Sie sind auf Seite 1von 8

View Article Online / Journal Homepage / Table of Contents for this issue

Analyst, May 199.5, Vol. 120 1357

Analysis of Diamonds and Indicator Minerals


for Diamond Exploration by Laser Ablation-
Inductively Coupled Plasma Mass
Spectrometry*
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

R. John Watling, Hugh K. Herbert, Ian S. Barrow and Allen G. Thomas


Chemistry Centre ( W A ) , 125 Hay Street, East Perth, Western Australia 6004

A method has been developed, using laser ablation- However, the analytical method employed in this study
inductively coupled plasma mass spectrometry (LA-ICP- requires only that instrument response functions and lasing
MS), for the semiquantitative determination of 43 elements conditions are as near identical as possible from sample to
in diamonds, chromites and garnets. Samples of diamonds sample and from day to day because trace element associa-
from five countries have been analysed and the tions, rather than element concentrations, are the basis of the
interrelationship between the trace elements gives distinctive interpretative protcol applied to the data.
patterns for each source. The data indicate that this The classification of garnets, into groups indicative of the
technique could form the basis of a provenance identification provenance of the mineral, has been used successfully to
protocol for diamonds. This would produce an analytical identify the diamondiferous potential of the host rock.1.2
technique to assist police forces in tracing the origin of stolen Analyses, associated with this exercise, have centred on
material. LA-ICP-MS can also be used to establish the trace quantification of the partitioning of nickel between chrome-
element distribution in indicator minerals for diamond pyrope and olivine, the so-called ‘nickel thermometer’ effect.
exploration, such as garnets and chromites. Of particular This methodology, based on proton microprobe analysis,3
significance in this data set in the comparison of the relative accurately identifies whether the garnet under investigation is
distributions of the rare earth elements, tantalum and formed in the so-called ‘diamond window’4.5 of the Earth’s
hafnium, which indicates a consistent inter-element mantle and hence whether the kimberlite or lamproite that is
relationship for garnets associated with diamondiferous being assessed has a significant potential for diamond produc-
kimberlites. This relationship is considerably different from tion.
that found in garnets obtained from non-diamondiferous and Initial research, using LA-ICP-MS, was undertaken to
other sources. Application of the principle of trace element assess the potential of the technique to produce quantitative
partitioning pattern recognition to chromites has indicated data for comparison with proton microprobe results for
the potential of the technique to identify also a kimberlitic or garnets. However, while the data obtained for selected
non-kimberlitic provenance for this mineral. The data elements using both techniques compared favourably, and will
presented here are the result of preliminary investigations. A be the subject of a future paper, it was also possible to obtain
more detailed study is currently being undertaken and will data for a wide selection of other analyte elements and thereby
be the subject of future papers. identify trace element distribution patterns that appeared to
correlate with the diamondiferous potential of the host rock. It
Keywords: Laser ablation-inductively coupled plasma mass
is these data that have the potential of being used to establish a
spectrometry; diamond; indicator mineral
novel method of geochemical exploration for diamondiferous
rocks. The same approach was applied to chromites from a
known kimberlitic and non-kimberlitic environment. Here
Introduction again, trace element distribution patterns, especially for the
The illegal appropriation of diamonds at all points in their rare earth elements (REEs), indicated the potential of the
production from source mine to final product nets a significant ‘fingerprint’ concept to identify the provenance of the
income for all those involved. Once on the ‘open market’ material.
diamonds are notoriously difficult to source and their small In additional to its obvious potential for use in diamond
size, easy transportability and large value make them a superb exploration, LA-ICP-MS has considerable additional poten-
international Black Market currency. Attempts to ‘finger- tial as a means of identifying the primary or secondary nature
print’ diamonds have, thus far, met with limited success. and metamorphic history of garnets and chromites and their
However, the use of laser ablation-inductively coupled host rocks. As the use of LA-ICP-MS becomes more
plasma mass spectrometry (LA-ICP-MS) to obtain element widespread, and the analysis of discrete mineral grains
compositional data for diamonds heralds a new age in the becomes more practical,6-*3it is possible that the elemental
search for an analytical technique that can establish the associations and the recognition of inter-element association
provenance of recovered stones. patterns, such as those described in this paper, will feature
Although LA-ICP-MS has the potential to provide quanti- more significantly in geochemical exploration initiatives.
tative data, this necessitates the use of accurately calibrated
standards, a situation that is difficult to achieve in this instance Experimental
as certified standards for diamonds would be extremely
Sample Preparation
difficult to produce and of course none exist at this stage.
Sample preparation for this type of analysis can be as simple or
* Presented at Geoanalysis 94: An International Symposium on the Analysis o f
as complex as the analyst requires. I n general, provided that
Geological and Environmental Materials, Amhleside, England, September 18- the sample fits directly inside the laser cell and is no more than
22, 1994. 5 mm above the sample stage, analysis can be performed
View Article Online

