J. Am. Ceram. Soc.

, 98 [12] 36243627 (2015)

DOI: 10.1111/jace.13866
2015 The American Ceramic Society

Journal Rapid Communication

Onset Criterion for Flash Sintering

Yanhao Dong and I.-Wei Chen
Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6272

 
When the sample temperature is warm enough so that the dT 1 V2
4 
 erS T  Tf
4
(2)
Joule heating takes over the environments radiation heating as dTf KC R0 expEa =kB T
the dominant heating means, thermal runaway follows and
flash sintering is triggered. This condition accurately predicts
the reported onset temperature Ton of all the constant-field (E) It may be cast into the following dimensionless form
experiments on flash sintering, performed under a constant using T~ kB T=Ea , T~f kB Tf =Ea , a = V2/KCR0, and b
heating rate. The predicted linear ln(E2/Ton4) versus Ton 1 erSEa 4 =KCkB 4
relation determines the activation energy and the relative rank-
ing of the electrical conductivity of the tested materials.  
dT~ 1
bT~f  bT~4
4
a exp  (3)
dT~f T~
I. Introduction
1

F LASH sintering, featuring a sudden increase in densica-

tion rate and conductivity, is known to happen to a
ceramic green body under a constant voltage while the fur-
where the terms on the right hand side are the rates
of, respectively, Joule heating, radiation heating from
furnace, and radiation cooling from insulator to environ-
nace temperature is ramped up. Much attention has been ment.
given to exploring the underlying sintering mechanisms,29 The analytic form of Eq. (3) immediately lends itself to
but there is no denial that thermal runaway caused by Joule the following approximate solution, in four segments. Ini-
heating is of critical importance. While Grasso et al.4 used tially, Joule heating is insignicant so radiation heating
nite element simulation to estimate a maximum of 1640C, from T~f causes T~ to rise in locked-step (dT=d
~ T~f  1) with a
local melting suggesting a much higher temperature has been
small lag, D 1=4bT~f . (ii) As T~f reaches T~  , Joule heat-
3
observed.5 Meanwhile, in situ reading of average sample tem-
perature using impedance spectroscopy has provided direct ing takes over causing T~ to run away. Here, T~  may be
evidence of thermal runaway.6 Therefore, it is of essential approximately estimated by equating Joule heating to radia-
tion heating from T~  T~  , that is, a exp1=T~  bT~  .
4
interest to ascertain the condition that triggers thermal run-
away and, hopefully, to analytically predict it in a straight- (iii) As T~ approaches its upper limit T~  , T~  is so high
forward way. Here, we present a simple yet accurate solution that Joule heating reaches its (innite T) ~ limit a and the
to the problem by solving the heat transfer equation both temperature does not rise further. This limit, set by equat-
analytically and numerically. The key features of the solution ing a and radiation cooling, is approximately given by
are substantiated by an extensive body of literature on the
a bT~  . (iv) As T~f catches up with T,~ radiation heating
4
onset temperature Ton of ash sintering.13,7,1025
and cooling once again equilibrates with each other so the
solution returns to that of (i). In summary, the approxi-
II. Solution of Onset Temperature for Thermal Runaway mate solution is
The temperature (T) evolution of an electrically loaded insu-
lator of a heat capacity C and a resistance R = R0 exp(Ea/  9
For 0 \T~f \T~ ; T~  T~f  1
 T~f >
>
kBT) (R0 is the pre-exponential factor, Ea the activation 4bT~f
3
>
=
  
energy, and kB Boltzmanns constant) is given by for T~f T~ ; T~ \T\~ T~ (4)
   >
for T \Tf  T ; T~ T~
~ ~ ~ >
>
   ;
dT 1 V2
 for T~f [ T~ ; T~  T~f
 erS T4  T4f (1)
dt C R0 expEa =kB T
Since T~ a=b1=4 in real temperature (of the order of
In the above, V is the voltage, e is the emissivity, r is the 5 9 105 K) well exceeds the melting point of all materials,
StefanBoltzmann constant, S is samples surface area, and Tf only segments (i-ii) are of practical interest. The numerical
is the environment (furnace) temperature. Typically, initial solution for Eq. (3) shown in Fig. 1 for one representative
heating requires auxiliary heating raising Tf at a constant set of a and b values (a = 7.32 9 108 and b = 1.63 9 107)
ramping rate K. Under this condition, Eq. (1) becomes conrms the bilinear shape of (i-ii): segment (i) is very close
to T~ T~f and segment (ii) runs away essentially vertically at
T~f T~ . To further understand the (i-ii) transition, we exam-
O. Guilloncontributing editor ine the switchover condition suggested above: at T~ , Joule
heating equals radiation heating
 
