Proximate analysis by automated

thermogravimetry

John P. Elder
institute for Mining & Minerals Research, Kentucky Center for Energy Research
Laboratory, PO Box 13015, Lexington, KY 40512 USA
(Received 17 August 1981; revised 7 December 1982)

A study has been made of the use of the Perkier-Elmer thermogravimetric instrument TGS-2, under the
control of the System 4 microprocessor for the automatic proximate analysis of solid fossil fuels and
related matter. The programs developed are simple to operate, and do not require detailed temperature
calibration of the instrumental system. They have been tested with coals of varying rank, biomass
samples and Devonian oil shales all of which were of special importance to the State of Kentucky.
Precise, accurate data conforming to ASTM specifications were obtained. The simplicity of the
technique suggests that it may complement the classical ASTM method and could be used when this
latter procedure cannot be employed. However, its adoption as a standardized method must await the
development of statistical data resulting from interlaboratory testing on a variety of fossil fuels.

(Keywords: coal; oil; biomass; instrumental methods of analysis)

Fossil fuel proximate analysis; the determination of

moisture, volatile matter, fixed carbon and non-
combustible, inorganic material content by pyrolysis and
combustion, comprises several methods, governed by
prescribed procedures ‘. Although a number of samples
can be analysed concurrently according to the several
methods the process is tedious and time consuming.
Classical proximate analysis does not lend itself to the
rapid determination of one sample. Furthermore,
required sample sizes, % log, are relatively large.
Thermogravimetric analysis (TGA) is a logical
alternative, as originally pointed out by Fyans’.
Relatively simple t.g.a. procedures have been devised for
various commercial instruments2p3. Recent developments
in microprocessor control of multi-stage furnace heating
and environment, in conjunction with an electronic
microbalance, enables the continuous monitoring of
sample weight as a function of either time oor temperature
in a sequence of heating steps. Programs were developed by
Elder4 for use with Perk&Elmer thermogravimetric
equipment for the proximate analysis of high fved carbon
substances, such as coal, high volatile matter content
materials, such as biomass, and fossil fuels diluted with
excess inert and thermally sensitive inorganic material,
namely oil shale.
Continuining studies have emphasized the accuracy
and precision obtainable with this automated procedure,
and have verified that elaborate temperature
programming of the TGA instrument is not required. The
increasing interest in automated procedures for
proximate analysis is exemplified by the recent
applications study by Earnest and Fyans5, and the
initiation of a new ASTM method by the E37 Thermal
Measurements committee. It is the purpose of this Paper
to share experiences gained in developing the automated
procedure, to help those who wish to use the thermogra-
vimetric technique.
~1~2361/83/05058~5%3.~
@ 1983 Butterworth& Co. (Publishers)Ltd
580 FUEL, 1983, Vol62, May
EXPERIMENTAL
The Perkin-Elmer TGS-2 thermogravimetric apparatus
under the control of the System 4 microprocessor was
used. Weight change and temperature/time data were
monitored on the microprocessor-operated X-Y,- Y,
recorder. Pure, dry nitrogen was passed through the
balance housing and Pyrex furnace tube continuously at a
rate of 100 ml min- *. Nitrogen or oxygen impinged on the
suspended platinum crucible containing the sample via a
side arm in the furnace tube at a rate of 30 ml min- ‘. The
microprocessor automatically switches the gas inlet valve
from nitrogen to oxygen at the appropriate time to initiate
combustion. The gas tank low-pressure regulators were
set such that the switch-over was smooth, and the balance
was not disturbed.
The experimental four-step control program is recalled
from permanent storage in memory for active use in the
program mode. The sample, x 5-10 mg, is placed in the
previously tared crucible, the furnace tube raised to close
the system, the weight per cent control set at 100% full
scale deflection on the recorder chart, and the
microprocessor start button is depressed. This unit
controls all the automatic functions of the recorder, as
well as temperature programming the furnace.
At the conclusion of the run, the pen is raised and
returns to zero, and the program temperature returns to
the sample Ioad temperature, nominally room
temperature, in preparation for the next run. In
accordance with ASTM specifications, all samples were
ground to -60 mesh (US standard sieve <250pm).
System operation
Although originally the internal temperature
calibration feature of the System 4 microprocessor was
employed with the TGS-2 apparatus for proximate
analysis4, a more recent study6 has shown that this so-
 Proximate analysis by automated thermogravimetry: J. P. Elder

