CHAPTER - 11 A

ANALYSIS OF VARIANCE
 INTRODUCTION

Geochemical studies and related investigations are becoming more

and more dependent on Mathematical analysis and Statistical

processing of data over the last few years. This is because huge

bulk of data needs to be processed for reasoning and

■; interpretation which can be performed due tothe easy

availability of Computers.

The purpose of the present Statistical analysis "ANOVA TEST" is

to select best analytical methods for specific element/elements.

The work involved chemical analysis of a large number bore hole

samples of three different sites using different methods. Coal

samples are analysed by Wet chemical, A.A.S., Emission

Spectrograph, and XRF techniques for high precision requirements.

It is found that same samples show different concentration values

of a specific element when analysed by different methods.

Obviously, it may be a matter of confusion that which of the

methods are more dependable i.e., to be more close to each other

and which one or ones are widely different or deviating from the
i mentioned cluster. If a single cluster out ofdifferent

techniques producing more closer/nearing values are detected,

- ' then convenience of analysis is stressed upon on the basis of

economy of an analysis i.e. time and financial involvment. This

is actually an optimization technique which has been attempted.

28
ANALYSIS OF VARIANCE OR ANOVA

Analysis of Variance or Simply ANOVA is a statistical method that

determines whether the nature of two or more populations under

study are similar to each other or have a significant difference

between them. For example, sample's collected from three bore

holes, after proper chemical analysis, may be tested through

ANOVA to determine whether the chemical data of the samples hold

similarity amongst them or different. This is the problem of the

simpliest ANOVA technigue called "one way Analysis of Variance".

At first, it is assumed that the samples are of same nature and

after proper statistical calculations, it is ascertained whether

the assumption is correct or not, i.e. whether there is a

significant variance between the samples or not. However this is

not all and to deal with. More complex variance analysis i.e.

"Two way analysis of variance" technigue is also applied. For

example, if the samples collected from three different boreholes

are subjected to chemical analysis by using three different

techniques, then, it will not only be necessary to know if any

significant difference exist between the samples, but also it

will show, whether there exists any difference between the

results obtained by different techniques employed. The second

part is of utmost important to a chemical analyst because, if it

is found that results obtained by different techniques used on

same specimens are different, then it becomes quite difficult to

ascertain the correct data sets for quantitative interpretation.

In the present work, the "Two way analysis of variance" technique

has been applied to check the validity of two assumptions

29
(a) The chemical nature of the individual samples are same (b)

The results obtained by using the different techniques are also

same. In all, 9 elements are considered for investigations which

show more variability. The elements are copper, manganese,

chromium, lead, vanadium, nickel, cobalt, iron and magnesium. The

samples are subjected to elemental analysis by four different

techniques viz. X-Ray Fluoresence Spectrometry (XRF), Atomic

Absorption Spectroscopy (AAS), Optical Emission Spectrography

(OES) and Wet chemical analysis. However due to technical

reasons, not all the elements are subjected to analysis by all

the four methods. For example Cu content of the samples are

analysed by XRF, AAS and OES methods. While Ni content of the

samples are analysed by AAS and OES methods only. The following

chart will show the list of the elements and the analytical

techniques with the help of which their contents in the sample

has been determined.

. Table - 2

Methods used for the determination of different elements.

Elements Methods

1. Copper XRF, AAS, OES •

2. Manganese XRF,. AAS , OES

3. Chromium XRF, AAS, OES

4. Vanadium XRF, AAS , OES

5. Nickel AAS , OES

6. Lead AAS , OES

7. Cobalt AAS, OES

8. Iron Wet Chemical, AAS

9. Magnesium Wet Chemical, AAS

30
Statistical Analysis of the Samples

As stated earlier, some of the samples considered in the present

case study belong to Lajkura Seam while the rest are from Rampur

Seam. The chemical analysis of each samples for the nine elements

under consideration were taken for "ANOVA". After that, 'ANOVA"

test was applied for each of the elements using a statistical

programme called EPISTAT.

The test was applied in two modes :

(a) All the samples were considered irrespective of the Seam.

(b) The samples were separated Seam-wise and then the test was

applied for the Seam having the largest number of samples.

The main purpose for such a consideration is that :-

(i) To see whether the different methods give the same result

for the same set of samples and for the same element consi­

dered.

{ii) To ascertain whether individual analytical methods can bring

out the similarities in Chemical data of the samples of the

same Seam.

31
Results:

Results of the Statistical test (ANOVA) applied for nine element

is summarised below from the detailed Table {Table - 3).

Table ._r_3a

Analytical data of Cu obtained by XRF, AAS and OES *

in different coal samples.