1
bT~f
4
Manuscript No. 37073. Received June 19, 2015; approved July 27, 2015. a exp  (5)

Author to whom correspondence should be addressed. e-mail: iweichen@seas. T~
upenn.edu

3624
December 2015 Rapid Communications of the American Ceramic Society 3625

predicted by (a) the numerical solution of either the

heat-transfer equation or (b) the modied switchover condi-
tionJoule heating rate equal to radiation heating rate
such as Eq. (8).

III. Comparison with Literature Data

Equation (8) is used to compare the predicted Ton with the
data in the ash sintering literature, which commonly reports
Ton and the electrical eld E (E = V/d, where d is sample
length). In Fig. 2(a) we plot all the collected data in the liter-
ature in a form motivated by Eq. (8)
   
E2 Ea erSd2 R0
ln ln (9)
T4on kB Ton b

We nd data of each experiment follow a straight line

with a slope that identies Ea/kB listed in Table I. The
Fig. 1. Numerical solution of Eq. (3) (black curve) and Eq. (5) apparent activation energies are mostly in broad agreement
(purple curve) using a = 7.32 9 108 and b = 1.63 9 107. The bottom with the reference activation energies of the sample materi-
and left axis show dimensionless temperature dened in the text; top als,7,8,15,2539 though they tend to be on the high side. This
and right axis show real temperature. Inset: onset furnace is understandable since, during sintering, density improves
temperatures from exact solution Ton (corresponding to T~ ) and with temperature, thus lowering the resistance and impart-
approximate ones: Tf|3,5 from intersect of black and purple curves, ing it steeper temperature dependence, hence a higher
Ta from Eq. (7) and Tb from Eq. (8).
apparent activation energy. A simplied form of Eq. (9),
without the Ton4 term, was previously proposed by us based
on an incomplete analysis.9 It also t the literature data of
or, in its explicit form,
zirconia well but provided a lower activation energy by 0.1
  0.5 eV. This simplied form worked because most studies
1 a V2 kB 4
T~f exp involved Ton between 800 and 1300 K, which is in a rela-
4
(6)
~
T b erSR0 Ea 4 tively small range in the logarithmic scale.
The materials covered in Fig. 2(a) range from relatively
In Fig. 1, Eq. (5) corresponds to a gradually rising curve conducting semiconductors [including some electrode
that intersects the bilinear solution of Eq. (3) at a point
slightly to the right of the vertex (Joule heating exceeds (a)
radiation heating above the curve, and vice versa). Since (ii)
is virtually vertical, the intersection T~f j3;5 is nearly identical
to T~  as illustrated in the inset of Fig. 1 for a variety of T~ 
solutions of various (a, b) covering the range of Ea = 12 eV,
R0 = 0.010.03 , V = 1264725 V, K = 520 K/min,
C = 0.0020.003 J/K, and e = 0.61. (Here, T~  and T~f j3;5 are
expressed in real temperature as Ton and Tf3,5, respec-
tively. They dier by no more than 2 K.) This conrms that
Eq. (5) is satised at the transition T~f T~  : the switchover
condition is correct.
Moreover, we nd an approximate solution to T~ can be
obtained by using either of the following two modications
of Eq. (5)
 
