Analytical methods in coal chemistry

ANALYTICAL METHODS IN

COAL CHEMISTRY

PROEFSCHRIFT

TER

VERKRIJ

GING

VAN DE GRAAD VAN

DOCTOR IN DE TECHNISCHE WETENSCHAP

AAN DE TECHNISCHE HOGESCHOOL

TE

DELFT OP GEZAG VAN DE RECTOR

MAGNIFICUS DR. R. KRONlG,

HOOGLERAAR

IN

DE

AFDELING

DER TECHNISCHE

NATUURKUNDE,

VOOR

EEN

COM

MISSIE

UIT

DE

SENAAT TE

VERDEDIGEN

OP

WOENSDAG

29 JUNI 1960

DES

NAMIDDAGS

TE

4

UUR

DOOR

LEENDERT BLOM

GEBOREN

TE ARNHEM

DRUKKERIJ EN UITGEVEI?IJ JOH. LUIJK C. V. EINDHOVEN

--

BIBLIOTHEEK

364

DIT PROEFSCHRIFT IS GOEDGEKEURD

DOOR DE PROMOTOR

Aan de nagedachtenis van

Dr. H

.

W

.

Deinum

.

Aan mijn ouders.

Aan

Rie.

CONTENTS

Infroduction

CHAPTER I

Methods for ultimate analysis of coal

§

1. Introduction

§

2. Carbon and Hydrogen

§

3. Nitrogen

§

4. Sulfur

a. Organie sulfur b. Total sulfur

§

5. Oxygen and water

a. Purification of carrier gas b. Blank value

c. Interference by sulfur d. Oxydation catalyst e. Influence of mineral matter f. Correction for moisture

g. Rapid method by direct titration of carbon dioxide.

References.

CHAPTER 11

Methods of analysis of oxygen-groups in coal.

§

1. Introduction 2. Carboxyl

a. Exchange with calcium acetate b. Iodometrie determination c. Decarboxylation

d. Methylation with diazomethane

§

3. Hydroxyl

a. Methylation with diazomethane b. Reaction with dinitrofluoro benzene c. Acetylation in pyridine

d. Acetylation in the presence of boron acetate e. Deuterium exchange

§

4. Carbony!

a. Oxime formation

b. Reduction of quinones with titanous chloride c. Reductive acety!ation of quinones

§

5. Che1ate rings

§

6. Other groups. References

CHAPTER III

ApplicatioD of the methods to a series of coals of different rank and discussioD of results

§

1. Ultimate analysis

§

2. Carboxyl

§

3

.

Hydroxyl

§ 4

.

Carbonyl

§ 5.

Chelate rings

§

6. Discussion on chelate rings and carbonyl groups a. The hydroxy-quinone model

b. The o-quinone model

§ 7

.

Miscellaneous oxygen groups

§

8. Genera! discussion of the results. References

Summary Samenvatting

Korte Levensbeschrijving

ANALYTICAL METHODS IN COAL CHEMISTRY

Introduction

As in all constitution studies, analytical methods play an important part in the study of the constitution of coa!. One might even say that "the constitution of coal" is a purely analytical problem.

Owing to the difficulty of this problem a great variety of methods have been used for attacking it. However, all investigations started by determ~ ining the elementary composition, which is the basis of our knowIedge.

Chapter I deals with the methods used in ultimate analysis, particularly with the determination of oxygen, which is still the subject of many investigations and disçussions.

The elementary composition being known, several main routes can be followed.

1. Development of a hypothesis concerning the formation of coal in nature; this too is done with the aid of ultimate analysis as this permits determination of the composition of all the hypothetical intermediate compounds. This work has contributed greatly to the classification of coals and suchlike products. However, the results with regard tot the constitution of coal are of a highly speculative nature.

2. Conversion of coal to products more accessible to chemical and physical investigation. This was done by extraction, hydrogenolysis, oxidation and, to a certain extent, also by pyrolysis and ozonolysis. Also in the investigation of these products ultimate analysis has been of paramount importance; in addition, use was made of methods for determination of functional groups. The methods developed for analyses of coal samples described in Chapter II are to a large extent also applicable to these substances.

3. Comparative study of the unaltered co al and model substances by physical and physico~chemical methods. The graphical~statistical method developed by D. W. van Krevelen and co~workersl for determining the fraction of aromatic carbon contained in coal is a well~known example of this kind of approach.

Knowledge of the distribution of organically bound oxygen over the various functional groups is needed for this method, to permit application of adequate corrections. For, the correction factors differ markedly with the nature of the oxrgen groups.

4. Study of the unaltered coal by chemical methods; this study is confined mainly to functional group analysis. The functional group contents

as such provide very important structural data and, as already stated, play a part in the statistical study of the carbon skeleton. The results found on a series of coals of different rank by means of the methods described in Chapter II are summarized and discussed in Chapter

lIl.

Organically bound oxygen deserves attention also from a more technical point of view.

It

is generally accepted now, that this oxygen forms part of the fundamental units of which coal is composed and thus plays a prominent part in the coking of coal. Therefore, a more detailed know-ledge of the contents of different oxygen groups in coal might be useful for obtaining a better understanding of coke formation.

Reference.

1) van Krevelen, D. W. and Schuyer,

J.

"Coa! Science", part. I1, E!sevier Pub!. Cy 1957.

CHAPTER I

SEMI-MICRO METHODS POR ULTIMATE ANALYSIS OP COAL

§

1. INTRODUCTION.

The significance of ultimate analysis for the classification of coals was appreciated al ready in 1837 by Regnault, who introduced the oxygen content as a parameter.

Since then many investigators have discussed the merits and demerits of classifications based on elementary composition. It was recognized from the beginning that the C, Hand 0 contents were essential in this respect, the S and N contents being more or less in significant.

The application of these classification systems, however, has been unfavourably influenced by the following factors.

a. The frequent use of microscopically unhomogeneous material instead of pure co al components.

b. The indirect determination of the oxygen figure, which plays such an important part in classification work. With this differential method all the errors in the analysis of the other elements accumulate in the generally low oxygen figure, th us giving rise to large relative errors. c. The fa ct th at the ultimate analysis is made unreliable by the mineral matter contained in the coal. The effect on the figures is both direct (e.g., decomposition of carbonates, falsifying the carbon and oxygen figures) and indirect (ash contents which differ markedly from mineral matter content).

d. Unreliable methods for determination of moisture and elimination of interferences caused by it.

In this chapter only a brief discussion will be given on the well elabor~

ated methods, which differ little from wel~known published methods. Mention will be made of the parts which have not yet been published, or which, in our opinion, have not received the attention they deserve. The determination of oxygen is treated rather extensively as it is still in the course of development and also because oxygen plays such an important part in the two following chapters of this thesis.

§

2. CARBON AND HYDROGEN.

For the determination of carbon and hydrogen in coal well-established methods of complete combustion have in most countries been accepted as standard methods. N early all of these are on a 20~ 1 00 mg scale, as a compromise between the (accurate) macro analysis technique, which is very laborious, and the micro-technique, which is unsuited for products like coal owing to difficulties in sampling.

