- Email: [email protected]
Quantitative analysis of glasses used within Australia
Forensic Elsevier
Science Scientific
International, 25 (1984) Publishers Ireland Ltd.
QUANTITATIVE
K.W. TERRY,
ANALYSIS
A. VAN RIESSEN,
19-34
19
OF GLASSES USED WITHIN AUSTRALIA
B.F.
LYNCHa
and D.J.
VOWLES
School of Physics and Geosciences, W.A. Institute of Technology, Kent Street, (Western Australia) and aForensic Science Laboratory, Government Chemical tories, Plain Street, Perth (Western Australia) (Received (Revision (Accepted
August 8th, 1983) received December January 25, 1984)
29,
Bentley Labora-
1983)
Summary An initial 134 glasses have been collected from eleven classifications of glass used within Australia. These include both local and imported glasses. Quantitative elemental analyses of the glasses have been determined using a scanning electron microscope equipped with an energy dispersive X-ray spectrometer. The current program provides for an elemental analysis for Na, Mg, Al, Si, P, S or Pb, Cl, K, Ca, Ba or Ti, V, Cr, Fe, Mn and Zn expressed as oxides, and has a sensitivity down to approximately 0.1%. The data for the six most commonly occurring elements, namely, Na, Mg, Al, Si, K and Ca, together with the refractive index are presented for each class of glass in terms of their mean value and standard deviation from the mean, and also in histogram form. Key words: probe
Glasses;
Australian;
Quantitative
analysis;
Energy
dispersive
electron
micro-
Introduction
There have been many attempts to classify glass fragments based upon the refractive index of the glass or its elemental composition. Two recent sum.. maries of these schemes are by Catterick and Hickman [l] and Hickman et al. [Z] . It is seen that much of the effort to date has been directed to the analysis of trace elements in glasses using such techniques as neutron activation analysis [3], emission spectrography [4], atomic absorption spectrometry [ 51, spark source mass spectrometry [6] and X-ray fluorescence spectrometry [ 71. Our approach has been to analyse quantitatively for the major and minor elements that constitute the glass network formers and glass modifiers. This has been achieved using a scanning electron microscope (SEM) equipped with an energy dispersive X-ray spectrometer (EDS) which results in sensitivities down to approximately 0.1%. The advantage of such a scheme is that all elements are analysed simultaneously in an analysis time of 50 s. Furthermore, the glass fragment is mounted in a resin block so is still available as a court exhibit if necessary. 0379-0738/84/$03.00 o 1984 Elsevier Scientific Publishers Printed and Published in Ireland
Ireland
Ltd.
20
The glasses used within Australia are largely locally made. However, there is a significant volume of imported glass from a wide range of manufacturers and countries. A summary of the origin of glass imports obtained from the Australian Bureau of Statistics for 1979 is given in Table 1. An initial collection by the authors of 134 glass samples have been catalogued into 11 categories namely, float, sheet, patterned, wired, windscreen, container, lightbulb, spectacle, tableware, labware and headlamp. The origin of these samples is given in Table 2. Our existing holding of glasses is far from being a full representation of the sources of Australian glasses, but is based upon those commonly available in Western Australia. The number of categories selected contrasts with the classifications reported in the above reviews. Previous workers have typically grouped all the flat glasses together or else as two groups of window and vehicle glasses. A statistical analysis of the appropriateness of these 11 categories in a classification system for glasses is detailed in a following paper [ 81. In this paper we present the mean, and standard deviation from the mean, of the six most commonly occurring elements, namely, Na, Mg, Al, Si, K and Ca, together with the refractive index for each class of glass and their country of origin subdivisions. In addition the range of certain classifications are presented in histogram form. It is of interest to compare these results of the Australian scene with the British ones previously published by Catterick and Hickman [ 11. TABLE
1
SOURCES OF Statistics 1979)
GLASSES
IMPORTED
INTO
AUSTRALIA
(Extracted
from
Bureau
Glass type
Country
Float Sheet Wired Patterned Windscreen Spectacles Headlamps Lightbulbs Containers Tableware
3, 4, 9, 10, 15, 24, 25, 31, 32 3, 5, 9, 10, 13, 16, 17, 19, 21, 22, 25, 26, 31 3, 10, 16, 22, 25, 31 3, 10, 16, 22, 25 10, 16, 24, 26, 31, 32 10, 16, 31, 32 4, 9, 10, 16, 26, 31, 32 2, 3, 4, 9, 10, 11, 12, 15, 16, 18, 28, 31, 32 3,6, 7, 9, 10, 12, 15, 16, 17, 18, 19, 25, 27, 31. 32 1, 2, 3, 4, 5, 6, 7, 8,9, 10, 11, 14, 15, 16, 17, 19, 20, 22, 23, 25, 26, 27, 28, 29, 30, 31, 32, 34 3, 5, 6, 7, 9, 10, 12, 16, 18, 19, 27, 31, 32
Labware
of
of origina
a 1: Argentina, 2: Austria, 3: Belgium-Luxembourg; 4: Canada; 5: China; 6: Czechoslovakia; 7: Denmark; 8: Finland; 9: France; 10: W. Germany; 11: Hongkong; 12: India; 13: Indonesia 14: Ireland; 15: Italy; 16: Japan; 17: Korea; 18: Netherlands; 19: New Zealand; 20: Norway; 21: Philippines; 22: Poland; 23: Portugal; 24: South Africa; 25: Spain; 26: Sweden; 27: Switzerland; 28: Taiwan; 29: Thailand; 30: Turkey; 31: UK; 32: USA; 33: Yugoslavia; 34: Hungary.
21 TABLE 2 ORIGINS OF GLASS MUSEUM SAMPLES Country of origin
Argentina Australia Belgium Canada China Czechoslavakia Italy Japan Philippines Poland Portugal South Africa Spain Thailand U.K. U.S.A. W. Germany Unknown TOTAL NO.
