Analytical

Methods
View Article Online
PAPER View Journal | View Issue

Heavy metals in powder-based cosmetics quantified by

Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

ICP-MS: an approach for estimating measurement

Cite this: Anal. Methods, 2013, 5, 402
uncertainty†
Beatrice Bocca, Giovanni Forte, Anna Pino and Alessandro Alimonti*

A method based on microwave digestion and high resolution mass spectrometry was developed for the
determination of Cd, Co, Cr, Ni and Pb in powder cosmetic samples. The procedure was validated in-
house at the native concentration found in the cosmetic and at three fortification levels. Measurement
uncertainties were calculated using data generated from repeatability, recovery and calibration linearity
studies. The mean repeatability estimates were #11% for Cd, Co, Ni and Pb and #25% for Cr. The
average recoveries ranged between 70% (Cr) and 112% (Pb). The expanded uncertainties (k ¼ 2) of the
Received 16th August 2012
Accepted 2nd October 2012
method were between 34% (Cd) and 63% (Cr) at the native concentration and from 14% (Cd and Pb)
to 34% (Cr) at the highest fortification level. The results accompanied by their uncertainty statements in
DOI: 10.1039/c2ay25914a
the powder cosmetic were as follows (in mg kg1): Cd, 0.026  0.009; Co, 1.61  0.60; Cr, 3.00  1.88;
www.rsc.org/methods Ni, 6.87  2.53; Pb, 0.25  0.09.

1 Introduction dermatitis.11 Moreover, diffusion of metals through damaged

skin was found to be greater than through intact skin and the use
This paper describes the process for validating an analytical of a common cleanser (for example sodium lauryl sulphate)
procedure and estimating the measurement uncertainty for the increased metal penetration.12,13 From a normative point of view,
determination of metals in powder cosmetic samples. The focus the European Union legislation listed these metals as substances
is on make-up products because they represent an important that are prohibited in any amount in cosmetics but trace
source of skin sensitization being used every day and are applied amounts of metals are permitted as “impurities” if unavoidable
to the thinnest areas of facial skin where absorption can be high.1 under conditions of good manufacturing practice.14 Limits for
The selected metals are those with known signicant toxicolog- Pb (20 ppm) and Cd (5 ppm) have been established in Germany
ical and allergological properties: Cd, Co, Cr, Ni and Pb. and are based on levels that could be technically avoided, rather
Cadmium has been classied as a human carcinogen by the than on a risk-based approach.15 For Ni, Co and Cr the scientic
National Institute for Occupational Safety and Health and Cd literature proposed a value of lower than 1 ppm for each metal to
occupational dermal exposure can cause irritant dermatitis.2 minimize the risk of sensitization in particularly sensitive
Lead impairs the renal, hemopoietic and nervous systems and subjects.16 However, there is a paucity of information about the
the use of leaded eye cosmetics in children have been observed to amount of these metals that should be dened as an “impurity”
be strongly correlated with elevated blood Pb levels.3–5 Nickel, Co and there are no validated methods for quantifying such traces.
and Cr are common contact allergens and cosmetics containing In order to be of support to the exposure assessment of
these metals have been linked to dermatitis and allergic reac- metals from cosmetic products, this study validated an analy-
tions, including perfumes, children’s bath products, mascaras, tical methodology to detect Cd, Co, Cr, Ni and Pb in powder
hair dyes, body creams and colours for tattooing.6–8 In general the cosmetics by using the microwave digestion (MW) of samples
absorption of metals through the skin is low but it can be of and the determination of metals by the high resolution induc-
concern when concentrated solutions are in contact with the tively coupled plasma mass spectrometry (HR-ICP-MS).
skin for several hours or longer as in the case of cosmetics.9,10 The Following well established guidelines, most notably the AOAC,
chemical form of a metal can also change the skin absorption as ISO and Eurachem documents, the uncertainty of measure-
in the case of Cr(VI) that exhibits a higher permeability than the ments was estimated using the validation data, namely repeat-
Cr(III) resulting in greater susceptibility to elicit contact ability, recovery and calibration linearity.17–21

Department of Environment and Primary Prevention, Italian National Institute of

Health, Rome, Italy. E-mail: [email protected]; Fax: +39 06 49902052; Tel: 2 Methodology
+39 06 49902080
† This article is part of a themed issue on Cosmetic Ingredients, Guest Edited by Samples of face powder of 0.08–0.12 g were weighed in Teon
Alberto Chisvert and Amparo Salvador. vessels and a mixture of 4 ml of super-pure HNO3 (Romil,