It

is general practice now to allow the samples to come into equilibrium with the moisture in the atmosphere before weighing them. Simultaneously with the analysis of the elements, the water content of the coal is determined.

In the past decade several new means have been proposed for absorbing nitrogen oxides from the gasstream. The lead peroxide originally used by Pregl had several disadvantages; for example, it showed a memory effect for water and did not allow the use of high flow rates.

We first tried manganese dioxide1, which removes nitrogen dioxide very effectively. However, nitrogen monoxide is not absorbed and there~ fore had to be oxidized be forehand. This involved the need of using a rather long contact period between carrier gas and nitrogen monoxide, before the combustion products are introduced into the absorption tube2_• Moreover, the exhaustion of the reagent could not be easily detected. A more suitable reagent, absorbing both nitrogen monoxide and dioxide, was proposed by V. A. Klimova and M. O. Korshun3_{• We used this} reagent for several years both in micro~ and semi~micro~techniques. It consist of a very porous silicagel, impregnated with dichromate~sulfuric acid mixture, and has a number of outstanding advantages.

1. High flow~rates of carrier gas are permitted (up to at least 9 1 per hour).

2. The capacity of the reagent is fairly high.

3. The extent to which the rea gent is consumed is readily seen (colour change from orange to green).

The silicagel used is purified by heating it which nitric acid (1.4) on a steam plate for several days and subsequently washing it with distilled water until the effluent is free of acid.

The preparation of the absorbent is described below. Preparation of the absorbent

Heat Si02 (partic1e size 0.5- 1 mm) in a tube at 200°C for 1 hour whilst passing a current of dry oxygen, or heat the material in a drying oven at the same temperature for 3 hours.

Dissolve about 7 9 of K2Cr207 in 50 mI H2S04 with gentIe heating. Transfer the Si02 into a fIask provided with a ground-in stopper and add the K2Cr207~ solution, supplying several drops at a time. Aftel. each addition close the fIask and shake vigorously. Continue adding the K2Cr207-solution until the silicagel starts sticking to the wal!. The freshly prepared reagent is yellowish-brown in colour; during absorption of the nitrogen oxides the colour gradually changes into green. Bring so much reagent into a tube of about 10 mm diameter until the length of the WIed section is about 9 cm for semimicro_ and about 3 cm for microwork. These lengths must not be exceeded. Enelose the silica filIing between two layers of anhydrone.

§

3. NITROGEN.

The determination of nitrogen in coal is generally carried out by two principally different methods, i.e., by the Dumas method and by the Kjeldahl method. Both methods are well~known and need not be discussed here in detail.

However, according to various authors they do not always give reliable results and in addition have some other special drawbacks.

Applied to coal. the Dumas method gives results which tend to he 0.1 - 0.2 per cent too low; this is mainly ascribed to formation of comhustion~resistant cokes. On the other hand methane formation mav cause positive errors. Moreover, the blank value obtained with the variabie copper oxide filOOg is not always constant, thus giving rise to inaccurate results.

The Kjeldahl method which has the advantage that it is particularly adaptable to series analysis, also gives of ten low results; the destruction of high rank coals and coke is extremely difficult and formation of elementary nitrogen is reported by different authors.

Depending on the type of catalyst used and on the destruction conditions,

the loss of nitrogen may vary to some extent.

In the past decade a modification of the Dumas method has been adopted in the micro analysis of organic compounds. In this method, introduced by

J

.

Unterzaucher4_{,5; the sample is bumt in an oxygen atmos~} phere (without copper oxide). W orking on a micro scale we found that samples in which nitrogen could hardly be determined with the original method (e.q. brucine), did not give any difficulty with the modified method.

Therefore we tried the method on a semimicro scale on coals. However, the quantitative reduction of the ex ce ss oxygen, which on the micro scale was carried out with copper activated with some iron, could not he realized on the semi micro scale. Better results were obtained with nickel activated with 2 per cent of thoria. This rea gent was prepared by decom~

posing a mixture of the carbonates at 1000

oe

in a stream of hydrogen. Used at 1000

oe

it proved to lose its activity af ter a few experiments, presumably through sintering of the material.

According to P. Mars* this trouble might be overcome by using the large surface area of kieselguhr as a support for the nickel~thoria mixture.

This suggestion could be confirmed and the new catalyst ramained active during at least a hundred experiments.

The oxygen used for the combustion was prepared by electrolysis of a sodium hydroxide solution with the aid of monel electrodes. The blank found with this device was almost negligible (0.01 mi).

It

was a definite advantage that the oxygen could be supplied in amounts adapted to the sample to be analysed. Rapid exhaustion of the reduction agent could th us be prevented.

The method was tested with pure silver nitrate, which was decomposed to form nitrogen oxides. Table I, 1 shows some results which prove that quantitative reduction is achieved even under these unfavourable con~ ditions.

TABLE

I.

1

REDUCTION OF NITROGEN OXIDES

Amount of Nitrogen content Carrier gas silver nitrate _found

I

calc. mg _{% %}

Carbon dioxide

44

.

1

8.36

8

.

23

54.4

8

.

13

48.5

8

.

18

Carbon dioxide +

52

.

6

8

.

16

Oxygen

41.5

8.23

Difference

I

%

+0.13

-0

.

10

-0

.

05

-0

.

07

0

.

00

The results found with a series of pure organic compounds a

r

e shown

in table I

,

2.

TABLE

I.

2

DETERMINATION OF NITROGEN IN ORGANIC COMPOUNDS

Weight of Nitrogen content

Difference Sample sample found

I

calc. % mg % %

Brucine

20

.

6

7.06

7.11

-0

.

05

30.8

7

.

21

+0.10

Cinchonine

19

.

9

9

.

49

9.51

-0

.

02

Cyclohexanone~

27

.

3

13.17

13.23

-0.06

~oxime

19.3

13.07

-

0.16

Dinit

r

ochlorobenzene

18

.

7

13.73

13

.

82

-0.09

22

.

8

13.94

+0.12

Cystine

27

.

8

11

.

85

11

.

67

+0

.

18

34

.

4

11.64

-

0

.

03

Sulfanilic acid

32.3

8

.

13

8.09

+0.04

31.8

8

.

21

+0

.

12

Azo benzene

15.3

15.38

15.38

0

.

00

15.3

15

.

38

+0

.

20

Some

high~rank

coals and two coke samples were analyzed with the

method described here and with a Kjeldahl method

.

The results are shown

in

table I

,

3

.

TABLE

I.

3

DETERMINA TION OF NITROGEN IN COALS AND COKES Nitrogen content found

with

Difference Sample Amount of

I

sample, mg Combustion Kjeldahl % method method % % Coking coal

I

92.6

1

1.82

I

1.7

1

+0.2

102.6

1.93

Anthracite

1

I

100.5

I

1.75

1

1.6

I

+0.2

102.6

1.89

Anthracite

2

I

102.8

I

1.97

I

1.9

I

+0.1

106.7

2

.