Element_-_Cu (reports in ppm)

DATA FILE A: DA 1A DATAFILE Ai: DA - IB

SAMPL NO XRF AAS OES SAMPL NO XRF AAS OES

1.CMPDI-5 89 180 150 1.CMPDI-5 89 180 150

2.CMPDI-16 78 100 90 2.CMPDI-16 78 100 90

3.CMPDI-17 80 150 100 3.CMPDI-17 80 150 100

4.CMPDI-20 92 150 125 4.CMPDI-20 92 150 125

5.CMPDI-25 44 90 70 5.HGK-15/1 51 45 35

6.CMPDI-36 48 50 50 6.HGK-15/14 64 100 75

7.CMPDI-45 25 40 35 7.HGK-15/36 71 131 80

8.HGK-15/1 51 45 35 8.HGM-6/1 72 40 35

9.HGK-15/14 64 100 75 9.HGM-6/10 66 101 90

10.HGK-15/36 71 131 80
•
11.HGM-6/1 72 40 35

12.HGM-6/10 66 101 90

Since most of the samples (9 out of 12) are from Lajkura Seam, so ■

first, all the 1 2 sample s were studied and then only samples

belonging to Lajkura Seam were considered.

32
The results may be represented as follows :

Cu DA-1A Cu DA- IB

Seam combined(Lajkura & Rampur) Single Seam (Lajkura only)

XRF AAS OES XRF AAS OES

••

XRF - SD SD XRF -
SS SS
MD MS MD MS

AAS - SD AAS _ SD
MD MD

OES - OES -

(See Table - 3)

Where SS = Samples similar

SD = Samples different

MS = Methods similar

MD = Methods different

The study on the combined sample shows that there is no

significant different between the results obtained by XRF and OES

techniques, but the results obtained by AAS vary from those of

XRF and OES. Moreover, all the three methods can identify that

the samples are different i.e. belong to two different seams with

respect to copper (Cu) content.

On the other hand, when the Lajkura samples were considered, it

is again seen that the results obtained by XRF and OES do not

vary significantly but that of AAS is different from the rest.

Although OES and XRF can reveal that the data of the samples are

similar, AAS shows a significant difference.

In this way, the results obtained for the other elements have

been documented and are as follows :

33
 Table - 3b

Analytical data of Mn obtained by XRF, AAS and OES

in different coal samples.

E1 e ments_z_Mn(reports in ppm)

DATAFILE A:DA-4

SAMPL. NO. XRF AAS OES

1. CMPDI-12 60 75 50

2. CMPDI-17 319 225 300

3. CMPDI-23 106 75 60

4. HGM-6/2 62 60 50

5. HGK-15/14 209 275 200

6 . HGK-15/15 62 20 35

7 . HGK-15/36 124 40 100

Mn DA-4

(Only Lajkura Seam Considered)

XRF AAS OES

XRF _ SD SD
MS MD

AAS _ SD
MD

OES -

(See Table - 3)

34
 Table - 3c

Analytical data of chromium obtained by XRF,

AAS and OES in different coal samples

Elements ' Cr'(reports in ppm)

DATAFILE A : DA--5A DATAFILE A : DA-!5B •

SMPL NO. XRF AAS OES SMPL NO. XRF AAS OES

1. CMPDI-5 111 140 '150 1. CMPDI-5 111 140 150

2. CMPDI-16 103 160 130 2. CMPDI-16 103 160 130

3. CMPDI-20 110 150 150 3. CMPDI-20 110 150 150

4. CMPDI-23 47 70 50 4. CMPDI-23 47 70 50

5. CMPDI-30 .76 140 100 5. HGK-15/7 95 80 110

6. CMPDI-45 31 75 55 6. HGK-15/11 66 75 60

7. HGK-15/7 38 60 70 7 . HGK-15/15 48 60 80

8 . HGK-15/11 95 . 80 110 8 . HGM-6/1 79 90 70

9. HGK-15/15 66 75 so 9 . HGM-6/10 76 140 100

10. HGM-6/1 48 60 80

11. HGM-6/10 79 90 70

Cr (DA--5A) Cr (DA-5B)

Combined (Lajkura & Rampur Seam) Separated {Only Lajkura Seam)

XRF AAS OES XRF AAS OES •

XRF SD SD XRF - SD SD
MD MD MD MD

AAS — SD AAS — SD
MS MS

OES — OES —

{See Table - 3)

35
 Table - 3d

Analytical data of Vanadium obtained by XRF,

AAS and OES in different_coal_samp1es

Element - V(reports in ppm)

DATAFILE A:: DA - 6A DATAFILE A:DA - *6B

SMPL NO XRF AAS OES SMPL NO. XRF AAS OES

1.CMPDI-4 127 350 100 1. CMPDI-25 135 450 200

2.CMPDI-17 151 200 150 2. CMPDI-30 176 200 200

3.CMPDI-23 164 200 150 3. CMPDI-32 194 450 175

4.CMPDI-25 135 450 200 4. CMPDI-40 84 200 120

5.CMPDI-30 176 200 200

6.CMPDI-32 194 450 175

7 . CMPDI-40 84 200 120

V (DA-6A) V (DA-6B)

Combined seam (Lajkura & Rampur) Single i(Rampur Only)

XRF AAS OES XRF AAS OES

XRF __ SS SS XRF _. SS SS
MD MS MS MS

AAS - SS AAS ___ SS

MD MS

' OES — OES •

—

(See Table 3)