1358 Analyst, May 1995, Vol. 120

directly on the original material with very little preparation. It the use of an internal standard is neither advocated nor, in our
must be remembered, however, that laser ablation of original opinion, necessary. Nevertheless, it is necessary to establish
material can result in the incorporation of surface debris, an optimum compromise series of parameters for laser
derived from the host material or a variety of other ablation for each sample type to ensure that a robust
anthropogenic sources. Analytical data will be compromised if comparison mechanism is achieved for different samples of
this occurs. Accordingly, unpolished garnets and chromites the same mineral.
are pre-cleaned by heating in concentrated hydrochloric acid The two variables of major significance, which were chosen
for 10 min, washing in de-ionized water and acetone and for this in.vestigation, were laser power and repetition rate of
drying at 60 "C prior to ablation. Diamonds are cleaned by laser shots. Preliminary studies indicated that analysis of
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

heating in a mixture of concentrated hydrochloric and garnets and chromites would be undertaken using a free
+
hydrofluoric acids (1 1) for 1 h at 120 "C, washing in water running lasing mode, whereas analysis of diamonds would be
and acetone and also drying at 60 "C prior to lasing. undertaken using Q-switched lasing conditions. Whenever
Obviously, if the sample is pre-cut or is in the form of a possible the free running lasing mode was used as it facilitated
polished mount this type of cleaning is both impractical and the development of the smallest craters. This was possible for
unnecessary. Unpolished material is mounted directly on both garnets and chromites; however, while it was possible to
plastic blocks using cyanoacrylate glue, with up to 25 samples couple a free running laser pulse to dark, inclusion-rich
being mounted in an appropriate array on the same block. diamonds, it was impossible to couple to clear diamonds.
These samples can also be mounted in small holes in the plastic Consequently, in order to ensure uniformity of analytical
block which are then backfilled with quick-setting plastic. The conditions, all diamonds were analysed using a Q-switched
mount is then polished to expose the surfaces of the samples. laser.
Samples prepared in this way are suitable for additional
investigation using scanning electron microscopic energy Diamonds
dispersive X-ray analysis and/or electron microprobe analysis.
For the analysis of diamonds, three sites on the surface of the
In most instances, it is difficult to prepare polished mounts of
in-house reference samples were pre-ablated, in Q-switched
diamonds and consequently they are usually analysed after
mode, for 15 s at 750 V laser power and a frequency of 6 Hz.
surface-mounting on the plastic blocks.
The coordinates of each ablated area were recorded by
computer in order that the sites could be re-located and
Instrumentation ablated when establishing the final lasing conditions.
Although the period of pre-ablation is not as long as
A VG Turbo Plus inductively coupled plasma mass spec-
traditionally accepted to ensure a steady-state signal, the
trometer fitted with an 's'-Option sensitivity upgrade and fine
concept of pre-ablation in this experimental protocol was not
focus LaserProbe accessory were used for the experimental
simply to establish uniform analyte release conditions, but
work. The LaserProbe is built around a 500 mJ pulsed
more importantly to clean the surface of any attached or
Nd :YAG laser operating at a wavelength of 1064 nm. The
incorporated extraneous material. In this way it was possible
fine focus optics facilitate the production of laser craters of less
to ensure that element association patterns resulted from the
than 30 pm in the free running mode and less than 100 pm in
ablation of matrix rather than from incorporated material.
the Q-switched mode for the sample matrices analysed.
Laser power was varied between 500 and 750 V and shot
frequency between 1 and 10 Hz.
Laser Optimization
Garnets and Chromites
The use of LA-ICP-MS could imply a degradation in precision
of results, when compared with equivalent solution ICP-MS In order to determine the optimum lasing conditions for the
data, simply because of variations in volatilized sample mass analysis of garnets and chromites, laser power was varied
between samples. This would be of serious concern if the use between 600 and 1000 V and repetition rate between one and
of the trace element partitioning pattern protocol required the five shots s-l for 1s ablation at each specific site. A total of 10
production of quantitative data. The concepts associated with sites, across the surface of each sample, was used for each
interpretation of data produced under this experimental experiment. The samples were not pre-ablated prior to
protocol have already been described in detail14 and rely on analysis.
the presence, association or lack of association, of specific The study was repeated on six separate occasions and results
suites of analyte elements. Consequently, quantitative data were averaged to produce response intensities for each
are not as important as they are in more conventional analyte. The response characteristics of 45 analytes were
analytical methodologies. investigated. The experimental matrix, designed to establish
Establishment of the optimum laser ablation conditions, for the optimum set of analytical conditions, was essentially
all mineral types analysed, is undertaken using the same similar to that established for previous investigations into gold
general experimental matrix. Significant variation in the fingerprinting.*5 Results indicate a trend in the analyte
characteristics of the physical matrix exists between samples of response under the different conditions that is essentially
the same mineral. Variations in colour, surface texture and related to the melting and boiling characteristics of the
sample orientation with respect to crystal planes all facilitate compounds of the individual analyte elements and the
variability in laser coupling. 15 Hence, the amount of material selective partitioning of these elements between vapour, melt
ablated will vary from sample to sample and, therefore, the and solid phases present during the ablation event.13 When
apparent analyte concentration will also vary. Consequently, establishing the optimum compromise set of ablation par-
it is again stressed that quantitative analysis, using the present ameters, it is necessary to ensure that variations in response
laser configuration, is not practical at this time but is also not between two sets of experimental conditions are as small as
required for data interpretation because the association of possible. This increases the robustness of the method and
elements, and not their actual concentrations, is used to ensures that variations in the trace element partitioning
establish the element distribution pattern. patterns are minimal when the analysis is undertaken on
It has already been established that there can be selective different instruments or on different days.
volatilization of elements from given matrices. 13716 Therefore, Spot size should be as small as possible, while still providing
at this preliminary stage of analytical method development, sufficient material for analysis. In this way small samples can
View Article Online