02

Anthracite

3

I

102.2

I

1.46

I

1.4

I

+0.15

98.9

1.67

Coke

1

I

104.5

I

1.28

1

1.0

I

+0

.

3

104.6

1.34

_•

Coke

2

I

108

.

0

1.22

1 1.1

+0.2

103

.

6

1.38

Obviously for these samples there exists a systematic discrepancy be-tween the figures, which proves again th at the Kjeldahl method is Ie ss suitable for analyzing these higly aromatic structures.

Procedure for the determination of nitrogen a. Preparation of nickel-catalyst

Dissolve 1000 9 nickel nitrate (6 H20) and 41 9 thorium nitrate in 2 I water; heat to 20°C. Add sodium carbonate solution until the pH reaches a value of abt. 7.5. Add 780 9 kieselguhr and heat to 95°C.

Stir the mixture for one hour, cool slightly and filter on a büchner filter. Wash the precipitate with a solution of ammonium carbonate to remove sodium. Heat at 150 °C until almost dry and press the cake (40 t.). Grind the pieces obtained and use the partic1es of 1-3 mm diameter (Presumably also the powder may be used). Reduce the catalyst with hydrogen at 1000 oe.

b. Analysis of a sample.

Set up the apparatus shown in figure I, 1. Pass carbon dioxide through the apparatus and wait until micro-bubbles are formed in the azotometer. Heat the furnaces. Weigh a suitable amount of sample in a platinum boat. Stop the carbon dioxide stream and introduce the boat into the tube at a distance of 22 mm from the large furnace. Stopper. the tube and pass a fairly rapid stream of carbon dioxide through the system during one minute, to completely flush out the air. Check wether micro bubbles are obtained; if not, repeat the flushing.

Pass oxygen from the electrolyser into the combustion tube (flowrate 1.5 mi

per minute). Heat the sample with the movable furnace until the combustion is finished. Subsequently replace the oxygen by carbon dioxide, maintaining the same flowrate. Redouble the flowrate as soon as the si ze of the gasbubbles in the azotometer decreases markedly. Read the azotomet"'r-burette when micro-bubbles are obtained.

C-1-2 Z LU 0

0

~ f

-Z

LL.

0

Z _~

0

E i= ..:!

«

Z ~ ~ LU f -LU Cl LU I f -~ _c

0

G G

'"

LL. c

_"

gl

.2 'w Vl _~-~

=>

• 0 c f - _~ .&

«

~ t1.

5

<

_~ a... a...

~

<

G C Lij 0 öii ~-

.

c ~ • 0 ~ ~

_a

.t

.

_~

li

_~LI)

_'"

1

~ _~ 0 § E

~

'ê ~ iJ:

§

4. SULFUR. a. Organic sulfur.

The determination of "organic sulfur" in coal is very complicated if

substantial amounts of pyrite and sulfates are present in the mineral matter. Most of the inorganic sulfur will be converted to sulfur oxides during the combustion process. On the other hand, if the co al contains certain

inorganic constituents, e.g., alkali and alkaline earth carbonates the sulfur oxides originating from organic sulfur, may be retained in the ash. There-fore, a low ash content is a prerequisite for a reliable direct determination of organic sulfur by combustion.

When the above inorganic sulfur compounds are present they should be determined separately and substracted from the total sulfur content (see below).

First a brief discus sion wil! be given on a method which, although not suited for coals of high ash content, can be applied to the majority of samples that have been pretreated to reduce the ash content to below 5 per cent.

The sample is bumt in an oxygen atmosphere (flow rate 3 1 per hour) at about 900

oe

.

The apparatus is fitted with the ignition device described by G. de Vries and

E

. van Dalen;

6 this device is particülarly suited for effecting rapid combustion of coals of high volatile matter content. The vapours are continuously ignited by a spark, struck between two platinum wires. The combustion in this apparatus takes no more than Bve minutes.

The heated zone of the quartz tube is charged with platinum wool to ensure complete combustion.

The sulfur oxides were determined by the method proposed by

E

.

W. D.

Hufmann7 _{and further developed by Zinneke8• The sulfur oxides are} absorbed on silver gauze, which is heated, to 450-500

o

e

.

According to

M. Vecera and D. Snobl9 _{the surface contains silver oxide, which effects} the quantitative oxidation of sulfur dioxide to silver sulfate. The gauze absorbs a much larger amount of sulfur than corresponds with a silver sulfate monolayer, which indicates that a rapid diffusion of silver to the surface takes place. For this reason a minimum temperature of 450

oe

must be maintained.

We found that the protection of the silvergauze against abrasion during introduction into and removal from the quartz tube is essential, particularly when the gauze is to be used repeatedly. Wrapping it round with a 0.2

mm platinum wire was effective.

Gravimetrie determination of the silver sulfa te' deposited on the gauze did not suffice. Therefore the sulfate was washed off in the apparatus proposed by Zinneke and titrated with standard potassium iodide. The

TABLE

I.

4

DETERMINATION OF SULFUR BY COMBUSTION AT 900

o

e.

Sample

I

SuIfur No. %

1 (coal)

0.59-0

.

63

2 ( coal)

0.62-0.66

3 (coke)

0.94-0

.

96

4 (active carbon)

0.18-0.17

5 (coal)

7.19-7

.

16-7

.

22-7.14

6 (coal)

1

.

79-1.84

7 (pitch)

0.67-0

.

72

end point was determined potentiometrically. The electrodes used were a 1 mm silver wire and a nickel - nickel sulfate electrode.

The latter is a very simple reference electrode, convenient in use and particularly suited for titration with silver salts*. Even with 0.0001 N potassium iodide the titration end~point is easily detected in a volume of 50 mI.

Some results which give an impression of the duplicability of the method on coal samples are listed in table I, 4.

Table I, 5 shows the results obtained on a number of pure compounds, both on centigram and milligram scale.

TABLE

I.

5

DETERMINATION OF SULFUR IN PURE COMPOUNDS % suifur found

Sample _centigram

I

milligram % suifur calculated

scale scale 18.38 18.26 Sulfanilic acid 18.59 18.27 18.50 18.64 18.31 14.63 14.63 Diphenylsulfone 14.44 14.70 14.68 14.50 Cystine

I

26.23

I

26.42

I

26.67 26.24 Methionine

I

21.48

I

-

I

21.49 21.63 Ammonium sulfate

I

24.00

I

-

I

24.24 24.38 Thiophenol

_I

29.2

_I

-I

29.1 Diphenyldisulfide

_I

- _{29.16 29.36} b. Total sulfur.

The détermination of total sulfur in coa1 was done with the method of W. Radmacher10, a modification of the method of Seuthe. The sample is burnt at about 12000

_C

_{with addition of ferric phosphate or other}

suitable agents to remove all the sulfur from the ash.

This method was modified by us to some extent. The sulfur oxides we re absorbed on silver gauze and determined as described above. In this way the method becomes more specific and simpier.