Classification
9 1
7 5
1
25
4
1 1
1 1
5 1 1
1
1 2
2 1 1
7
1 1 1 1 3
1 3 2
10 1 14 20
7
14
8
14
37
4
1
4
8
1 1
1 1 1
1 7
9
6
7
1 47 6 1 2 2 4 14 2 1 1 1 1 2 7 13 3 26 134
Experimental Sample preparation A piece of each glass sample was mounted using a polyester resin in a halfinch diameter acrylic tubing. The sample was then ground and polished prior to being coated with a vacuum evaporated thin film of carbon. Glass fragments down to 0.2 mm size, or 20 pg, have been prepared successfully using this method. Refractive index (RI) determination The RI of each glass sample was determined using a Mettler FP5 heating stage controlled by a Mettler FP52 temperature programmer in conjunction with an optical microscope. Fragments of each glass were immersed in Dow Coming DC 710 silicone oil and the mean temperature obtained of the disappearance and reappear-
22
ante of the Becke line in sodium light. Reference the oil provided the value of the refractive index.
to the calibration
chart for
SEMIEDS instrumentation and procedure The elemental analyses have been carried out using a JEOL JSM 35C scanning electron microscope equipped with an United Scientific lithium drifted silicon detector and associated electronics. The spectra are accumulated in a Tracer Northern TN 1705 multi-channel analyser (MCA). For elemental analysis an accelerating voltage of 20 kV and a current of 5 X 10-l’ amps was used at normal incidence onto a 40 X 40 I.tm raster on a polished sample and a detector take-off angle of 35”. A typical count rate of 5000 cps is obtained during an analysing live time of 50 s. The choice of voltage is a compromise between the 25 kV required to maximise the peakto-background ratio in our system [9] and a low voltage to minimise the A minimum voltage of 15 kV is magnitude of the matrix corrections required for adequate excitation of zinc. It has been found that when using 20 kV negligible variations in analysis occur upon varying the carbon thickness on a sample. The precision and accuracy of SEM/EDS systems is discussed by Reed and Ware [lo] and Ware [ 111. The choice of a small raster rather than a spot mode analysis was to minimise the loss of sodium counts during analysis. The spectra are collected in the MCA and are then stored on a stereo cassette for subsequent batch processing on a PDP ll/lO computer. The elemental analysis is subsequently computed using the peak integration with background subtraction (PIBS) technique of Ware [ 111. The current program provides for an elemental analysis for Na, Mg, Al, Si, P, Cl, S or Pb, K, Ca, Ba or Ti, V, Cr, Fe, Mn and Zn expressed as oxides and has a sensitivity down to approximately 0.1%. The final analysis adopted is the mean of two analyses obtained on two different occasions. During each run of analyses NBS Standard Glass SRM 621 has been included as an internal control. Results and discussion Accuracy that has been used to calculate the The PIBS program of Ware [ll] quantitative elemental analysis of each glass requires certain calibration factors to be determined using standard samples. Four glasses from the range of NBS Standard Reference Materials and nine mineral samples have been used to calibrate the program. The calibration has been checked against some glass samples provided by Pilkington ACI. Their analyses and ours are given in Table 3. These data indicate a satisfactory level of accuracy for the technique. Precision The duplicate analyses of the 134 glass museum samples has taken place over a period of 3 months. During each analysing session NBS glass SRM 621
23 TABLE
3
COMPARISON
OF ANALYSIS
FROM
PILKINGTON
AC1 GLASSES
-
Sample
Analyst
Na,O
MgO
Al,O,
SiO,
SO,
K,O
AC1 1
AC1 WAIT
13.70 13.4
3.80 3.7
0.94 0.9
74.01 72.8
0.25 0.3
0.16 0.2
9.52 8.8
AC1 2
AC1 WAIT
14.06 14.6
0.16 0.1
1.72 1.6
71.92 72.1
0.03
11.53 11.4
0.03
0.05
0.2
AC1 WAIT
13.73 14.0
0.16 0.2
1.77 1.7
72.93 73.1
0.27 0.3
10.51 10.4
0.03
0.10
0.3
AC1 WAIT
13.23 13.7
0.28 0.2
1.77 1.7
72.95 73.3
0.15 0.1
10.99 10.8
0.05
0.07
0.2
AC1 WAIT
13.97 14.4
0.29 0.3
1.74 1.7
72.51 72.6
0.03
10.87 10.6
0.04
0.05
0.3
AC1 3
AC1 4
AC1 B
Co0
TiO,
Fe0 0.16
has been included as an internal control. The mean, standard deviation and coefficients of variation of the 15 sets of analysis from NBS 621 over the collection period are presented in Table 4 for each element detected. The large coefficients of variation for MgO and SO3 are due to their content being at the limit of detection for the technique. The values for the other elements indicate that the technique has a satisfactory level of precision. Museum catalogue For each of the glasses in our collection the following information has been recorded: museum catalogue number, origin of sample, classification, colour, thickness, fluorescence, refractive index, and elemental composition for up to 15 elements expressed as oxides. A summary of the data is presented in Table 5 in which the refractive index values and the values for the six most commonly occurring oxides are tabled. Although the data in many cases are not normally distributed the mean and standard deviation from the mean is provided for the appropriate subgroups of each
TABLE
4
PRECISION
Mean Std.Dev. C.V. (%) NBS data
OF RESULTS
ON NBS 621 OVER
3 MONTHS
Na,O
MgO
40,
SiO,
SO,
K,O
CaO
12.62 0.30 2.4 12.71
0.2 0.06 28.6 0.27
2.77 0.05 1.8 2.77
71.34 0.28 0.4 71.14
0.13 0.08 61.5 0.13
2.02 0.04 2.0 2.01
10.76 0.18 1.7 10.71
Country
Australia Belgium USA Total
Japan Philippines Poland Total
Australia Belgium Japan W. Germany Total
Australia Japan Total
Glass class
Float
Sheet
Patterned
Wired
FROM
OF DATA
5
SUMMARY
TABLE
1 7 8
7 5 1 1 14
9 1 10 20
No.
