J Forensic Sci, May 2012, Vol. 57, No.

3
doi: 10.1111/j.1556-4029.2011.02025.x
TECHNICAL NOTE Available online at: onlinelibrary.wiley.com

CRIMINALISTICS

Regina Eckardt,1 Dipl.-Min.; Erik Krupicka,2 Ph.D.; and Wolfgang Hofmeister,1 Ph.D.

Validation of Powder X-Ray Diffraction

Following EN ISO⁄IEC 17025

ABSTRACT: Powder X-ray diffraction (PXRD) is used widely in forensic science laboratories with the main focus of qualitative phase
identification. Little is found in literature referring to the topic of validation of PXRD in the field of forensic sciences. According to EN ISO ⁄ IEC
17025, the method has to be tested for several parameters. Trueness, specificity, and selectivity of PXRD were tested using certified reference materi-
als or a combination thereof. All three tested parameters showed the secure performance of the method. Sample preparation errors were simulated to
evaluate the robustness of the method. These errors were either easily detected by the operator or nonsignificant for phase identification. In case of
the detection limit, a statistical evaluation of the signal-to-noise ratio showed that a peak criterion of three sigma is inadequate and recommendations
for a more realistic peak criterion are given. Finally, the results of an international proficiency test showed the secure performance of PXRD.

KEYWORDS: forensic science, powder X-ray diffraction, validation, EN ISO ⁄ IEC 17025, laboratory proficiency test, trueness, specificity,
selectivity, robustness, preparation errors, detection limit

Little is found in literature referring to the topic of validation and software used is Diffrac Plus Release 2005 Eva Version 11.0 from
powder X-ray diffraction (PXRD), especially relating to the field of Bruker AXS GmbH. Reference database used is ICDD PDF2 of
forensic sciences (1). This paper gives a summary of how required 2005 and self-established user databases. If not stated otherwise,
parameters can be examined and used for validation report. diffractograms are evaluated by import of raw files in Eva, fol-
Validation according to EN ISO ⁄ IEC 17025 is ‘‘(…) the confir- lowed by background subtraction and Fourier transformation.
mation by examination and the provision of objective evidence that
the particular requirements for a specific intended use are fulfilled’’ Micro X-Ray Fluorescence
(2, p. 41). Point 5.4.5.3 in EN ISO ⁄ IEC 17025 states that ‘‘the
range and accuracy of the values obtainable from validated meth- To support the evaluation of diffractograms, most laboratories
ods (…) as assessed for the intended use, shall be relevant to the apply elemental analysis as well. In this work, elemental analysis is
customer’s needs’’ (2, p. 41). The main intended use of PXRD in performed by micro-X-ray fluorescence (l-XRF). The device used is
forensic science laboratories is qualitative phase identification (3,4). Eagle II (Rçntgenanalytik Messtechnik GmbH, Taunusstein,
To meet the expected standards, the PXRD method is tested for Germany) with an Si(Li)-detector. With the help of mono-capillary
parameters of trueness, specificity, selectivity, robustness against optics, small sample quantities can be analyzed. Measurement
preparation errors, and detection limit. parameters and evaluation of spectra are carried out with the soft-
ware Eagle II Vision 32. The sample preparation is the same as with
the PXRD, so samples analyzed with both methods do not change
Materials and Methods
preparation between the two measurements. The following parame-
Powder X-Ray Diffraction ters are used for elemental analyses if not stated otherwise (Table 2).
The device used for PXRD analysis is the D8 Advance (Bruker
AXS GmbH, Karlsruhe, Germany) with transmission, h ⁄ 2h geome- Weighing
try, and a position-sensitive Vntec detector. The sample rotates with Two balances were used: Mettler model AE 163 (Mettler
constant angular speed throughout the measurement. Different aper- Toledo GmbH, Gießen, Germany) and a Cahn-25 balance (Fisher
tures are available with a diameter of 3, 6, and 15 mm. The following Scientific, Schwerte, Germany).
measurement parameters are used if not stated otherwise (Table 1).
Powders are prepared between two Kapton foils (8 lm; Spex-
Certiprep, Stanmore, U.K.) on a specimen holder (Fig. 1). Sample Materials
The specimen holder is inserted in a sample holder which is Zinc oxide, calcite, sodium chloride, potassium chloride, lithium
placed into a magnetic holder of the D8 Advance. The evaluation fluoride, lead carbonate, galactose, lactose, arabinose, xylose,
1 sorbose, glucose, fructose, mannitol, talc, and cellulose were pur-
Institute of Geosciences, Johannes Gutenberg University, D-55099
Mainz, Germany.
chased from Merck (Darmstadt, Germany); caffeine and lactose
2
Forensic Science Institute, BKA, D-65173 Wiesbaden, Germany. monohydrate from Fluka Chemie GmbH (Buchs, Switzerland); lead
Received 28 July 2010; and in revised form 16 Feb. 2011; accepted 12 sulphate from Riedel de Han (now Fluka Chemie GmbH); SRM
Mar. 2011. 2686 Portland Cement Clinker from the National Institute of