We found that ferric phosphate was very suited and gave reproducib1e results. However, depending on the composition of the ash, difficulties,

which we believed to be due to a low melting point of the ash, were *) The author is indebted to Mr. G. de Vries, Vrije Universiteit, Amsterdam for

this information.

sometimes observed. If in such a case the combustion is carried out at 1200 oC, part of the molten ash is absorbed in the porcelain, and, in

consequence, is kept from reacting with ferric phosphate.

These difficulties could be overcome by a slightly altered procedure: combustion of the organic material at about 800°C, addition of ferric phosphate to the ash and subsequent heating to 1200o

C.

This had two additional advantages: the contents of the porcelain boat are not driven out by the sweJling coal as of ten happens when the ferric phosphate is added before the combustion, and no organic material is kept from combustion by the ferric phosphate.

In table I, 6 some results obtained with the method under discus sion are compared with results found on the same samples in the laboratories of Radmacher and partly also with the Eschka values.

Sample no. 1 2 3 4 5 6 7 8 9 10 TABLE

I.

6

DETERMINA TION OF TOTAL SULFUR IN COAL RESULTS IN PERCENTAGE SULFUR

I

Combustion at 1200 0C

I

Eschka

Laboratory 1

_I

Laboratory 2 Laboratory 2

I

1.10

I

1.11

I

-1.26 1 1.04

I

1.07

I

1.03 1.09

I

1.02

I

106

I

-1.06

I

0.90

I

0.93

I

0.95 0.99

I

1.03 0.93 1.00

-1.08 1.09

I

0.86

I

0.87

I

-0.88

I

10.32

I

10.32

I

10.90 10.66

I

-

I

0.38

I

0.35 0.38

I

-

I

2.00

I

2.09 2.15

-I

2.75

I

-2.87

Laboratory 1: Results from Dr. Lange and Dr. Mohrhauer,

Gemeinschaftsorga-nisation Ruhrkohle, Essen. Laboratory 2: authors' results.

The table shows that the reproducibility and the repeatability are satis-factory. The agreement with the Eschka method is also good.

W . Schuhknecht and H. Kunzl l _{stated that ferric phosphate gives}

irregular results when the sulphuric acid is determined by indirect titration (addition of excess barium hydroxide and back-titration of the excess with lithium sulfate). Interference through phosphoric acid is not likely to occur when the silvergauze-method is used.

N 14 v> _w zQ g~ u'" ~5 0::> Uv> z w

g

o

z

o

~

ii ::> Q.

O

l

z o 5i ::>

'"

L o U 1'rC===d(.~ E

J

w N ::> ~

'"

w >

...

ij;

o

\.!) z :I: Vl « ~ C-1-7

From the total~sulfur content found with the method described the sulfur present as pyrite and sulfate should be subtracted. These two figures can be determined with the methods described by Radmacher and Mohrhauer. 12,13

Procedures for the determination of suUur.

1. Sulfur of combustion.

Set up the apparatus shown in figure J, 2. Pass oxygen through the system, flowrate 3 I per hour.

Weigh a suitable amount of sample into a quartz ampul1a and introduce it into the tube (see figure). Switch on the ignition system and heat the sample gradual1y by increasing the current through the heating coil (max. abt 10000 _{C) until}

combustion has been completed.

Remove the silvergauze from the tube and introduce it into the extraction apparatus heated with boiling isopropylalcohol. Add water until the gauze is completely immersed and wait for 3 minutes. Siphon the extract into a 150 mi beaker and repeat the washing three times. (2, 2 and 1 minutes resp.). Add 2 mi nitric acid 2 N to the combined washings.

Titrate the silversulfate potentiometrically with standard potassium iodide solution, using a silver and a nickel~nickelsulfate electrode.

Activate the silvergauze now and then by immersing it in 2 N nitric acid and removing it as soon as gasbubbles are observed. Subsequently immerse it in ammonia (10%), wash with water and dry.

2. Total Sulfur.

Weigh a suitable amount of sample into a clean porcelain boat and intro duce it into a quartz tube the foremost part of which may be heated with a coil to abt.

900° C, the middle part, Wied with platinum woel. with a fumace to 1250 °c and the back part, containing silvergauze, with a furnace to 500°C.

Heat the sample with the heating coil until the organic matter has disappeared.

Remove the boat from the tube, add 1 9 of ferric phosphate and introduce it again.

Shove it careful1y forward - with the aid of a pusher - until it is completely in the 1250 °C-zone. Af ter 10 minutes remove the silvergauze and determine the silversulfate with the method described above.

§

5. OXYGEN.

The direct determination of the organic oxygen content of co al has been the subject of many investigations. Two types of principally different methods will be mentioned here.

The one group of methods is based on the work done by H. ter Meulen and

J.

Heslinga. 14,15 In 1936

T.

Inaba and Y. Abe16 applied the method

to coal and soon afterwards, B. N. Afanas'ev17 _{made a detailed study of it.}

The principle of the method is as follows :

the sample is pyrolized at 900-1000

°c

in a hydrogen atmosphere, the gases formed are cracked on platinized quartz, or asbestos, and sub~

sequently reduced at 300-3500_{C on a hydrogenation catalyst. (Ni,} activated with Th). The water formed is determined gravimetrically.

In his thesis C. Georgadis18 _{described a modification of this method.}

He paid special attention to the influence of the temperature during pyrolysis on the quantitavity of the conversion of organic oxygen.

(2,5 to 3,5 hours) and that frequent regeneration of the catalyst was necessary. Although modifications of the method were reported in literature up to 1956,19 the work of Spooner,20 published in 1947 which was based on the well-known Schütze-Unterzaucher21,22 method, became more and more the basis for further investigation. The sample is pyrolized at 1200

oe

in nitrogen, the oxygen being converted to carbon monoxide on wood charcoal. Af ter removal of interfering substances, the carbon monoxide is converted to carbon dioxide and determined as such.

H. Guérin, M. Bastick and P. Marce123 compared the two types of methods on eleven different coals and found the latter method to be preferabIe.

Georgadis,24 who also adopted this method, seems to have arrived at the same conclusion.

Recently, W. Radmacher and A. Hoverath25 proposed a modification of the Unterzaucher method. Simultaneously with the volumetrie deter-mination of the iodine they did a gravimetrie deterdeter-mination of the carbon dioxide.

In this way the occurrence of interferences may of ten be readily detected. We started by using the method of H. W. Deinum and A. Schouten26 whieh was a modification of the Spooner's method. The main alterations were:

a. the carbon monoxide was oxidized on mercurie oxide at 150

oe;

b. the carbon dioxide formed was determined by absorption in excess

baryta and back-titration of the excess with hydrochlorie acid.

Our experience with the above method and the additional alterations will be discussed briefly. A more detailed discussion will be given of a new method using direct titration of the carbon dioxide and allowing separate determination of water. This method has distinct advantages over all other methods and is particularly suitable for hygroscopie substances like coal.

a. Purification of the carrier gas.

The nitrogen was originally purified by passing it through an ammoniacal solution of ammonium chloride in which copper shavings are immersed. The maintenance of this device was rather laborious and we therefore tried another method.

The gas is led in succession over copper oxide at 750

oe,

soda asbestos, magnesium perchlorate, activated copper at 450

oe

and again over magnesium perchlorate.

