SME Annual Meeting

Feb. 23-25, Denver, Colorado

Preprint 04-43
QUANTITATIVE ANALYSIS OF PLATINUM GROUP METALS
USING X-RAY FLUORESCENCE SPECTROMETRY

H. Yoon, C. Park, C. Yoon and J. Hong

Korea Basic Science Inst. - Seoul Branch
Seoul, Korea

N. Kim and K. Han

South Dakota School of Mines and Technology
Rapid City, SD

ABSTRACT spectrometry ICP-MS (Coedo, et al., 1996; Borisov, et al., 1997B;

The contents of three platinum group metals (PGM-platinum,
palladium and rhodium), from the spent automotive catalyst, and two
NIST SRM (National Institute of Standard and Technology Standard
Reference Material, Used Auto Catalysts) were determined by the
X-ray fluorescence (XRF) spectrometry. The XRF technique has
been proven to be a fast, effective and accurate method in deter-
mining not only PGM but also other solid materials of less soluble for
wet digestion analysis. The detection limit of the XRF analysis is
better than that of the inductively coupled plasma mass spectrom-
etry which usually requires significant dilution to avoid high dis-
solved solid concentration. However, the XRF technique suffers
from a strong matrix dependent problem.
The objective of this study was to examine the applicability of
the XRF analysis where the sample matrix was not exactly the same
as the standard materials with particular reference to PGM analy-
sis. This study enabled to compare the effect of various matrices
including alumina, iron, copper and nickel on determination of the
contents of PGM.
Jarvis, et al., 1997; Vlasankova, et al., 1999).
It has been established (Yoon et al., 2003; Borisov et al.,
1997A) that the major problems encountered during the PGM analy-
sis were due to the spectral interference in AAS and ICP-AES
analysis and the formation of interfering monoatomic or polyatom-
ic species and the occurrence of matrix-induced interferences in
ICPMS analysis. Spectral interference resulted in false reading of
Au and PGM in ICP-AES and AAS analysis in the presence of high
iron (Yoon et al., 2003).
Therefore, it is important to select the appropriate analytical
method, the choice being usually a wet-chemical method such as
AAS, ICPAES, and ICP-MS or a dry method, namely the XRF tech-
nique. Figure 1 presents a schematic selection procedure for such
analytical methods.

INTRODUCTION

Accurate, reproducible and cost effective methods in determin-

ing the chemical composition of various metal elements are urgent-
ly needed in the fast growing metal industry especially with regard
to treatment of recycled metals industry. To meet such demands in
accurate and meaningful measurement of metal contents in various
sources, the advancement in hard-ware equipment as well as sta-
tistical analysis to help improve reliable analysis of measured val-
ues (Pereire and Brandao et al., 2001) has been made in recent
years.
Because PGM have gained a considerable importance in their
industrial and commercial value, the correct and fast evaluation of
PGM contents from industrial materials as well as recycling products Figure. 1. A schematic showing the selection process between
has become critical to help this fast growing industry. The determi- a wet-chemical method and a dry analytical technique.
nation of PGM contents from industrial scrap including spent auto-
motive catalysts is complex because not only they occur together
with other interfering metals, but also they exhibit similar physico- The wavelength-dispersive X-ray fluorescence spectrometer,
chemical behaviors as already presented previously (Yoon et al., another name for XRF has advanced its usefulness in application
2003; Ruiz et al., 2002). due to the relatively simple sample preparation as well as its ability
Numerous new instruments have been introduced for the use of rapid analytical determination. Fast, efficient and direct analytical
of quantitative analysis since the early 1970s. These include atom- techniques such as the XRF spectrometry are being developed in
ic absorption spectrometry, AAS (Haines and Robert, 1982), induc- a rapid rate.
tively coupled plasma atomic emission spectrometry, ICP-AES However, it should be noted that XRF is not free from short-falls
(Vlasankova, et al., 1999) and inductively coupled plasma mass in its applications. The accuracy of the technique is questionable