402 | Anal. Methods, 2013, 5, 402–408 This journal is ª The Royal Society of Chemistry 2013
 View Article Online

Paper Analytical Methods

Cambridge, UK), 0.5 ml of supra-pure HF (Merck, Darmstadt, Table 3 Linear regression data
Germany) and 2 ml of supra-pure H2O2 (Merck) was added. The
Concentration
samples were digested in a MW oven (Ethos 900-Mega II, FKV
range (mg L1) y ¼ a + bx r
Milestone, Milan, Italy) using the following MW program: 10 min
at 250 W; 10 min at 400 W; 10 min at 600 W. The resultant digested Cd 0.1–2.5 y¼ 0.027 + 0.132x 0.99992
solutions were transferred into polypropylene liners and lled up Co 1–25 y¼ 2.912 + 0.493x 0.99990
to a nal volume of 50 ml with high-purity deionised water Cr 10–250 y¼ 7.469 + 0.642x 0.99947
Ni 10–250 y¼ 3.360 + 0.130x 0.99986
(Barnstead EASYpure II, Dubuque, IA, USA). Reagent blanks Pb 1–25 y¼ 1.114 + 0.809x 0.99986
underwent the same digestion cycle in order to control impurities
Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

in the chemicals and containers used. The HR-ICP-MS (Element2,

comparing isotopic ratios for multi-isotopic metals and results
Thermo-Electron, Bremen, Germany) was equipped with sampler
obtained at two different instrumental resolutions (i.e. LR and
and skimmer cones in Pt, a Meinhard nebulizer, a water cooled
MR). Under the conditions reported in Table 2 the interferences
spray chamber (Scott-type) and a guard electrode device. Argon
(listed in the last column of the table) were resolved and the
ow rates, torch position, lenses, radiofrequency power were
measurements were specic.
optimized every day by measuring a multi-element daily perfor-
mance solution at three different mass resolutions, namely low 3.2 Calibration linearity
resolution (LR ¼ m/Dm 300), medium resolution (MR ¼ m/Dm
4000) and high resolution (HR ¼ m/Dm 10 000). The sensitivity, The addition calibration approach, by adding the cosmetic with
oxides and double charged ions formation, and stability were increasing concentrations of the metals, was used for quanti-
checked daily and the minimum criteria considered appropriate cations. Calibration standards were prepared by appropriate
to start the analysis are reported in Table 1. dilutions of 1000 mg L1 mono-element stock standard solutions
(High-Purity Standards, Charleston, SC, USA). Six calibration
3 Validation levels were used in the range of the concentration of interest and
each standard was run in triplicate. A suitable internal standard
The methodology was validated on a face powder that is (IS), namely 1 mg L1 of 103Rh, was added to calibration standards
commonly used as a base before applying other make-up. It was (and test samples) to correct for instrumental dris, so the
selected as a representative commodity for other powder make- regression curves are based on the ratio of the analyte response to
up products, such as eye shadows or blushes. The concentration that of the IS. For each metal, the relationship between the
ranges covered by the validation study were: Cd 0.02–1.25 mg instrument response (y) and metal level (x) was described by the
kg1; Co 1.25–12.5 mg kg1; Cr 2.5–125 mg kg1; Ni 5.0–125 mg equation of the line, i.e., y ¼ a + bx, where a is the intercept with
kg1; Pb 0.25–12.5 mg kg1. The method was validated for the the y-axis and b is the gradient of the line. The correlation coef-
following performances: specicity, calibration linearity, limit cient (r) was used as a measure of the quality of the correlation
of detection (LoD) and limit of quantication (LoQ), repeat- between y and x values and b represented the sensitivity of the
ability and recovery.17–21 method. The regression data, reported in Table 3, showed r to be
very close to 1 for all the metals. Moreover, the linearity was
3.1 Specicity
tested by the examination of a plot of residuals produced by the
The specicity was investigated by analyzing the cosmetic linear regression and calculated by the difference:
samples and verifying the presence of interfering ions by
Residuals ¼ yi  ŷi
Table 1 HR-ICP-MS operative conditions
where yi was the value of y observed for the ith calibration point
Parameter Value and ŷi was the value of y calculated for the ith calibration point
by means of the equation of the line.22,23 For all the metals the
Sensitivity: 115In in LR $8  105 cps/1 mg L1 residuals were scattered randomly around zero (Fig. 1 shows the
Sensitivity: 115In in MR $7  104 cps/1 mg L1
example of Cr). The lack of t due to a nonlinear calibration
Sensitivity: 115In in HR $1.5  104 cps/1 mg L1
Oxides: 137Ba16O/137Ba in LR #0.02 function was also checked by means of a test of signicance
Double charged ions: 137Ba2+/137Ba in LR #0.07 (ANOVA) and for all the metals the calibration curve gave a large
Stability: 7Li, 115In and 238U in LR #1.5% F value showing a good linear relationship.23