1 a V2 kB 4
T~f exp
4
(7)
aT~f b erSR0 Ea 4

  (b)
1 ba bV2 kB 4
T~f exp
4
(8)
T~f b erSR0 Ea 4

Here, a and b are numerical constants that may be empiri-

cally set to force an agreement between the solutions of these
modied forms and T~  . Indeed, we nd a in the range 1.11
1.17 will allow a perfect agreement between the solution of
Eq. (7) and T~  , and if an average a = 1.14 is used, the onset
furnace temperature Ta calculated from Eq. (7) agrees rea-
sonably well with the furnace temperature Ton corresponding
to T~  as shown in the inset of Fig. 1. Likewise, we nd to
produce a perfect agreement b will vary from 7 to 9, and if
Fig. 2. (a) Flash sintering data from literature on ln(E2/Ton4) vs.
an average b = 8.2 is used, the calculated onset furnace tem- Ton1 plot with slope indicating Ea/kB. Electronic semiconductors
perature Tb is again very good as shown in the same inset. are circled in green, oxygen ion conductors in blue, and insulating
The accuracy for the latter is better than 4%. oxides in red. (b) Ranking by material index
In summary, the onset furnace temperature Ton that signi- lnerSd2 R0 expEa =kB T=b. Materials become more resistive from
es the abrupt transition to thermal runaway is accurately left to right.
3626 Rapid Communications of the American Ceramic Society Vol. 98, No. 12