T&e 1 Proximate analysis thermogravimetric programs

Heating time T/C indicated Iso. period Elapsed time ASTM

Step (min) iso. temp. (“C) (min) Atmosphere (min) standard

1 - -30 5 Nitrogen 5
2 0.5 110 4.5 (9.5)= Nitrogen 10 (15jb D3173
3 5 990 (950) 5 Nitrogen 20 (25) D3175
4 - 990 (950) 10 Oxygen 30 (35) D3174

Program 1 Program 2 Program 3

Zero 0 0 -20
Range 0 0 0
Rate parameter 20 20 41.6-0.03037
Upper limit (“C) 900 950 900

Step 2 Step 3 Step 2 step 3 Step 2 Step 3

qso (“Cl 30.0 102.5 30.0 102.5 50.0 121.5

p (“C min-‘1 145.0 161.6 145.0 173.0 143.0 162.5
qso (“Cl 102.5 911.5 102.5 967.5 121.5 934.0

a High (>lO%l moisture content material

b Gas inlet valve time

called calibration is not necessary. The instrument may be
operated normally for precise, accurate analyses. In
developing a suitable thermogravimetric proximate
analysis program, the sample environment temperature
should agree with that prescribed in the classical
procedures’. The environmental temperature is
monitored by a chromel-alumel thermocouple located
immediately below the crucible. It has been shown6 that
under isothermal conditions, between a program
temperature, 7; from 200 to 950°C the thermocouple
indicated temperature is given by an expression of the
form of equation (1).
T&o = K, + K,.T- K,. T2 (1)
The coefficients, which should be determined for each
individual furnace, are of the order, K, = 1.03 to 1.04,
Ks=0.5t00.8.10-4andK,=Z+K4,whereK4=9t011
and Z is the chosen setting of the analog zero control on
the heater control module of the TGS-2. For T < 200°C
the TClso value has to be determined by actual
measurement.
Under dynamic conditions, the thermocouple
indicated temperature does not mirror the actual sample
temperature, especially at high heating rates and high
temperatures. It has been shown6, by using magnetic
transition standards for in-situ temperature calibrations,
that the sample and program temperature may be made to
agree within +2”C over a wide range of temperatures and
heating rates. This is carried out by manually varying the
rate parameter’ via the microprocessor keyboard during
the analysis.
Thermogravimetric proximate analysis programs
Table 2 lists the sequences for three different programs
used in this study. Each step is defined in terms of a fixed
time interval. The program temperature levels, and
therefore the heating rates are governed by the furnace
characteristics as described above. In step 1, the system is
held at approximately 30°C for 5 min, which allows the
sample to lose any adsorbed moisture while the furnace
tube is purged of all air. In step 2, the sample is heated to
110°C in 30 s and held isothermally for a further 4.5 min,
to effect the removal of absorbed water.
For high moisture content (2 10%) materials, for
example, biomass, a 9.5min hold time is recommended.
The sample is then heated over a 5min period to either
900°C or 95O”C, and held isothermally at this upper limit
for a further 5min to effect the removal of all volatile
matter. Although the ASTM standard D3174i for the
determination of residual ash strictly calls for combustion
at 750°C it was found unnecessary to cool the system
from the upper limit to this temperature before initiating
combustion. Oxygen is admitted at the upper limit after
an elapsed time of 20 (or 25) min for a further lOmin,
which time has been found quite adequate for the
complete bum-off of all fixed carbon material.
RESULTS AND DISCUSSION
Figure 1 shows the percentage weight loss and
temperature profiles as a function of time for the 30 min
proximate analysis, program 1 on samples of (a) pure
graphite, SP-3, (Union Carbide Corp., Carbon Products
Division, NY 10017); (b) a standard medium volatile
bituminous coal, AR216, (Alpha Resources, Inc., Xenia,
OH 45385). The solid temperature profile refers to the
program temperature, T, and the dashed line is the
thermocouple indicated temperature for the graphite
sample. In this case, there is no volatile matter in the
region of the crucible and underlying thermocouple. On
admitting oxygen into the system, there is a steady linear
weight loss.
The case for bituminous coal is different. Before
admitting oxygen through the furnace tube side-arm,
there is a heavy atmosphere of volatile matter in the
crucible-thermocouple environment. Because of the
relatively poor heat transfer characteristics of the vapour,
the environmental temperature, as monitored by the
thermocouple, increases, as shown by the dotted line. As
the vapour slowly dissipates, the environmental
temperature falls. On introducing oxygen, combustion of
the fixed carbon commences and is complete after
* 7 min.
Figure 2 shows the coal weight-loss profile as a function
of program temperature for programs 1,2 and 3 (see Table
I). Although the profile during step 3, loss of volatile

