Thermal degradation of ionic liquids at elevated temperatures

Article abstract of DOI:10.1071/CH03221

Ionic liquids based on the imidazolium cation are found to degrade, yielding volatile degradation products, at temperatures significantly lower than previously reported and thus a parameter T_z/x (the maximum operating temperature) is developed to provide a more appropriate estimate of thermal stability.

Full text of DOI:10.1071/CH03221

Rapid Communication
CSIRO PUBLISHING
www.publish.csiro.au/journals/ajc
Aust. J. Chem. 2004, 57, 145–147
Thermal Degradation of Ionic Liquids at Elevated Temperatures
Krisztian J. Baranyai,A Glen B. Deacon,A Douglas R. MacFarlane,A Jennifer M. Pringle,A
and Janet L. ScottA,B
A
Centre for Green Chemistry and School of Chemistry, Monash University, Melbourne VIC 3800, Australia.
B Author to whom correspondence should be addressed (e-mail: janet.scott@sci.monash.edu.au).
Ionic liquids based on the imidazolium cation are found to degrade, yielding volatile degradation products, at
temperatures significantly lower than previously reported and thus a parameter Tz/x (the maximum operating
temperature) is developed to provide a more appropriate estimate of thermal stability.
Manuscript received: 29 August 2003.
Final version: 10 October 2003.
O
S
O
S
O
S
O
S
Ionic liquids, such as those based on the imidazolium cation
1+, Scheme 1, have been widely exploited in applications
ranging from reaction solvents[1] to battery electrolytes.[2]
The utility of this novel class of compounds relies on
their wide liquid range,[3] high thermal and electrochemi-
cal stability,[3–5] and their unusual solvent properties. The
relative stability imparted by different cation/anion combina-
tions, as determined by rising temperature thermogravimetric
analysis (TGA),[6] can produce ionic liquids that are stable
to in excess of 400◦C. This limit of thermal stability has been
studied across a wide range of ionic liquids and reported to
correlate loosely with anion hydrophilicity.[4]
R
N
N
F3C
N
CF3
H3C
N
CH3
O
O
O
O
1ϩ
2Ϫ
3Ϫ
Scheme 1. The methylethylimidazolium (emIm) cation 1+ (R =
CH2CH3,) with bis(trifluoromethanesulfonyl)amide (TFSA) anion 2−
forms a hydrophobic ionic liquid while emIm with bis(methanesulfonyl)
amide 3− forms a water-miscible ionic liquid.
a specific extent at each temperature. The results show that
degradation, significant in the context of applications and
issues mentioned above, takes place at temperatures much
lower than previously thought.
However, while the use of the ‘step tangent method’
for TGA-determined degradation onset temperatures allows
ready comparison and reproducibility of measurements, it
results in a significant overestimation of thermal stability,
as significant mass loss (implying significant evolution of
volatile degradation products) takes place well below this
temperature.[6] From a practical point of view, a more detailed
picture of ionic liquid thermal stability is needed, including
(a) maximum operating temperature, below which degrada-
tion and thus production and evolution of volatile degradants
may be considered to be negligible; (b) how rapidly an
ionic liquid degrades at a specific temperature; and (c) what
decomposition products are likely to result. These issues
are particularly important as ionic liquids are increasingly
employed as potential solvent replacements, a role in which
high-temperature operations, such as distillation of solutes,
may be considered. In this role the ionic liquid must also
be substantially recyclable, to satisfy growing requirements
for environmentally friendly processes, as well as being eco-
nomically competitive, as demanded when the replacement
solvent is many orders of magnitude more expensive than
typical molecular solvents.[7] Both of these requirements are
reliant on minimal degradation of the liquid at its maximum
operating temperature. In the work reported here, a parame-
ter is developed and measured that describes breakdown of
Methodology, based on precise TGA measurements and
developed for the analysis of decomposition of samples
(particularly solids[8]), that undergo an A(s) → B(s) + C(g)
decomposition, allows determination of a point as close to the
‘procedure-independent decomposition temperature’(PIDT)
as possible.[9] Mass loss curves are converted to percent-
age mass loss, α, ꢀversus time, t, curves which, for both ionic
liquids N-ethyl N -methylimidazolium bis(trifluoromethane-
sulfonyl)amide 1 · 2 and N-ethyl Nꢀ-methylimidazolium bis-
(methanesulfonyl)amide 1 · 3, were found to be straight lines
over a wide decomposition range, implying a zeroth-order
rate law. This is as might be expected given the conditions
of the experiment, where volatile degradants are removed
immediately and there is thus no increase in the concentra-
tion of dissolved gaseous degradants during the course of
decomposition. Rate constants, k, derived from the gradient
of this straight line by least-squares regression, are employed
in the determination of the value of t0.99, the time taken for
1% degradation to occur. Each experiment was carried out in
triplicate in both nitrogen and air and values of t0.99 at each
isothermal temperature plotted versus temperature, as illus-
trated in Fig. 1. The transformation temperature, Tz/x, is
defined as the temperature at which the degree of conver-
sion z (=α/100; 0 < z < 1) occurs over a specified amount of
© CSIRO 2004
10.1071/CH03221
0004-9425/04/020145

