Thermal stability of solid C60

Article abstract of DOI:10.1103/PhysRevB.55.127

We have used differential scanning calorimetry, optical microscopy, and x-ray diffraction to study the thermal stability of pristine C₆₀ in the solid state. To minimize contamination and/or reaction of the sample with the container, sublimed si

Full text of DOI:10.1103/PhysRevB.55.127

PHYSICAL REVIEW B
VOLUME 55, NUMBER 1
1 JANUARY 1997-I
Thermal stability of solid C60
M. R. Stetzer and P. A. Heiney
Department of Physics and Astronomy and Laboratory for Research on the Structure of Matter,
University of Pennsylvania, Philadelphia, Pennsylvania 19104
J. E. Fischer
Department of Materials Science and Engineering and Laboratory for Research on the Structure of Matter,
University of Pennsylvania, Philadelphia, Pennsylvania 19104
A. R. McGhie
Laboratory for Research on the Structure of Matter, University of Pennsylvania, Philadelphia, Pennsylvania 19104
͑Received 29 August 1996͒
We have used differential scanning calorimetry, optical microscopy, and x-ray diffraction to study the
thermal stability of pristine C60 in the solid state. To minimize contamination and/or reaction of the sample
with the container, sublimed single crystals and powder derived from resublimed C60 material were heated in
hermetically sealed Au pans. Crystals heated at 1260 K for more than 10 min decomposed into amorphous
carbon without losing their faceted morphology. ͓S0163-1829͑97͒02702-1͔
I. INTRODUCTION
quantified these decomposition kinetics by heating a
fullerene mix ͑85% C , 15% C ) sealed in evacuated
60
70
quartz ampoules to several temperatures ranging from 873 to
Much of the recent interest in C , buckminsterfullerene,
can be attributed to the nearly spherical molecular architec-
ture exhibited by this caged carbon compound, an architec-
ture which profoundly influences the solid-state structures
and dynamics of these molecules.1,2 The existence and prop-
60
1
273 K for durations of 5 min to 6 h.16 Under the assumption
of unimolecular decay, LGAA determined an activation en-
ergy of 266Ϯ9 kJ/mol for the thermal decomposition of
C , significantly lower than prior estimates. Moreover,
60
those samples heated to temperatures greater than 1273 K
were insoluble in toluene and displayed no Bragg peaks, thus
indicating that the samples had decomposed into amorphous
carbon. Surprisingly, these samples still retained their exter-
nal morphologies.
erties of a stable liquid phase of C have been the subject of
60
numerous theoretical studies.3 There is considerable theo-
–8
retical disagreement about whether a triple-point and high-
temperature liquid-vapor coexistence are to be expected in
C , and on the value of the putative triple-point tempera-
6
0
We have extended the work of LGAA by using pure
C60 rather than a fullerene mix and by using gold pans rather
than quartz tubes to contain the sample. We thus attempted
to ensure that both the C60 sample and the heating vessel
were free of any fullerene and nonfullerene impurities which
could potentially nucleate premature decomposition. The
presence of C70 in the sample complicates the story, both
because the kinetics of C70 decomposition may not be the
same of those of C , and because C molecules may act as
nucleation sites of molecular or crystalline disruption in the
C60 lattice. Therefore, we used samples consisting of either
high-quality sublimed single crystals of C60 or powders de-
rived from resublimed C60 material in order to minimize both
higher-fullerene and solvent impurities. In addition, hermeti-
cally sealed gold pans were adopted as heating vessels to
eliminate the possibility of the fullerene material reacting
with O2 from its quartz container at the elevated tempera-
tures examined in these studies. Differential scanning calo-
rimetry ͑DSC͒ was used to determine the purity and integrity
of high-quality pristine C60 samples after various thermal
treatments by monitoring the enthalpy change associated
with the orientational ordering transition at ϳ260 K.17
ture. Even if precise calculations indicate the existence of a
true C60 liquid phase, it may be experimentally inaccessible
if the molecules themselves decompose below the triple-
point temperature. Hence, a thorough understanding of the
thermal stability of solid C60 is imperative if one ultimately
wishes to characterize the high-temperature regime of the
C60 phase diagram.