1 Copyright 2004 by SME

 SME Annual Meeting
Feb. 23-25, Denver, Colorado

when the exact composition of the matrix is unknown. The XRF
spectral line intensity of an element, analyte, should be proportion-
al to the concentration of the analyte. Due to the matrix effect of most
abundantly present in the sample specimen, the intensity is usually
not proportional (Whiston, 1991). In addition, it requires homogene-
ity in the sample matrix along with the same size and surface area
in reference to the standard sample matrix (Whiston, 1991) to be
calibrated against.
In this study, the applicability of the XRF analysis will be dis-
cussed. The effect of different matrix modifying materials such as
y- Al2O3, copper, iron, zinc, and nickel on PGM analysis will be
examined and discussed. Since the PGM analysis is frequently
concerned with spent automotive catalysts where PGM are deposit-
ed on y- Al2O3 surface, instead of using aluminum metal powder for
the source of Al y- Al2O3 was used and matrix corrections are
made only for the presence of y- Al2O3. This study enables the com-
parison of different matrix modification allowing high X-ray intensity
by dilution of materials of heavy absorber. Advantages of using the
XRF technique for solid samples are also presented and discussed.
EXPERIMENTAL
XRF (X-ray Fluorescence Spectrometry)
The X-ray fluorescence (XRF) spectrometry as a non-destruc-
tive analytical method is a reliable technique used to identify and
determine the concentrations of the elements present in solids, pow-
ders and liquids. Compared to the other analytical methods that
require sample digestion prior to analysis, XRF is known as one of
the simplest methods for the determination of the chemical com-
position of geological materials including ores, ceramics and even
metallurgical materials.
In the XRF analysis three electron shells are involved in emis-
sion of x-rays: K, L, and M shells. An emission spectrum for a given
metal has several intensity peaks arisen from the emission of K, L,
or M shell electrons. The most commonly used x-ray emission lines
are K and L shells and metals whose atomic number greater than 57
have been measured by M shell emissions (EPA Method 6200). K
line is produced by a vacancy in the K shell filled by an L shell elec-
tron, while K line is produced by a vacancy in the K shell filled by
an M shell electron. Often the K transition is 6 to 7 times more
probable than the K transition and therefore the intensity of the K
line for an element of interest is usually 7 times greater than that of
the K line (Whiston, 1991).
Parameters and operating conditions of XRF used during this
study are given in Table 1. The detection limit of XRF analysis
depends on several factors including analyte, detector type, excita-
tion and type of source, count time for irradiate sample, physical and
chemical matrix effect. The detection limits of PGM (Pt, Pd, and Rh)
are also presented compared with other analytical methods in Table
2.
XRF analysis on the determination of PGM. For this reason, the XRF
analysis has been carried out for samples whose matrix was modi-
fied by the matrix materials.
Matrix materials were prepared and used for the dilution of
both pre-concentrated automotive catalysts and the two NIST SRM
samples to have several PGM concentrations. Each matrix modifier
was selected based on previous studies and in most cases y- Al2O3
was used as the primary matrix modifier because it exhibited bet-
ter absorption of X-ray fluorescence compared to the other metals
used in this study (Yoon et al ., 2003).
Preparation of solid pellet samples for XRF analysis
A series of standard SRM and sample pellets were prepared
from the original pre-concentrated spent automotive catalysts and
the two NIST Standard Reference Materials by dilution with matrix
modifies y- Al2O3, copper, iron, zinc, and nickel.
To make pellet samples for the XRF analysis, the automotive
catalysts were ground in an agate mortar for 5 minutes prior to mix-
ing and saved. Then, 0.001 or 0.002 grams of those well grounded
automotive catalysts were added to 3 grams of each mixed matrix
modifiers prepared. Again they were well mixed in an agate mor-
tar. The finely ground samples were mixed with cellulose binder
(Spectroblend powder, cat # 660, Chemliex Inc. USA) at 1 to 1
ratio. Then, the exactly weighed and well-mixed powder samples
were pelleted at 15,000 Kpa with a hydraulic press. All the remain-
ing reagents used were of analytical grade. Two concentration lev-
els, namely PGM 1 and PGM 2 were introduced in relation to the
concentration of Pr, Pd and Rh. PGM 2 has two times higher PGM
content than PGM 1.
Each NIST SRM sample was used to make the standard pel-
lets for the evaluation of the matrix effect on the XRF analysis.
Five different matrix modifications (Al, Al;Fe, Al:Cu, Al:Ni, and Al:Zn)
were used in this case. All matrix modifying metals were mixed by
weight to satisfy one to one ratio.