722  2012 American Academy of Forensic Sciences

 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 723

TABLE 1—Measurement parameters for PXRD analysis. The problem with discrimination between a single phase and a
mixed crystal comes up with the examination of the phase corun-
Radiation Cu Ka1 1.540598 dum. The evaluation software Eva proposes besides pure Al2O3 a
Voltage 40 kV mixed crystal Al1.98Cr0.2O3, whose pattern can only be differenti-
Current 40 mA ated from pure Al2O3 pattern with the support of elemental analysis
Measurement range 5–80 2h information (Fig. 2).
Time ⁄ step 0.1 sec
2h range ⁄ step 0.0086
A greater complexity is shown by the examination of the cement
Overall measurement time 14.55 min and soil standards, which consist of multiple crystalline phases
Rotation 30 r ⁄ min (Tables 3 and 4).
Aperture 6 mm The main diffraction peaks of alite (PDF no. 055-0740) and belite
PXRD, powder X-ray diffraction. (PDF no. 033-0302) overlay each other (Fig. 3). The only free dif-
fraction peaks of belite are the ones around 44.1 < 2h < 44.8 or
c. 37.3. The lower intensity of the belite phase (Ca2SiO4) compli-
cates the identification, and furthermore, the presence of a solid solu-
tion, for example, (Ca1.99Mg0.01)SiO4, cannot be excluded. Periclase,
which in total shows only five diffraction peaks in the range of
5 < 2h < 80, can only be identified with the support of elemental
analysis and the knowledge that it may be a typical cement phase.
As can be seen in Fig. 4, even with a measurement time of 4 h,
chlorite, which holds nine weight percent, cannot be identified.
FIG. 1—Sample preparation of powder X-ray diffraction (PXRD). Referring to illite ⁄ muscovite, only a hint of presence of a mica
group mineral can be given. The matrix absorption, crystallinity,
TABLE 2—Measurement parameters for l-XRF analysis. and stray radiation prevent the identification of crystalline phases
with lower weight percentage. An operator can see diffraction
Anode Material Rhodium peaks (e.g., 2h = 33.1 or 2h = 39.2) that are still left over, but
Voltage 40 kV even with the support of elemental analysis no secure identification
Current Variable (2500 cps) of other crystalline phases is possible.
Atmosphere Vacuum
Spot size 300 lm
Overall measurement time 300 sec Specificity and Selectivity
Type of measurement 10-point analysis
If a qualitative analysis result is to be given with a certain
l-XRF, micro-X-ray fluorescence. method, the parameters’ specificity and selectivity have to be tested.
The criterion of specificity is achieved if the relevant analyte can be
Standards and Technology (Gaithersburg, MD); SRM 674 a-alu-
identified or quantified. The criterion of selectivity is fulfilled if all
minium oxide and titanium oxide from the National Bureau of
analytes can be identified separately. Both parameters can be tested
Standards (Gaithersburg, MD); the soil standard ‘‘Light alluvial-de-
with a synthetically generated phase composite of certified reference
luvial meadow soil’’ PS-3 COD No 312a-98 from MBH Analytical
materials. Referring to PXRD, the success of identifying one or
Ltd. (Barnet, UK); silver material from Ventron GmbH (Karlsruhe,
more phases in a phase mixture is strongly depending on the quality
Germany); and gold material from Balzers Union (now Leica
of phase crystallinity, the presence of amorphous compounds, and
Microsystems, Wetzlar, Germany). Commercially available primers
similar crystal structures (e.g., multiple mica phases, multiple cubic
(Magtech Cleanrange [Magtech Ammunition Company, Inc., Lino
metal phases, or mixed crystal systems). If a phase mixture consists
Lakes, MN], Geco-Sinoxid [RUAG Ammotec GmbH, Frth,
of multiple compounds whose diffraction peaks overlay each other,
Germany], and S&B nontox [Sellier Bellot J.S.C., Vlaim, Czech
secure identification can be problematic.
Republic]) were provided by KT23, BKA, Wiesbaden, Germany.
The ability of PXRD to differentiate between crystallographic
Heroin sample was seized by KT34, BKA. No certificate of analy-
similar patterns or to reach this ability with the support of elemen-
sis was available of the explosives, trinitrotoluene, pentaerythritol
tal analysis is tested. Other compound mixtures consist of multiple
tetranitrate, and trinitro-2, 4,6-phenylmethylnitramine, and so they
phases and are analyzed for example to test how many phases can
were examined by nuclear magnetic resonance spectroscopy.
be detected in one mixture.
Although the following sample mixture is not normally analyzed
Survey and Results by the method of PXRD in real casework scenario, it is a good
example to show the method’s limitations. The analysis of a mix-
Trueness
ture of gold and silver with PXRD is problematic. Multiple phases
The parameter trueness should be tested by analyzing certified are suggested by the search ⁄ match routine of the Eva software
reference material or material that is already analyzed with another (e.g., OV, LiH, Pd3Gd, and CuBr), all consisting of different ele-
validated method in case no certificate is given. In qualitative anal- ments (Fig. 5).
ysis, the only conclusion drawn is the phase identification. All phases belong to the crystallographic space group Pm3m or
Materials chosen to test this parameter are lactose monohydrate, Fm3m, and it is impossible to say which pattern could be used for
aluminium oxide, the soil standard PS-3, the Portland cement clin- the identification because all of them match the diffractogram. On
ker, trinitrotoluene, pentaerythritol tetranitrate, and trinitro-2,4,6- the contrary, with the help of elemental analysis, it is clear that the
phenylmethylnitramine. mixture consists of silver and gold.
Identification of a single phase of crystalline material is in gen- A mixture of seven sugars (consisting of galactose, lactose
eral unproblematic. The identification of lactose hydrate and the hydrate, arabinose, xylose, sorbose, glucose, and fructose to 1 ⁄ 7
three explosives was without difficulties. weight percent each) is analyzed. With a measurement time of 1 h
724 JOURNAL OF FORENSIC SCIENCES