Molecular-dynamics simulations9
,10
have predicted gas-
60
70
phase decomposition of C60 near 5000 K, possibly preceded
by ‘‘melting’’ of the individual molecules. There have been
several experimental studies of the thermal decomposition of
C60 both in the gas phase1
1–14
15,16
and in the solid state,
and
even the gas-phase experiments have shown decomposition
at temperatures well below those predicted theoretically.
1
5
Sundar et al. used x-ray powder diffraction ͑XRPD͒ along
with uv-visible absorption spectroscopy and resistivity ex-
periments in order to show that solid C60 begins to decom-
pose into amorphous carbon when heated for 24 h at tem-
peratures in excess of 973 K. In these thermal studies, C60
powder pellets of greater than 99% purity were heated for 24
h in evacuated quartz ampoules. Although x-ray patterns of
samples heated to 973 and 1073 K exhibited a few broad
Bragg reflections superimposed on an amorphous back-
ground, the x-ray pattern of a sample heated to 1173 K re-
vealed only diffuse features characteristic of amorphous car-
bon. Leifer, Goodwin, Anderson, and Anderson ͑LGAA͒
II. EXPERIMENTAL PROCEDURES
Three samples were studied. The first two were sublimed
single crystals of C , and the third was a C powder.
60
60
0
163-1829/97/55͑1͒/127͑5͒/$10.00
55
127
© 1997 The American Physical Society

128
STETZER, HEINEY, FISCHER, AND McGHIE
55
Samples 1 and 2 had masses of 11.2 and 4.1 mg, respec-
tively, each sample consisting of several small ͑volume Ͻ 2
3
mm ) single crystals. Sample 3 consisted of a powder ob-
tained by mortaring resublimed C60 material, thus allowing
for x-ray powder diffraction ͑XRPD͒ both before and after
heat treatment. The mortaring of the C60 was performed un-
der an inert argon atmosphere to minimize contamination by
impurities such as O . 10.2 mg of this powder were sub-
2
jected to the heat treatments described below, and the re-
maining powder was used for an XRPD study of the material
before heat treatment. In all cases, the samples were loaded
under an inert argon atmosphere into hermetically sealed
gold DSC pans of greater than 99.95% purity manufactured
by TA Instruments.
For the XRPD studies, the C60 powder was transferred
into 0.5 mm glass capillaries which were then sealed under
vacuum. An incident beam of wavelength 1.5406 Å was col-
limated by a Gr͑002͒ monochromator, and the diffracted
beam was measured using an INEL position-sensitive detec-
tor; each diffraction profile was collected during 3600 s of
detector live time. Single-crystal x-ray diffraction ͑XRD͒
measurements used an incident beam of wavelength 0.7107
Å collimated by a Gr͑004͒ monochromator. The diffracted
beam was collimated by Soller slits and detected with a NaI
scintillation counter.
In order to monitor the decomposition of C60 as a function
of successive heat treatments in a nearly in situ fashion,
sample integrity was studied via DSC using a TA Instru-
ments DSC 2920. The sample remained in the same hermeti-
cally sealed container at all times, thus minimizing handling
and contamination. We performed an initial DSC scan on
each sample prior to heat treatment in order to establish a
baseline enthalpy change and onset temperature for the ori-
entational ordering transition. After placing the sample into a
Seiko Instruments TG/DTA 320 under a flow of high-purity
argon, the sample temperature was then ramped at 20 K/min
from room temperature to the elevated temperature of inter-
est, and the sample was held at this annealing temperature
for a specified time ͑usually 10 min͒. Upon completion of the
anneal, the sample was allowed to cool naturally to room
temperature before another DSC measurement was per-
formed.