36
 Table - 3e

Analytical data of nickel obtained by AAS

a nd _ O ES in d if fere n t coa 1 _ .samp 1 e s

Element - Ni(reports in ppm)

DATAFILE A: DA-7A DATAFILE A: DA-7TB

SMPL NO. AAS OES SMPL NO AAS OES
l.CMPDI-1 225 175 1. CMPDI-1 225 175
2.CMPDI-5 70 60 2 . CMPDI-5 70 60
3.CMPDI-10 130 80 3 . CMPDI-10 130 80
4.CMPDI-20 100 80 4. CMP.DI-20 100 80
5.CMPDI-23 125 100 5 . CMPDI-23 125 100
6.CMPDI-33 75 90 6 . HGK-15/1 55 45
7.CMPDI-45 50 40 7 . HGK-15/7 100 85
8. HGK-15/1 55 45 8 . HGK-15/14 50 65
9. HGK-15/7 100 85 9 . HGK-15/34 220 275
10.HGK-15/14 50 65 10 . HGK-15/40 300 260
ll.HGK-15/34 220 275 11. HGK-15/50 260 210
12.HGK-15/40 50 65 12 . HGM-6/1 75 35
13.HGK-15/50 260 210 13. HGM-6/6 100 65
14.HGM-6/1 75 35 14 . HGM-6/15 90 60
•

15.HGM-67 6 100 65

16.HGM-6/15 90 60

Ni (DA - 7A) Ni (DA - 7B)

Combined seam (Lajkura & Rampur) Single Seam (Only Lajkura)
AAS OES AAS OES
AAS '- SD AAS __ SD
MD MD
OES — OES —

[See Table 3]

37
 Table 3f

Analytical data of lead obtained by AAS and

OES in different coa1_samples

Element - Pb(reports in ppm)

DATAFILE A : DA-8A DATAFILE A : DA-81?

SMPL NO. AAS OES SMPL NO., AAS OES

l.CMPDI-1 130 100 1.CMPDI-1 130 100

2.CMPDI-5 65 50 2.CMPDI-5 65 50

3.CMPDI-10 80 30 3.CMPDI-10 80 30

4.CMPDI-17 300 250 4.CMPDI-17 300 250

5.CMPDI-22 85 50 5.CMPDI-22 85 50

6.CMPDI-23 70 45 6.CMPDI-23 70 45

7.CMPDI-30. 75 85 7.HGK-15/7 70 60

8.HGK-15/7 70 60 8.HGK-15/14 100 100

9.HGK-15/14 100 100 9.HGK-15/15 20 15

10.HGK-15/15 20 15 10.HGK-15/19 25 40

11.HGK-15/19 25 40 11.HGK-15/34 75 100

12.HGK-15/34 75 100 12.HGK-15/49 225 275

13.HGK-15/49 225 275 13.HGK-15/50 310 250

14.HGK-15/50 310 250 14.HGM-6/6 90 50

9

15.HGM-6/6 90 50

Pb (DA - 8A) Pb (DA - 8B)

Combined seam (Lajkura & Rampur Seam) Single seam (Only Lajkura Seam)

AAS OES AAS OES

AAS _ SD AAS SD 9

MD MS

OES — OES —

(See Table - 3)

38
 Table - 3q

Analytical data of cobalt obtained by

AAS and OES in different coal samples

Elements 'CO'(reports in ppm)

DATAFILE A : DA-9A DATAFILE A : DA-9B

SMPL NO. AAS OES SMPL NO. AAS OES

1.CMPDI-1 55 40 1.CMPDI-1 55 40

2.CMPDI-10 65 35 2.CMPDI-10 65 35

3.CMPDI-17 60 35 3.CMPDI-17 60 35

4.CMPDI-22 65 50 4.CMPDI-22 65 50

5.CMPDI-27 65 50 5.HGK-15/7 90 75

6-CMPDI-36 60 45 6.HGK-15/14 85 35

7.CMPDI-47 20 15 7.HGK-15/24 85 120

8.HGK-15/7 90 75 8.HGK-15/34 100 125

9.HGK-15/14 85 35 9.HGK-15/49 50 65

10.HGK-15/24 85 120 10.HGM-6/19 60 70

11.HGK-15/34 100 125 11.HGM-6/22 90 80

12.HGK-15/49 50 65

13.HGM-6/19 60 70

14.HGM-6/22 90 80 •

CO (DA-9A) CO_.(DA-9B)

Combined SeamfLajkura & Rampur) Single Seam (Only Lajkura Seam)

AAS OES AAS OES

AAS _ SD AAS SD
MS MS

OES — OES —

(See Table - 3)

39
 Table - 3h

Analytical data of iron obtained by wet

chemical and AAS in dif f er ent_coa I s ampule

Element_-_Fe1(reports in ppm)