1.5196 1.5212 1.5210
1.5192 1.5238 1.5229 1.5252 1.5215
1.5145 1.5136 1.5164 1.5147
1.5194 1.5194 1.5208 1.5201
(21) (20)
(29)
(12) (23)
(9)
(2)
(20) (16)
(6)
USED WITHIN
Refractive index
GLASSES
13.9 13.1 13.2
13.6 13.5 12.6 14.1 13.6
13.4 14.4 13.6 13.6
13.9 13.7 13.7 13.8
Na,O
(7) (7)
(4)
(2) (4)
(4)
(3)
(2) (2)
(1)
AUSTRALIA
(1)
(1)
(2) (1)
(1)
3.7 2.2 (12) 2.4 (12)
3.4 (2) 3.9 (9) 0.7 2.3 3.3 (10)
3.6 3.6 3.4 3.6
3.7 3.6 3.6 3.6
MgG
0.6 1.6 1.5
1.2 0.4 1.6 0.8 0.9
1.8 1.6 1.2 1.7
0.1 0.1
Al&‘,
(1) (3)
(5)
(3) (3)
(2)
(1)
(1) (1)
72.8 72.0 72.1
72.7 71.9 72.3 71.5 72.3
73.1 72.4 73.4 73.1
73.2 73.3 73.1 73.1
SO,
(5) (6)
(7)
(5) (8)
(4)
(3)
(8) (6)
(3)
0.1 0.8 (1) 0.7 (2)
0.1 (1) 0.2 (2) 0.6 0.5 0.2 (2)
0.7 (3)
0.9 (0) 0.3
40
(4)
(2)
(2) (2)
(2)
8.6 10.0 (14) 9.8 (14)
8.5 (3) 9.8 (5) 11.5 10.8 9.3 (10)
7.0 7.0 8.1 7.1
8.9 8.8 9.0 8.9
CaO
1.5201 1.5214 1.5222 1.5213 1.5154 1.5208 1.5207 1.5196 1.4772 1.5626 1.5126 1.5232 1.5234 1.5174 1.5175 1.5488 1.5325 1.5146 1.4776
7 6 3 6 2 13 37
2 5
1 3
1 7
4 7 1 1
1 5
Aust. NSW Au&. SA Aust. TAS Aust. WA Aust. Vie Various Total
Soda-lime Borosilicate
Lead Others
Alumina/lead Others
Australia Total Lead High alumina
Soda-lime Borosilicate
Headlamp
Lightbulb
Spectacle
Tableware
Labware
1.5195
Container
63
Total
1.5182
Flat
14
Total
Windscreen
(91)
(11) (33)
(7)
(26)
(11) (14)
(10) (8) (12) (5) (0) (30) (23)
(30)
(26)
(3) (9) (2) (2) (0) (12) (12)
(5)
(4)
(4)
13.8 4.1
15.5 15.0 1.0 1.3
(7)
(2) (6)
2.5 9.9 (14)
7.7 16.1
14.5 (11) (3) 4.5
13.3 13.5 13.8 14.0 17.1 13.7 13.8
13.5
13.1 (8)
(5)
(3)
(3)
2.8 0.1
8.9
(1)
0.8 (11)
0.2
3.3
1.7 (14) 0.2 (2)
0.2 (2) 0.3 (1) 0.1 (1) 0.6 (1) 3.4 (0) 1.4 (12) 0.8 (11)
3.3
3.3 (7)
(5)
(0) (1)
(8)
(3)
(3) (2)
3.3 2.9 (14)
19.7
1.4 1.4
12.2 1.1
1.6 1.8
1.6 2.1
1.6 (1) 1.8 (6) 1.4 (0) 1.6 (1) 1.5 (0) 2.2 (11) 1.8 (7)
0.8
1.0
70.5 74.6
73.3 72.9 65.9 62.8
71.1 70.5
62.5 72.0
70.8 73.8
73.3 72.6 72.8 72.4 70.4 71.6 72.2
72.7
72.7
(34)
(4) (8)
(14)
(9)
(7) (35)
(5) (10) (4) (2) (3) (17) (14)
(7)
(6)
(1)
(3)
1.9 0.8 (11)
0.2 14.0
7.2 (18)
0.7 (2)
2.2 (7) 0.2 (2)
0.9 (7) 0.5 (6)
0.1
0.2 (1) 0.6 (8)
0.3 (4)
0.6 (4)
(7)
5.2 0.3
(2)
9.5 (2) 9.0 (12) 0.2 6.8
8.1 (12)
8.9 5.9
6.7 (14) 0.1 (1)
11.1 (1) 11.0 (4) 11.5 (5) 10.9 (2) 5.3 (1) 9.7 (14) 10.3 (16)
8.9 (11)
8.9 (10)
26
classification such as country of origin, or type of glass. This choice of presentation provides a compact listing of the seven variables of the various sources of glass products available within Australia. It is recognised that the range of composition of each classification is of interest to the forensic scientist and is partly provided in histograms later in this paper. A full listing of the catalogue is available from the authors. An inspection of the data for each sample enables it to be typed as sodalime-silica glass, borosilicate glass, lead glass, or high alumina glass. Characteristic of the borosilicates is their refractive index value of less than 1.50. A further characteristic of all the borosilicate glasses examined is a soda content of less than 5% and a lime content of less than 0.6%. The lead glasses have, besides their lead content, a refractive index value greater than 1.54. The two samples typed as high alumina glasses both have an alumina content greater than 10%. Discussion There
are many
significant
features
highlighted
in Table
5. The
float
glasses from the three countries represented have a remarkable consistency about their data including the virtual absence of alumina and potash. The sheet glasses from Japan have consistent data and hence dominate the mean results for this classification. Each source of patterned glass has its own characteristic data. The Australian wired glass differs from the Japanese imports. The 14 windscreens have come from eight different world wide car manufacturers so a common composition is not expected. Most containers used within Australia are manufactured locally in each State by the same national company. The two Victorian containers are repeated samples from a pharmaceutical pack and have an unusual composition. The 13 overseas containers originate from nine countries and hence again there is a variability of data. The borosilicate headlamps have a fairly uniform set of data but the two soda-lime-silica headlamps are quite different. It can be seen that the borosilicate labware differs mainly in potash and lime to the borosilicate headlamp. A characteristic of the soda-lime-silica spectacle glasses is their potash content in excess of 5.0%, the only other glasses with such a potash composition being lead glasses. It is difficult to make further comments upon the light bulbs, spectacles, tableware and labware due to the lack of samples in each subgroup. The data obtained are also presented in histogram form so as to visually display the similarities and differences of the various chosen glass classifications. The differences in the various flat glasses are displayed in Fig. 1 in which the refractive index is plotted for all the flat glasses, and for each of the five subgroups as well aa for the entire glass museum. The float glasses have all very constant values apart from three American samples. The
21
4030
-
20
-
a
e 10 -
~1.510
1.512
1.514
1,516
1.516
1.520
1.522
1,524
1,526
1,526
.I.526
Refractive Index
Fig. 1. Histograms of refractive indices for glass: (a) total glass museum; (b) total flat glasses; (c) float; (d) sheet; (e) wired; (f) patterned;(g) windscreen.