Table 2 Specificity

Masses selected Resolution adopted Interferences resolved

Cd 114 LR Not interfered

40 19
Co 59 MR Ar F, 43Ca16O, 41K18O, 42Ca16O1H
40 12
Cr 52 MR Ar C, 36Ar16O, 38Ar14N, 35Cl17O, 37Cl15N, 35Cl16O1H
44
Ni 60 MR Ca16O, 23Na37Cl, 36Ar24Mg, 120Sn2+
Pb 208 LR Not interfered

This journal is ª The Royal Society of Chemistry 2013 Anal. Methods, 2013, 5, 402–408 | 403
 View Article Online

Analytical Methods Paper

Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

Fig. 1 Residuals plot: Cr.

3.3 Limit of detection and limit of quantication Repeatability RSD values were #11% for Cd, Co, Ni and Pb and
#25% for Cr and they were found to be dependent on the metal
To calculate the LoD and LoQ of the method, 20 analyses of the
level with a decreasing trend with increasing concentration. The
cosmetic and 10 analyses of the cosmetic added with the lowest
SD value was then useful in determining the repeatability limit
level of the calibration curve were analyzed. The LoD and LoQ
(Lrep) for each metal calculated as SD  tcrit where tcrit is the
values, expressed as 3  SD and 10  SD on the matrix, are
critical Student t value for n  1 degrees of freedom at 95%
reported in Table 4.24 The LoD values obtained were below the
condence (where n is the number of results used to estimate
rst calibration level and lower than their respective levels
found in the native samples (see Table 6, rst column). the repeatability).

3.4 Repeatability 3.5 Recovery

Because the dispersion of results may change with metal The recovery of the procedure was investigated using the forti-
concentration, the repeatability was checked at the level found ed samples because a cosmetic-based certied reference
in the native cosmetic and at 3 fortication levels obtained by material (CRM) was not commercially available. The inherent
spiking the native cosmetic with the concentrations (cspike) problem with this procedure is that metal introduced in such a
reported in Table 5. These spikes represented the lower (level I), way would probably not be held as strongly as that naturally
middle (level II) and higher (level III) parts of the linear range.25 present in the matrix. It is however the most common way of
Twenty analyses for both native and fortied samples were determining recovery efficiency in the absence of a suitable
performed and each analysis represented a complete applica- CRM and is recognised as an acceptable way of doing so.18,26,27
tion of the method, including sample preparation steps. Table 6 The recovery was tested at the same 3 fortication levels used
presents the mean concentration and the SD on the native for the repeatability study. The formula for calculating the
cosmetic (
cnative) and on the fortied samples (
cfortied). recovery was:

Table 4 Method detection limits in mg kg1 Table 5 Spike levels (cspike) in mg kg1

LoD LoQ I II III

Cd 0.004 0.012 Cd 0.047 0.24 1.18

Co 0.17 0.57 Co 0.47 2.36 11.8
Cr 0.23 0.78 Cr 4.72 23.6 118
Ni 0.30 1.00 Ni 4.72 23.6 118
Pb 0.03 0.10 Pb 0.47 2.36 11.8

404 | Anal. Methods, 2013, 5, 402–408 This journal is ª The Royal Society of Chemistry 2013
 View Article Online