DATAFILE A : DA-3A DATAFILE A : :DA-3B

SMPL NO WET CHM AAS SMPL NO. WET CHEM AAS

1. HGM-6/11 6 4.29 1. HGM-6/11 6 4.29

2. HGM-6/5 2 1.33 2. HGM-6/5 2 1.33

3. HGK-15/15 2 1.20 3. HGK-15/15 2 1.20

4. CMPDI-44 8 5.88 4. HGK-15/48 1 0.67 •

5. HGK-15/48 1 0.67

Fe_(DA-3 A.]i Fe (DA- 3B)

Combined seam*(Lajkura & Rampur) Single Seami (Only Lajkura

Seam Considered)

Wet Chemical AAS Wet Chemical AAS

Wet SD Wet SD
Chemical MD Chemical MS

AAS — AAS —

[See Table - 3]

40
 Table_-._3i

Analytical data of magnesium obtained by wet chemical

and AAS in_different coal samples

Element-Mg(reports in ppm)

DATAFILE A : DA-2A DATAFILE A : DA-2B

SMPL NO. WET CHEM AAS SMPL NO. WET CHEM AAS

1. CMPDI-24 1 0.60 . CMPDI-39 3 4.2

2. CMPDI-34 3 2.17 . CMPDI-24 1 0.60

o
CM

3 . CMPDI-39 3 . CMPDI-34 4 . 2.17

4 . HGM-6/11 0.27 0.57

Mg (DA-2A) Mg (DA-2B)

Combined seam (Lajkura & Rampur) Single Seam (Only Rampur

Seam)

Wet Chemical AAS We.; Chemical AAS

Wet - SS Wet — SS
Chemical MS Chemical MS

AAS — AAS _

{See Table - 3)

Compilation of Results

From the mentioned statistical analyses, the relevant conclusions

have been drawn with respect to different elements which are

described sequentially.-
1. Copper

(i) the results obtained by XRF and OES are similar in

nature

(ii) Results by AAS vary from the other two methods.

(iii) All the three methods can identify the disimilar

nature of the samples wheh combined seam is considered.

(iv) When a single seam is taken, XRF and OES can identify

their similarity but not in the case of AAS

2. Manganese

(i) All the methods depict that the samples are different.

(ii) The results obtained by -XRF amd OES vary from each

other, but that between AAS & OES and XRF & AAS are

simialr.

3. Chromium

(i) When the combined seam are considered, a,ll the three

methods report that chromium content of the samples ar^

different.

(ii) Result obtained by AAS and OES are similar.

(iii) For the separated seams, it is seen that all the three
methods foretell that chromium content of the samples

are different.

(iv) Results obtained by AAS and OES are similar.

42
4.Vanadium

(i) When the combined seam are considered, samples are

similar in nature.

(ii) Results obtained by XRF and OES are similar. •,

(iii) In the case of separated seams, it is seen that

samples are similar in nature.

(iv) Results obtained by all the three methods are similar.

m
5. Nickel

Here analysis was carried out by two methods viz. AAS & OES

(i) When the combined seams are considered, samples and

methods are different.

(i"i) For the separated seams, samples and methods are different.

6. Lead

Analysis was carried out by two methods AAS and OES

(i) For the combined seams, samples and methods are different.

(ii) When the separated seam is considered, lead content of the

samples are different but the results obtained by AAS & OES

are same.

7. Cobalt
(i) When the combined seams are considered, cobalt content of

the samples are found to be significantly different but

the results obtained by AAS and OES are similar.

43
(ii) For the separated seam, the same result obtained as in the

combined seams.

8.Iron

The methods applied for the analysis of iron are wet chemical# and

AAS.

(i) When the combined seams are considered, iron content of the

samples are found to be different and the results

obtained by AAS and wet chemical methods are different.

(ii) For the separated seam, iron content of the samples are

found to be different, and the results obtained by wet

chemical and AAS are of similar i.e. different techniques

produce similar data.

9•Magnesium

{i5 For the combined seam, magnesium content of the samples are

same and the results obtained by the methods are similar.

(ii) For separated seam, same results are found as that of

the combined seams.

 INTERPRETATION OF THE RESULTS

From the above summary, it may be concluded that

(i) For the elemental analysis of copper, vanadium and

cobalt, XRF and OES give better results as both can

reveal the disimilarities between samples for a combined

seam analysis and identify that the samples are similar

when samples are collected from the single seam. Moreover,

the results given by the two methods are similar.

(ii) As for chromium, all the methods can identify the chemical

disimilarities when samples are collected from different

seams but cannot do so for a single Seam samples.

Moreover there is a similarity between the results obtained by

AAS and OES.

(iii) For manganese, though all the samples were taken from a

single seam, but it is not reflected in the result. Here,

the results obtained by XRF and AAS are similar.

(iv) For nickel, AAS and OES the results vary significantly and

also neither of the two methods can establish whether sam­

ples are taken from the same seam or not.

{v} As for lead, the results vary significantly for combined

analysis but not so for single seam samples.

From the above discussion, it may be concluded that OES and XRF

are the best methods so far as the trace elemental analyses are
concerned, because both can identify the chemical dissimilarities

of samples of different seam and the chemical similarities of the

samples of the same seam, the only exception is in the case of Pb.