distribution of the data for sheet glasses reflects the country of origin, besides being all lower than the value of float glass. The wired glasses despite being mainly fro& one country indicate a bimodal distribution, which could suggest two sources from within Japan. The patterned glasses again reflect the fact that the origin of the samples is mainly from two countries. The
28
windscreens have a wider distribution due to their multi-national origin. A check across classifications suggests that the single Japanese patterned glass is from the same source as one set of the Japanese wired glasses. The small number of samples make this cross interpretation suspect but the histograms themselves support the concept of maintaining the five classifications of flat glasses. Recognising the lack of samples in each classification the histograms for the other parameters have been restricted to those classifications or groups with 20, or more, samples. This limits the plots to four categories, namely, float, container, all the flat glasses and the total collection. The data for the soda content do not reveal in the initial plots (Figs. 2ad) much difference between flat, float and containers. With the expanded plots in Figs. 2e and f the uniformity of the float glass compositions is again emphasised. The data in Fig. 3 demonstrate the usual magnesia content in flat glasses and its absence in containers, although of course exceptions exist. The magnesia content in floats does not appear to differ much from the bulk of other flat glasses. The alumina data are plotted in Fig. 4. These histograms clearly reveal float glasses as being unique in having very low alumina content. However, other flat glasses have a wide range of alumina content while many containers are seen to have a fairly restricted range of alumina content. The silica plots of Fig. 5 do not help distinguish between flat, float or container classifications. There is a wide range of potash content in the various glasses as shown in Fig. 6a. However most of the glasses have a potash content of less than 1% so the accompanying plots of Figs. 6b-e are for this restricted range. The floats all have a potash content below the detection limit but are not alone in that feature as some other flats and some containers are similar. Finally Fig. 7 displays the lime data and indicates that the lime content in soda-lime-silica glass is in the range 4-13s. A reasonably wide range of lime composition occurs in the flat and container classifications but its content in float glasses is again in a very restricted range.
Conclusions A fast non-destructive elemental analysis to assist in the classification of glass fragments can be achieved on a SEM/EDS system. The simultaneous analysis of all commonly present elements in glasses, apart from boron and oxygen, down to a composition of 0.1% may be seen to be advantageous compared to other analytical techniques. The results have been presented of the analysis of the six most commonly occurring components and the refractive index of glasses used within Australia. The selection of classifications based upon the use and/or method of manufacture presents data for the range and sources of products which
29
a 60 -
40 -
b
30 -
20 -
10 -
2O- c 10 -
2or
d
10 -
0
I
I
I
,
I
6
7
6
9
10
1,
12
13
14
15
16
I 12.2
I 12.5
I 12.6
I
I
13.1
13.4
13.7
14.0
14.3
s14.6
w17
20- e 10 -
2o
f
10 0
I 11.9
% Na20
Fig. 2. Histograms of percentage soda in glass: (a) total (c) float (d) container (e) total flat glass (f) float.
glass museum
(b) total
flat glass
30
JO 30
-
20
L
a
-
10 -
1
40
-
30
-
20
-
b
10 -
%MgO Fig. 3. Histograms of percentage magnesia total flat glass; (c) float; (d) container.
a
50-
40
-
-
content
in glass:
(a) total
glass museum;
(b)
31
20- c
;IdyA . _ _ 0
0.1
05
I.0
1.5
2.0
2.5 %
Al2
3.0
3.5
4.0
4.5
5.0
03
Fig. 4. Histograms of percentage alumina in glass: (a) total glass museum; (b) total flat glass; (c) float; (d) container.
4030
-
20
-
10
-
2010
a
c
-
I
10
-
d
0
69.0
69.5
70.0
70.5
71-O
71.5
72.0
72’5
73.0
I
I
I
73.5
74.0
.74.5
% SI 02
Fig. 5. Histograms of percentage silica content in glass: (a) total glass museum; (b) total flat glass; (c) float; (d) container..
32
ao-
a
70 -
60
-
50
-
40
7
30
-
20
-
10 -
I 0
40
--
01
I
2
3
4
5
0.1
0.2
0.3
0.4
0.5
0.6
6
7
0 7
0.8
8
9
>lO
b
e
10
I 0
%
Fig. 6. Histograms of potash glass; (d) float; (e) container.
1 o-9
'1 0
K20
content
in glass:
(a,b)
total
glass museum;
(c)
total
fla
33 40
30
20
10
rb
40
30
20
10
2or c 10 -
20- d 10 C
0.1
1
2
3
4
5
6
7
6
9
10
11
12
13
%CaO
7. Histogram of lime content (c) float; (d) container.
Fig.
in glass: (a) total glass museum; (b) total flat glass;
may be met in case work. Although the collection is still small certain differences in the variables between these classifications can be seen. A further statistical analysis of the data is given in a paper by Terry and Reid [ 81. Acknowledgements We thank Pilkington AC1 for their cooperation in providing a set of analysed glasses and permission to publish their values. This work has been supported by a grant from the Australian Criminology Research Council. References 1 T. Catterick and D.A. Hickman, The quantitative analysis of glass by inductively coupled plasma-atomic-emission spectrometry : a five element survey. Forensic Sci. ht., 17 (1981) 253-263.
34 2 D.A. Hickman, G. Harbottle and E.V. Sayre, The selection of the best elemental variables for the classification and discrimination of glass samples. Forensic Sci. Int., 23 (1983) 189-212. 3 G.C. Goode, G.A. Wood, G.A. Brooke and R.F. Coleman, Multielement analysis of glass fragments by neutron activation and the application to forensic science. Atomic Weapons Research Establishment Report No. 024/71, (1971). 4 B. Beattie, B. German and K.W. Smalldon, The analysis of small glass fragments by emission spectrography. Home Office Central Research Establishment Report No. 245, (1977). 5 J.C. Hughes, T. Catterick and G. Southeard, The quantitative analysis of glass by atomic absorption spectroscopy. Forensic Sci., 8 (1976) 217-227. D. Morgans, A. Butterworth and A. Scaplehorn, A survey of British con6 B. German, tainer glass using spark source mass spectrometry with electrical detection. J. Forensic Sci.. 18 (1978) 113-121. 7 CR. Howden, R.J. Dudley and K.W. Smalldon, The analysis of small glass fragments using energy dispersive x-ray fluorescence spectrometry. Home Office Central Research Establishment Report No. 241, (1977). 8 K.W. Terry and D.W. Reid, Statistical analysis of major and minor oxides as a basis of glass classification. To be published. 9 D.J. Vowles, K.W. Terry and A. van Riessen. Optimisation of accelerating voltages for Micron, 13 (1982) 289-290. EDS analysis in the scanning electron microscope. 10 S.J.B. Reed and N.G. Ware. Quantitative electron microscope analysis using a lithium drifted silicon detector. X-ray Spectroscopy, 2 (1973) 69-74. 11 N.G. Ware, Computer programs and calibration with the PIBS technique for quantitaComput. Geosci.. tive electron probe analysis using a lithium-drifted silicon detector. 7 1981) 167-184.