Paper Analytical Methods

Table 6 Repeatability data Data showed that Cr was the only metal with low recovery. This
might be ascribed to a mixture of factors: (i) incomplete disso-
Fortied Fortied Fortied
lution of chromite (FeCr2O4); (ii) presence of refractory
Native I II III
compounds (TiO2, Al or other oxides) that may not be totally
Cd dissolved and may sequester Cr; (iii) high concentration of

c (mg kg1) 0.026 0.071 0.27 1.26 elements (Ca, Mg, Na, Fe, K, Zn) that may decrease the Cr
SD (mg kg1) 0.002 0.004 0.009 0.04 instrumental response due to ionization suppression; (iv) high
RSD (%) 9.0 5.8 3.4 3.0
Lrep (mg kg1) 0.007 0.012 0.03 0.11
concentration of carbonates (ZnCO3, CaCO3, MgCO3) that may
Co not be totally decomposed and may entrap Cr; and (v) adsorp-
Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

c (mg kg1) 1.61 2.11 4.05 14.4 tion of Cr on silica residues that may not be totally dissolved.
SD (mg kg1) 0.18 0.19 0.31 0.6 In general, the recovery for metals satised requisites
RSD (%) 11.4 8.5 7.6 4.2 reported in the Decision 2002/657/EC that accepted a recovery
Lrep (mg kg1) 0.55 0.55 0.92 1.8
Cr
bias between 50% and +20% for concentrations #1.0 mg kg1

c (mg kg1) 3.00 6.95 19.5 95.1 (Cd and Pb) and between 30% and +10% for concentrations in
SD (mg kg1) 0.75 1.33 1.4 6.4 the range 1.0–10 mg kg1 (Co, Cr and Ni).28
RSD (%) 25.1 21.1 7.8 6.8
Lrep (mg kg1) 2.24 3.94 4.1 19.1
Ni 4 Uncertainty

c (mg kg1) 6.87 12.1 29.5 117
SD (mg kg1) 0.74 1.01 2.19 7 The uncertainties of measurements were estimated using the
RSD (%) 10.7 8.3 7.4 5.7 in-house validation data. The uncertainty contributions
Lrep (mg kg1) 2.18 3.0 6.5 20 considered were those coming from the following studies:
Pb  repeatability;

c (mg kg1) 0.25 0.68 2.90 12.1
 recovery;
SD (mg kg1) 0.03 0.03 0.08 0.6
RSD (%) 10.1 4.1 2.8 4.6  calibration linearity.
Lrep (mg kg1) 0.07 0.08 0.24 1.6 Other uncertainty components not adequately covered by the
aforesaid studies were considered only if they deserved signi-
cfortified  cnative cant attention. Each component was quantied and incorpo-
rec ¼
cspike rated into the combined uncertainty that was calculated both
for the native cosmetic and for the 3 fortication levels used in
The mean recoveries, reported in Table 7, were: Cd, 96–
the validation study. Practical applications in the calculation of
104%; Co, 102–108%; Cr, 70–84%; Ni, 92–109%; Pb, 91–112%.
measurement uncertainty have been previously described by
Table 7 Recovery data the authors with reference to procedures for the determination
of metals in human blood and serum.29–31 The equations used
Spike for the calculation of the uncertainty components are described
Metal I II III in the sections below.

Cd
4.1 The relative uncertainty of the repeatability (urep)

cfortied 
cnative (mg kg1) 0.045 0.25 1.23
SD (mg kg1) 0.004 0.009 0.05 The urep for each metal was the RSD% on native and fortied
RSD (%) 8.9 3.6 4.1 samples calculated during the repeatability study as reported in
rec (%) 95.7 104.2 104.2 Table 6.
Co

cfortied 
cnative (mg kg1) 0.48 2.42 12.7
SD (mg kg1) 0.07 0.20 0.60 4.2 The relative uncertainty of the recovery (urec)
RSD (%) 14.5 8.3 4.7
rec (%) 102.1 102.5 107.6 The urec comprised both the contributions from the analytical
Cr measurements performed during the recovery estimation as

cfortied 
cnative (mg kg1) 3.97 16.6 90.5 well as from the concentration of the spike levels.32 So, the urec
SD (mg kg1) 1.02 1.28 7.0 was given by:
RSD (%) 25.7 7.7 7.7 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rec (%) 84 70 77 u   2ffi
u SD 2
=n u c
urec ¼ rec %t
spike
Ni 2 þ

cfortied 
cnative (mg kg1) 5.15 22.5 109 cfortified  cnative cspike
SD (mg kg1) 0.47 1.5 12
RSD (%) 9.1 6.7 11 where SD was obtained from the replicated analyses performed
rec (%) 109 95 92 during the recovery study (see Table 7). The u(cspike) was the
Pb
uncertainty in the preparation of the spike, which was a

cfortied 
cnative (mg kg1) 0.43 2.65 11.9
SD (mg kg1) 0.03 0.06 0.6 combination of the uncertainties related to the purity of the
RSD (%) 7.0 2.3 5.0 metals and to the volumetric pipettes, both reported in the
rec (%) 91 112 101 supplier’s certicate.