45
Of the two methods i.e. XRF and OES, XRF is comparatively cost

intensive and it is not possible to carry out analysis of a huge

number of samples by this' technique in a short time which can

effectively be done with the help of OES more rapidly. Since an

excellent agreement has been found through statistical studies’ of

the analytical results obtained from OES, XRF and AAS, it may be

recommended that OES may also be used where accurate quantitative

«

estimation of trace element analysis is needed. Such a study has

not been attempted earlier. However, this technique cannot be

applied for the elements like iron and magnesium which ai*e

present in the samples in large amount. Generally, OES technique

is not suitable when the concentration of the element is in

percentage level. In the present study, analyses of iron and

magnesium were done by both wet chemical and AAS methods. It is

seen that the results given by the two methods are more or less

the same for both Fe and Mg, except when iron is considered for

combined seam samples. It is also found that iron content of the

samples are significantly different for both combined and

separated•seam samples but magnesium content is similar for both

the set of samples considered. However, wet chemical method is a

classical and one of the most reliable methods.

Instrumental analysis is no doubt a better choice where a large

number of analysis is required in a short time. Since in the.

present study (for iron and magnesium) only a small number of

samples have been analysed, wet chemical analysis should be the

better choice.

46
CHAPTER -MB

FACTOR ANALYSIS
 INTRODUCTION

Statistical techniques were applied to chemical data obtained on

10 samples of coal containing 31 major, minor and trace elements.

•
The purpose of the present work was to apply the statistical

techniques of multi-variate correlation and factor analysis on

data accumulated for 10 samples of coal (Table 5) with an attempt

to quantify the relationship that exists between chemical

variables and to determine the number and nature of underlying

•
geologic factors which monitored these variables. The present

approach has been aimed to arrive at better logistic conclusions.

CORRELATION

The computed symmetric correlation matrix for all data is

included in Table - 6. The diagonal where all values are 1.000,

represents the perfect correlation of each variable with itself.

Both negative and positive correlations are present, with the

confidence factor at 5% significance level of probability equal

to 0.335 and at the 1% level equal to 0.430. Those values which

are of greater than 5% level are only high light and considered

in Table 6. The 5% level of probability means that there are only

five chances in 100 that the correlations observed are by

coincidence or accident, and the more precise 1% level means

there is only one chance in 100.

47
 Table-4

FACTOR ANALYSIS PROCEDURE FOR

10 COAL SAMPLES

data matrix
(10 samples x 31 variables)

*
TRANSFORMATION
(to achieve normality)

I
CORRELATION MATRIX
(31x31)

{
FACTOR LOADINGS

ROTATED FACTOR LOADINGS

*
INTERPRETATION
OF RESULTS

FINDINGS.
AND
CONCLUSION
PRINCIPAL COMPONENT ANALYSIS

AND R-MODE FACTOR ANALYSIS :

This section is an exploration into the possibilities of using

factor analysis techniques in order to simplify relations

•

existing between certain chemical variables of coals, which

hopefully, would provide some understanding of the underlying

nature of these variables. This analysis is based upon

mathematical methods of principal components which established an

Orthogonal Co-ordinate scheme and upon subsequent rotation of

«

these orthogonal axes to a new position, by the "Varimax method"

(Harman 1960, P. 301)•■ Rotation was performed to help in the

interpretation of underlying factors. The details and criticisms

of the method are available in Krumbein and Graybill (1965) and

Matalas and Reiher (1967) who listed papers on the original

development and application of the technique.

CORRELATIONS BETWEEN ELEMENTS :

The pattern of association of SiC>2 content shows a high posi­

tive correlation with AI2 O3 (0.97) and Ti02 (0.94), with K9 0

' (0.16) and with Cr (0.49) ; but a high negative correlation co­

efficient with F2 O3 (-0.992) and pyritic sulphur content (-0.54)

(Table - 6).

The AI2 O3 content varies between 22 and 27% with the exception
of one sample containing 3.64% of AI2 03, with a mean of 22.85%

(See Table - 5). Both values lie within the ' range given by

previous authors (Gorman and Walker, 1971, Ruch et al 1974). When

48
the values of AI2 O3 content are plotted against SiC>2 content,

they show a systematic variation indicating the same origin,

which strengthens the inference of a possible clay mineral source

more particularly when presence of Kaolinite and Illite were

confirmed by X-ray diffraction study. (Fig - 1).

Titanium oxide content varies between 0.7 and 4.6% with a mean of

3.67%. The world-wide average Ti02 content in Coal ash is

believed to be 0.05% (Volkovic, 1983). According to Finkelman

(1980), most of the titanium oxide is found mainly in many

authigenic clays, supposedly in kaolinite minerals. In support of

the later suggestion, the titanium Oxide contents of IB-river

coals show similar distribution trends with that of both Si02

content (r = 0.94} and AI2 O3 content (r = 0.96) suggesting its

presence in clay minerals.