Analyst, M a y 1995, Vol. 120 1359

be ablated under the same conditions as large samples, Standard Reference Materials (SRMs) NIST SRM 610, 612
thereby maintaining the relevance of intercomparison of the and 614 glass standards. The choice of this material (and not
fingerprint. Lower laser power and reduced shot numbers are one of the actual experimental matrices) is based on the
to be preferred as this significantly reduces cross-contamina- premise that experimental optimization for this type of
tion and memory effects. analysis requires only the production of a reproducible set of
Ablation conditions were established for the three sample analytical and instrumental conditions which can be re-establi-
types as follows: shed on a daily basis. To this end, the use of a reference
material, rather than an inhomogeneous sample matrix, was
Diamonds: Q-switched mode; 625 V and 6 Hz
preferred. The fact that glass and the minerals under study are
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

Garnets: free-running mode; 750 V with a double pulse at 'similar' matrices is an advantage. The use of NIST SRM
each site glasses has proved to be robust and convenient and facilitates
reproducible instrument optimization.
Chromites: free-running mode; 700 V with a single pulse at
Once optimization had been achieved, in-house reference
each site
samples (three diamonds, four garnets and three chromites)
Where possible, samples should be ablated at a number of were used where appropriate to ensure data intercomparabil-
sites per sample. However, it may only be possible to use one ity on a daily basis and to provide a cross-reference
site when dealing with very small samples. Acquisition time comparison with previously obtained data. The data thus
for each mass spectrum is usually set at a minimum of 100 s. In produced facilitated maintenance of long-term quality control
this way it is possible to use the multichannel analyser to watch and ensured intercomparability of fingerprint data. Because
the fingerprint accumulate and reject data that have resulted the technique was not designed to provide quantitative data, it
from ablation of inclusions. This is evidenced by sudden was only necessary to confirm comparability of the element
increases in analyte response. A long potential acquisition associations within the sample with data from previous
time also allows the analyst to move the sample and ablate analyses of the same sample prior to undertaking the analysis
different spots, thereby ensuring better precision and provid- of a new series of samples. This cross-calibration procedure
ing a more intense spectrum for comparison. usually required about 1 h to complete.