FIG. 1. Differential scanning calorimetry data collected upon
heating for sample 1, the first of two single-crystal C60 samples
studied. ͑a͒ shows the ordering transition endotherm exhibited by
the material prior to heat treatment, while ͑b͒–͑f͒ show the
endotherm after sequential anneals. The temperature and time of
anneal are given for each scan, and the curves are offset for clarity.
Significant sample decomposition was observed for annealing tem-
peratures of 1235 K and above, with this decomposition manifest-
ing itself in a lowering of the ordering transition’s onset tempera-
ture and a decrease in the enthalpy change associated with the
endotherm. Values of the onset temperature and enthalpy change
for each scan are presented in Table I.
the decomposition mechanism appears to have destroyed
both the material’s crystalline order and the C60 molecules
themselves, suggesting that the enthalpy change at the tran-
sition provides at least a qualitative indication of the amount
of remaining crystalline C . Our procedure therefore was to
Any significant decreases in both the onset temperature
and the enthalpy change of the endotherm associated with
the sample’s ordering transition were attributed to the de-
composition of the crystalline C60 material. It is possible that
the decomposition mechanism is such that large regions of
the sample decompose completely while others remain in-
tact. In this case, the integrated enthalpy change through the
transition should be directly proportional to the amount of
remaining crystalline C . Alternatively, individual mol-
60
perform multiple cycles of high-temperature annealing fol-
lowed by calorimetry through the ϳ260 K ordering transi-
tion.
III. RESULTS
For sample 1 ͑single crystals͒, a preliminary DSC run
indicated an enthalpy change of 11.2 J/g and an onset tem-
perature of 261.81 K, both consistent with the values ex-
pected for high-purity single crystals. This baseline scan on
the first sample as well as scans obtained after various heat
treatments at temperatures ranging from 1086 to 1235 K are
presented in Fig. 1; our results for this sample are summa-
rized in Table I. Relatively little change in the ordering tran-
sition’s endotherm was observed after thermal treatments at
1086 and 1185 K. After 10 min at 1235 K, the sample had
decomposed significantly, as manifested in both the lowering
of the onset temperature to 258.70 K and the 14.5% reduc-
60
ecules destroyed by the heat treatment may permeate the
sample uniformly, functioning as point defects. In this case,
the enthalpy change would most likely not be linearly pro-
portional to the number of intact C60 molecules, although one
would expect a monotonic relationship. In our measure-
ments, the crystallinity of the sample was independently
monitored via Bragg diffraction, and the existence of C60
molecules was tested via solubility in toluene. Since our
samples failed to exhibit either strong Bragg reflections or
solubility in toluene after their endotherms had disappeared,

55
129
THERMAL STABILITY OF SOLID C60
TABLE I. Summary of phase transition onset temperatures and
associated enthalpy changes obtained from DSC data for sample 1
after various thermal treatments. Results are listed in the order in
which the anneals were performed.