FIG. 2—Diffractogram of the sample Al2O3.

TABLE 3—Phase composition of SRM 2686 Portland cement clinker as Robustness Against Preparation Errors
given by certificate.
A disadvantage of the focusing geometry of the powder diffrac-
Phase Mass Percent tometer is the high sensitivity for errors in sample height. These
Alite (C3S) 58.6 € 4.0 errors can appear through an improper fixation of the clamping
Belite (C2S) 23.3 € 2.8 rings or the preparation of the powder between the two Kapton
Ferrite (C4AF) 14.1 € 1.4 foils. To evaluate the consequences of such errors, test samples
Periclase (MgO) 3.3 € 1.9 were consciously prepared falsely.
Aluminate (C3A) 2.3 € 2.1
C, CaO; S, SiO2; A, Al2O3; F, Fe2O3.
Inhomogenic Particle Size Distribution
Primer composition samples show in general a heterogeneous
TABLE 4—Phase composition of soil standard PS-3 as given by certificate.
material with phases consisting of different particle size. Because
Phase Weight Percent of the sensitivity to friction and compression, grinding of such sub-
stances can be dangerous and has to be carried out with great care.
Quartz 28
Caused by the different particle size and the upright position of the
Albite 18
Orthoclase 14 sample during a measurement, there is a possibility that only bigger
Chlorite 9 particles are fixed by the strain of the two Kapton foils. As a con-
Calcite 8 sequence, material loss could occur with smaller particles falling
Illite + Muscovite 7 out of the irradiated area and the corresponding phases would not
Tremolite 6
Montmorillonite 2 be detected.
Goethite 2 To evaluate the consequences of such a problem, three primers
Gibbsite 2 were prepared in three different ways. The comparison of the three
Rhodochrosite 1 S&B nontox samples is shown for example (Fig. 8). The first way
Organic compounds 1
of preparation is as stated in the general sample preparation tech-
nique (Fig. 8A), the second preparation includes a grinding step
(Fig. 8B), and the third one uses the grinding step plus a powder
and the support of elemental analysis, the seven sugar phases could fixation on the lower Kapton foil with the help of silicon oil
be clearly identified (Fig. 6). (Fig. 8C). The silicon oil should improve the adhesion of particles
A likewise mixture consisting of 10 phases (zinc oxide, titanium to the foil and prevent particle loss.
oxide, aluminium oxide, silicon oxide, calcium oxide, lead carbon- No material loss or missing phases could be observed comparing
ate, lead sulphate, sodium chloride, potassium chloride, and lithium the three different preparation techniques.
fluoride) was prepared as an example of an inorganic mixture. All The Magtech samples showed a displacement of D2h = 0.02.
phases could be identified with a measurement time of 1 h and the During the grinding step, ungrindable spheres were observed, and
support of elemental analysis (Fig. 7). some were sorted out for further measurements. Scanning electron
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 725

FIG. 3—Diffractogram of SRM 2686 Portland cement clinker.

microscopy of the amorphous particles showed only oxygen, nitro- The best match with muscovite is given after pulverizing for
gen, and carbon. It is concluded that these particles consist of nitro- 20 min because more diffraction peaks are detected and can be
cellulose that is a typical composite of primers. assigned to the pattern of the reference database.