Results and Discussion

l::::p
Optimization of Long-term Data Intercomparability Diamonds
The instrument operating parameters were optimized using Obviously, only a limited number of samples (six from each
National Institute of Standards and Technology (NIST) deposit) have been analysed at this point in the research

1000000 , 1

100 000 100000 .


10 000

1000

100
I .. 100

t 10

000 000

100 000
10 000 000

1 000000
(4
w
10 000

1000

100

1 000000 I 1 10 000 000

1 000000
100 000
100 000
10 000
10 000
1000
1000
100
100

10

Fig. 1 Mass spectral element profiles of diamonds from five deposits. ( a ) Russia (R); ( h ) Australia (A); (c) Zaire ( Z ) ; ( d ) China (C);
( e ) South Africa (SA), and (f> composite average for each deposit.
View Article Online

1360 Analyst, May 199.5, Vol. 120

programme as uncut diamonds are extremely expensive and average data are compared directly. Of particular note are
difficult to obtain from international producers. However, it is differences in the relative abundances of copper, zinc,
anticipated that, once the potential of the technique is fully yttrium, zirconium, tungsten, lead and thorium as well as
recognized, international producers will be more co-operative within the KEE assemblage. In order to demonstrate these
in supplying diamonds and the problems associated with differences more effectively, conventional ternary diagrams
obtaining a significant number of stones from each deposit, were constructed. Because ternary variation diagrams use
with which to establish a robust methodology, will be ratio percentages, the plots do not require data to be
overcome. quantitative, only reproducible. They provide a means of
The mass spectral elemental profiles of diamonds from the comparing the relationship between individual sets of data.
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

five deposits so far investigated are shown in Fig. l(a)-(e). The Consequently, they are ideally suited to interpreting the type
ordinate scale is an indication of the abundance of the analyte of data produced using LA-ICP-MS under the experimental
isotope being measured. No attempt has been made to relate protocol described in this paper.
isotopic abundance to elemental abundance as the concept of Selected ternary plots are presented in Fig. 2 ( a ) - o .
inter-element association patterns relies on relative abun- Although there is some spread in the data points within a
dances of analytes and not on their concentrations. Conse- deposit, each deposit forms a relatively coherent cluster of
quently, provided that the same isotope of an element is points that are essentially isolated from those of a different
compared between profiles, and as there are no interfering deposit. It must be remembered that these data are of a
isobaric overlaps, spectral comparisons are readily facilitated. preliminary nature and further refining of the technique will
The reproducibility of the spectral pattern within a deposit is
evident in all profiles. This suggests a common origin for the
2000
analysed diamonds from each deposit. This is obviously of
fundamental importance as, if significant variations in element 1800
profile did occur, the potential of the technique to establish a 1600
definitive provenance of diamonds would be severely
1400
degraded.
Although the over-all elemental distribution pattern for 1200
each of these deposits appears, at first sight, to be similar, 1000
significant differences do occur between them. The signifi-
800
cance of these variations is illustrated in Fig. 1 0 where
600

400

200

0
60 80 100 120 140 160 180 200 220
2000

1800

1600

1400

g! 1200
Ga Zr Ba Mo.10 3
8 1000
w.102 Rb.10 s
2 800

600

400

200

0
60 80 100 120 140 160 180 200 1
220

2000

1800

Rb Zr.10 w.10~ Pb 1600


Sn. 10 1400

1200

1000

800

600

400

200

0
w.10~ Pb Ce Mo 60 80 100 120 140 160 180 200 220
U
Fig. 2 Sclcctcd ternary distribution diagrams for clcmcnt associa-
tions in diamonds from fivc deposits: A , Australia; 0 . South Africa; Fig. 3 Mass spectra of three garnets dcrivcd from diamondifcrous
0, China; 0, Zairc; and a, Russia. host rocks.
View Article Online