Tanneal ͑K͒
tanneal ͑min͒
Tonset ͑K͒
⌬H ͑J/g͒
Untreated
261.81Ϯ0.02
261.91Ϯ0.02
261.80Ϯ0.02
258.70Ϯ0.02
258.68Ϯ0.02
213 Ϯ2
11.2Ϯ0.4
10.8Ϯ0.4
10.3Ϯ0.4
9.6Ϯ0.4
9.4Ϯ0.4
1.2Ϯ0.2
1
1
1
1
1
086
185
235
084
235
10
10
10
10
20
tion of the enthalpy change to a value of 9.6 J/g. Annealing
the sample at a lower temperature of 1084 K failed to im-
prove the integrity and crystallinity of the sample. After 20
additional min at 1235 K, the onset temperature was lowered
to 213 K and the associated enthalpy change was reduced to
1.2 J/g. If, as discussed in the previous section, we associate
the enthalpy change with the quantity of remaining crystal-
line C , this is consistent with approximately 11% of the
6
0
sample remaining as crystalline C . At this point, the re-
60
maining sample was placed in toluene, which is known to be
a good solvent for C . The majority of the material did not
dissolve, but the appearance of a faint purple tint indicated
that some did, consistent with the hypothesis that a small
FIG. 2. Differential scanning calorimetry data collected upon
heating for sample 3, a powder derived from resublimed C60 mate-
rial. ͑a͒ shows the ordering transition endotherm exhibited by the
material prior to heat treatment, while ͑b͒–͑g͒ show the endotherm
after sequential anneals. The temperature and time of anneal are
given for each scan, and the curves are offset for clarity. The broad
precursor to the untreated sample’s endotherm, shown in ͑a͒, was
attributed to stacking faults introduced into the material as a result
of mortaring. Significant sample decomposition was observed for
annealing temperatures of 1232 K and above. Values of the onset
temperature and enthalpy change for each scan are presented in
Table II.
60
fraction of the sample remained as unmodified C . The in-
60
tegrity of our measurements on sample 1 was somewhat
compromised by the subsequent discovery of small translu-
cent lumps of material among the decomposed C60 crystals,
provisionally identified as arising from small glass shards
inadvertently loaded with the sample.
Sample 2, also consisting of single crystallites, was pre-
pared in a similar manner. An initial enthalpy change of 9.0
J/g, a measured onset temperature of ϳ255.52 K, and a split-
ting in the major endotherm indicated that this sample was
not quite as pure and defect free as sample 1. In addition, a
very small endotherm with an onset temperature of approxi-
mately 280 K indicated the presence of a small amount of
was observed after heat treatment at both 1212 and 1221 K.
After 10 min at 1232 K, the enthalpy change through the
transition was reduced to 4.5 J/g and the onset temperature
was dramatically lowered to 239.3 K. This enthalpy change
was further reduced to 1.3 J/g, approximately 16% of its
original value, after 10 min at 1240 K. The DSC scan per-
formed on the sample after a final heat treatment at 1250 K
18
C O in the sample. The sample was heated for 10 min at
6
0
1034, 1137, 1187, and 1237 K with no noticeable degrada-
tion. After 10 min at 1262 K, the enthalpy change associated
with the endotherm decreased from 8.9 to 7.3 J/g, this de-
crease being accompanied by a broadening of the endotherm.
After an additional 10 min at 1261 K, the major endotherm
disappeared entirely. At this point, material from this sample
was insoluble in toluene, consistent with the hypothesis that
all of the crystalline C60 had decomposed as a result of the
heat treatment.
TABLE II. Summary of phase transition onset temperatures and
associated enthalpy changes obtained from DSC data for sample 3
after various thermal treatments. Results are listed in the order in
which the anneals were performed.
Measurements on sample 3, the resublimed C60 powder,
yielded results similar to those obtained from the single-
crystal samples; these results are presented in Fig. 2 and
Table II. Before heat treatment, the endotherm had an en-
thalpy change of approximately 8.1 J/g, with an onset tem-
perature of 257.31 K. The relatively broad precursor below
the endotherm can most likely be attributed to the presence
of stacking faults in the sample resulting from the mortaring
Tanneal ͑K͒
tanneal
Tonset ͑K͒
⌬H ͑J/g͒
Untreated
883
1212
1221
1232
257.31Ϯ0.02
257.70Ϯ0.03
256.51Ϯ0.01
253.78Ϯ0.03
239.3Ϯ0.4
207Ϯ2
8.1Ϯ0.4
8.0Ϯ0.4
7.2Ϯ0.3
6.3Ϯ0.5
4.5Ϯ0.6
1.3Ϯ0.2
0.14Ϯ0.09
4 h
10 min
10 min
10 min
10 min
10 min
1240
1250
1
9
procedure. Annealing at 883 K for 4 h had no noticeable
effect on the endotherm. Minor degradation of the endotherm
214Ϯ2

130
STETZER, HEINEY, FISCHER, AND McGHIE
55
FIG. 4. X-ray powder-diffraction patterns for sample 3, a C60
powder, collected ͑a͒ prior to heat treatment and ͑b͒ after annealing
at 1250 K. Both patterns were counted for 3600 s of detector live
time. Note the change in vertical scale between the top and bottom
curves. Pattern ͑b͒ demonstrates the primarily amorphous nature of
the treated material, although the weak reflection at Qϭ0.752
Ϫ1
Å
may correspond to the 111 reflection in a dilated C60 lattice,
indicating that a small amount of crystalline C60 survived the an-
nealing procedure.