Science Scientific
International, 25 (1984) Publishers Ireland Ltd.
QUANTITATIVE
K.W. TERRY,
ANALYSIS
A. VAN RIESSEN,
19-34
19
OF GLASSES USED WITHIN AUSTRALIA
B.F.
LYNCHa
and D.J.
VOWLES
School of Physics and Geosciences, W.A. Institute of Technology, Kent Street, (Western Australia) and aForensic Science Laboratory, Government Chemical tories, Plain Street, Perth (Western Australia) (Received (Revision (Accepted
August 8th, 1983) received December January 25, 1984)
29,
Bentley Labora-
1983)
Summary An initial 134 glasses have been collected from eleven classifications of glass used within Australia. These include both local and imported glasses. Quantitative elemental analyses of the glasses have been determined using a scanning electron microscope equipped with an energy dispersive X-ray spectrometer. The current program provides for an elemental analysis for Na, Mg, Al, Si, P, S or Pb, Cl, K, Ca, Ba or Ti, V, Cr, Fe, Mn and Zn expressed as oxides, and has a sensitivity down to approximately 0.1%. The data for the six most commonly occurring elements, namely, Na, Mg, Al, Si, K and Ca, together with the refractive index are presented for each class of glass in terms of their mean value and standard deviation from the mean, and also in histogram form. Key words: probe
Glasses;
Australian;
Quantitative
analysis;
Energy
dispersive
electron
micro-
Introduction
There have been many attempts to classify glass fragments based upon the refractive index of the glass or its elemental composition. Two recent sum.. maries of these schemes are by Catterick and Hickman [l] and Hickman et al. [Z] . It is seen that much of the effort to date has been directed to the analysis of trace elements in glasses using such techniques as neutron activation analysis [3], emission spectrography [4], atomic absorption spectrometry [ 51, spark source mass spectrometry [6] and X-ray fluorescence spectrometry [ 71. Our approach has been to analyse quantitatively for the major and minor elements that constitute the glass network formers and glass modifiers. This has been achieved using a scanning electron microscope (SEM) equipped with an energy dispersive X-ray spectrometer (EDS) which results in sensitivities down to approximately 0.1%. The advantage of such a scheme is that all elements are analysed simultaneously in an analysis time of 50 s. Furthermore, the glass fragment is mounted in a resin block so is still available as a court exhibit if necessary. 0379-0738/84/$03.00 o 1984 Elsevier Scientific Publishers Printed and Published in Ireland
Ireland
Ltd.
20
The glasses used within Australia are largely locally made. However, there is a significant volume of imported glass from a wide range of manufacturers and countries. A summary of the origin of glass imports obtained from the Australian Bureau of Statistics for 1979 is given in Table 1. An initial collection by the authors of 134 glass samples have been catalogued into 11 categories namely, float, sheet, patterned, wired, windscreen, container, lightbulb, spectacle, tableware, labware and headlamp. The origin of these samples is given in Table 2. Our existing holding of glasses is far from being a full representation of the sources of Australian glasses, but is based upon those commonly available in Western Australia. The number of categories selected contrasts with the classifications reported in the above reviews. Previous workers have typically grouped all the flat glasses together or else as two groups of window and vehicle glasses. A statistical analysis of the appropriateness of these 11 categories in a classification system for glasses is detailed in a following paper [ 81. In this paper we present the mean, and standard deviation from the mean, of the six most commonly occurring elements, namely, Na, Mg, Al, Si, K and Ca, together with the refractive index for each class of glass and their country of origin subdivisions. In addition the range of certain classifications are presented in histogram form. It is of interest to compare these results of the Australian scene with the British ones previously published by Catterick and Hickman [ 11. TABLE
1
SOURCES OF Statistics 1979)
GLASSES
IMPORTED
INTO
AUSTRALIA
(Extracted
from
Bureau
Glass type
Country
Float Sheet Wired Patterned Windscreen Spectacles Headlamps Lightbulbs Containers Tableware
3, 4, 9, 10, 15, 24, 25, 31, 32 3, 5, 9, 10, 13, 16, 17, 19, 21, 22, 25, 26, 31 3, 10, 16, 22, 25, 31 3, 10, 16, 22, 25 10, 16, 24, 26, 31, 32 10, 16, 31, 32 4, 9, 10, 16, 26, 31, 32 2, 3, 4, 9, 10, 11, 12, 15, 16, 18, 28, 31, 32 3,6, 7, 9, 10, 12, 15, 16, 17, 18, 19, 25, 27, 31. 32 1, 2, 3, 4, 5, 6, 7, 8,9, 10, 11, 14, 15, 16, 17, 19, 20, 22, 23, 25, 26, 27, 28, 29, 30, 31, 32, 34 3, 5, 6, 7, 9, 10, 12, 16, 18, 19, 27, 31, 32
Labware
of
of origina
a 1: Argentina, 2: Austria, 3: Belgium-Luxembourg; 4: Canada; 5: China; 6: Czechoslovakia; 7: Denmark; 8: Finland; 9: France; 10: W. Germany; 11: Hongkong; 12: India; 13: Indonesia 14: Ireland; 15: Italy; 16: Japan; 17: Korea; 18: Netherlands; 19: New Zealand; 20: Norway; 21: Philippines; 22: Poland; 23: Portugal; 24: South Africa; 25: Spain; 26: Sweden; 27: Switzerland; 28: Taiwan; 29: Thailand; 30: Turkey; 31: UK; 32: USA; 33: Yugoslavia; 34: Hungary.
21 TABLE 2 ORIGINS OF GLASS MUSEUM SAMPLES Country of origin
Argentina Australia Belgium Canada China Czechoslavakia Italy Japan Philippines Poland Portugal South Africa Spain Thailand U.K. U.S.A. W. Germany Unknown TOTAL NO.