Table I. Calculated Activation Energy from Fig. 2(a) If the sample resistance is too low, then this voltage may be
Compared with Literature Value just an insignicant fraction of the applied voltage. This is
the case of the current compliance, which sets the Joule heat-
Calculated ing rate at a constant thus allows no thermal runaway. Sec-
Materials Ea (eV) Reference Ea (eV)
ond, the resistance/conductance of relevance to Eq. (1) is the
8YSZ 2
1.40 0.91.326,27 one at the onset temperature or higher. So, if the sample is
8YSZ3 1.56 0.91.326,27 highly conductive at low temperature but becomes less so at
8YSZ10 1.24 0.91.326,27 high temperature, as in the case of a metal, no thermal run-
3YSZ1 1.39 1.01.77,28 away is expected. Likewise, while BZCYYb is a good proton
3YSZ7 1.72 1.01.77,28 conductor in moist atmosphere, at high temperature the
GDC1011 1.16 0.50.829,30 atmosphere may be so dry that proton conduction is no
GDC2012 0.77 0.5-0.829,30 longer signicant. (A moist atmosphere should help trigger
Y2O313 1.88 1.452.0731,32 ash sintering, just as reduced atmosphere helps trigger ash
Al2O3/3YSZ14 1.92 / sintering in ZnOR.22) Third, while we have ignored heat con-
TiO215 1.99 1.615,33 duction loss from the sample, this contribution may be
BaTiO316 1.52 1.01.334,35 included in the analysis by adding a heat loss term propor-
SrTiO317 1.66 0.71.236 tional to TTf. This term does not signicantly aect the
Co2MnO418 0.38 0.1837 onset condition in Fig. 1 unless it limits the ability of radia-
Co2MnO419 0.58 0.1837 tion heating to heat up the sample to T~ . Fourth, the very
SiC20 2.68 / abrupt nature of the transition and the vertical character of
SiC20 2.76 / the runaway in Fig. 1 obviates the need for any elaboration
SiC20 2.18 / of the curvature of the vertex. This renders unnecessary
SiC20 2.63 / the previous analyses of Todd et al.7 and Zhang et al.,8
ZnO21 2.76 1.988 which considered the higher order derivatives of temperature
LSCF23 0.24 0.138 changes but lacked the simplicity of our solution. These anal-
LSCF/GDC11 1.23 / yses failed to predict Eq. (8), which proves most general and
BZCYYb24 1.32 0.6939 useful.
MgOAl2O325 5.04 1.62.325
V. Conclusions
materials such as (La,Sr)(Co,Fe)O3, abbreviated as LSFC] to (1) The furnace temperature Ton at the abrupt onset of ther-
ionic conductors (such as doped zirconia) to highly insulating mal runaway is accurately predicted by either the heat trans-
oxides (such as magnesia-doped alumina, abbreviated as fer equation or the switchover conditionthe Joule heating
MgOAl2O3). Broadly speaking, their relative positions in rate equals radiation heating rate. Using a simplied switch-
Fig. 2(a) follow the above sequence from right to left. Specif- over condition such as Eq. (8), the onset temperature can be
ically, there are two zirconias (8 and 3 mol% yttria-stabilized easily calculated within an accuracy of 4% by using experi-
zirconia, that is, 8YSZ2,3,10 and 3YSZ1,7), Gd-doped ceria mentally known parameters.
which has the same structure as zirconia but with Gd instead (2) The predicted linear relationship ln(E2/Ton4) versus
of Y as the acceptor dopant (GDC10 for Gd0.1Ce0.9O211 and Ton1 holds for all the reported ash sintering studies per-
GDC20 for Gd0.2Ce0.8O212), Y2O313 which has a structure formed under a constant voltage-constant heating rate set-
closely related to zirconia, a zirconia-alumina composite ting, providing an activation energy and a material
(3YSZ/Al2O314), three titanates (TiO2,15 BaTiO316 and conductivity ranking that are in broad agreement with the
SrTiO317), a spinel (Co2MnO418,19), silicon carbide (SiC20), reference values.
ZnO,21,22 a perovskite electrode ceramic La0.6Sr0.4Co0.2Fe0.8O3
(LSCF23) and its composite (LSCF/GDC11), and a complex
Acknowledgments
perovskite BaZr0.1Ce0.7Y0.1Yb0.1O3 (BZCYYb24) which is
a fast proton conductor. Although the position of This work was supported by the Department of Energy (BES grant no.