FIG. 3. A single grain of C60 from sample 2 after annealing at
261 K. Single-crystal XRD on this grain revealed only extremely
1
weak 111 and 200 reflections superimposed upon a significant
amount of diffuse scattering, and subsequent XRPD resulted in an
entirely amorphous spectrum. However, the slightly pitted surface
of the thermally treated grain retained its original faceted morphol-
ogy even after the decomposition of crystalline C60 , suggesting that
the decomposition took place in the solid state.
der, thus implying that, at most, a very small fraction of the
thermally treated sample was crystalline C . XRPD patterns
60
from sample 3, the powdered C60 sample, both before heat
treatment and after heat treatment up to 1250 K are shown in
Fig. 4. The rather large stacking-fault foot below the crystal-
line 111 reflection in the XRPD spectrum collected before
heat treatment ͓Fig. 4͑a͔͒ resulted from the mortaring of the
sample, and is consistent with the broad precursor observed
below the main endotherm in the sample’s preliminary DSC
revealed only an extremely weak endotherm. Given the error
bars of Ϯ0.09 J/g associated with the endotherm’s integrated
enthalpy change of 0.14 J/g, this result is consistent with the
complete decomposition of the sample.
19
scan. The XRPD profile collected after heat treatment at
1
250 K reveals a broad amorphous pattern, although one
In all of our work on both single crystals and powders,
significant decomposition of the C60 samples began when
Tу1232 K, and complete decomposition on the 10-min time
scale took place in the 1250–1262 K temperature range.
These results are qualitatively consistent with LGAA’s ob-
servation that fullerene samples heated to temperatures in
excess of 1273 K fully decomposed into amorphous
Ϫ1
very weak reflection appears to exist at Qϭ0.752 Å ͓Fig.
͑b͔͒. This feature is at too small an angle to be identified as
4
Ϫ1
the 111 reflection from pristine C60 ͑Qϭ0.772 Å ), but
may correspond to the 111 reflection after lattice dilation
resulting from heat treatment, an effect that Sundar et al.
observed in their studies of C60 pellets.15
1
6
carbon.
Both sample 1 and sample 2 were still faceted after heat
treatment ͑Fig. 3͒, although the faces of the samples were
somewhat pitted. This indicates that, although the C60 vapor
was certainly quite high at the treatment temperatures, most
of the sample remained in the solid state throughout the an-
nealing process, and that C60 decomposition took place in the
solid state. Single-crystal XRD performed on a crystal from
sample 2 revealed the presence of extremely weak 111 and
IV. CONCLUSIONS
We have examined the thermal stability of high-purity
C60 in the solid state using DSC and XRPD. On the time
scales used in these experiments, ‘‘total’’ decomposition of
the sample was observed to occur at temperatures between
1250 and 1262 K, manifesting itself through the disappear-
ance of the characteristic C60 ordering transition and its as-
sociated endotherm. Qualitatively, our results are similar to
those obtained by LGAA for fullerene mixes, thus ruling out
200 reflections as well as a significant amount of diffuse
scattering. An entirely amorphous pattern, however, was ob-
tained from XRPD upon crushing this same crystal into pow-

55
131
THERMAL STABILITY OF SOLID C60
the possibility that premature decomposition may have been
induced in the latter experiment as a result of C70 or O2
impurities. In light of the low activation energy suggested by
these studies for the solid-state decomposition of pristine
C , it is improbable that a liquid phase of C will ever be
ACKNOWLEDGMENTS
We thank M. Haluska and H. Kuzmany for providing the
samples used in these studies. This work was supported in
part by NSF Grant No. DMR-MRL-92-20668; acknowledg-
ment is also made to the Donors of the Petroleum Research
Fund, administered by the American Chemical Society, for
partial support of this research.