The activated copper is obtained in the following way : 300 9 copper (11) nitrate

+

5 9 iron (111) nitrate are dissolved in water, sodium hy-droxide is added in excess, the precipitate is filtered, washed and dried.

The dried mixture of oxides is ground and the fraction of 1-2 mm dia-meter is isolated.

It is reduced with hydrogen at 400

oe

and reoxidized and reduced, if necessary. With this method the oxygen content could be reduced to less than 10-5 per cent (by volume).

b. Blank value.

The Farnell càrbon used originally gave rise to rather high blanks (abt. 0.3 mg) and therefore was replaced by benzene soot prepared by incom-16

plete combustion of benzene. The blank values found with this reduction agent were low. However, its preparation was cumbersome and it attacked the quartz tube rather seriously.

Therefore it was recently replaced by gas~soot which is commercially available and proved to be much less detrimental to the quartz tube. The blank value decreases more rapidly to nearly zero than that obtained with benzene soot.

Applying a new method for determination of carbon dioxide described below to pure organic compounds we discovered another cause of blank values. Now and then, before titration of the carbon dioxide originating from the organic oxygen, a small amount of titrant corresponding with about 0.3 mg oxygen was consumed separately. We could establish that this was due to water adsorbed on the quartz pusher and that the amount fully depended on the area of the surface formed by ageing. By heating the pusher in an oxygen flame the phenomenon was completely suppres~ sed. In the original method this adsorption was the main cauSe of the residual blank value. It is obvious that the degree of adsorption depends to a large extent on the relative humidity of the atmosphere.

c. lnterference by sulfur.

It is a well~known fact that samples rich in sulfur may cause difficulties owing to formation of carbon disulfide and carbon oxysulfide. The method of I.

J

.

Oita and H. S. Conway27 which we used for several years, is very suitable for preventing this interference. The gas is led over copper at 900°C to decompose the above products.

d. The oxidation catalyst.

The use of copper oxide as an oxidation catalyst would be very attrac~

tive as copper oxide is a cheap, readily available substance which can be easily purified. However, We did not succeed in obtaining satisfactory results. At 300°C the blank va lues fluctuate in an unacceptable manner. The high va lues are presumably due to oxidation of volatile hydrocarbons (olefins ?). We also tried lower temperatures, bus this was at variance with the requirement of complete combustion of carbon monoxide.

Anhydroiodic acid, which is generally accepted now as the most suitable rea gent, was tried by us some five years ago. The preparation of a good~ quality product gave many troubles and we could not find any advantage of this product over mercuric oxide which we have used now for about ten years.

Following Deinum and Schouten's method we used mercuric oxide at 150°C and at the end of each determination raised the temperature to 250 oe. Using constant temperature of 200°C we obtained equally satisfactory results. This simplified the procedure.

Using the method resulting from the above modifications we analyzed a large number of coals. The other elements were determined simultan~ eously. All coals had an ash content below 5 per cent as was prescribed by our method. Of the 277 analyses made. 253 gave results . which were within the desired limits of 100.0 ± 0.5 per cent (on a dry ~nd ash~free

basis). Larger deviations may have been due to interfering components of the mineral matter.

e. lnfluence of mine ral matter.

Inorganic matter affects the oxygen figures to a greater extent than the carbon and hydrogen figures. The mineral matter in coal consits mainly of silicates of aluminium, calcium, magnesium and iron and further of quartz, carbonates and sulfates of calcium and magnesium, pyrite and ferrous oxide. The main interferences to be expected are those originating from incomplete dehydration of the minerals at 105°C and from decompo~

sition of carbonates and sulfates.

J

Incomplete dehydration will occur e.g. if Kaolinite is present, which

releases its water between 400 and 500°C. Also montmorillonite gives up its last molecule of water in this temperature region. Muscovite holds

its water more strongly; it is dehydrated above 1000 °C. The oxygen contents in these cases found will be too high.

If the mineral matter contained in a coal with 10 per cent ash consists

mainly of Muscovite which holds 4.5 per cent water, the oxygen figure

will be about 0.5 per cent too high.

It

may weil be that in the carbon and hydrogen determination (below 1000°C) this water remains in the ash In consequence, the total content of the five elements will exceed 100 per cent.

Calcium carbanate as weil as other carbonates are decomposed below 900°C. The carbon and oxygen contents found will be too high, the ash content too low.

In the pres en ce of substantial amounts of sulfur, sulfates may be formed, which is a complicating factor. The presence of ferrous oxide, which will

be found in the ash as ferric oxide, will give too high results for the elements.

As shown in Table I, 8, high ash contents aften have no harmful effect

on the total analysis. However, it should be borne in mind that ash

components which dehydrate between 100-900 °C give rise to high oxygen

figures, whereas the "total" result is not aHected.

TABLE

I.

8

INFLUENCE OF ASH CONTENT ON ULTIMA TE ANALYSES RESULTS EXPRESSED ON D.A.F. BASIS, PER CENT

Sample

_I

C H 0 N

_I

s

I

Total Ash Moisture

Peat 60.7 5.3 32.6 1.7 0.1 100.4 13.3 14.5

Brown Coal I 68.6 5.7 24.0 0.8 0.9 100.0 34.2 9.1

Brown Coal II 71.6 5.5 11.7 1.8 9.3 99.9 13.2 7.7

Brown Coal III 73.7 5.6 12.9 1.8 5.8 99.8 6.6 9.1

Black Lignite 79.5 5.5 11.1 3.2 0.8 100.1 7.5 9.0 Coke 96.7 0.2 0.4 1.7 1.1 100.1 10.0 0.3 Brown Coal IV 68.9 4.8 22.6 1.2 3.6 101.1 19.3 10.7 Brown Coal V 74.5 5.3 8.6 1.2 12.2 101.8 12.4 1.6 Brown Coal VI 78.4 5.5 8.3 1.7 9.4 103.3 12.3 1.2 Anthracite 95.1 2.1 1.9 1.1 0.7 100.9 6.0

By X~ray analyses of the ash we tried to detect which of the compo~

nents of the mineral matter might have been responsible for the high results found for the last four samples.

- - -

-The only difference noted as to the composition of the ash was that the ash, in contrast with that of the other samples did not contain quartz. An emission spectrographic analysis did not reveal a fundamental difference as to the main elements : aluminimum, silica, calcium, magnesium, iron and titanium. However, the potassium content of three of the four samples was much higher than that of the others. This suggested that muscovite (K20.3 AbOa. 6 Si02. 2 H20) was present, which might, at least partly, have been the cause of the erroneous results. The fourth sample had an exceptional high silver content (replaces potassium ?).

From the above considerations it may be concluded that errors in the oxygen content due to interference by mineral matter wil[ frequently not be detected, as the "tota!" result may weU be within the limits (ZOO ± 0.5). Therefore we decided to remove most of the mineral matter from

the samples by flotation with zinc chloride solution, aiming at an ash content below 5

per

cent. However, in some cases the amounts of sample available did not allow such a treatment.

f. Correction for moisture.