Materials
The quantitative analysis of PGM was studied in two sample
groups: The pre-concentrated spent automotive catalyst, which was
supplied from Remetech Ltd. Korea and two NIST (National
Institute of Standard and Technology) Standard Reference
Materials referred to as Used Auto Catalyst SRM #2556 and SRM
#2557.
The standard and sample pellets were prepared from a sam-
ple material as well as two NIST SRM samples. The process for
making the XRF pellets is shown in Figure 2. The chemical compo-
sitions of finely powdered spent automotive catalysts and NIST
SRM samples are presented in Table 3.
One of the main aims of this study is to compare the matrix
effect of different matrix modifiers such as y- Al2O3 (instead of Al), Figure 2. The process for the preparation of sample pellets for
XRF from the original spent automotive catalysts.
copper, iron, zinc, and nickel and to check the detection limits of the

2 Copyright 2004 by SME

 SME Annual Meeting
Feb. 23-25, Denver, Colorado

Table. 1. Operating conditions for the Phillips PW2404 XRF Spectrometer. A rhodium (Rh) tube is run at 60kV/60mA with full vacu-
um with using Scintillation detector for all elements.

Table 2. The detection limits of PGM for major analytical methods.

AAS, ICP-AES data from Thermo Jarrell Ash An elementary overview of elemental analysis. All data based on PGM concentra-
tion in solution is ppb (mg/dm3).
*: Data from Thermo Jarrell Ash An elementary overview of elemental analysis based on PGM concentration in solution.
**: From Pillips Analytical Information. XRF data is based on PGM concentration in solid samples ppm (g/g).

Table 3. Original concentrations of Pt, Pd, and Rh from pre-concentrated automotive catalysts and NIST SRM #2556 and #2557
(ppm of solid).

*: Analyzed by wet chemical method using ICP-MS (Yoon et al., 2003).

**: Data provided by NIST which was analyzed by ICP-MS (IDICP-MS). The analytical values for Cu and Ni in NIST SRM are not pro-
vided.

RESULTS AND DISCUSSION
Quantitative analysis of Pt, Pd, and Rh for different sample
matrices have been studied by the XRF technique in a single matrix
and their analytical results are presented in Figure 3. Figure 3
shows the XRF calibration curves for Pt, Pd, and Rh in [03b3]-Al2O3,
Cu, Fe and Ni matrices. It is noted that due to the matrix effect the
x-ray fluorescence intensity changes considerably. A strong matrix
effect resulted in approximately 20 times different values of x-ray
intensities for each PGM depending on the different sample matri-
ces (Fig. 3).
The x-ray fluorescence (XRF) intensities of Pt, Pd, and Rh
were also measured in different matrix mixtures (Fig. 4). To test the
best matrix modifier for analyzing PGM, several modifications have
been tested. Mixing ratio of each matrix modifying materials was var-
ied as it appeared in Figures 4 and 5. Since y- Al2O3 is the most
abundantly present phase in the spent automotive catalyst, y- Al2O3
was selected as a major matrix modifying material while the mixing
concentration of other metals was varied. Depending upon the
matrix modifiers used the measured absolute intensity value of x-
ray fluorescence changed considerably (Fig. 4). The Al:Fe and Al:Ni
combination modifiers show higher intensities especially for Pt
analysis than the Al:Cu and Al:Zn matrix modifiers. However no
tremendous changes were observed from other metals like Pd and
Rh.
In figure 5, the absolute intensities of x-ray vary considerably as
the Fe content increased where Al and Fe were used as major
matrix modifying materials. As the Al content in sample matrix is
increased the absorbed x-ray intensity is stabilized. In Al and Zn
combination matrix, the reverse situation is observed. As the Al
concentration in matrix increased the absolute x-ray fluorescence
intensity was increased and stabilized (Fig. 5).