6
0
60
observed since the solid will necessarily decompose at tem-
8
peratures well below the liquid regime predicted to exist
between 1600 and 1920 K.
1
12
P. A. Heiney, J. Phys. Chem. Solids 54, 1333 ͑1992͒.
J. E. Fischer and P. A. Heiney, J. Phys. Chem. Solids 54, 1725
A. M. Vassallo, L. S. K. Pang, P. A. Cole-Clarke, and M. A.
Wilson, J. Am. Chem. Soc. 113, 7820 ͑1991͒.
E. Kolodney, B. Tsipinyuk, and A. Budrevich, J. Chem. Phys.
100, 8542 ͑1994͒.
S. von Gersum, T. Kruse, and P. Roth, Ber. Bunsenges. Phys.
Chem. 98, 979 ͑1994͒.
C. S. Sundar, A. Bharathi, Y. Hariharan, J. Janaki, V. S. Sastry,
and T. S. Radhakrishnan, Solid State Commun. 84, 823 ͑1992͒.
S. D. Leifer, D. G. Goodwin, M. S. Anderson, and J. R. Ander-
son, Phys. Rev. B 51, 9973 ͑1995͒.
P. A. Heiney, J. E. Fischer, A. R. McGhie, W. J. Romanow, A.
M. Denenstein, J. P. McCauley, Jr., A. B. Smith III, and D. E.
Cox, Phys. Rev. Lett. 66, 2911 ͑1991͒.
G. B. M. Vaughan, P. A. Heiney, D. E. Cox, A. R. McGhie, D. R.
Jones, R. M. Strongin, M. A. Cichy, and A. B. Smith III, Chem.
Phys. 168, 185 ͑1992͒.
2
1
1
3
4
͑
1993͒.
3
4
N. W. Ashcroft, Europhys. Lett. 16, 355 ͑1991͒.
A. Cheng, M. L. Klein, and C. Caccamo, Phys. Rev. Lett. 71,
1200 ͑1993͒.
5
M. H. J. Hagen, E. J. Meijer, G. C. A. M. Mooij, D. Frenkel, and
H. N. W. Lekkerkerker, Nature ͑London͒ 365, 425 ͑1993͒.
L. Mederos and G. Navascu e´ s, Phys. Rev. B 50, 1301 ͑1994͒.
V. I. Zubov, N. P. Tretiakov, J. N. Teixeira Rabelo, and J. F.
Sanchez Ortiz, Phys. Lett. A 194, 223 ͑1994͒.
C. Caccamo, Phys. Rev. B 51, 3387 ͑1995͒.
B. L. Zhang, C. Z. Wang, C. T. Chan, and K. M. Ho, Phys. Rev.
B 48, 11 381 ͑1993͒.
15
16
17
6
7
8
9
1
1
8
9
1
1
0
1
S. Serra, S. Sanguinetti, and L. Colombo, J. Chem. Phys. 102,
2151 ͑1995͒.
C. I. Frum, R. Engleman Jr., H. G. Hedderich, P. F. Bernath, L.
D. Lamb, and D. R. Huffman, Chem. Phys. Lett. 176, 504
G. B. M. Vaughan, Y. Chabre, and D. Dubois, Europhys. Lett. 31,
525 ͑1995͒.
͑
1991͒.