FUEL, 1983, Vol 62, May 581

Proximate analysis by automated thermogravimetry: J. P. Elder

upper limit. Since, in many cases, certain small weight

fluctuations are evident during this interval, it is
90 4900) considered best to record data as in Figure 3, where the
pen position on the chart is clearly delineated just before
the oxidation of the fixed carbon.
A certain amount of heat is generated during the
combustion as shown by the indicated temperatures of
913°C and 956°C respectively. At the conclusion of
combustion, after an elapsed time of x27 min, the
indicated temperature drops to the programmed levels,
900°C and 950°C respectively. Table 2 summarizes the

90 -

t 70-
I
0 5 10 15 20 25 30 E
2 6Q-
- Time (m) -
8
Figure 1 Weight loss and temperature-time profiles a, Standard
; 5Q-
medium volatile bituminous coal, AR216; b, pure graphite, SP-3
$

3 4Q-

I
30-

1 I I I
015 06s
It I
0 250 500 750 ~“~~olOOO

- TX Indicated Temperature (OC) -

Figure 3 Weight loss as a function of sample environment

temperature, programs 1, 2 and 3, for a standard medium volatile
bituminous coal, AR21 6

Table 2 Thermogravimetric proximate analysis of coal and oil

shale

Standard mvb coal, AR216a (I 1 analvses)

20 Moisture-
free ASTM
(%I Mean B valu& valuea
10 - --

0
I I
250
I
500
I
750
034

P,lk!1.6
Moisture
Volatile matter
0.5
24.9
0.13
0.41 25.1 25.1
Fixed carbon 66.8 0.36 67.2 67.4
- Temperature (“0 - Residual ash 7.8 0.22 7.8 7.5

Figure 2 Weight loss as a function of program temperature for a

Sunburv Shale, East Kentucky (I 1 analysesId
standard medium volatile bituminous coal, AR216 according to
program 1, ----; 2, ______;
a,-,d3,
(%I Mean B
-

matter, depends upon both the heating rate and Moisture 0.9 0.23
instrument control mode, the final weight loss value is Volatile matter 15.6 0.29
constant at 74.8%. Figure 3 shows similar weight loss Fixed carbon 7.1 0.13
profiles for the same coal, subjected to the three programs, Residual ash 76.4 0.24

recorded as a function of the thermocouple indicated a Alpha Resources, Inc.

temperature. As can be seen, the profile is independent of b Results within intra-laboratory ASTM specifications
the program mode. The temperature overshoot and decay c Results within inter-laboratory ASTM specifications
are clearly seen during the 5 min isothermal hold at the d Sufficient sample not available to determine ASTM values

