laptev A, Hennicke J, Vleugels J, Van der Biest O (2005) Acta Mater 53:4379 9. Groza JR, Zavaliangos A (2000) Mater Sci Eng A 287:171 10. Carney CM, Mah TI (2008) J Am Ceram Soc 91:3448 11. Munir ZA, Anselmi-Tamburini U, Ohyanagi M (2006) J Mater Sci 41:763. doi:10.1007/s10853-006-6555-2 12. Chaim RJ (2006) J Mater Sci 41:7862. doi:10.1007/s10853006-0605-7 13. Olevsky E, Froyen L (2009) J Am Ceram Soc 92:S122 14. Hulbert DM, Anders A, Dudina D, Andersson J, Jiang D, Unuvar C, Anselmi-Tamburini U, Lavernia EJ, Mukherjee AK (2008) J Appl Phys 104:033305 15. Langer J, Hoffmann MJ, Guillon O (2009) Acta Mater 57:5454 16. Kun W, Zhengyi F, Weimin W, Yucheng W, Jinyong Z, Qingjie Z (2007) J Mater Sci 42:302. doi:10.1007/s10853-006-1013-8 17. Zhaohui Z, Fuchi W, Lin W, Shukui L, Osamu S (2008) Mater Lett 62:3987 18. Song X, Liu X, Zhang J (2006) J Am Ceram Soc 89:494 19. Demirskyi D, Borodianska H, Agrawal D, Ragulya A, Sakka Y, Vasylkiv O (2012) J Alloys Compd 523:1 20. Chaim R (2007) Mater Sci Eng A 443:25 21. Aman Y, Garnier V, Djurado E (2011) J Am Ceram Soc 94:2825 22. Gephart S, Singh J, Kulkarni A (2011) J Mater Sci 46:3659. doi: 10.1007/s10853-011-5283-4 23. Kang S-JL (2005) Sintering, 1st edn. Elsevier, Oxford 24. Frei JM, Anselmi-Tamburini U, Munir ZA (2007) J Appl Phys 101:114914 25. Raouafi F, Barreteau C, Desjonque`res MC, Spanjaard D (2002) Surf Sci 505:183 26. Kumeda K, Nakamura Y, Takata A, Ishizaki K (1999) J Ceram Soc Jpn 107:187
J Mater Sci (2012) 47:5766–5773 27. Stanciu L, Quach D, Faconti C, Groza JR, Raether F (2007) J Am Ceram Soc 90:2716 28. Alcock CB, Itkin VP, Horrigan MK (1984) Can Metall Q 23:309 29. Geiger F, Busse CA, Loehrke RI (1987) Int J Thermophys 8:425 30. Brewer L, Searcy AW (1951) J Am Chem Soc 73:5308 31. Lou VKL, Mitchell TE, Heuer AH (1985) J Am Ceram Soc 68:49 32. Levi G, Kaplan WD (2002) Acta Mater 50:75 33. Ho Oh S, Chisholm MF, Kauffmann Y, Kaplan WD, Luo W, Ru¨hle M, Scheu C (2010) Science 330:489 34. Lekner J (2010) J Electrostatics 68:299 35. Paschen F (1889) Ann Phys 273:69 36. Dhariwal RS, Torres JM, Desmulliez MPY (2000) IEE Proc 147:261 37. Slade PG, Taylor ED (2002) IEEE Trans Compon Packag Technol 25:390 38. Tirumala R, Go DB (2010) Appl Phys Lett 97:151502 39. Radmilovic´-Radjenovic´ M, Radjenovic´ B (2008) Plasma Sources Sci Technol 17:024005
5773 40. Go DB, Pohlman DA (2010) J Appl Phys 107:103303 41. Richardson OW (1901) Proc Camb Phil Soc 11:286 42. Brandes EA, Brook GB (eds) (1992) Smithells metals reference book, 7th edn. Butterworth-Heinemann Ltd., Bodmin 43. Timko H, Djurabekova F, Nordlund K, Costelle L, Matyash K, Schneider R, Toerklep A, Arnau-Izquierdo G, Descoeudres A, Calatroni S, Taborelli M, Wuensch W (2010) Phys Rev B 81: 184109 44. Cordier A, Kleitz M, Steil MC (2012) J Eur Ceram Soc doi: 10.1016/j.jeurceramsoc.2011.12.022 45. Demirskyi D, Borodianska H, Grasso S, Sakka Y, Vasylkiv O (2011) Scripta Mat 65:683 46. Holland TB, Thron AM, Bonifacio CS, Mukherjee AK, van Benthem K (2010) Appl Phys Lett 96:243106 47. Toyofuku N, Kuramoto T, Imai T, Ohyanagi M, Munir ZA (2012) J Mater Sci 47:2201. doi:10.1007/s10853-011-6026-2
123