Texture Effects Variations in Sample Geometry

Preparation of layered minerals, for example phyllo-silicates, can Consequences of variations in sample geometry need to be eval-
be difficult. Because these minerals show a layered crystallographic uated. It is common that different operators prepare a different
structure and are also elastic, grinding of such materials is compli- mass of material between the two Kapton foils or even one opera-
cated. When the material is fixed between the two Kapton foils, a tor if preparing the same material multiple times. Another error can
preferential orientation parallel to the foil occurs. This effect can be occur in case one clamping ring is not properly fixed (Fig. 10).
enhanced by the statical interaction between, for example, micas The specimen could also be inserted upside down into the mag-
and the Kapton foils. In consequence, an adequate number of sta- netic holder of the D8 Advance. A sample could be prepared too
tistically oriented crystals are not warranted. Selected diffraction thick or too thin. Consequences of these errors were tested on an
peaks are given a much higher intensity, whereas others will not be inorganic sample (the cement standard) and an organic sample
detected caused by their positioning in the measurement geometry. (heroin).
This causes a problem for the evaluation software, because the It is shown that a thinner (Fig. 11C) than normal preparation
resulting pattern does not match the ones in the reference database. (Fig. 11A) of organic material results in a strong intensity loss,
Pulverizing a mica mineral (Fig. 9A) for 2 (Fig. 9B), 8 (Fig. 9C), whereas a thicker preparation (Fig. 11B) shows a strong intensity
and 20 min (Fig. 9D) lowers the texture effect. gain.
726 JOURNAL OF FORENSIC SCIENCES

FIG. 4—Diffractogram of the soil standard PS-3.

Because inorganic materials have higher absorption coefficients, one after another without taking the sample holder out of the D8
a thicker (Fig. 12B) than normal preparation (Fig. 12A) results in Advance or change of sample preparation. This should give an
an extremely intensity loss, whereas the thinner preparation approximate range of measurement uncertainty which can then be
(Fig. 12C) does not show a great effect on diffracted intensity. compared to the intensity variations caused by operators.
A displacement of the pattern for about D2h = 0.003 when the Three different operators prepared the soil standard 10 times.
clamping ring is not properly fixed (Figs 11D and 12D) occurs One operator has a preparation experience for 30 years, the sec-
because of the aberration of the focusing geometry. This effect ond one has no preparation experience, and the third one has
can be observed more clearly when the specimen holder is about a 5-year preparation experience. To compare the diffracto-
inserted into the magnetic holder upside down (Figs 11E and grams on a statistical basis, the criterion Inetto, full width at
12E). The difference in height from the focusing plane to the half-maximum, and chord middle are determined and evaluated
sample surface plane is about 3.5 mm and results in a displace- with the Eva software for selected diffraction peaks in the raw
ment of about D2h = 0.6. The evaluation software gives the pos- data. Inetto is the difference between the absolute measured inten-
sibility to enlarge the accepted displacement of diffraction peaks sity of a diffraction peak and the intensity of the background at
for the search ⁄ match routine. Although with a displacement of that 2Theta position. Three diffraction peaks of the quartz phase
D2h = 0.6, there is actually no chance for the software to suggest ([101], [100], and [122]), two diffraction peaks of the calcite
any proper phases. phase ([104] and [116]), and one diffraction peak of the albite
[)201] were integrated with Eva software, followed by the deter-
mination of the arithmetic mean of the net intensities including
Operator-Dependent Errors
span widths.
Operator-dependent preparation errors were investigated by mul- It can be seen in Fig. 13 that the statistical device measurement
tiple measurements of a certified reference material. First of all, the uncertainty is relatively low compared to the one introduced by
soil standard was prepared once and measured 10 times directly operators and multiple preparation times.
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 727

FIG. 5—Diffractogram of Ag ⁄ Au mixture.

The relative errors of operator A are the smallest ones, so in this Applied to PXRD, the criterion for the presence of a diffraction
example a 30-year experience in preparation results in a more repro- peak is therefore given as:
ducible intensity distribution compared to the operators with lower p
Nbrutto  NB  x  N B
or no preparation experience. Nevertheless, although the highest
standard deviation of intensity was found to be 30%, no difference with Nbrutto being the total counts in the diffraction peak, and x
in phase identification was given by three different operators. being a factor of confidence. Choosing x = 3, for example,
means that the counts above this limit have a probability of
Detection Limit 99.87% of truly belonging to a diffraction peak and not to back-
ground noise.
The reproducibility of diffracted intensities is in qualitative phase It is tested whether the above formula is applicable to PXRD
analysis important in case diffraction peaks of phases with low and whether a criterion of x = 3 is adequate. Measurement tests are
weight percentage are identified in one measurement and not iden- carried out for samples of varying density and a fluorescent sample
tified in a repeated measurement or new preparation because of as well to see whether there are any implications for the peak crite-
intensity variation. For the examination of detection limits, it was rion. Test material includes Kapton foils, silicon oil, MgO, ZnO,
differentiated between the detection limit of an absolute sample ZnS, PbS, and Fe2O3 which were prepared between two Kapton
quantity minimum of a pure phase and a detection limit of a cer- foils and measurement with all available apertures. In case of back-
tain phase in a phase mixture. ground measurement of Kapton foils, 4 Kapton foils were pre-
The detection limit is typically determined as the lowest quantity pared between the two Kapton foils used for fixation. The
of a substance that can be distinguished from the absence of that identical sample was used for the measurement with the three
substance or from the background noise. According to He (5), the apertures.
standard deviation is given by After measurement, the diffractograms are imported into Eva
p software, followed by a background subtraction (curvature 1,
rB ¼ N B
threshold 1). The resulting files are transformed to an x–y file with
where NB is the background counts. the program Powder-Cell for Windows Version 2.4 (Bundesanstalt
728 JOURNAL OF FORENSIC SCIENCES