The standard procedurefor determination of water in coal samples are all based on the assumption that at

105°C

coallooses all its water.

This had to be verWed, however, by an independent method. For, it is not unlikely that part of the water is bound so strongly that distinctly higher temperatures are required to release it. A fuIly satisfactory method for checking this is difficult to find. We supposed that a determination of the water with Fischer reagent might weIl give an indication as to how far additional water is present.

It

is weIl known, for instance, that the water in a large number of (inorganic) hydrates decomposing at temperatures far above

105°C,

can be readily determined with Fischer reagent.28

In table I, 9 the results found with this method are compared with those of H. W . Deinum and M. L. Goedkoop's29 standard method.

TABLE I, 9

COMPARISON BETWEEN TWO METHODS POR DETERMINATION OP MOISTURE IN COAt Sample Fischer Method Ash Content %

I

Standard proc. 1

0

.

7

0.9

24

.

1

2

1.7

1.4

4

.

0

3

2

.

1

1.8

1.1

4

3

.

2

3.0

0

.

6

5

3.9

3

.

7

0

.

4

6

6

.

1

6.0

0.8

7

8.2

8

.

3

3.2

The results with Fischer reagent tend to be somewhat higher. In this method the coals were pretreated with pyridine at room temperature for four hours. Samples 5 and 7 were also analysed af ter being pretreated for

one and two hours. The figures found were 3.5~3.8 and 8.1~8.2 respectively, showing that even a one hour's treatment is almost sufficient.

The agreement between these results provide astrong indication, though no definite proof, that all of the water present is determined by the two methods. The hygroscopicity of coals necessitates the moisture to be determined simultaneously with the elements, in particular together with oxygen and hydrogen. As will be shown below, however ,the determination of oxygen may be made independent of the determination of water. g. Rapid method by direct titration ot carbon dioxide.

The method described above for determination of oxygen had one serious drawback: the procedure was rather lengthy, every determination lasting about 70 minutes. This was attributed to th ree factors: the rather low maximum flowrate of the carrier gas, the indirect estimation of the carbon dioxide and the great care to be taken in the back~titration of the bcuyt.a.

Therefore, we recently tried tó use the direct~titration method for carbon dioxide, which we described a few years ago,30 for determination of oxygen.

The carrier gas is passed through pyridine containing thymol blue (indicator). Carbon dioxide present in the gas dissolves in the pyridine, giving rise to local yellowing. The solution is kept blue by addition of standard sodium methylate dissolved in a l : 4 mixture of methanol and pyridine.

With this method the flowrate of the nitrogen may be markedly in~ creased owing to the high solubility of carbon dioxide in pyridine; moreover the determination can be finished as soon as the carbon dioxide is expelled. In this way a determination of the oxygen in organic com~ pounds can be carried out every twenty minutes. This also holds for the determination of total oxygen in co al samples, in which case the oxygen present in the moisture is determined together with the organic oxygen. If it is desired to establish the percentages of the two components separately, the sample is first heated at about 105°C for 20 minutes, whereafter the dried sample is pyrolysed at 900~1120 °C for another 15 minutes. FIOm the two portions of carbon dioxide titrated, both the water content and the organic~oxygen content can be calculated. The method is more sensitive and more accurate than the original one and has the added advantage that irregularities in the pyrolyses process are readily detected.

Table I, 10 shows the results obtained on pure calcium carbonate and on a number of pure organic compounds in the milligram and centigram scale.

TABLE

I.

10

DETERMINATION OF OXYGEN IN PURE COMPOUNDS

1% 0, calc· 1

% 0, found

I

difference Sample

micro

I

semi micro

I

micro

I

semi micro

31.8 - -0.2 -Calcium carbonate 32.0 31.9

-

- 0.1

-31.9 - -0.1

-32.3

-

+0.3

-I

I

11.00 11.10 00.00 .10 2-Naphthol 11.10 10.9 _11.0 11.11 _{11.01 -0.1} -0.2 _-+0.01 _0.09

I

11.2 11.11

I

+0.1 +0.01 11.3 11.22 +0.2 +0.12 Anthraquinone

I

15.37

I

15.4

I

15.43

I

0.0

I

+0.06 15.5 15.51 +0.1 +0.14

I

I

26.0

I

26.02

I

-0.2

I

-0.18 Benzoic acid

I

26.20

I

26.34 +0.14 26.4 26.26

I

+0.2 +0.06 Benzophenone

I

8.78

I

8.6

I

8.81

I

-0.2

I

+0.03 8.8 8.85 0.0 +0.07 Hydroquinone

I

29.06

I

29.1

I

29.02

I

0.0

I

-0.04 29.3 29.09 +0.2 +0.03 Terephthalic

I

33.0

I

33.2

I

-I

+0.2

I

-acid-dimethyl ester 33.2 - +0.2

-Saccharose

I

51.42

I

51.2

I

51.33

I

- 0.1

I

- 0.09 51.3 51.37 - 0.2 - 0.05 Acetanilide

I

11.84

I

12.1

I

11.93

I

+0.3

I

+0.09 12.05 +0.3 +0.21 Dinitro-chlorobenzene

I

31.60

I

31.5

I

31.64

I

+0.1

I

+0.04 31.6 31.55 0.0 - 0.05 Azoxy benzene

I

8.1

I

7.9

I

-I

- 0.2

I

-8.1 - 0.0 -Dinitro benzene

I

38.07

I

38.0

I

37.82

I

-0.1

I

-0.25 38.3 37.79 +0.2 - 0.28 Cystine

I

26.63

I

26.4

I

26.31

I

-0.2

I

- 0.32 26.5 26.84 -0.1 +0.21 Diphenylsulfone 14.7

I

14.8

I

-I

+0.1 -15.0 - +0.3

-Table I, 11 shows the repeatability of the results found with three coals

of different oxygen content.

TABLE

I.

11

REPEATABILITY OF THE RESULTS FOUND BY DIRECT TITRATION (RESULTS IN PERCENTAGE OXYGEN)

Sample I (89 % C) Sample II (87 % C) Sample III (80% C)

3.73

5.74

13.85

3.80

5.82

14.01

3.88

5.77

13.99

3.89

5.81

13,71

3.74

5.80

14.25

3.82

5.75

13.91

3.77

_I

14.26

5.78

14.00

These results are highly satisfactory.

The amounts of sample

111

varied from 15 to 30 mg, which is rather

little for coal. The weight of sample

11

va ried from 30 to 65 mg and of sample I from 50~100 mg.

Because of its sensitivity the method under discus sion is particularly suited for analysis of samples of low oxygen content and for micro analysis.

The method was recently applied to a series of coals of different rank (see table I, 12) used in the work described in the next chapters * )

Simultaneously with the oxygen, we determined the moisture content

of the samples. The results found were compared with those of the

standard method of Deinum and Goedkoop used so far28 _{and are also}

reported in table I, 12.

*) It was found that these finely devided samples had oxidized during several years storage. The aVf~rage oxygen percentage proved to have increased by 15 per cent of its original value. This will be discussed further in chapter

nr

.