MgOAl2O325 suggests a very insulating ceramic, which it is, DEFG02-11ER46814) and used the facilities (LRSM) supported by the U.S.
National Science Foundation (grant no. DMR-1120901).
the estimated Ea from this plot signicantly deviates from
the literature value perhaps because of inadequate number of
data (only two experimental values of Ton, of very similar References
values). Two studies of ZnO21,22 (a semiconductor) under 1
M. Cologna, B. Rashkova, and R. Raj, Flash Sintering of Nanograin Zir-
dierent atmospheres, which are known to aect resistivity, conia in <5 s at 850C, J. Am. Ceram. Soc., 93 [11] 35569 (2010).
are also compared in the plot: as expected, the one in 2
M. Cologna, A. L. G. Prette, and R. Raj, Flash-Sintering of Cubic Yttria-
reduced atmosphere22 (ZnOR) lies on the right since it is Stabilized Zirconia at 750C for Possible Use in SOFC Manufacturing, J.
more conductive. (Its activation energy cannot be accurately Am. Ceram. Soc., 94 [2] 3169 (2011).
3
J. S. C. Francis and R. Raj, Flash-Sinterforging of Nanograin Zirconia:
estimated from the slope because, just as density, reduction Field Assisted Sintering and Superplasticity, J. Am. Ceram. Soc., 95 [1] 138
proceeds during ash sintering causing the appearance of 46 (2012).
4
a higher activation energy.) Assuming (e, S, d) do not vary S. Grasso, et al., Modeling of the Temperature Distribution of Flash Sin-
signicantly between experiments, we see erSd2 R0 exp tered Zirconia, J. Ceram. Soc. Jpn., 119 [1386] 1446 (2011).
5
M. C. Steil, D. Marinha, Y. Aman, J. R. C. Gomes, and M. Kleitz, From
Ea =kB T=bwhich is no more than the sample resistance Conventional AC Flash-Sintering of YSZ to Hyper-Flash and Double Flash,
R = R0exp(Ea/kBT) multiplied by a numerical constantmay J. Eur. Ceram. Soc., 33 [11] 2093101 (2013).
6
be used as a material index for ranking resistivity. In J. Park and I. W. Chen, In Situ Thermometry Measuring Tempera-
Fig. 2(b), we show such ranking by the material index, com- ture Flashes Exceeding 1700C in 8 mol% Y2O3-Stablized Zirconia
Under Constant-Voltage Heating, J. Am. Ceram. Soc., 96 [3] 697700
puted from the slopes and the intercepts in Fig. 2(a) while (2013).
setting T = 1000 K, seems quite reasonable over more than 8 7
R. I. Todd, E. Zapata-Solvas, R. S. Bonilla, T. Sneddon, and P. R. Wil-
orders of magnitude. shaw, Electrical Characteristics of Flash Sintering: Thermal Runaway of
Joule Heating, J. Eur. Ceram. Soc., 35 [6] 186577 (2015).
8
Y. Zhang, J. Jung, and J. Luo, Thermal Runaway, Flash Sintering and
IV. Discussion Asymmetrical Microstructural Development of ZnO and ZnO-Bi2O3 Under
Direct Currents, Acta Mater., 94, 87100 (2015).
9
Several notes are mentioned below in connection to our anal- Y. Dong and I. W. Chen, Predicting the Onset of Flash Sintering, J. Am.
ysis. First, the voltage in Eq. (1) is the voltage on the sample. Ceram. Soc., 98 [8] 23335 (2015).
December 2015 Rapid Communications of the American Ceramic Society
10
J. A. Downs and V. M. Sglavo, Electric Field Assisted Sintering of Cubic
Zirconia at 390C, J. Am. Ceram. Soc., 96 [5] 13424 (2013).
11
A. Gaur and V. M. Sglavo, Flash Sintering of (La, Sr)(Co, Fe)O3-Gd-
Doped CeO2 Composite, J. Am. Ceram. Soc., 98 [6] 174752 (2015).
12
X. M. Hao, Y. J. Liu, Z. H. Wang, J. S. Qiao, and K. N. Sun, A Novel
Sintering Method to Obtain Fully Dense Gadolinia Doped Ceria by Applying
a Direct Current, J. Power Sources, 210, 8691 (2012).
13
H. Yoshida, Y. Sakka, T. Yamamoto, J. M. Lebrun, and R. Raj, Densi-
cation Behaviour and Microstructural Development in Undoped Yttria Pre-
pared by Flash-Sintering, J. Eur. Ceram. Soc., 34 [4] 9911000 (2014).
14
K. S. Naik, V. M. Sglavo, and R. Raj, Field Assisted Sintering of Cera-
mic Constituted by Alumina and Yttria Stabilized Zirconia, J. Eur. Ceram.
Soc., 34 [10] 243542 (2014).
15
S. K. Jha and R. Raj, The Eect of Electric Field on Sintering and Elec-
trical Conductivity of Titania, J. Am. Ceram. Soc., 97 [2] 52734 (2014).
16
J. C. MPeko, J. S. C. Francis, and R. Raj, Field-Assisted Sintering of