Analyst, May 1995, Vol. 120 1361

undoubtedly lead to closer associations of points. To date, it Garnets


has not been possible to find a suitable element with which to
normalize data internally. Consequently, only uncorrected A series of 11 garnets from two diamondiferous deposits and
raw data are plotted. This means that differences in element 86 garnets from known igneous and metamorphic environ-
release, caused by variation in ablation conditions between ments were prepared for analysis. The mass spectra of three
samples, remain uncorrected. representative garnets typical of diamondiferous deposits are
From the data shown in Fig. 2(u)-(f), it is clear that tungsten presented in Fig. 3(a)-(c), while those from the non-diamon-
is a major discriminatory element for the Russian deposit, diferous host rocks are shown in Fig. 4(a)-(c).
whereas lead, molybdenum and rubidium are the predomi- The mass spectra of garnets from the diamondiferous
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

nant discriminatory elements for the South African deposit deposits all indicate a similar trace element pattern. Zinc (64,
investigated. The occurrence of high concentrations of lead in 66,67,68 and 70 u), gallium (69 and 71 u) and zirconium (90,
diamonds from the South African deposit is surprising in that 91, 92, 94 and 96 u) are relatively enhanced, whereas yttrium
this element is extremely volatile and usually not associated (89 u) has been removed from the spectral scan as its
with ultramafic magmas. One possibility is that the lead has concentration is often so high that it trips the ion multiplier,
been incorporated during transport and emplacement of the resulting in the loss of data. The element distribution pattern
deposit during the geological evolution of host eclogite and/or for garnets from non-diamondiferous host rocks contrasts
garnet peridotite, or that it results from sub-cratonic metaso- sharply with that for garnets associated with diamondiferous
matic processes. Significant lasing of these diamonds only deposits, while the elemental assemblage also appears much
marginally reduces the lead content and confirms that the lead more variable. Elements such as uranium (235 and 238 u)
is pervasive throughout the diamonds. thorium (232 u), lead (204,206,207 and 208 u), mercury (196,
l98,199,200,201,202and204u), tin (112,114,115,116,117,
118, 119, 120, 122 and 124 u), strontium (84, 86,87 and 88 u)
and rubidium (85 and 87 u), which are traditionally associated
with crustal rocks, are significantly more abundant. As with
lead from the South African diamonds investigated in this
1600 study, these elements may reflect metasomatic events or they
may be inherited from host rock evolutionary processes.
Of particular interest in all the spectra is the REE (139-176
1200
u) distribution pattern. The distribution of these elements, in
1000 garnets from the diamondiferous deposits, when chondrite-
800 normalized,l7 indicates a significant depletion in the lighter
600
REE and enhancement in the heavier REE. The characteristic
pattern exhibited by the indicator garnets from the two
400
diamondiferous deposits is shown in Fig. 5(a) and (6). The two
200 profiles, while different in detail, are remarkably similar, both
0 within and between deposits. Conversely, the chondrite-
60 80 100 120 140 160 180 200 220 normalized REE profile of garnets from a high temperature
2000 and pressure eclogitic environment [Fig. 5(c)] exhibits much
1800
more variable characteristics with a far less pronounced
depletion in the lighter REEs compared with the heavier
1600 REEs. In most samples there is also a suggestion of minor
1400 europium depletion, while one sample illustrates europium
5
v)
1200 enrichment. The REE profiles of garnets from a low
temperature metamorphic environment indicate no enhance-
8 1000
ment of heavy REEs or depletion in light REEs [Fig. 5 ( d ) ] .In
3
2 800 fact, some of the garnets indicate that there has been a
600 significant enhancement in the lighter REEs. Hafnium and
400 tantalum also do not show the characteristic downward slope
pattern that characterizes their association in garnets from
200
diamondiferous deposits and in garnets from eclogites. In
0 order to discriminate better between the types of garnets,
60 80 100 120 140 160 180 200 220 ternary diagrams were again prepared [Fig. 6(a)-(f)]. From
2000
these diagrams the identification of garnets derived from two
1800 diamondiferous and two non-diamondiferous host rocks is
1600 readily apparent. Plots of elemental associations in garnets
from the two diamondiferous deposits are essentially overlap-
1400
ping whereas the data from the non-diamondiferous host
1200 rocks are scattered almost over the entire plotting area.
1000 However, it is interesting that while there is considerable
800 scatter of the non-diamondiferous data, very few data points
overlap into the diamondiferous field.
600
400 Chromites
200
Six chromites each from kimberlitic and non-kimberlitic host
0 rocks were analysed and the elemental association patterns
60 80 100 120 140 160 180 200 220 are presented in Fig. 7(a)-(d). The reproducibility of the
U elemental signature for both types of chromite was essentially
Fig. 4 Mass spectra of three garnets derived from non-diamondifer- excellent [Fig. 7(a) and (c)]. However, there appeared to be
ous host rocks. an anomalous sample in the kimberlitic chromite suite [Fig.
View Article Online