146
K. J. Baranyai et al.
O
S
O
S
O
S
O
S
60
50
40
30
20
10
0
N
N
N
NH
F3C
N
R
CF3 H3C
N
R
CH3
R
O
O
O
O
R
5
Rϭ CH3
4a
4b Rϭ CH2CH3
6
7
Scheme 2. Major degradants detected using pyrolysis–mass
spectrometry.
1 · 2
1· 3
300
320
340
360
380
400
1 · Br
T (ºC)
Fig. 1. Time taken for 1% degradation of 1 · 2 at a series of isother-
mal temperatures under nitrogen flow. A first-order exponential equa-
tion, of the form t = h exp(−x/j) was fitted. If t is replaced by the
designated time in which 1% decomposition is deemed acceptable
then x is T0.01/x, the temperature at which 1% decomposition occurs
in the designated time.
4a
4b
5a
5b
Fig. 2. Normalized ratios of cation degradation products. Alkyl
migration products 5a and 5b are identified in the fluorinated anion
2− ionic liquid only.
Table 1. Maximum temperatures at which 1% degradation occurs,
T
0.01/x [◦C]
Ionic liquid
x = 1 h
x = 10 hA
pyrolysis–mass spectrometry. The major cation degradants
detected were neutral mono-N-alkylated imidazoles 4 as
described by Chan et al. in a systematic study of thermolysis
of quaternary imidazolium halide salts.[10]
N2
1 · 2
307
199
187
251
167
159
1 · 3
1 · BrB
Air
1 · 2
295
198
184
234
167
158
Pyrolysis of the non-fluorinated anion salt 1 · 3 yields 4a
and 4b, Scheme 2, almost exclusively as cation degradation
products and the alkylated species 7a and 7b as anion degra-
dation products along with a highly volatile species, dimethyl
sulfone, arising from the decomposition of the methylated
anion. Pyrolysisofthefluorinatedanionsalt1 · 2yieldscation
degradation products 4a and 4b as well as some 2-alkylated
products 5, and these are presumed to arise from the thermally
induced rearrangement of 4, described previously.[11] No
alkylated anions 6a and 6b are detected at 300◦C, although
small quantities are noted on raising the pyrolysis tempera-
ture to 700◦C. Clearly, even at 300◦C, the anion must degrade
or form some neutral species, but no volatile acid 2 · H is
detected, and it appears that anion degradation follows an as-
yet undetermined route. While quantification of degradants
is rendered difficult by the lack of baseline separation in
the GC/MS traces, normalized peak areas provide relative
quantities of similar species, Fig. 2.
The predominance of N-ethylimidazole 4b as the major
cation degradant from both 1 · 3 and 1 · Br implies a similar
mechanism of decomposition to the SN2 process postulated
by Chan et al.[10] The change in distribution of degradant
type on anion fluorination may reflect the decrease in anion
nucleophilicity inferred from the acidity of the acids from
which these are derived: pKa (H · 2) = 1.7[12] and pKa
(H · 3) = 2.85.[13] The marked change in the ratio of 4a to
4b, the appearance of 5a and 5b, and the lack of alkylated
anion species 6a and 6b implies that a different mechanism
of decomposition predominates in 1 · 2 and the removal of the
ethyl group, by elimination of ethylene, cannot be discounted
as a significant contributor to degradation. The poor nucleo-
philicity of the bis(trifluoromethanesulfonyl)amide anion
1 · 3
1 · BrB
A Extrapolation to longer times yields only a slight decrease in T0.01
.
B Decomposition of bromide salts is catalyzed by aluminium sample
holders, as reported previously.[5]
time, x, (e.g. T0.01/10 h is the temperature at which α = 1%
conversion occurs over ten hours) and this is determined by
extrapolation (or interpolation) of the first-order exponential
curve determined experimentally, Fig. 1.
The ionic liquids 1 · 2 and 1 · 3 were compared in this
manner and the bromide salt, from which these are syn-
thesized, is included for comparison with previous studies
of decomposition of similar species.[10] T0.01/x values for
each of the ionic liquids are presented in Table 1. These
values represent estimates of maximum operating tempera-
tures for each of the ionic liquids and it is quite clear that
these are significantly lower than degradation temperatures
arrived at by step-tangent analysis of rising temperature
scanning differential thermal analysis (SDTA) experiments.
For example, the decomposition temperature of N-ethyl Nꢀ-
methylimidazolium bis(trifluoromethanesulfonyl)amide 1 · 2
is described as approximately 440[3] or 455◦C,[4] a value
almost 150◦C higher than that calculated to result in 1%
degradation after only 1 h of continuous use.
Both rising temperature and isothermal TGA experiments
(conducted under nitrogen flow) yield close to 100% weight
loss, indicating that, under the conditions of this experi-
ment, complete degradation to volatile products occurs. To
allow identification of the major components of the vapours
evolved, samples of each ionic liquid were subjected to