3 Copyright 2004 by SME


Figure 3. Calibration curves of XRF analysis for Platinum (Pt),
Palladium (Pd), and Rhodium (Rh) in y- Al2O3, Cu, Fe and Ni
matrices. Pt was analyzed using L line and Pd and Rh were
analyzed using K line.
Figure 4. Measured absolute intensity of x-ray fluorescence
for Pt, Pd and Rh in different matrix mixtures. The PGM con-
centration in PGM1 sample: Pt 36.28ppm, Pd 13.72ppm, and
Rh 7.27ppm. PGM concentration in PGM2 sample: Pt
72.57ppm, Pd 27.43ppm, and Rh 14.53ppm. The zero repre-
sents samples with no PGM in it.
4 Copyright 2004 by SME
SME Annual Meeting
Feb. 23-25, Denver, Colorado
Figure 5. Measurement results of absolute intensities of x-ray
fluorescence for Pt, Pd and Rh while the PGM concentrations
in sample pellets were same (fixed to have PGM1 composi-
tion). The matrix y- Al2O3 concentration changed from 0.5, 1,
2, 3 while the concentration of Fe and Zn was fixed as 1. `
For the verification of using particular matrix modifier in PGM
analysis, the results of PGM analysis from the pellet samples which
were prepared from the two NIST SRM using y- Al2O3, copper,
iron, zinc, and nickel as matrix modifiers were plotted against certi-
fied values provided by NIST (Fig. 6). Each Pt, Pd, and Rh concen-
tration has been compared to XRF analysis data of this study
(KBSI XRF).
It is apparent that data are widely scattered in measured PGM
by XRF especially for Pt. The Pt analysis by XRF may not have
been corrected well for enhancement or depression by matrix
modifying materials. It is noted that the most accurate Pt, Pd, and
Rh values were achieved from the samples modified with using y-
Al2O3. The presence of alumina (y- Al2O3) in the matrix does not
affect the analytical results of PGM (Fig. 6). Deviation of measured
XRF data from NIST data is fairly big and in most cases PGM con-
centration was under estimated. The discrepancy increased high-
ly especially for Pt analysis in the Al-Fe and Al-Zn matrices although
they exhibited very strong x-ray fluorescence intensities during the
PGM analysis (Fig. 4). The main problem with using Al-Fe and Al-
Zn matrices for modifying matrix for the analyte could be not
because of the poor precision of XRF measurement but because
of the strong matrix effect of combined metals, or in-homogeneity of
sample pellets such as irregular grain size effect.
The deviation of measured XRF data from the NIST certified
values was also checked by a linear fitting procedure. The calculat-
ed R2 value of alumina for NIST SRM #2556 is 0.9998 while the R2

value of alumina for SRM #2557 is 0.9999. In relation to this, the
other analytical result of PGM by XRF in different matrix modifies
such as Fe, Cu, Ni and Zn showed strong deviation from the
observed values (Fig. 6).
The calculated percent deviation of the XRF analysis results
of NIST SRM #2556 and #2557 in different matrixes are presented
in Figure 7. According to Figure 7, the percent deviation from
expected values zero was considerably small in the alumina (y-
Al2O3) matrix (10%) compared to the other matrices such as Al-Fe,
Al-Cu, Al-Ni, and Al-Zn. The deviations from the expected value
of PGM, 0 for both SRM #2556 and #2557, getting higher in the
following order, Al- Cui<Al-Ni<Al-Fe<Al-Zn.
Figure 6. KBSI-XRF (this study) analysis data for Pt, Pd and
Rh in various matrix modifiers compared to NIST ICP-MS
data. Matrix effect has been corrected during XRF analysis. Al
denotes y- Al2O3.
The XRF analysis results for each platinum group metal were
statistically accurate within the concentration range of the sample
materials. According to the previous study the detection limit of XRF
is much higher than the other common analytical instruments such
as AAS, ICP-AES, and ICP-MS (Yoon et al., 2003). Because of low
dilution from the original sample material the XRF analysis gives
better results especially when solid samples are dealt with. The
XRF analysis results demonstrated that the quantitative analysis of
Pt, Pd, and Rh using the calibration curves for the same matrix
materials is valid for a wide concentration range. However, due to
the strong matrix dependent problem, the maximum difference of
PGM between different sample matrices resulted in considerably dif-
ferent XRF intensities for same element when the matrix correction
has not been made properly (Fig. 4 and 5).
5 Copyright 2004 by SME
SME Annual Meeting
Feb. 23-25, Denver, Colorado
Figure 7. Calculated deviation of XRF analysis results of
NIST SRM in different matrixes.
The matrix effect during modification of matrix materials in
PGM analysis is compared and presented to check the best combi-
nation of matrix mixtures to get less matrix dependent calibrations
for the XRF analysis. To satisfy this requirement, the applicability of
the XRF analysis where the sample matrix was not exactly the same
as the standard materials were studied. The most accurate Pt, Pd,
and Rh values were achieved from the NIST SRM samples modified
by y- Al2O3 matrix modifier (Fig. 6). Analytical accuracy was also
proved by checking the linear fitting procedure and the calculated
R2 values 0.9998 for SRM #2556 and 0.9999 for SRM #2557.
However, the XRF analysis results in different matrices showed
strong deviation from the observed values. The calculated percent
deviation of NIST SRM #2556 and #2557 analysis results in differ-
ent matrixes showed clear differences. The deviation from the
expected value zero was considerably small in the y- Al2O3 matrix
(10%) than in the other matrices such as Al-Fe, Al- Cu, Al-Ni, and
Al-Zn. The deviation from expected value was increased in the fol-
lowing order Al-Cui<Al- Ni<Al-Fe<Al-Zn.
CONCLUSIONS
It was demonstrated that the XRF technique could be used as
a useful tool for analyzing PGM in various materials. Almost straight
lines of the calibration slop as well as zero point interception for all
PGM metals studied in this work enabled XRF qualified as a power-
ful tool for analyzing PGM from many matrix materials. Low dilution
with matrix modifiers during sample pre-treatment for the XRF analy-
sis, the detection limit of the XRF analysis is frequently better than
ICP-MS that is usually accompanying large dilution to avoid high
dissolved solid concentration. From the points of easiness of sample
preparation, the XRF technique was certainly proven to be one of