582 FUEL, 1983, Vol62, May

 Proximate analysis by automated thermogravimetry: J. P. Elder

Table 3 Coal proximate analysis: thermogravimetry versus classical ASTfv? method

TGA values Mean moisture-free values

Sample Moisture Volatile Fixed Ash Volatile Fixed Ash

I.D. (%) matter carbon (%I matter carbon (%)
(%) (%) 1%) (%)
-
a b a b a b

Ml901 2.3 2.4 36.7 36.9 50.2 49.8 10.8 10.9 37.7 38.6 51.3 50.1 11.1 11.3
Ml902 2.6 2.3 31.9 31.7 43.7 43.5 21.8 22.5 32.6 32.3 44.7 44.7 22.7 23.1
Ml903 3.2 3.0 36.6 36.5 54.2 54.6 6.0 5.9 37.7 39.0 56.2 54.9 6.1 6.1
Ml904 2.7 2.8 34.5 34.7 49.7 49.5 13.1 13.0 35.6 35.9 51.0 50.7 13.4 13.4
M2091 3.4 3.2 35.5 36.0 51.3 51.9 9.8 8.9 37.0 38.1 53.4 52.7 9.7 9.3
tvl2092 4.2 4.2 37.1 37.6 53.5 53.0 5.2 5.2 39.0 40.4 55.6 54.4 5.4 5.2
M2093 3.5 3.3 32.1 32.8 47.3 47.0 17.1 16.9 33.6 34.0 48.8 48.6 17.6 17.4
M2094 5.3 5.3 37.7 37.9 50.5 50.3 6.5 6.5 39.9 41.2 53.2 51.7 6.9 7.1
M2995 3.7 3.7 33.3 33.5 37.7 38.0 25.3 24.8 34.7 34.7 39.4 39.3 26.1 26.0

a T.g.a. values
b Classical ASTM values

Table 4 Biomass proximate analysis by thermogravimetrv weight loss of approximately 20% of sample weight, the
weight suppression feature of the TGS-2 is employed,
Sample Moisture Volatile Fixed enabling the use of more sensitive weight ranges. Since
I.D. (%) matter (%) carbon (%) Ash (%)
certain minerals undergo thermal decomposition, it is
Ml008 9.9 10.0 76.0 76.6 12.7 11.9 1.4 1.5 useful to record concurrently the first time derivative of
Ml012 5.6 5.5 76.2 76.2 16.7 17.0 1.5 1.3 the weight loss during the analysis. Figure 4 shows such
Ml017 6.0 6.1 74.9 74.4 17.3 17.8 1.8 1.7 TGA and DTG curves for two oil shales, a Sunbury shale
Ml021 5.0 4.9 75.2 75.8 18.1 17.6 1.7 1.7 from East Kentucky and a standard USGS black
Devonian shale, SDO-1, from Southern Indiana,
subjected to program 1. Approximately 50% of the sample
weight was suppressed, and 100% full scale chart
deflection corresponds to 5mg. The Sunbury shale
results of 11 analyses using the three different programs,
two different modes of temperature axis recording, and exhibits one weight loss process, with a maximum rate of
two different furnaces and thermocouples. The standard weight loss at 494°C ascribed to loss of volatile organic
deviations from the means lie within ASTM specifications matter.
Eleven analyses of this shale gave consistent results, as
for intra-laboratory comparisons, and on a moisture-free
shown in Table 2. Again, the precision obtained is within
basis agree within specifications for inter-laboratory
comparisons with the classical ASTM data. (Although the
size of the samples used in the present procedure is quite
small, the excellent agreement with the ASTM results
suggests that sampling errors or the effects of possible
sample inhomogeneities are not serious.)
Since, in step 3, program 1 and 3, the environmental
temperature rises to ~950°C before decaying slowly to
9OO”C,it is not considered necessary to operate the system
according to program 2, where the environmental
temperature rises to 990°C. Therefore, it is recommended
that proximate analyses be run according to program 1
with the data recorded as a function of the thermocouple
indicated temperature. In Table 3 the results of duplicate
measurements on nine differently ranked coals, analysed
according to program 1 are shown. As can be seen, the
moisture-free values are in excellent agreement with those
obtained by classical methods.
It should be emphasized that step 1 is very important.
80
Surface moisture can ‘account for up to 3% of the total
content of some materials, particularly biomass, which
can contain up to 10% moisture. Table 4 lists some
duplicate results on four maize samples analysed according
to the 35min program 1. With this material it was 60 I
observed that, during the 5 min isothermal hold in step 3, .94 s*9 751
t t t t
occasionally weight fluctuations occurred. By recording 1 I II I I I I J

the weight loss data as a function of the indicated rather 0 250 500 750 1000

than the program temperature, potential difficulties in - Temperature (OC) -

identifying the onset of the fixed carbon combustion were Figure 4 Weight loss and derivative weight loss profiles for A,
obviated. Sunbury Shale (SUN-001); 6, US Geological Survey Standard
In analysing mineral-rich oil shales, exhibiting a total Black Devonian Shale (SDO-1) under program 1 operation