The highest chromium concentration recorded in the IB-river coal

samples was 150 ppm (within the 10 samples) with a mean of 85 ppm

Cr. The highest concentration in South Wales coals was reported

to be 500 ppm Cr with a mean of 147 ppm cr (Chatterjee and

Pooley, 1977). It was also reported that in British coal asjhes,

Cr ranges between 114 and- 177 ppm (Abernathy and Gibson, 1963).

The relation of Cr with both SiC^ and Al2 O3 contents seems to

support the view that Cr is largely associated and/or sub­

stituted in the clay-mineral fraction ; kaolinite is thought

to be the most likely site. The weak correlation of chromium with

Potassium (r = -0.04) dismisses the possibility of an illite

association. (Asuen, 1987)

49
Although the correlation coefficient between K2 0 content and S*#

Si02 content reveals low weightage (r = 0.16), its correlation

with the ash content (r = - 0.12) shows a similar trend. The range

of potash for these coals lies between 0.09 and 9.60% K2 O with a

mean of 1.84%. The K2 O values show a low positive correlation

with Si02 content, indicating a possible clay mineral source.

Potash content in coal is highly dependent upon the abundance of

clay minerals and its presence generally suggests an important

source of illite since kaolinite possesses no available lattice

site open to potassium. The lack of control of potash by

Kaolinite is shown by the low correlation with alumina in these

coals. The low correlation Coefficient between K2 O content and

ash content appears to be related tc a non-sedimentational phe­

nomenon. It is apparent that the variation and distribution of

potash is a reflection of a Swamp paleo-environment and/or

source-area conditions.

The total iron content, which ranges between 1.00 and 6.8% Fe2 03

{high'value of one sample, 64.8% of Fe2 03 which is not considered

and taken as local variability) with a mean of 3.45%, shows a

significant association of Fe with pyritic sulphur (r = 0.54).

This pattern of variation appears to be influenced by the

abundance of sulphide minerals as the total iron content

increases, the content of pyritic sulphur also increases. The

relationship suggests a genetic association involving a

contribution of iron of more than 70% towards the sulphide

minerals formation. The calcium content in coal is most

frequently referred to an association with calcite, dolomite and

50
ankerite (Pringle and Bradburn, 1958, Gorman and Walker, 1972,

Ruch et al. 1974). The range for calcium in these coals is

between 0.47 and 12% CaO with a mean of 5.9%.

The relationship between CaO with Fe2 03 is 0.136. The range ft>jr

manganese is between 0.01 and 0.06% MnO within the 10 samples of

IB-river which taken for factor analysis. The fact that the two

variables showing a markedly positive trend, indicates that they

have some genetic connection. The relationship between the MgO

with Fe2 02 is 0.76. Evidence of both a MnO and a MgO relationship

with CaO content is an indication of the presence of mixed

carbonate minerals. The modes of association of these oxides are

probably the result of divalent characters common to both, which

is likely to have favoured a considerable interchange of their

elements.

Relationship between carbon and inorganic elements have also been

examined to determine the extent of chelating and/or the extent

of a contribution by inorganic to coal constituents. Curties

(1966) observed that V, Cr, Mn, Ni, Cu, and Mo show correlation

with organic carbon in Carboniferous Coal Measures Strata , whj.le

Gad et al.(1969) stated that Cu, Ni, V, Ag and Mo are associated

with organic carbon in the Whitbian rocks of the Yorkshire coast.

The work presented here accounts the elements Ni, Zr, Sc, Mo, Cu,

Ga, Mn, Nb, Y, Ba, Pb, Cr, La, Co, Sr, Ge possibly associated
with the organic fraction (i.e. forming humates or chelates).

Based on the correlation coefficients, Mo and Cu in this study

show a positive correlation with carbon. Additional evidence for

this relationship possibly lies in the high negative correlation

51
with the ash content (i.e. for Mo = -0.41, Cu = -0.44). it is

thought that organic matter produced complexes with the elements

(i.e. chelates). The association of these elements with carbon

may have resulted from their direct incorporation into fhe

organic Constituents or perhaps as numates (Guy et. al. 1975). In

a nutshell, it would appear that the retention of Mo and Cu in

coal basin could be due to the formation of stable and insoluble

complexes (Zubovic, 1966). Despite the evidence that the above

elements are organically bound, there is an indication that softe

are inorganically bound in view of the levels of correlation

coefficient quoted.

PRINCIPAL COMPONENTS ANALYSIS SOLUTION

Four components accounted for 79.4% of"the total variance in the

original data set for the IB-River coals. The vector loadings

relate strengths of association of only 31 major and minor

elements which are either present in the coal as detritus,

organically bound to the plant material, and/or inorganically

bound. Table 7 and 9 shows the maximum vector loadinqs of the

variables on these four components.

Component 1 shows high positive loadings for the major Oxides

Si09, Al2 O3 and TiQ2 and the minor element Zr, Sn, Mgo and Mno.
This group is controlled by the inorganic phases in the

depositional peat-forming environment as indicated by the clay

mineral and detrital fraction. The clay minerals, dominated by

kaolinite, confirmed by XRD which formed under stable conditions

/

have promoted sorption of detrital fractions. Negative vector

52
loading for Fe2 0-j and pyritic sulphur indicate euxhinic chemical

conditions which is quite different from those for the clay­

forming elements. High value of one sample of Fe2 0^, accounts, for
■

local variation.