This journal is ª The Royal Society of Chemistry 2013 Anal. Methods, 2013, 5, 402–408 | 405
 View Article Online

Analytical Methods Paper

In addition, to determine whether the mean recovery for calibration point and
x was the mean of the xi values of the
each metal was signicantly different from 1.0, the following calibration points. The ucalib was nally expressed as % of the
equation was applied: prediction interval for the x concentration found in the test
samples.
j1  recj
tcalc ¼
urec
4.4 Other uncertainty contributions
and tcalc was compared with the Student tcrit value.32 If tcalc < tcrit
The uncertainty related to the weight of the test sample was
the recovery was not signicantly different from 1 and there was
considered and it was calculated using the uncertainty of the
no reason to correct the results for the recovery. However,
balance declared in the manufacture’s certicate and applying a
because there is still an uncertainty associated with the esti-
Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

rectangular distribution. Anyway, this contribution was very

mation of rec as the signicance test could not distinguish
small (i.e. 0.4%) compared to the other sources of uncertainty
between a range of values about 1.0, the uncertainty of the
mentioned above.
recovery (urec0 ) was calculated by:
tcrit  u rec
u rec0 ¼ 4.5 Combined (ucomb) and expanded uncertainty (U)
1:96
The relative uncertainties of repeatability, recovery and cali-
bration linearity were combined to obtain the full measurement
If tcalc > tcrit (and this happened for most of the metals and
uncertainty. The ucomb was calculated by:
concentration levels), the recovery is signicantly different from qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 and data should be corrected for the recovery efficiency. But, u comb ¼ u rep 2 þ u rec 2 þ u calib 2
since in the normal application of this method the recovery
correction was not included in the calculation of the nal result,
it was necessary to increase the uncertainty of the recovery U was the ucomb multiplied for the appropriate coverage
(urec0 0 ) to take account of this uncorrected bias. The relevant factor (k ¼ 2). Table 8 reports the contributions urep, urec and
equation was: ucalib and the U values at the native concentration and at the 3
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 fortication levels.
1  rec The expanded uncertainty ranged between 34 and 63% at the
u rec00 ¼ þu rec 2
2 native concentration, 31 and 54% at level I, 19 and 47% at level
II, and 14 and 34% at level III. The overall uncertainty was found
where 2 was the k coverage factor used to calculate the expanded
to be dominated by the terms of repeatability and calibration
uncertainty.
linearity for all the metals analyzed. In the case of Cr also the
Moreover, the uncertainty of recovery was propagated to the
recovery represented a signicant component of the total
measurements performed on the native cosmetic assuming that
the recovery bias observed for a spiked sample (in particular at
the lowest spike level) was truly representative of that for the Table 8 Summary of uncertainties (in%)
unspiked sample.
Metal Native I II III

4.3 The relative uncertainty of the calibration linearity Cd urep 9.0 5.8 3.4 3.0
urec 2.4 2.4 2.4 2.4
(ucalib)
ucalib 14.5 14.5 8.7 5.9
The residual standard deviation (SDresiduals) was calculated by ucomb 17.2 15.8 9.7 7.0
the formula: U 33.9 31.1 19.0 13.8
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Co urep 11.4 8.8 7.7 4.2
uP
un 2 urec 3.6 3.6 2.2 4.1
u y  y^i ucalib 14.5 14.3 8.3 5.8
ti¼1 i
SDresiduals ¼ ucomb 18.8 17.2 11.5 8.3
n2 U 37.2 33.8 22.9 16.3
Cr urep 25.1 19.1 7.2 6.8
where n was the number of the calibration points.23 SDresiduals
urec 11.3 11.3 21.2 15.1
was then used as an estimate of the prediction interval for the ucalib 14.4 15.2 9.2 5.8
sample that gives a y signal by the equation: ucomb 31.1 26.9 24.2 17.6
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi U 62.7 53.8 47.5 34.5
u
SDresiduals u 1 1 ðy  yi Þ2 Ni urep 10.7 8.3 7.4 5.7
u þ þ
b tN n P n urec 4.8 4.8 2.9 4.5
b2 ðxi  xÞ2 ucalib 14.6 14.4 8.7 6.0
i¼1
ucomb 18.8 17.3 11.8 9.4
where: b was the gradient of the calibration line, N was the U 37.0 34.1 23.4 18.6
Pb urep 10.1 3.9 2.9 4.6
number of measurements made on the test sample, n was the
urec 5.4 5.4 4.9 1.5
number of the calibration points,
y was the mean of N ucalib 14.4 14.5 8.6 5.6
measurements of y for the test sample,
yi was the mean of the yi ucomb 18.4 15.9 10.3 7.4
values of the calibration points, xi was the x value for the ith U 36.2 31.3 20.2 14.6