Thermal Degradation of Ionic Liquids at Elevated Temperatures
147
thus appears to confer far greater stability on the ionic liquid.
Requaternization processes[14] will compete with any anion
mediated dealkylation and thus the boiling points (and rates
of evaporation of alkylated anion species) as well as the rates
of (re)quaternization reactions will be important.
[2] A. Webber, G. E. Blomgren, in Advances in Lithium Ion
Batteries (Eds W. van Schalkwijk, B. Scrosati) 2002, pp.
185–232 (Plenum: New York, NY).
[3] P. Bonhôte, A.-P. Dias, N. Papageogiou, K. Kalyanasundaram,
M. Grätzel, Inorg. Chem. 1996, 35, 1168. doi:10.1021/
IC951325X
Factors such as sample holder composition and the pres-
ence of contaminants may have a significant effect on the rate
and mechanism of decomposition of salts. We find no effect
(within experimental error) on the measured rate of decom-
position of salts 1 · 2 and 1 · 3, when comparing aluminium,
alumina, and platinum pans, although, in line with previously
reported results,[5] the bromide salts are found to decompose
more rapidly when in contact with aluminium.
[4] J. G. Huddlestone, A. E. Visser, W. M. Reichert, H. D. Willauer,
G. A. Broker, R. D. Rogers, Green Chem. 2001, 3, 156.
doi:10.1039/B103275P
[5] H. L. Ngo, K. LeCompste, L. Hargens, A. B. McEwan, Thermo-
chim. Acta 2000, 97. doi:10.1016/S0040-6031(00)00373-7
[6] Z. Zhang, R. G. Reddy, Proc. TMS Annu. Meeting, Seattle,
2002, p. 33.
[7] J. D. Holbrey, W. M. Reichert, R. P. Swatloski, G. A. Broker,
W. R. Pitner, K. R. Seddon, R. D. Rogers, Green Chem. 2002,
4, 407. doi:10.1039/B204469B
In conclusion, application of methodology designed to
analyze decomposition of solids to ionic liquids yields rate
data from which a parameter T0.01/x (for a given time x) may
be estimated.This allows comparison of real operational tem-
perature ranges and decomposition temperatures for different
ionic liquids, and these are shown to be significantly lower
than might be inferred from decomposition temperatures
derived, by step tangent methods, from rising temperature
thermal analysis. Volatile degradants, which might have been
inferred as likely from a suitable retrosynthetic analysis, are
identified and it is of note that these are both potentially reac-
tive and moderately volatile and thus are of great importance
when considering ionic liquids as green replacements for
volatile organic compounds in bulk processes or in applica-
tions requiring extensive recycling and reuse of ionic liquids.
The ability of the anion to participate in dealkylation and of
the cation to undergo alkyl migration or elimination reactions
are major factors in the degradation of ionic liquids and
further studies to expand on this are ongoing.
[8] While it might be argued that an ionic liquid is not a solid
and that this treatment does not distinguish between diffusion-
controlled processes and processes under chemical control,
the approach may be justified as follows: The negligible Pvap
of ionic liquids below the temperature of decomposition and
the relatively rapid diffusion expected in a liquid sample of
low viscosity (ionic liquid viscosity decreases sharply with
increased temperature), coupled with the rapid removal of
degradant vapours by a high flow rate sweep gas, allows very
close approximation of the rate of mass loss with the rate of
degradant evolution. The rate of degradant evolution cannot
be faster than the rate of degradation and overestimation of
maximum operational temperature is minimized by including
low-temperature degradation experiments.
[9] J. Czarnecki, J. Šesták, J. Therm. Anal. Calorim. 2000, 60, 759.
doi:10.1023/A:1010187019979
[10] B. K. M. Chan, N.-H. Chang, M. R. Grimmett, Aust. J. Chem.
1977, 30, 2005.
[11] C. G. Begg, M. R. Grimmett, P. D. Wethey, Aust. J. Chem.
1973, 26, 2435.
[12] J. Foropoulos Jr, Inorg. Chem. 1984, 23, 3720.
[13] V. A. Blaschette, E. Wieland, G. Seurig, D. Koch, Anorg. Allg.
Chem. 1983, 506, 75.
[14] A. J. Jeapes, R. C. Thied, K. R. Seddon, W. R. Pitner,
D. W. Rooney, J. E. Hatter, T. Welton, Int. Patent Appl. WO
0115175, 2001.
References
[1] (a) H. Olivier-Bourbigou, L. Magna, J. Mol. Catal. A: Chem.
2002, 182/183, 419. doi:10.1016/S1381-1169(01)00465-4
(b) P. Wasserscheid, W. Keim, Angew. Chem. Int. Ed. 2000,
39, 3772. doi:10.1002/1521-3773(20001103)39:21<3772::AID-
ANIE3772>3.0.CO;2-5