22

TABLE

I.

12

SEPARATE DETERMINATION OF OXYGEN AND MOISTURE IN THE OXYGEN APPARATUS

Sample %0 %H20 % H20 acc. to standard _method

Vitrinites

1

24.24

_24.36

10.17

_10.34

10

.

2

2

18

_18.95

.

64

6

_6.47

.

34

6.2

3

13.96

_14.11

5.34

_5.45

5.4

4

9

_9.58

.

46

₂

2

_.

.

22

₂₈

2

.

4

5

3.27

_3.31

_1.30

1.29

1.3

6

_3.17

3.14

1.01

1.03

7

2

_2.17

.

21

1.05

1.05

Aach. Kreide

9

23.59

_23.59

9.29

₉

_.

₆₆

8.7

a)

Faulquemont

10

13.85

6.61

6

.

8

14.01

6

.

72

a) Amount of sample too smalI.

These data demonstrate that the duplicability of both the oxygen and the moisture determination is very satisfactory. The agreement of the latter method with the standard method is also satisfactory.

Summarizing, it may be stated that the new method has the following features.

a. lts accuracy is at least equal to that of the original method (indirect titration) .

b.

It

is much more rapid in those cases where the total oxygen content (water included) is to be determined.

c. Water may be determined separately, and if this is done the analysis takes still less time than the analysis by the original method.

d. The sensitivity is better; the method is much more suited for micro and tra ce analysis.

e. If desired the course of the decomposition of an oxygen compound with time may be followed.

N ..;..

l'''~'w--:f

.

-Titrant

~

=-=-

.

""di ..

:

.

P1ston burette ·t1tration veesal Figure J. 3.

Apparatus for the determination of oxygen.

200 °c Red mercur1c oxide Ascar1~ Copper

900

°c gauze

/URIPICATION

cr

'1F.E NITROOEN ..

0, . COPPer oxide ADhydr'one

.~

L

N1trcgen

r--~

.9.Jbble counter. Carbon

Procedure for the determination of oxygen and water by direct titration.

Set up the apparatus shown in figure I, 3. Pass nitrogen through the purification assembly and apparatus at a fIowrate of 3 1 per hour to remove traces of air. (note 1). Charge the titration vessel with a suitable amount of purified pyridine and add sodium metbylate solution (0.05 N) in methanol-pyridine 1 : 4 dropwise until the colour turns blue.

Remove the rubber stopper form the quartz tube, place the platinum boat containing a weighed amount of sample (a mg) on the quartz pusher (note 3 and

4) and stopper the tube (note 5). a. Determination of water (note 6).

If the sample has a high water content (}. 5 per cent). heat first at 500_{C for}

5 minutes and subsequently at 105 °C, otherwise heat immediately to 105°C. Titrate the carbon dioxide absorbed in the pyridine, continuously, taking care that the pyridine solution at A stays blue; (titrant consumption Vi mI). b. Determination of oxygen (note 6).

As soon as all carbon dioxide orginating from the water has been titrated, slowly move the sample into the 1120 °C zone with the aid of the electro-magnet. Increase tbe current through the heating coil to bring the tempera tu re at about 1000 °C and titrate the evolving carbon dioxide as mentioned above (v2 ml ).

Calculate the amounts of water and oxygen in per cent by weight by means of the following formulae :

1802. Vl. t.

a

1600 . V2 • t.

per cent O.

a

Notes: 1. All temperatures are to be continuously kept at the values indicated in the figure, except the temperature of the mercuric oxide, which is allowed to drop overnight. When the apparatus is not in use, the fIowrate of nitrogen is kept at about 1 1 per hour.

2. Commercially available pyridine is of ten contaminated with interfering substances. If so, it should be purificd by distillation. In some cases heating with active carbon proved to be sufficient.

3. Samples of high oxygen content (> 25 per cent) and rapidly decomposing samples require special attention. The solution should be kept blue carefully.

4. The pusher should be heated in an oxygen fIame now and then to remove the carbon-black and reduce its surface area.

5. In all these manipulations care should be taken to keep atmospheric oxygen from entering the tube.

6. The determination of water lasts 15-25 minutes depending on tbe water content.

REFERENCES TO CHAPTER I.

1. Be/cher, R &' Ingram, G.

Ana\. Chim. Acta 4 (1950), 401. 2. Ingram, G.

Mikrochemie v.m. Mikrochim. Acta 36/37 (1951) 690. 3. Klimova, V.A. & Korshun, M.O.

Zhur. Ana\. Khim 6 (1951), 230.

4. Unterzaucher,].

Chem. Ing. Technik 22 (1950) 128. 5. ibid.

Mikrochemie v.m. Mikrochim. Acta 36/37 (1951), 715. 6. de Vries, G. & van Dalen, E.

Ana\. Chim. Acta 7 (1952), 274. 7. Hufmann, E. W. D.

Ind. Eng. Chem. Ana!. Ed. 12, (1940), 53. 8. Zinneke, F.

Z. f. Ana\. Chem. 132 (1951), 175. 9. Vecera, M. & Snobl, D.

Chem. Listy 50 (1956), 1941. 10. Radmacher, W.

Brennst. Chemie 33 (1952), 129. 11. Schuhknecht, W. & Kunz, H.

Brennst. Chemie 37 (1956), 78. 12. Radmacher, W.

Glückauf 89 (1953), 503. 13. Radmacher, W. & Mohrhauer, P.

Brennst. Chemie 34 (1953), 33. 14. ter Meu/en, H. & Heslinga, ].

Chem. Weekb\. 19 (1922), 191. 15. ter Meu/en, H. & Hes/inga, ].

Nouvelles Methodes d'Analyse Chim. Organique Ed. Dunod, Paris 1932. 16. lnaba, T. & Abe, Y.

J.

Soc. Chem. Ind. Jap, 39 (supp\.) (1936), 91. 17. Afanas' ev, B.N.

Zavodskaya Lab. 6 (1937), 551. 18. Georgadis, C.

Thesis, Université de LiIle, 1946.

19. Smith, R.N., Duffield, ]., PieTotti, A & Mooi, ]. Ana\. Chem. 28 (1956), 1161. 20. SpooneT, C. E. Fuel 26 (1947), 15. 21. Schütze, M. Naturwiss. 27, (1939), 822. 22. Unterzaucher,]. Ber. 73 B, (1940), 391. 23. Guérin,]., Bastick, M. & Marcel, P.

Chaleur et Ind. 33 (1952), 177. 24. Georgadis, C.

J.

Usines Gaz. 76 (1952), 188. 25. Radmacher, W. & Hoverath, A.

Z. f. Ana\. Chem. 167 (1959), 336. 26. Deinum, H. W. & Schouten, A.

Ana\. Chim. Acta 4 (1950), 286. 27. Oita, I. ]. & Conway, H. S.

Ana\. Chem. 26, (1954), 600. 28. Mitchell,]. & Smith, D. M.

"Aquametry", lst ed. (1948), p. 231 Intersc. Pub\., N. York. 29. Deinum, H. W. & Goedkoop, M. L.