Component 2 shows high positive loading for the minor elements

Cu, Pb, and La, representing the elements coexisting in time and

place with organic matter, it was shown earlier that several, if

not all of these elements, were concentrated during decay of the

peat-forming plant remains when the organometallic complexes were

formed. If they are to remain stable, they are usually insoluble

complexes which are also more rapidly precipitated in peat

environments {Zubovic 1966).

Component 3 dominantly controls high negative loading of ash with

high positive loading of Moisture and Fixed Carbon along with

moderately high V.M. value. These are related to organic source

material and diagenesis of coal humas. Subdominant sulphur valde

has also revealed anaerobic bacterial activity in peat swamp

leading to precipitation of free Sulphur.

Component 4 shows the major oxide CaO, together v?ith MgO and MnO

that.indicate a mixed carbonate.fraction. It is worth noting that

the oxides are present in significant amounts in this mixed

Carbonate.

53
 Tcx\dL-C - %Q^

Vo-.'f CcxN^c-e -€->« pl.oJt»v\«.cL \o^ Covv%po'A4£.wt/i ,

ft
1 2 •_'> 4 5

11.169 7.655 3,597 2.190 "..99?

•
6 -7 8 9 10

1.801 1.464 1 O'. 733 0..395 0.000

11 12 13 -14 15

0.000 0.000 0.000 0.000 0.000

16 17 18 .19.
•
0.000 0.000 0.000 0.000

TexU-e- 8b

cxivs-^d. ,
Befoev^t of k.okoX Yccr\.<xv\c«_

1 2 3 4

36.028 24.693 11.604 7.063 6.442

6 7 ' 8 9 10

5.810 4.721 2.365 1.274 0.000

11 12 13 14

y 0.000 0.000 0.000 0.000 0.000

16 17 18 | 19 20

0.000 0.000 0.000 0.000 0.000

 ~\ Q-laVe, - ^
Roto-Wd- R- Cow^povrewt 'YvxcxtvCx of fc^c-bov IocxACyw>
joY lO OocxX /^CX-Vv-n\R-R> ,

- p A CT O R, S> -
VAR'AfittS
r
F, ‘ “
Ft *4
«ASHk O' 3B2i r 0’ /87 - O’ 9/A n- 003
MOISTURE -O'Sid 0 * 004 0 • 93? 040/
V. M ~0'43o * ^3£~ . 0'<o0? n' no
4. F.c 0-089 O’ A70 O'8^8 - o-113
0; 7-43 0 ■ //?. - 0' 093 0' 0 84
S vOj_
rV /S? 9’ 060- 0’ U3'
A^t.Q-3 ' 0’?&r
- 0' /34 O' 041___ - O' 008
- 0*
0.' 2,61 0'4S2 - 0 '144
!^0 -O'^ei
’0vooa - O'^o'y
CjxO -O' f$4 - 0r 302,
-O' let
'• ■ Ho^D 0 * 0 A3 - 04 64 a-oer
’ K2.O O'2333 - 0‘l%f 0' /*j 0 - teo ;
TIO2,' 0--J& O' /eb - 0- /re O' /$c
MyO -O'ew O' /-f3 0' 3^2, O' /?2-
P-2_0^- -O'QGf - O' 32$~ O' 3,4-4 0-040
S -O'T/o -0'/03 O'OfS' O'^Si
Cu O'064 O' 864 0'3}0 -O'Of 3 .
JsiC
O'309 ■ O'So 3 O' 0J3 - O’O//
Co O' IZ<f ' 0’30$~ o- ATC - 0'037'
Cy O' 3re 1 O' $$2 O’ 3^2, - 0’ .3^7
-..t.iTf - O'2,<tsj
V _ • O' $or O' lid
__________ -T Q *■ Oil ___ OAjrs - O'o\8o O • f7 <nr
R-a. ___ 0-^0 04^4
_ 0’ S=f* 0’/r9
Sy O'£66 ____aioyt O' 0^4-
O'0 16
Mb O'$00 _ ^'37^ - O' 48$ 0'3Gi'
_______ _y - 0" Rf ____ P ‘! SB 0 ■ 6SS~ O'/ £3,
* Lxx» O'300 ____ Cll.^ 0'/84 O' /03
..._2y Q-Qlz ____ C ' 4 $6k__ - 0 • /ra 0*2,04
Mo - C$72 ____Q* 0/3 0'SS9 -O'SOS,
Sc_ 0'3JSb : O'30! 0-3*13 - 0 * J4-R
j~ ’. Gx -O' /$$ O’3d 2, a'4-24
—t—/— O'<210
j S'n - O'60S' - ..O'$$4 - O'2^3 O '302,

5 /■ 4 /a^4
"VV\ 4-\ cov,c v • 3

1
ji

1 UvAi ^yvAp C. o^v^p p

'
*9 3o
FACTOR ANALYSIS SOLUTION

In order to further strengthen and make the idea clearer, the

significant relationships between the geochemical parameters,

factor ("Varimax") analysis has also been used. The pattern* of

the elements are explained below

F The factor accounts for the environment of deposition of

1
clay mineral and detrital fractions comprising SlC> f AI 2 2 O ,
3

Ti02 and Cr. The negative vector loadings for MnO, Fe 2 O 3

and Sulphur show inverse relationships with them an.d

indicates that they were formed under reducing conditions.