Classification
9 1
7 5
1
25
4
1 1
1 1
5 1 1
1
1 2
2 1 1
7
1 1 1 1 3
1 3 2
10 1 14 20
7
14
8
14
37
4
1
4
8
1 1
1 1 1
1 7
9
6
7
1 47 6 1 2 2 4 14 2 1 1 1 1 2 7 13 3 26 134
Experimental Sample preparation A piece of each glass sample was mounted using a polyester resin in a halfinch diameter acrylic tubing. The sample was then ground and polished prior to being coated with a vacuum evaporated thin film of carbon. Glass fragments down to 0.2 mm size, or 20 pg, have been prepared successfully using this method. Refractive index (RI) determination The RI of each glass sample was determined using a Mettler FP5 heating stage controlled by a Mettler FP52 temperature programmer in conjunction with an optical microscope. Fragments of each glass were immersed in Dow Coming DC 710 silicone oil and the mean temperature obtained of the disappearance and reappear-
22
ante of the Becke line in sodium light. Reference the oil provided the value of the refractive index.
to the calibration
chart for
SEMIEDS instrumentation and procedure The elemental analyses have been carried out using a JEOL JSM 35C scanning electron microscope equipped with an United Scientific lithium drifted silicon detector and associated electronics. The spectra are accumulated in a Tracer Northern TN 1705 multi-channel analyser (MCA). For elemental analysis an accelerating voltage of 20 kV and a current of 5 X 10-l’ amps was used at normal incidence onto a 40 X 40 I.tm raster on a polished sample and a detector take-off angle of 35”. A typical count rate of 5000 cps is obtained during an analysing live time of 50 s. The choice of voltage is a compromise between the 25 kV required to maximise the peakto-background ratio in our system [9] and a low voltage to minimise the A minimum voltage of 15 kV is magnitude of the matrix corrections required for adequate excitation of zinc. It has been found that when using 20 kV negligible variations in analysis occur upon varying the carbon thickness on a sample. The precision and accuracy of SEM/EDS systems is discussed by Reed and Ware [lo] and Ware [ 111. The choice of a small raster rather than a spot mode analysis was to minimise the loss of sodium counts during analysis. The spectra are collected in the MCA and are then stored on a stereo cassette for subsequent batch processing on a PDP ll/lO computer. The elemental analysis is subsequently computed using the peak integration with background subtraction (PIBS) technique of Ware [ 111. The current program provides for an elemental analysis for Na, Mg, Al, Si, P, Cl, S or Pb, K, Ca, Ba or Ti, V, Cr, Fe, Mn and Zn expressed as oxides and has a sensitivity down to approximately 0.1%. The final analysis adopted is the mean of two analyses obtained on two different occasions. During each run of analyses NBS Standard Glass SRM 621 has been included as an internal control. Results and discussion Accuracy that has been used to calculate the The PIBS program of Ware [ll] quantitative elemental analysis of each glass requires certain calibration factors to be determined using standard samples. Four glasses from the range of NBS Standard Reference Materials and nine mineral samples have been used to calibrate the program. The calibration has been checked against some glass samples provided by Pilkington ACI. Their analyses and ours are given in Table 3. These data indicate a satisfactory level of accuracy for the technique. Precision The duplicate analyses of the 134 glass museum samples has taken place over a period of 3 months. During each analysing session NBS glass SRM 621
23 TABLE
3
COMPARISON
OF ANALYSIS
FROM
PILKINGTON
AC1 GLASSES
-
Sample
Analyst
Na,O
MgO
Al,O,
SiO,
SO,
K,O
AC1 1
AC1 WAIT
13.70 13.4
3.80 3.7
0.94 0.9
74.01 72.8
0.25 0.3
0.16 0.2
9.52 8.8
AC1 2
AC1 WAIT
14.06 14.6
0.16 0.1
1.72 1.6
71.92 72.1
0.03
11.53 11.4
0.03
0.05
0.2
AC1 WAIT
13.73 14.0
0.16 0.2
1.77 1.7
72.93 73.1
0.27 0.3
10.51 10.4
0.03
0.10
0.3
AC1 WAIT
13.23 13.7
0.28 0.2
1.77 1.7
72.95 73.3
0.15 0.1
10.99 10.8
0.05
0.07
0.2
AC1 WAIT
13.97 14.4
0.29 0.3
1.74 1.7
72.51 72.6
0.03
10.87 10.6
0.04
0.05
0.3
AC1 3
AC1 4
AC1 B
Co0
TiO,
Fe0 0.16
has been included as an internal control. The mean, standard deviation and coefficients of variation of the 15 sets of analysis from NBS 621 over the collection period are presented in Table 4 for each element detected. The large coefficients of variation for MgO and SO3 are due to their content being at the limit of detection for the technique. The values for the other elements indicate that the technique has a satisfactory level of precision. Museum catalogue For each of the glasses in our collection the following information has been recorded: museum catalogue number, origin of sample, classification, colour, thickness, fluorescence, refractive index, and elemental composition for up to 15 elements expressed as oxides. A summary of the data is presented in Table 5 in which the refractive index values and the values for the six most commonly occurring oxides are tabled. Although the data in many cases are not normally distributed the mean and standard deviation from the mean is provided for the appropriate subgroups of each
TABLE
4
PRECISION
Mean Std.Dev. C.V. (%) NBS data
OF RESULTS
ON NBS 621 OVER
3 MONTHS
Na,O
MgO
40,
SiO,
SO,
K,O
CaO
12.62 0.30 2.4 12.71
0.2 0.06 28.6 0.27
2.77 0.05 1.8 2.77
71.34 0.28 0.4 71.14
0.13 0.08 61.5 0.13
2.02 0.04 2.0 2.01
10.76 0.18 1.7 10.71
Country
Australia Belgium USA Total
Japan Philippines Poland Total
Australia Belgium Japan W. Germany Total
Australia Japan Total
Glass class
Float
Sheet
Patterned
Wired
FROM
OF DATA
5
SUMMARY
TABLE
1 7 8
7 5 1 1 14
9 1 10 20
No.