FIG. 6—Diffractogram of the organic mixture.

fr Materialforschung und -prfung, Berlin, Germany) (W. Kraus background examination. The determination of the factor x was
and G. Nolze, programmers), so the data can be read by Excel. carried out in Excel by calculating the arithmetic mean of measured
Three ranges of 2h = 5 (581–582 single measured intensities) intensities within the 5 range, its standard deviation, and the differ-
without any diffraction peaks or contribution from amorphous com- ence in the arithmetic mean (IMW) to the maximum (Imax) and min-
pounds were selected in each diffractogram in the range between imum (Imin) single measurement. It was then calculated how often
10 £ 2h £ 70. In the case of a-Fe2O3, only 650 single measure- the standard deviation is contained in the maximum difference to
ments could be selected in the range of 60 £ 2h £ 70 for retrieve the factor x.
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 729

FIG. 7—Diffractogram of inorganic mixture.

Imax  IMW IMW  Imin the factor x, the particular highest value is selected. Table 5 shows
x¼ x¼ a summary of the results of the particular highest value for the par-
r r
ticular aperture used and the corresponding 2h range.
An example of the background examination of the 4 Kapton The calculations show that a factor of x = 3 is not high enough
foils measured with the 6-mm aperture is shown (Fig. 14). to adequately include the statistical scattering of background
For any measured material, there are nine calculations of the noise. The calculated factors of 3.98 and 3.95 indicate that even
kind shown above (3 apertures · 3 5 ranges). For calculation of a factor of 4 could be too low in some cases. It should be paid
730 JOURNAL OF FORENSIC SCIENCES

FIG. 8—Diffractograms of the three differently prepared S&B nontox samples.

attention to the fact that a background calculation to identify a marked ‘‘peaks’’ which are even above a five-sigma criterion (an
diffraction peak later on is not carried out over such a large 2h arithmetical mean of 50 counts leads to a five-sigma criterion of
range. Because the examination in this work should lead to a fun- 85.4 counts). If transforming the diffractogram by the function
damental conclusion, a large 2h range was selected for statistical ‘‘smooth,’’ the degree of periodicity fades (Fig. 15B). Although the
analysis. marked intensities are not differentiable from background noise in
the raw data, a mathematical transformation generates an unfavor-
able situation for phase identification. To overcome such a prob-
Pitfalls in Mathematical Smoothing of Raw Data
lem, the original raw data should always be imported for the
When evaluating a diffraction pattern, diffractograms are in evaluation of possible diffraction peaks near the limit of detection.
general prone to a background subtraction (curvature 1, threshold The presence of a true diffraction peak is most probably given if
1) and a Fourier transformation (factor 20 · 1). The background the ‘‘peaks’’ can be seen in the raw data as well and if multiple fol-
subtraction does not change the ratio of the measured intensities lowing measurement values lie above a (in this case) five-sigma
at all. Although a Fourier transformation, which is a way to facil- criterion.
itate phase identification, is not an ideal tool for the examination It should be kept in mind that the total diffraction pattern is
of diffraction peaks near the limit of detection, the sine terms in examined. The presence of one single diffraction peak usually
the mathematical formula of the Fourier transformation automati- cannot be used for the secure identification of a phase. In addi-
cally create a peak form. The arrows in Fig. 15 indicate where tion to overcome such a problematic situation, the detection limit
the mathematical transformation automatically creates a peak form can be doubled when measurement time is increased by a factor
above the five-sigma criterion without the presence of diffraction of four. If there is a chance to identify more phases with longer
peaks. measurement time, it should always be preferred if enough time
The Fourier-transformed diffraction pattern (Fig. 15A) shows a is given.
clear periodicity traceable through the wave function. Independent For the calculations of detections limits, a diffraction peak crite-
of the smooth function the operator used, he may question the rion of Nbrutto ) NB ‡ 5 · NB is used.
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 731

FIG. 9—Diffractogram of muscovite pulverized for 0, 2, 8 and 20 min.