F Although chelate fractions comprising Cu, Pb and -La bonded

2
to coal material are indicated by this factor, evidence

of inorganic association has also been shown for these

elements. Negative vectors for ash and K90 represent detri­

tal influx of inorganic matter.

F Factor accounts for organic source materials having

3
antipodal relationship with inorganic materials (Ash).

This clearly indicates coal formina condition in peat

•
swamp and later change in rank by metamorphism where

V.M shows inverse relationship with fixed carbon.

F Carbonate minerals are demonstrated by the positive

4
loadings for CaO, MgO. These minerals might have taken in

Uranium by sorption. The mechanism by which Uranium is

concentrated in coal by the association with carbonate has

been proposed (Breger, 1974).

54
 Toble-10. Diagonalized and simplied mair»x rotated,
factor loading for R mode.

FACTORS
VARIABLES F, F2 F3 Ft

Fe203 + + + +

ai2o, 0 O O O

Si02 O O O O
Ti02 O 0 O 0
P2O5 + + + +

V. M O O
MgO
MnO
Zr
Sn

Cu OO OO
Pb oo oo
La oooo

Sc oo

F.C
Ash
Moisfure

S oo
Y oo
CaO
-Ve +V&
oooo ).0O
oo .60 — -BO * *

The -most significant, aspect of Table-10 is the loading

values - whether there are two or four pluses ana zeros*
The pluses or zeros represents positive or negetive
values carried through mathematically ff om the corre­
lation chart. Whether they are actually positive or
negetive values is not important. All signs present
in -table-10 have been rtiversecJ from thu calculated
values far ease of discussion of the proposed factors.
This can be done 03 long as all pluses or zeros calcula­
ted are changed to the opposite sign. This is in effact
inverting the matrix or "Looking at i tr from another
positi on . "
 SUMMARY

Chemical considerations

The associations of major and minor elements throughout the coals

•

have frequently been shown to be complicated by the fact that

sometimes an element may be associated with more than one

mineral. The correlation coefficient data do not confirm a total

dependency of the elements on either the minerals or the organic

matter. The association of major elements in coals has been shown

•
in this study to have occurred predominantly during the second

phase of coalification. All those elements, which are related to

one another in a number of ways, are derived from the clay mine­

rals, carbonates, pyrites etc. found in these coals. Although

their mode of formation is obvious, interest also lies in the

associations of the considerable number of minor elements.

Throughout these coals, the principal modes of association of the

.minor elements Sn, Y and Ag have developed through mineral

sorption. Consequently, the degree of enrichment of the minor

elements is related to the absorptive and/or adsorptive

capacities of minerals. .

The distribution of other minor elements i.e. Ni, Zr, Sc, Mo, Cu,

Ga, Mn, Nb, V, Ba, Pb, Cr, La, Co, Sr and Ge particularly those

that can be refered to the first phase of coalification, tend to

constitute a more satisfactory reference because of the

sensitivity of these elements to swamp paleo-environment. Their

association with the organic matter of the coal (i.e.

organometallic complexes) is believed to be associated rather

55
differently, hence the difference in level of correlation

coefficients is obvious.

Statistical considerations

Principal component analysis and factor analysis given for major

oxides and minor elements do confirm in part, the mode of

formation and in places, leads to providing the basis for

suggesting three significant modes of concentration and

association : (1) biological fixation, (2) minerals sorption, and

(3) localized concentrations.

(1) Biological fixation covers those inorganic components which

accumulated either through detrital and/or authegenic

processes during the peat stage. Their concentrations and

stabilities are both linked with the • elements held by the

stage of organic matter and inorganic species. The organic

and inorganic parameters are perhaps of equal importance in

the final distributions since statistical analysis did not

disclose, the extent to which the minor elements were

associated (i.e. with either organic matter or inorganic

matter).

(2) Mineral sorption has been shown to be linked with major mine­

ral components, i.e., Si02 - AI2 0^ - TiC>2, Fe2 O3 - S, and

MgO - CaO - MnO, whose adsorptive, and/or cation exchange

capacities are apparently dependent on factors such availabi­

lity.of the inorganic source and other physico-chemical

conditions such as Eh, pH etc.

56
From the swamp paleo-environmental point of view, the clay­

forming constituents and their associated minor elements

fall into different classes which may relate well to the

different localities. Clay-mineral control of minor elements

is believed to be related to the influence of localized

mineralization episodes.