1.5196 1.5212 1.5210
1.5192 1.5238 1.5229 1.5252 1.5215
1.5145 1.5136 1.5164 1.5147
1.5194 1.5194 1.5208 1.5201
(21) (20)
(29)
(12) (23)
(9)
(2)
(20) (16)
(6)
USED WITHIN
Refractive index
GLASSES
13.9 13.1 13.2
13.6 13.5 12.6 14.1 13.6
13.4 14.4 13.6 13.6
13.9 13.7 13.7 13.8
Na,O
(7) (7)
(4)
(2) (4)
(4)
(3)
(2) (2)
(1)
AUSTRALIA
(1)
(1)
(2) (1)
(1)
3.7 2.2 (12) 2.4 (12)
3.4 (2) 3.9 (9) 0.7 2.3 3.3 (10)
3.6 3.6 3.4 3.6
3.7 3.6 3.6 3.6
MgG
0.6 1.6 1.5
1.2 0.4 1.6 0.8 0.9
1.8 1.6 1.2 1.7
0.1 0.1
Al&‘,
(1) (3)
(5)
(3) (3)
(2)
(1)
(1) (1)
72.8 72.0 72.1
72.7 71.9 72.3 71.5 72.3
73.1 72.4 73.4 73.1
73.2 73.3 73.1 73.1
SO,
(5) (6)
(7)
(5) (8)
(4)
(3)
(8) (6)
(3)
0.1 0.8 (1) 0.7 (2)
0.1 (1) 0.2 (2) 0.6 0.5 0.2 (2)
0.7 (3)
0.9 (0) 0.3
40
(4)
(2)
(2) (2)
(2)
8.6 10.0 (14) 9.8 (14)
8.5 (3) 9.8 (5) 11.5 10.8 9.3 (10)
7.0 7.0 8.1 7.1
8.9 8.8 9.0 8.9
CaO
1.5201 1.5214 1.5222 1.5213 1.5154 1.5208 1.5207 1.5196 1.4772 1.5626 1.5126 1.5232 1.5234 1.5174 1.5175 1.5488 1.5325 1.5146 1.4776
7 6 3 6 2 13 37
2 5
1 3
1 7
4 7 1 1
1 5
Aust. NSW Au&. SA Aust. TAS Aust. WA Aust. Vie Various Total
Soda-lime Borosilicate
Lead Others
Alumina/lead Others
Australia Total Lead High alumina
Soda-lime Borosilicate
Headlamp
Lightbulb
Spectacle
Tableware
Labware
1.5195
Container
63
Total
1.5182
Flat
14
Total
Windscreen
(91)
(11) (33)
(7)
(26)
(11) (14)
(10) (8) (12) (5) (0) (30) (23)
(30)
(26)
(3) (9) (2) (2) (0) (12) (12)
(5)
(4)
(4)
13.8 4.1
15.5 15.0 1.0 1.3
(7)
(2) (6)
2.5 9.9 (14)
7.7 16.1
14.5 (11) (3) 4.5
13.3 13.5 13.8 14.0 17.1 13.7 13.8
13.5
13.1 (8)
(5)
(3)
(3)
2.8 0.1
8.9
(1)
0.8 (11)
0.2
3.3
1.7 (14) 0.2 (2)
0.2 (2) 0.3 (1) 0.1 (1) 0.6 (1) 3.4 (0) 1.4 (12) 0.8 (11)
3.3
3.3 (7)
(5)
(0) (1)
(8)
(3)
(3) (2)
3.3 2.9 (14)
19.7
1.4 1.4
12.2 1.1
1.6 1.8
1.6 2.1
1.6 (1) 1.8 (6) 1.4 (0) 1.6 (1) 1.5 (0) 2.2 (11) 1.8 (7)
0.8
1.0
70.5 74.6
73.3 72.9 65.9 62.8
71.1 70.5
62.5 72.0
70.8 73.8
73.3 72.6 72.8 72.4 70.4 71.6 72.2
72.7
72.7
(34)
(4) (8)
(14)
(9)
(7) (35)
(5) (10) (4) (2) (3) (17) (14)
(7)
(6)
(1)
(3)
1.9 0.8 (11)
0.2 14.0
7.2 (18)
0.7 (2)
2.2 (7) 0.2 (2)
0.9 (7) 0.5 (6)
0.1
0.2 (1) 0.6 (8)
0.3 (4)
0.6 (4)
(7)
5.2 0.3
(2)
9.5 (2) 9.0 (12) 0.2 6.8
8.1 (12)
8.9 5.9
6.7 (14) 0.1 (1)
11.1 (1) 11.0 (4) 11.5 (5) 10.9 (2) 5.3 (1) 9.7 (14) 10.3 (16)
8.9 (11)
8.9 (10)
26
classification such as country of origin, or type of glass. This choice of presentation provides a compact listing of the seven variables of the various sources of glass products available within Australia. It is recognised that the range of composition of each classification is of interest to the forensic scientist and is partly provided in histograms later in this paper. A full listing of the catalogue is available from the authors. An inspection of the data for each sample enables it to be typed as sodalime-silica glass, borosilicate glass, lead glass, or high alumina glass. Characteristic of the borosilicates is their refractive index value of less than 1.50. A further characteristic of all the borosilicate glasses examined is a soda content of less than 5% and a lime content of less than 0.6%. The lead glasses have, besides their lead content, a refractive index value greater than 1.54. The two samples typed as high alumina glasses both have an alumina content greater than 10%. Discussion There
are many
significant
features
highlighted
in Table
5. The
float
glasses from the three countries represented have a remarkable consistency about their data including the virtual absence of alumina and potash. The sheet glasses from Japan have consistent data and hence dominate the mean results for this classification. Each source of patterned glass has its own characteristic data. The Australian wired glass differs from the Japanese imports. The 14 windscreens have come from eight different world wide car manufacturers so a common composition is not expected. Most containers used within Australia are manufactured locally in each State by the same national company. The two Victorian containers are repeated samples from a pharmaceutical pack and have an unusual composition. The 13 overseas containers originate from nine countries and hence again there is a variability of data. The borosilicate headlamps have a fairly uniform set of data but the two soda-lime-silica headlamps are quite different. It can be seen that the borosilicate labware differs mainly in potash and lime to the borosilicate headlamp. A characteristic of the soda-lime-silica spectacle glasses is their potash content in excess of 5.0%, the only other glasses with such a potash composition being lead glasses. It is difficult to make further comments upon the light bulbs, spectacles, tableware and labware due to the lack of samples in each subgroup. The data obtained are also presented in histogram form so as to visually display the similarities and differences of the various chosen glass classifications. The differences in the various flat glasses are displayed in Fig. 1 in which the refractive index is plotted for all the flat glasses, and for each of the five subgroups as well aa for the entire glass museum. The float glasses have all very constant values apart from three American samples. The
21
4030
-
20
-
a
e 10 -
~1.510
1.512
1.514
1,516
1.516
1.520
1.522
1,524
1,526
1,526
.I.526
Refractive Index
Fig. 1. Histograms of refractive indices for glass: (a) total glass museum; (b) total flat glasses; (c) float; (d) sheet; (e) wired; (f) patterned;(g) windscreen.