the better techniques considered for PGM analysis in various source
materials
However, the strong matrix dependent problem during PGM
analysis resulted in considerable differences in absorbed x-ray fluo-
rescence intensity between different sample matrices even for the
same element analyzed.
In order to avoid the strong matrix dependent problems during
the XRF analysis, various matrix combinations were tested. Several
modifications in making matrix have been tested to find a best matrix
modifier for analyzing PGM from spent automtive catalyst (industrial
materials). Form this study, Alumina (y- Al2O3) has proven to be the
best matrix modifying material for PGM analysis.
REFERENCES
Borisov, O. V., Coleman, D. M., Kristine A., Oudsema and
Carter III, R. O., (1997A) Determination of platinum, palladium,
rhodium and titanium in automotive catalytic converters using
Inductively Coupled Plasma Mass Spectrometry with liquid nebuliza-
tion. Journal of Anal. Atom. Spectrometry, Feb. Vol. 12, 239-246.
Borisov, O. V., Coleman, D. M., and Carter III, R. O., (1997B)
Determination of vanadium, rhodium and platinum in automotive cat-
alytic converters using Inductively Coupled Plasma Mass
Spectrometry with Spark Ablation. Journal of Anal. Atom.
Spectrometry, Feb. Vol. 12, 231-237.
Coedo, A. G., Dorado, M. T., Padilla, I., Alguacil, F., (1996)
Preconcentration and matrix separation of precious metals in geo-
logical and related materials using metalfix-chelamie resin prior to
6 Copyright 2004 by SME
SME Annual Meeting
Feb. 23-25, Denver, Colorado
inductively coupled plasma mass spectrometry. Analytica Chim.
Acta. 340, 31-40.
Haines, J. and Robert, R. V. D. (1982). The determination by
atomic absorption spectorphotometry using electrothermal atomiza-
tion of platinum, palladium, rhodium, ruthernium, and iridium.
Counc. Miner. Technol. (MINTEK) Randburh. Rep. No. M34.
Jarvis, K. E., Williams, J. G., Parry, S. J., and Bertalan, E.,
(1995) Quantitative determination of the platinum-group elements
and gold using NiS fire assay with laser ablation-inductively coupled
plasma-mass spectrometry. Chem. Geology 124, 37-46.
Pereira A. M. T. and Brandao P. R. G. (2001) Statistical valida-
tion of standardless and standard-based analysis by x-ray fluores-
cence spectrometry in iron ores characterization. Min. Eng., Vol. 14,
No. 12, 1659-1670.
Ruiz M. del. C., Rodriguez M. H., Perino E., and Olsina R. A.
(2002) Determination of Nb, Ta, Fe and Mn by x-ray fluorescence.
Min. Eng., Vol. 15, 373-375.
Vlasankova, R., Otruba, V., Bendl, J., Fisera, M., and Kanicky,
V., (1999) Preconcentration of platinum group metals on modified sil-
icagel and their determination by inductively coupled plasma atom-
ic emission spectrometry and inductively coupled plasma mass
spectrometry in airborn particulates. Talanta, 48, 839-846.
Whiston C. (1991) X-ray Methods. John Wiely & Sons. 228-261.
Yoon H., C. Yoon, C. S. Park, T. Ko, N. S. Kim and K. N. Han
(2003) Quantitative Analysis of PGM using ICP-MS, ICP-AES, AAS,
and XRF. The 132 SME Annual Meeting, Cincinnati, OH, February
24-27, 2003. Preprint 03-63.