aperture. Diffractograms were background-subtracted (curvature 1,

threshold 1) and Fourier-transformed (factor 20 · 1). The tool
‘‘area’’ of the Eva software calculates information such as net
intensities of selected diffraction peaks, which were used for the
examination of the peak criterion. A substance is clearly identi-
fied if more than four diffraction peaks fulfill the criterion
Nbrutto ) NB ‡ 5 · NB.
The detection limits for single-phase crystalline components as
lactose monohydrate, quartz, and lead oxide are found to be 61.6,
28.1, and 15.4 lg, respectively. This is easily explained as scatter-
ing power rises owing to electronic density of the irradiated mate-
rial. As samples were analyzed for only 15 min and the aperture
FIG. 10—Improper fixation of the upper clamping ring. of 15 mm was used, it is probable that detection limits are low-
ered by longer measurement time and application of a smaller
aperture. Although the data give a good estimation of detection
limits of pure phases measured with the D8 Advance, the effect
Detection Limit of an Absolute Sample Quantity Minimum of crystallinity, crystal size, and matrix composition should be
The detection limit of a pure single crystalline phase is obeyed.
affected by ‘‘background noise,’’ scattering power, and absorption
coefficients of the phase irradiated. The determination of the Detection Limit of a Certain Phase in a Phase Mixture
detection limit of a pure phase is tested with lactose hydrate,
quartz, and lead oxide. The materials were weighed directly onto Four mixtures of forensic relevance are prepared in the concen-
the Kapton foil that is already fixed to the metal specimen tration range of interest.
holder (the Cahn-25 balance was used) in the range from 15 to The two organic mixtures contain the agent MDMA-HCl
60 lg for each phase. After the weighing step, the sample was (3, 4-methylenedioxymethylamphetamine hydrochloride) in an ideal
taken out of the balance, and the second foil was directly fixed matrix. Mixture 1 consists of about 75 weight percent cellulose, 6
to the sample. The samples were measured with the 15-mm weight percent talc, 6 weight percent magnesium stearate, and 12
732 JOURNAL OF FORENSIC SCIENCES

FIG. 11—Diffractogram of heroine: results of possible preparation errors of an organic sample.

weight percent caffeine. Matrix 2 consists of about 66 weight per- The composites were weighed with the Mettler balance
cent cellulose, 7 weight percent of talc and magnesium stearate and homogenized for 15 min in an agate mortar. The agent
each, and 21 percent of lactose hydrate. Table 6 shows the compo- MDMA-HCl was added to these matrices in different
sition of the matrices. amounts (Table 7). Again, a phase is identified if more than
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 733

FIG. 12—Diffractogram of cement standard: results of possible preparation errors of an inorganic sample.

four diffraction peaks match the pattern of the reference data- The inorganic mixtures were weighed together using the Mettler
base. Results show that detection limit of MDMA-HCl in balance as Table 8 shows and homogenized with a swing mill (Rets-
matrix one lies around 3 weight percent and in matrix two ch GmbH, Haan, Germany) for 2 min. Detection limit of calcite in
around 2 weight percent. quartz is found to be around 2 weight percent, and the detection limit
734 JOURNAL OF FORENSIC SCIENCES

FIG. 13—Comparison of intensity variation in selected diffraction peaks.

FIG. 14—Background examination of 4 Kapton foils measured with 6 mm aperture.

of quartz in calcite lies around 3 weight percent. This is easily With the help of blind tests, false-positive and false-negative
explained as calcite has a higher scattering power than quartz. findings can be examined. A false-positive results mean that a
As the test results of the MDMA-HCl in matrix 2 resulted in 2 phase is identified that is not contained in the sample. A false-nega-
weight percent of MDMA-HCl being detected and 3 weight percent tive finding states that a phase that is actually present is not identi-
not being detectable, blind test was carried out. fied. The test was carried out with two operators. Operator A
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 735

TABLE 5—Calculation of the factor x. TABLE 6—Composition of the two tablet matrices.