406 | Anal. Methods, 2013, 5, 402–408 This journal is ª The Royal Society of Chemistry 2013
 View Article Online

Paper Analytical Methods

Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

Fig. 2 Single sources of uncertainty at various concentration levels (n ¼ native; I; II; III ¼ fortification levels).

powder cosmetic was validated in-house by evaluating the most

important performance capabilities such as specicity, detec-
tion limits, repeatability, recovery and linearity. The measure-
ment uncertainty was then expressed by using the previously
estimated characteristics and the law of propagation of uncer-
tainty. All the uncertainty sources resulted in being included in
the three components of repeatability, recovery bias and line-
arity. The validation of this method and the estimate of its
uncertainty were key steps to rationalize the experimental
efforts, improve the quality of the data, and help making
decisions regarding toxic contaminants in cosmetics.
Fig. 3 Expanded uncertainty (U) profile for the validated concentration
range: Cd.
Acknowledgements
The authors are grateful to Federico Pecoraro for assistance in
uncertainty. The metals had a smaller uncertainty at the highest the uncertainty calculation.
concentration (see Fig. 2).

References
4.6 Reporting the expanded uncertainty
The Eurachem Guide stated that each result x should be stated 1 A. C. de Groot, E. G. Beverdam, C. T. Ayong, P. J. Coenraads
together with its expanded uncertainty U(x).20 The following and J. P. Nater, Contact Dermatitis, 1988, 19, 195.
form was recommended: x  U(x). In the case of the powder 2 R. H. Guy, J. J. Hostynek, R. S. Hinz and C. R. Lorence, Metals
cosmetic, the results accompanied by their uncertainty were as and the Skin: Topical Effects and Systemic Absorption, Marcel
follows (in mg kg1): Cd, 0.026  0.009; Co, 1.61  0.60; Cr, Dekker, Inc., New York, USA, 1999.
3.00  1.88; Ni, 6.87  2.53; Pb, 0.25  0.09. 3 P. A. Meyer, M. J. Brown and H. Falk, Mutat. Res., 2008, 659,
Furthermore, it was possible to dene the U prole for the 166.
validated range of concentrations (U vs. conc.). The calculated 4 R. V. Sprinkle, J. Fam. Pract., 1995, 40, 358.
regression lines were: Cd, y ¼ 0.130x + 0.011 (see Fig. 3 as an 5 G. M. Mojdehi and J. Gurtner, Am. J. Public Health, 1996, 86,
example); Co, y ¼ 0.135x + 0.403; Cr, y ¼ 0.330x + 1.603; Ni, y ¼ 587.
0.170x + 1.762; and Pb, y ¼ 0.142x + 0.111. These equations can 6 G. Forte, F. Petrucci and B. Bocca, Inammation Allergy: Drug
be used to assess the U for any powder cosmetic sample having Targets, 2008, 7, 145.
a metal concentration within that range. 7 G. Forte, F. Petrucci, A. Cristaudo and B. Bocca, Sci. Total
Environ., 2009, 407, 5997.
5 Conclusions 8 B. Bocca, G. Forte, F. Petrucci and A. Cristaudo, J. Pharm.
Biomed. Anal., 2007, 44, 1197.
The present methodology, based on MW digestion and HR-ICP- 9 J. Stauber, T. Florence, B. Gulson and L. Dale, Sci. Total
MS, for the determination of Cd, Co, Cr, Ni and Pb in the face Environ., 1994, 145, 55.