distribution of the data for sheet glasses reflects the country of origin, besides being all lower than the value of float glass. The wired glasses despite being mainly fro& one country indicate a bimodal distribution, which could suggest two sources from within Japan. The patterned glasses again reflect the fact that the origin of the samples is mainly from two countries. The
28
windscreens have a wider distribution due to their multi-national origin. A check across classifications suggests that the single Japanese patterned glass is from the same source as one set of the Japanese wired glasses. The small number of samples make this cross interpretation suspect but the histograms themselves support the concept of maintaining the five classifications of flat glasses. Recognising the lack of samples in each classification the histograms for the other parameters have been restricted to those classifications or groups with 20, or more, samples. This limits the plots to four categories, namely, float, container, all the flat glasses and the total collection. The data for the soda content do not reveal in the initial plots (Figs. 2ad) much difference between flat, float and containers. With the expanded plots in Figs. 2e and f the uniformity of the float glass compositions is again emphasised. The data in Fig. 3 demonstrate the usual magnesia content in flat glasses and its absence in containers, although of course exceptions exist. The magnesia content in floats does not appear to differ much from the bulk of other flat glasses. The alumina data are plotted in Fig. 4. These histograms clearly reveal float glasses as being unique in having very low alumina content. However, other flat glasses have a wide range of alumina content while many containers are seen to have a fairly restricted range of alumina content. The silica plots of Fig. 5 do not help distinguish between flat, float or container classifications. There is a wide range of potash content in the various glasses as shown in Fig. 6a. However most of the glasses have a potash content of less than 1% so the accompanying plots of Figs. 6b-e are for this restricted range. The floats all have a potash content below the detection limit but are not alone in that feature as some other flats and some containers are similar. Finally Fig. 7 displays the lime data and indicates that the lime content in soda-lime-silica glass is in the range 4-13s. A reasonably wide range of lime composition occurs in the flat and container classifications but its content in float glasses is again in a very restricted range.
Conclusions A fast non-destructive elemental analysis to assist in the classification of glass fragments can be achieved on a SEM/EDS system. The simultaneous analysis of all commonly present elements in glasses, apart from boron and oxygen, down to a composition of 0.1% may be seen to be advantageous compared to other analytical techniques. The results have been presented of the analysis of the six most commonly occurring components and the refractive index of glasses used within Australia. The selection of classifications based upon the use and/or method of manufacture presents data for the range and sources of products which
29
a 60 -
40 -
b
30 -
20 -
10 -
2O- c 10 -
2or
d
10 -
0
I
I
I
,
I
6
7
6
9
10
1,
12
13
14
15
16
I 12.2
I 12.5
I 12.6
I
I
13.1
13.4
13.7
14.0
14.3
s14.6
w17
20- e 10 -
2o
f
10 0
I 11.9
% Na20
Fig. 2. Histograms of percentage soda in glass: (a) total (c) float (d) container (e) total flat glass (f) float.
glass museum
(b) total
flat glass
30
JO 30
-
20
L
a
-
10 -
1
40
-
30
-
20
-
b
10 -
%MgO Fig. 3. Histograms of percentage magnesia total flat glass; (c) float; (d) container.
a
50-
40
-
-
content
in glass:
(a) total
glass museum;
(b)
31
20- c
;IdyA . _ _ 0
0.1
05
I.0
1.5
2.0
2.5 %
Al2
3.0
3.5
4.0
4.5
5.0
03
Fig. 4. Histograms of percentage alumina in glass: (a) total glass museum; (b) total flat glass; (c) float; (d) container.
4030
-
20
-
10
-
2010
a
c
-
I
10
-
d
0
69.0
69.5
70.0
70.5
71-O
71.5
72.0
72’5
73.0
I
I
I
73.5
74.0
.74.5
% SI 02
Fig. 5. Histograms of percentage silica content in glass: (a) total glass museum; (b) total flat glass; (c) float; (d) container..
32
ao-
a
70 -
60
-
50
-
40
7
30
-
20
-
10 -
I 0
40
--
01
I
2
3
4
5
0.1
0.2
0.3
0.4
0.5
0.6
6
7
0 7
0.8
8
9
>lO
b
e
10
I 0
%
Fig. 6. Histograms of potash glass; (d) float; (e) container.
1 o-9
'1 0
K20
content
in glass:
(a,b)
total
glass museum;
(c)
total
fla
33 40
30
20
10
rb
40
30
20
10
2or c 10 -
20- d 10 C
0.1
1
2
3
4
5
6
7
6
9
10
11
12
13
%CaO
7. Histogram of lime content (c) float; (d) container.
Fig.
in glass: (a) total glass museum; (b) total flat glass;
may be met in case work. Although the collection is still small certain differences in the variables between these classifications can be seen. A further statistical analysis of the data is given in a paper by Terry and Reid [ 81. Acknowledgements We thank Pilkington AC1 for their cooperation in providing a set of analysed glasses and permission to publish their values. This work has been supported by a grant from the Australian Criminology Research Council. References 1 T. Catterick and D.A. Hickman, The quantitative analysis of glass by inductively coupled plasma-atomic-emission spectrometry : a five element survey. Forensic Sci. ht., 17 (1981) 253-263.
34 2 D.A. Hickman, G. Harbottle and E.V. Sayre, The selection of the best elemental variables for the classification and discrimination of glass samples. Forensic Sci. Int., 23 (1983) 189-212. 3 G.C. Goode, G.A. Wood, G.A. Brooke and R.F. Coleman, Multielement analysis of glass fragments by neutron activation and the application to forensic science. Atomic Weapons Research Establishment Report No. 024/71, (1971). 4 B. Beattie, B. German and K.W. Smalldon, The analysis of small glass fragments by emission spectrography. Home Office Central Research Establishment Report No. 245, (1977). 5 J.C. Hughes, T. Catterick and G. Southeard, The quantitative analysis of glass by atomic absorption spectroscopy. Forensic Sci., 8 (1976) 217-227. D. Morgans, A. Butterworth and A. Scaplehorn, A survey of British con6 B. German, tainer glass using spark source mass spectrometry with electrical detection. J. Forensic Sci.. 18 (1978) 113-121. 7 CR. Howden, R.J. Dudley and K.W. Smalldon, The analysis of small glass fragments using energy dispersive x-ray fluorescence spectrometry. Home Office Central Research Establishment Report No. 241, (1977). 8 K.W. Terry and D.W. Reid, Statistical analysis of major and minor oxides as a basis of glass classification. To be published. 9 D.J. Vowles, K.W. Terry and A. van Riessen. Optimisation of accelerating voltages for Micron, 13 (1982) 289-290. EDS analysis in the scanning electron microscope. 10 S.J.B. Reed and N.G. Ware. Quantitative electron microscope analysis using a lithium drifted silicon detector. X-ray Spectroscopy, 2 (1973) 69-74. 11 N.G. Ware, Computer programs and calibration with the PIBS technique for quantitaComput. Geosci.. tive electron probe analysis using a lithium-drifted silicon detector. 7 1981) 167-184.