Chemisch Weekblad 48 (1952), 170. 30. Blom, L. & Ede/hausen, L.

Ana\. Chim. Acta 13 (1955), 120.

CHAPTER 11

METHODS OF ANALYSIS OF OXYGEN GROUPS IN COAL

§

1. INTRODUCTION.

The methods of analysis described in the foregoing chapter enable us to determine the elementary composition of the coat the main elements being carbon, hydrogen and oxygen. Starting from this knowledge two important fields for additional chemical investigation can be distinguished: a) the study of the carbon skeleton and b) the study of the periphery of the "coal molecule" i.e. the distribution of the oxygen over different functional groups. Our study is restricted to the second field.

At the time the present study was started (1952), littIe was known about methods suitable for determining oxygen groups in high rank coal. Information was obtained mainly from work done on humic acids, brown coal and low rank bituminous coal. W. Fuchs and co-workerslj2j3j4 made a detailed study of the methods for determining acid groups in humic acids and brown coal (both carboxyl- and hydroxyl groups) and applied them to low rank coals.5. F. Heathcoat and

R.

V. Whee1er6 _dit some preliminary work on the determination of hydroxyl groups in coal.

K.

I. Syskow and T. A. Kukharenko7 _{developed procedures for the} determination of carboxyl and acid hydroxyl groups in brown coal by means of ion exchange with barium acetate or barium hydroxide (in the latter case estimates on the contents of carboxyl plus phenolic hydroxyl are obtained).

In the course of Our study the work do ne by A. Ihnatowicz8 _became known to us. This author described methods for determining carboxyl, hydroxyl and carbonyl groups and employed them for examining a large number of coal samples of different oxygen content. He distinguished between two types of oxygen groups : reactive and non-reactive groups, the first type comprising carboxyl, hydroxyl and carbonyl groups and the second ether groups (occasionally in cyclic structures).

This distinction is highly questionable, however, as it is quite possible that exceptionally reactive ether oxygen occurs by the side of carbonyl groups of very low reactivity.

The work of

J.

O. Brooks and T. P. Maher9 _{on the direct titration of} acid groups in coal suspended in ethylene diamine, initiated an intensive study in this field during the past six years. The results will be mentioned in the discussion.

Finally, it should be stated here that, by the side of chemical methods, also physical methods have come to contribute to our knowledge of the functional groups in coal. The results are interesting mainly from a qualitative point of view. Particularly the infrared-absorption of co al has been studied extensive1y.

The oxygen groups which, a priori, may be expected to exist in coal are:

carboxyl,

hydroxyl (acid and non-acid), carbonyl (ketonic and quinonic), ester (mainly lactonic), peroxide (in different forms), ether (linear or cyclic).

The methods by which We intended to determine the various groups when starting upon this study will be discussed in the above order af ter which a description will be given of the development of the methods and their application to model substances. The ?pplication to coal will be described in chapter lIl, and the results mentioned there wilt as far as possible, be compared .with those obtained with different methods by other investigators.

Right from the start it was considered desirabIe to have at least two independent methods for determining each of the functional groups in order to have a check on the results.

It

is quite obvious that in the applic,ation of chemica I reactions to a substance like coat specific difflculties were to be expected. Primarily, there were the difficulties concerning the poor accessibility of the material to the reagents employed.

It

proved that in general very long reaction periods and thorough grinding of the material we re essential. Moreover, it appeared that a few solvents only were suitable for enhancing the accessibility. The solvent which, in our opinion, is to be preferred to all others is pyridine.

A second problem was the choice of representative model substances needed for checking the methods of analysis, As several of the methods chosen make use of the insolubility of the coal samples, the model sub-stances should also be insoluble. In many cases ho wever, these insoluble model-substances are much less accessible to the reagents than coal.

In the third place it had to be borne in mind, that co al, on account of its large surface area possesses definite adsorptive properties.

Finally, as We decided to study pure macerals - some of which are very hard to obtain - the amounts of samples available we re of ten very smalt As a result, we were of ten compelled to work on a semi-micro ·scale.

§ 2. CARBOXYL.

Although it was assumed that carboxyl-groups occur only in coals of very low rank, this had still to be proved. To this end an extensive investigation was made of the possibilities for determining carboxyl groups, together with a simultaneous study of oxidation products of coat The methods published in the literature were developed mainly for the analysis of humic acids and brown coat

The indirect titration methods proposed by W. Fuchsl!2t3,

I. Ubaldini

10 and H. H. Lowryll, as weU as the methods for direct titration developed by L. F. Brown and A. R. Collet12 _{and by}

1.

_{D. Brooks and T. P. Maher}9 have se rio us drawbacks in our opinion. When the co al is heated with alcoholic potassium hydroxide solution, as is done in the first group of methods, acid groups may readily form by hydrolysis and oxidation reactions. Direct titration in ethylene diamine mayalso give rise to

28

oxidation. But our main objection concerned the risk of incomplete reaction with the suspended coal.

a. Exchange with calcium acetate

Several methods are based on a process of ion exchange with salt solutions, the principle of which was, as far as I know, first mentioned by Francis and Wheeler13_{, who used various cations (Ba, Fe, Ag and Cu)} in their experiments. The insolubility of the sample is an essential con~

dition for the applicability of these methods. The carboxyl content is calculated from the number of cations disappearing from the solution. Fuchsl used calcium acetate as areagent and described a well~elaborated

method. He assumed that one calcium atom taken up should correspond with two carboxylic acid~groups (or carboxylate groups) and calculat,~d the number of carboxyl groups on this basis. His IIlethod was also applied by Ihnatowicz8, who derived the number of "free" carboxyl groups from the amount of ace tic acid liberated and the sum of the free and alkali~

bound carboxyl from the decrease in Ca

+ +

~content. Apparently both authors did not realize that in the caSe of an undissolved sample this calculation is valid only for special structural configurations in which the carboxyl groups occur in pairs. I~ the case of an isolated carboxyl group the product of.the ion exchange will have the following structure

-,/0 <[l

)c ....

oCa-O-C-cH3

which shows that one carboxyl group corresponds with one calcium atom.

Accordingly, we succeeded in establishing from experiments on some samples, that part of the carboxyl groups indeed occur in isolated positions.

At the same time we found that at least part of the carboxyl groups in coals are present as salts. Therefore, we included a pre~treatment of the coal with hydrochloric acid in our standard method, as we were only interested in the total oxygen contained in the two groups. A detailed description of the ion exchange method with calcium acetate is given below.

TABLE 11, 1

RESULTS OF CARBOXYL DETERMINATION BY ION EXCHANGE ON A NUMBER OF SAMPLES OBTAINED BY OXIDATION OF A 87%

CCOAL WITH AIR AFTER DIFFERENT PERIODS OF TIME MsC coal oxidized for

0.5 h

2

4

6

12

24

48

Carboxylgroups found, mg eq. per g.

0.15 0

.

14

0

.

27 0

.

26

0.55 0.44

0.72 0.67

0.82 0.74

1.41

1.39

1.80 1.84 1.93

3

.

19