Sample Name x 2-Theta-Range Lactose

Cellulose Talc Mg-Stearate Caffeine Hydrate
Scatter 4 Kapton foil 3-mm aperture 3.56 12–17 (wt%) (wt%) (wt%) (wt%) (wt%)
Scatter 4 Kapton foil 6-mm aperture 3.23 12–17
Scatter 4 Kapton foil 15-mm aperture 3.29 26–31 Matrix mixture 1 75.0 c. 6.3 c. 6.3 12.5 0.0
Scatter silicon oil 3-mm aperture 3.67 10–15 Matrix mixture 2 c. 64.3 c. 7.1 c. 7.1 0 c. 21.4
Scatter silicon oil 6-mm aperture 3.55 30–35
Scatter silicon oil 15-mm aperture 3.98 50–55
Scatter MgO 3-mm aperture 3.43 10–15 As the operator examined the diffractogram for a known sub-
Scatter MgO 6-mm aperture 3.70 10–15 stance, it is obvious why 0.4 weight percent of calcite can give a
Scatter MgO 15-mm aperture 3.66 10–15 hint to the substance, whereas the detection limit lies around 2
Scatter ZnS 3-mm aperture 3.95 39–44 weight percent.
Scatter ZnS 6-mm aperture 3.77 38–43
Scatter ZnS 15-mm aperture 3.34 50–55 It should be kept in mind that a mixture is rarely totally homo-
Scatter ZnO 3-mm aperture 3.20 39–44 geneous, especially if the prepared sample amount lies around only
Scatter ZnO 6-mm aperture 3.61 50–55 a few tens of milligrams. Every substance or mixture is individual,
Scatter ZnO 15-mm aperture 3.50 10–15 so there can be no general statement as ‘‘PXRD ⁄ D8 Advance has a
Scatter PbS 3-mm aperture 3.95 10–15
Scatter PbS 6-mm aperture 3.33 55–60
detection limit of x weight percent.’’ The tests carried out should
Scatter PbS 15-mm aperture 3.52 36–41 only give a general idea for a detection limit for PXRD and the
Scatter Fe2O3 3-mm aperture 3.53 42–47 device used.
Scatter Fe2O3 6-mm aperture 3.89 10–15
Scatter Fe2O3 15-mm aperture 3.40 42–47
Laboratory Forensic Proficiency Test
Another method to prove the trueness of the results of a method is
prepared samples that were preselected from operator B and tested to take part at proficiency tests. Because such a procedure is
for the presence of the selected compound. A false identification recommended and proficiency tests carried out by scientific commu-
has especially in forensic laboratories fatal juristic consequences. nity are rare, such a test was organized by the German Federal Crimi-
As Tables 9 and 10 show, no false-positive or false-negative nal Police Office. One operator generated the samples and organized
results were reported. the test, whereas another operator tested the unknown samples.

FIG. 15—Comparison of mathematical smooth functions member.

736 JOURNAL OF FORENSIC SCIENCES

TABLE 7—Composition and fulfillment of diffraction peak criterion of MDMA-HCl in the two organic mixtures.

MDMA-HCl Matrix 1 Matrix 2 MDMA MDMA MDMA

Sample Name (wt%) (wt%) (wt%) >4 Peaks 2–4 Peaks 1 Peak
MDMA3Matrix1-1_99 1 99 – – +
MDMA3Matrix1-2_98 2 98 – + +
MDMA3Matrix1-3_97 3 97 + + +
MDMA3Matrix1-4_96 4 96 + + +
MDMA3Matrix1-5_95 5 95 + + +
MDMA3Matrix1-10_90 10 90 + + +
MDMA3Matrix2-1_99 1 99 – – +
MDMA3Matrix2-2_98 2 98 + + +
MDMA3Matrix2-3_97 3 97 – + +
MDMA3Matrix2-4_96 4 96 + + +
MDMA3Matrix2-5_95 5 95 + + +
MDMA3Matrix2-10_90 10 90 + + +

TABLE 8—Composition and fulfillment of diffraction peak criterion of the In total, 15 laboratories took part and 13 laboratories provided
inorganic mixtures. their results. The samples tested consisted of:
CaCO3 SiO2 CaCO3 CaCO3 CaCO3 • 25 mg Sicotan (BASF, Ludiwigshafen, Germany), a yellow-
Sample Name (wt%) (wt%) >4 Peaks 2–4 Peaks 1 Peak ish pigment (NixSbyTi1-x-yO2) on rutile basis
CaCO3SiO2-0.3_99.7 0.3 99.7 – – + • a 200 mg mixture of 10% Cu, 15% NH4NO3, 20% S8, 35%
CaCO3SiO2-0.5_99.5 0.5 99.5 – + + Pb3O4, and 20% lactose hydrate
CaCO3SiO2-1_99 1 99 – + + • a 300 mg mixture of 1 ⁄ 3 mannitol, 1 ⁄ 3 acetaminophen, and
CaCO3SiO2-2_98 2 98 + + + 1 ⁄ 3 caffeine
CaCO3SiO2-3_97 3 97 + + +
CaCO3SiO2-4_96 4 96 + + + The samples were weighed with a Mettler balance and homoge-
CaCO3SiO2-5_95 5 95 + + + nized in an agate mortar for 1 h.
SiO2 SiO2 SiO2
>4 peaks 2–4 peak 1 peak
The Sicotan sample was probably the most difficult one,
because no matching pattern exists in the reference database ICDD
CaCO3SiO2-99.5_0.5 99.5 0.5 – + + PDF2. By the forensic scientific background and the elemental
CaCO3SiO2-99_1 99 1 – + + analysis information of Ni and Sb, an operator should be able to
CaCO3SiO2-98_2 98 2 – + + conclude that pure TiO2 is a white pigment, so Ni and Sb are
CaCO3SiO2-97_3 97 3 + + +
CaCO3SiO2-96_4 96 4 + + + responsible for the yellowish color. Examination of the powder dif-
CaCO3SiO2-95_5 95 5 + + + fractogram with all that information should lead to the result of a
rutile structure with additional nickel and antimony elements incor-
porated into the crystal lattice.
The proficiency test was evaluated by a point system (Table 11).
As the Sicotan was the most difficult sample, 0.5 points were
TABLE 9—Results of blind tests for MDMA-HCl in organic mixture. given for the identification of rutile phase. Another half point was
added if the elements nickel or antimony had been taken into
MDMA-HCL Matrix 1 Matrix 2 Report
Sample Name (wt%) (wt%) (wt%) Operator B account. Another full two points were given if the pigment Sico-
tan was identified. Other samples were evaluated by giving two
Blindtest A 3 97 ++ points for each identified phase. False-identified phases and not
Blindtest B 2 98 ++
Blindtest D 1 99 o identified phases were not included in the point system as some
Blindtest F 100 – laboratories would have received negative points.
Blindtest E 3 97 + The forensic laboratory of the German federal criminal police
Blindtest G 1 99 o office, the Bundeskriminalamt, is Lab J. The secure phase identifi-
Blindtest H 2 98 o
cation and the ‘‘confirmation by examination and the provision of
Blindtest C 100 –
objective evidence that the particular requirements for a specific
++, identification; +, detection; o, hint to a substance; –, not identified. intended use are fulfilled’’ were proven once more, although the
laboratory proficiency test shows that more tests are advisable.