1362 Analyst, May 1995, Vol. 120

7 ( a ) ] .The ablation characteristics for this sample were also Lu Nd


different from the other chromites in this suite. When the
element profile for this sample was removed and the data were
replotted [Fig. 7 ( h ) ] ,a much more reproducible elemental
pattern resulted.
A comparison of the average element profile for the two
suites of samples is shown in Fig. 7(d). Here, the differences
between the two sample suites are clear. Kimberlitic chro-
/ 0 ,-
mites contain significantly more rubidium, strontium, yttrium,
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

zirconium, niobium, barium, hafnium, tantalum, thorium and


uranium and the entire REE suite than do chromites from the
non-kimberlitic rocks. Ternary plots of these data (Fig. 8) Ce. 10 Nd.10 La Ce
elegantly confirm the definitive identification of the two suites Sm Pr
of chromites. Furthermore, the one ‘suspect’ chromite from
the kimberlite is, in all instances, intimately associated with
data from the non-kimberlitic chromites. It is felt that this
chromite represents material entrained from higher crustal
levels during ascent of the kimberlitic magma.

Conclusions
In Australia, as in other diamond-producing countries, stolen
diamonds represent major losses of revenue to producing
companies through direct loss and reduced sales, of dividends
to investors, of tax to Government and of loss of foreign La.10 Gd La.10 Ce
revenue. More dramatically, diamond theft is illegal and can Ta. 10 Sm
result in death or injury to innocent bystanders and should be
discouraged. However, until now, identification of the prove-
nance of diamonds, as a means of assisting the authorities in
criminal investigations, has not been possible.
Studies so far have shown that it is possible to identify the
provenance of different diamonds on the basis of their unique
trace element association patterns. The method is rapid and
relatively simple, although interpretation of spectra requires
considerable experience. Ultimately, it is hoped that the
identification of a particular diamond will be achieved using a
computerized database of mass spectra.
Application of the technique to the analysis of garnets and La.10 Hf La.10 Ce.10
chromites has produced data for a wide range of analyte Fig. 6 Selected ternary element distribution diagrams tor element
elements which, until now, have been at concentrations below associations in garnets from two diamondiferous deposits (0)and two
non-diamondiferous host rocks (0).

1 000 000 110 000000 I 1

100 000 1oooo0o

10 000 100 000

10 000
1000
I \$ 1000

-
100
100 p
t
h

.- 10
3 lo
>
-2
.-
a
1 I l l l l l l l l l l l l l l

;;i 1 000 000


v

a
5
73
100000
10 000
3
2 10000
1 1000

100

10
loo
1
1
La C e
I I
Pr
I ’
Nd Sm ELI Gd
l I I
Tb
I
Dy Ho
I I I
Er Tm
I I
Yb
l
Lu Hf
I J
Ta
’”, ,
,,,
,
La,I Ce Pr Nd Sm ELI Gd Tb Dy Ho Er Tm Yb Lu Hf Ta

Fig. 5 Chondrite normalized REE profilcs of garncts derived from two diamondiferous deposits (a and b ) , a non-diamond producing eclogite ( c )
and a low-grade metamorphic host rock (d).
View Article Online

Analyst, M a y 1995, Vol. 120 1363

the analytical sensitivity for most conventional techniques. Bi.10 Tb.10


The data for this increased range of analytes have facilitated
the identification of trace element association patterns that
have not previously been available to geologists and geoche-
mists and which can be used for provenance identification of
the minerals concerned. Such data have the potential of being
applied in association with conventional geochemical data for
improving the use of indicator minerals in the exploration for
diamond deposits.
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

The methodology described in this paper further demon-


strates the application of LA-ICP-MS to the area of forensic
mineralogy and provides the experimental design to transfer La.10 Nb Dy.10 Nd
this technology to other aspects of the analysis of solid Zn cu
samples. It is not designed to provide quantification of analyte
concentrations in a sample, but rather to provide a distribution
pattern of elements present within a particular sample, under
conditions that are robust and reproducible between the same
type of sample on a day-to-day basis. Interpretation of results
is based on the association of elements within a sample and
comparison of these associations with reference samples or
samples in a database.
Although it is possible to discriminate between diamonds
from the deposits studied, between garnets from diamondife-
Nb.102 Ce.102 Ga
rous and non-diamondiferous host rocks and between chro-
mites from kimberlitic and non-kimberlitic host rocks using Nd. 1O2 Nb
simple ternary diagrams, the potential of improved discrimi-
nation, using tetrahedral plots and more complicated multi-
variate statistical software, is being investigated. Again, it
must be stressed that all data presented in this paper are
preliminary and represent the analysis of a limited sample
base. Significant further research and development must be
undertaken before the true potential of the technique can be
demonstrated. However, the initial data for all sample types
are extremely encouraging.