FUEL, 1983, Vol62, May 583

Proximate analysis by automated thermogravimetry: J. P. Elder
ASTM specifications. However, for the black shale,
SDG-1, three contiguous processes are seen with
maximum rates of weight loss at 494°C 569°C and 731°C.
From a magnetic transition standards calibration6, it is
estimated that the actual sample temperature
corresponding to these program temperatures are 531°C
588°C and 756”C, respectively. A DSC scan shows an
endothermic shift in the range, 55&59O”C for both shales.
The endotherm is more intense and clearly defined in the
case of the black shale, with the endotherm peak at 560°C.
The high sulphur content, 5.2x, of this particular shale
could mean a high pyritic level in the predominantly clay
mineral portion. If illite is present, its dehydroxylation
(5~550°C’) occurs during the release of volatile matter
from the kerogen. The second weight loss peak in the
weight loss curve of the black Devonian shale could
signify kaolinite dehydroxylation’, pyrite-pyrrhotite
transformation8’g or both. The third, low intensity weight
loss process could be due to the dehydroxylation
montmorillonite*.
SUMMARY
Automatic proximate analysis by thermogravimetry is a
useful analytical procedure, which can be used as a viable
alternative to classical methods in certain circumstances.
Since it requires small sample sizes in the l-10 mg range, it
is extremely useful when sample availability is at a
premium. It has utility when immediate results on one
sample are required. However, since it does not offer itself
to multiple sample handling, it should be considered as
complementing the classical ASTM procedure. Using the
Perkin-Elmer TGS-2 apparatus under the control of the
System 4 microprocessor, the procedure is simple, rapid
and yields precise, accurate values.
The instrument system may be operatred without the
necessity for a detailed or complicated temperature
calibration. No special precautions need be followed with
coals. For high moisture content materials, such as
584 FUEL, 1983, Vol82, May
biomass, it is advisable to use the 35 min elapsed time
program. Weight loss data, as a rule, should be monitored
as a function of the thermocouple indicated temperature.
Because of the possible interference due to mineral matter
thermal decomposition, care should be taken in the
interpretation of the data on oil shales. Simultaneous
monitoring of the derivative weight loss curve is
recommended. Allowance should be made in the
calculations for clay dehydration and dehydroxylation,
and pyrite transformation in the eastern Devonian shales,
and carbonate decomposition in the western Green River
shales.
ACKNOWLEDGEMENT
The author is grateful to his associate M. B. Harris for her
assistance in acquiring the oil shale data.
of
REFERENCES
ASTM Standards, Gaseous Fuels, Coal am Coke, Part 26, Methods
D3172-75, 1973
Fyans, R. L. Thermal Analysis Application Study No. 21, Perkin-
Elmer Corp., Norwalk, CT, 1977
Hassel, R. L. Thermal Analysis Application Brief 54, DuPont Co.,
Instrument Products. Scientific and Process Division, Wihnintzton.
_
DE
Elder, J. P. Proc. ‘10th North American Thermal Analysis Society
Conference’, Boston, MA, 1980, p.247
Earnest, C. M. and Fyans, R. L. Thermal Analysis Application Study
No.32, Perkir-Elmer Corp., Norwalk, CT, 1981
Elder, J. P. ‘Proc. 11th North American Thermal Analysis Society
Conference’, New Orleans, LA, October 1981.._paper _ 124: ‘Ihermo-
chimica Act; 1982 52,235.
System 4 Operating Manual No. 993-9446, Appendix 2, Perkin-
Elmer Corp., Norwalk, CT
Warne. S. St. J. in ‘Analytical Methods for Coal and Coal Products.
Volume III’, (Ed. C. Kz&r, Jr.), Academic Press, New York, 1979,’
p.459
Wen, C. S. and Yen, T. F. in ‘Thermal Hydrocarbon Chemistry, Adv.
in Chem. Series 183, (Eds. A. G. Oblad, H. G. Davis and R. T.
Eddinger), p. 350