Undoped BaTiO3: Microstructure Evolution and Dielectric Permittivity, J.
Eur. Ceram. Soc., 34 [15] 365560 (2014).
17
A. Karakuscu, et al., Defect Structure of Flash-Sintered Strontium Tita-
nate, J. Am. Ceram. Soc., 95 [8] 25316 (2012).
18
A. L. G. Prette, M. Cologna, V. Sglavo, and R. Raj, Flash-Sintering of
Co2MnO4 Spinel for Solid Oxide Fuel Cell Applications, J. Power Sources,
196 [4] 20615 (2011).
19
A. Gaur and V. M. Sglavo, Flash-Sintering of MnCo2O4 and Its Relation
to Phase Stability, J. Eur. Ceram. Soc., 34 [10] 2391400 (2014).
20
E. Zapata-Solvas, S. Bonilla, P. R. Wilshaw, and R. I. Todd, Preliminary
Investigation of Flash Sintering of SiC, J. Eur. Ceram. Soc., 33 [1314] 2811
6 (2013).
21
C. Schmerbauch, J. Gonzalez-Julian, R. Roder, C. Ronning, and O. Guil-
lon, Flash Sintering of Nanocrystalline Zinc Oxide and its Inuence on
Microstructure and Defect Formation, J. Am. Ceram. Soc., 97 [6] 172835
(2014).
22
Y. Zhang and J. Luo, Promoting the Flash Sintering of ZnO in Reduced
Atmospheres to Achieve Nearly Full Densities at Furnace Temperatures of
<120C, Scripta Mater., 106, 269 (2015).
23
A. Gaur and V. M. Sglavo, Densication of La0.6Sr0.4Co0.2Fe0.8O3 Cera-
mic by Flash Sintering at Temperature Less Than 100C, J. Mater. Sci., 49
[18] 632132 (2014).
24
T. Z. Jiang, Y. J. Liu, Z. H. Wang, W. Sun, J. S. Qiao, and K. N. Sun,
An Improved Direct Current Sintering Technique for Proton Conductor
BaZr0.1Ce0.7Y0.1Yb0.1O3: The Eect of Direct Current on Sintering Process,
J. Power Sources, 248, 706 (2014).
3627
25
M. Cologna, J. S. C. Francis, and R. Raj, Field Assisted and Flash Sin-
tering of Alumina and its Relationship to Conductivity and MgO-Doping, J.
Eur. Ceram. Soc., 31 [15] 282737 (2011).
26
I. Kosacki, T. Suzuki, V. Petrovsky, and H. U. Anderson, Electrical
Conductivity of Nanocrystalline Ceria and Zirconia Thin Films, Solid State
Ionics, 136, 122533 (2000).
27
O. H. Kwon and G. M. Choi, Electrical Conductivity of Thick Film
YSZ, Solid State Ionics, 177 [3536] 305762 (2006).
28
R. Ramamoorthy, D. Sundararaman, and S. Ramasamy, Ionic Conduc-
tivity Studies of Ultrane-Grained Yttria Stabilized Zirconia Polymorphs,
Solid State Ionics, 123 [14] 2718 (1999).
29
R. O. Fuentes and R. T. Baker, Structural, Morphological and Electrical
Properties of Gd0.1Ce0.9O1.95 Prepared by a Citrate Complexation Method, J.
Power Sources, 186 [2] 26877 (2009).
30
H. Inaba and H. Tagawa, Ceria-Based Solid ElectrolytesReview, Solid
State Ionics, 83 [12] 116 (1996).
31
T. Norby and P. Kofstad, Electrical-Conductivity and Defect Structure
of Y2O3 as a Function of Water-Vapor Pressure, J. Am. Ceram. Soc., 67 [12]
78692 (1984).
32
N. M. Tallan and R. W. Vest, Electrical Properties and Defect Structure
of Y2O3, J. Am. Ceram. Soc., 49 [8] 4014 (1966).
33
A. L. Linsebigler, G. Q. Lu, and J. T. Yates, Photocatalysis on TiO2 Sur-
faces - Principles, Mechanisms, and Selected Results, Chem. Rev., 95 [3] 735
58 (1995).
34
N. Hirose and A. R. West, Impedance Spectroscopy of Undoped BaTiO3
Ceramics, J. Am. Ceram. Soc., 79 [6] 163341 (1996).
35
J. C. MPeko, Dynamics of the Electrical Response of Ceramic Dielectric
Materials in the Presence of Interfacial Blocking Eects, J. Mater. Sci. Lett.,
19 [21] 19257 (2000).
36
N. H. Chan and D. M. Smyth, Nonstoichiometry in SrTiO3, J. Elec-
trochem. Soc., 125 [3] 17629 (1978).
37
R. Kumar, S. K. Arora, I. V. Shvets, N. E. Rajeevan, P. P. Pradyumnan,
and D. K. Shukla, Structural and Transport Properties of Bi-Substituted
Co2MnO4, J. Appl. Phys., 105 [7] 07D910 (2009).
38
L. W. Tai, M. M. Nasrallah, H. U. Anderson, D. M. Sparlin, and S. R.
Sehlin, Structure and Electrical-Properties of La1xSrxCo1yFeyO3. Part 2.
The System La1xSrxCo0.2Fe0.8O3, Solid State Ionics, 76 [34] 27383
(1995).
39
X. L. Zhou, et al., Ionic Conductivity, Sintering and Thermal Expansion
Behaviors of Mixed Ion Conductor BaZr0.1Ce0.7Y0.1Yb0.1O3-d Prepared by
Ethylene Diamine Tetraacetic Acid Assisted Glycine Nitrate Process, J.
Power Sources, 196 [11] 50006 (2011). h