The authors thank the Director of the Research Centre for Ba Zr Ce Zr


Advance Mineral and Materials Processing, University of Fig. 8 Selected ternary distribution diagrams for element associa-
Western Australia, for financial support in running the LA- tions in chromites from kimberlitic and non-kimberlitic host rocks. 0 ,
ICP-MS facility and the Director of the Chemistry Centre Kimberlitic and H,non-kimberlitic. X , Garnet from a kimberlitic host
(WA) for permission to publish this work. rock showing non-kimberlitic characteristics.

10 000 000

1 000 000

100 000

10 000

1000
h

.-v)
I

c 100
S

-
iii
0
10 000000

5
-0
1000000

2 100000
a
10 000

1000

100

Fig. 7 Mass spectral element profiles of chromites from kimberlitic and non-kimberlitic host rocks. ( u ) Kimberlitic host rock containing a
possible non-kimberlitic chromite ( X ) . ( b ) kimberlitic chromites from ( a ) with the possible non-kimberlitic chromite profile removed, ( c )
non-kimberlitic and ( d ) average comparison. K, Kimberlitic and NK, non-kimberlitic.
View Article Online

1364 Anulyst, Muy lo%, Vol. 120

References 9 Imai, N . , A n d . Chim. Actu, 1990. 235, 381.


10 Pearcc. N. J . G., Perkins, W. T., Abcll, I.. Duller. G. A. T.,
1 Griffin, W. I,.. Couscns, D. R.. Ryan, C. G., Sic. S. H.. and
and Fugc, R., J. Anal. At. Spectrom., 1992, 7, 53.
Sutcr, G . F . , Cnntrih. Minerd. Petrol.. 1989, 103, 199.
1 1 Pcrkins, W. T., Pcarcc, N . J. G., and Jcffrics, T. E., Geockim.
2 Griffin. W. L., Gurney. J . J . . and Ryan. C. G.. Contrih.
Cosmociiim. Arta, 1993, 57, 475.
Minerul. Petrol., 1992, 110, 1.
12 Rccd, S. J . B., Ciiem. Geol.. 1990, 83, 1.
3 Griffin, W. L., Jaqucs, A . L.. Sic. S. H.. Ryan, C. G . , Couscns,
13 Watling, R. J.. Herbert, H . K. and, Abell, I . D.. Chern. Geol.,
D. R., and Sutcr, G. F., C’onrrih. Minerul. Petrol., 1988, 99,
in the press.
143.
14 Watling. R. J . , Hcrbcrt, H. K . , Dclev, D., and Abell, I . A.,
4 Kennedy, C. S., and Kennedy. G. C., J . C;eophy.\. R c J . , 1976,
Spectrodiim. Actu, P u t R, 1994, 49, 205.
Published on 01 January 1995. Downloaded by University of Michigan Library on 26/10/2014 11:25:35.

81, 2467.
1s Jcffrics, T., Pcrkins, W. T.. and Pearcc, N . J . G . , Analyst, 1995,
5 Griffin, W. L., and Ryan, G. C., in Mineralogy in the Service of
120. 1365.
Munkind. Program and Abstracts International Congress on
16 Chencry, S., Hunt, A., and Thompson, M., J . Anal. At.
Applied Mincralogy, ICAM ‘93, Pcrth. Wcstern Australia. cd.
Spectrom., 1992, 7. 647.
Griffin, B.. Graham, J., and Linge, H., pp. 280-284.
17 Nakamura, N., Geochim. Cosrnochim. Acta, 1974, 38, 757.
6 Broadhcad, M., Broadhcad, R.. and Hager, J . W.. At.
Spectro.sc*., 1990, 11. 20s.
7 Broadhcad, M., At. Spectrosc., 1991. 12, 45. Paper 4106997H
8 Crain, J . S., and Gallimorc, D. L., J . Anal. At. Spectrom.. 1992, Received November 16, 1944
7. 605. Accepted January 3, 1995

Das könnte Ihnen auch gefallen