This journal is ª The Royal Society of Chemistry 2013 Anal. Methods, 2013, 5, 402–408 | 407
 View Article Online

Analytical Methods Paper

10 F. Larese Filon, G. Maina, G. Adami, M. Venier, N. Coceani, 22 ISO 11095, Linear Calibration Using Reference Materials, ISO,
R. Bussani, M. Massiccio, P. Barbieri and P. Spinelli, Int. Geneva, Switzerland, 1996.
Arch. Occup. Environ. Health, 2004, 77, 85. 23 LGC/VAM/2003/032, Preparation of Calibration Curves, A
11 R. S. Shelnutt, P. Goad and D. V. Belsito, Crit. Rev. Toxicol., Guide to Best Practice, LGC, Teddington, UK, 2003, pp. 1–27.
2007, 37, 375. 24 ISO 11843-1, Capability of Detection – Part 1 Terms and
12 A. Emilson, M. Lindberg and B. Forslind, Acta Derm.- Denitions, ISO, Geneva, Switzerland, 1997.
Venereol., 1993, 73, 203. 25 ISO 3534-1, Statistics – Vocabulary and Symbols – Part 1:
13 F. Larese Filon, M. Boeniger, G. Maina, G. Adami, P. Spinelli General Statistical Terms and Terms Used in Probability, ISO,
and A. Damian, J. Occup. Environ. Med., 2006, 7, 692. Geneva, Switzerland, 2006.
Published on 02 October 2012. Downloaded by University of Newcastle on 18/04/2017 12:24:28.

14 Council Directive 76/768/EEC of 27 July 1976, Official Journal 26 M. Thompson, S. L. R. Ellison, A. Fajgelj, P. Willetts and
L 262 of 27/09/1976, pp. 169–200. R. Wood, Pure Appl. Chem., 1999, 71, 337.
15 Bundesgesundheitsblatt (Federal Health Journal, Germany), 27 ISO 5725, Standard Accuracy (Trueness and Precision) of
1985, vol. 28, p. 216. Measurement Methods and Results, Part 1–6, ISO, Geneva,
16 D. A. Basketter, G. Angelini, A. Ingber, P. S. Kern and Switzerland, 1994.
T. Menné, Contact Dermatitis, 2003, 49, 1. 28 Commission Decision 2002/657/EC implementing Council
17 AOAC Peer Veried Methods Advisory Committee, AOAC Peer Directive 96/23/EC concerning the performanceof
Veried Methods Program – Manual on Policies and Procedures, analytical methods and the interpretation of results. Off. J.
AOAC International, Gaithersburg, USA, 1998, pp. 1–35. Eur. Communities: Legis., 2002, 221/8.
18 Eurachem Guide. The Fitness for Purpose of Analytical Methods. 29 B. Bocca, D. Mattei, A. Pino and A. Alimonti, Rapid Commun.
A Laboratory Guide to Method Validation and Related Topics, Mass Spectrom., 2011, 25, 453.
LGC, Teddington, UK, 1998. 30 B. Bocca, D. Mattei, A. Pino and A. Alimonti, Curr. Anal.
19 ISO/IEC 98, Guide to the Expression of Uncertainty in Chem., 2011, 7, 269.
Measurement (GUM), ISO, Geneva, Switzerland, 1995. 31 B. Bocca, D. Mattei, A. Pino and A. Alimonti, Rapid Commun.
20 Eurachem/Citac Guide CG 4. Quantifying Uncertainty in Mass Spectrom., 2010, 24, 2363.
Analytical Measurement (QUAM), ed. S. L. R. Ellison, M. 32 V. J. Barwick and S. L. R. Ellison, VAM Project 3.2.1
Rosslein and A. Williams, LGC, Teddington, UK, 2000. Development and Harmonisation of Measurement Uncertainty
21 ISO/IEC 17025, General Requirements for the Competence of Principles. Part (d): Protocol for Uncertainty Evaluation from
Testing and Calibration Laboratories, ISO, Geneva, Validation Data, Report no: LGC/VAM/1998/088, LGC,
Switzerland, 2005. Teddington, UK, 2000, pp. 1–90.

408 | Anal. Methods, 2013, 5, 402–408 This journal is ª The Royal Society of Chemistry 2013