TABLE 10—Results of blind tests for CaCO3 and SiO2 in inorganic matrix. Discussion

CaCO3 SiO2 Report It is shown that the required parameters of trueness, selectivity,
Sample Name (wt%) (wt%) Operator B and specificity were fulfilled. Only phases below the detection limit
could not be identified with PXRD although this problem can
Blindtest J 100 0 –
Blindtest K 99.6 0.4 +
sometimes easily be solved by longer measurement time. Tested
Blindtest I 99 1 ++ preparation errors showed to be nonsignificant for phase identifica-
Blindtest M 0 100 – tion or are detected easily by an operator with scientific back-
Blindtest L 0.5 99.5 o ground, so in no case wrong phase identification should occur. No
Blindtest N 1 99 o false-positive or false-negative results were reported at blind tests.
++, identification; +, detection; o, hint to a substance; –, not identified. Examination of background scattering showed that the generally
 ECKARDT ET AL. • VALIDATION OF POWDER X-RAY DIFFRACTION 737

TABLE 11—Results of the proficiency test.

Lab-Code

Sample Phases ⁄ Elements A B C D E F G H I J K L M

1 Rutile 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
Element Antimony 0.5 0.5 0.5 – 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
Element Nickel 0.5 0.5 0.5 – 0.5 0.5 0.5 – 0.5 0.5 0.5 0.5 –
(Ti, Ni, Sb)O2 ⁄ Sicotan – – 2 – – 2 1.5 – 2 1.5 – 2 –
Gelb (Pigment Yellow 53)
2 Lead oxide (Pb3O4) 2 2 2 2 2 2 2 2 2 2 2 2 2
Sulfur 2 – 2 – 2 2 2 – 2 2 2 2 2
Ammonium-nitrate – – 2 – – 2 – – 2 2 – – –
Lactose hydrate – – 2 2 – 2 2 – 2 2 2 – –
Copper – – 2 – 2 2 2 – 2 2 2 1 2
3 Mannitol 2 – 2 2 2 2 2 2 2 2 2 2 –
Acetaminophen – – 2 2 2 2 2 2 2 2 2 2 2
Caffeine – 2 2 2 2 2 2 2 2 2 2 2 –
… of max. 19.5 points 7.5 5.5 19.5 10.5 13.5 19.5 17.5 9 19.5 19 15.5 14.5 9
False-identified phase(s) 0 3 0 0 2 0 0 0 0 0 0 1 0
Not identified phase(s) 6 7 0 4 3 0 1 5 0 0 2 2 5

used peak criterion of three sigma is inadequate to test for the real 3. Rendle DF. X-ray diffraction in forensic science. Rigaku J 2003;19(2)
presence of an X-ray diffraction peak. Recommendations for a 20(1):11–22.
4. Kugler W. X-ray diffraction analysis in the forensic science: the last
more realistic criterion are given. The execution of a laboratory resort in many criminal cases. Adv X-Ray Anal 2003;46:1–16.
proficiency test showed that more tests should be carried out for 5. He BB. Two-dimensional s-ray diffraction. Hoboken, NJ: John Wiley &
scientific community. Sons, 2009.

References Additional information and reprint requests:

Regina Eckardt, Dipl.-Min.
1. Berti G, D’Acunto M, De Marco F. A quarter century of round robin Institute of Geosciences
tests on XRD. Adv X-Ray Anal 2002;45:60–5. Johannes Gutenberg University
2. ISO Committee on Conformity Assessment (CASCO). General require- D-55099 Mainz
ments for the competence of testing and calibration laboratories: EN Germany
ISO ⁄ IEC 17025:2005. Brussels, Belgium: CEN ⁄ CENELEC, 2005. E-mail: [email protected]