Rapid screening of arsenic species in urine from exposed human by inductively 1 coupled plasma mass spectrometry with germanium as internal standard 2

10 In the present work, internal standardization based on species-unspecific isotope dilution 11 analysis technique is proposed in order to overcome the matrix effects and signal drift originated in 12 the speciation of As in urine by HPLC-ICP-MS. To this end, Ge has been selected as a pseudo13 isotope of As. The resulting mass flow chromatogram of the element allows the calculation of the 14 corrected overall species concentrations without requiring any methodological calibration, 15 providing high-throughput sample processing. The validation was carried out by analyzing a blank 16 human urine fortified at three concentration levels and an unspiked human urine sample containing 17 different species of arsenic. In all cases, recoveries ranging from 90 to 115% and RSD below 10% 18 were attained with this approach. Furthermore, the proposed method provided results in excellent 19 agreement with those obtained using standard additions and internal standard calibration, allowing a 20 fast way to assess human exposure to arsenic species. 21


Introduction
2][3][4] Urine is regarded as an important biomarker of arsenic intake.][7][8][9][10] However, signal drift and matrix effects are observed due to urine matrix, hampering the quantification of such species. 11 use of species-unspecific isotope dilution analysis allows the correction for those errors, providing accurate and precise determinations of the sought element.In the case of arsenic, there is only one isotope available to be measured (m/z 75); therefore, an internal standard of an element close to the analyte mass has to be selected to follow the isotope dilution procedure. 12This approach, which we have called "pseudo-unspecific isotope dilution analysis", could make possible to obtain the accurate concentrations of the above mentioned species in a single run.
The developed method was applied for the analysis of human urine samples fortified at different concentration levels and compared with internal calibration and the standard additions method.

Instrumentation
The HPLC system consisted of an Agilent 1100 Series (Agilent, Waldbronn, Germany) binary pump and auto injector with a programmable sample loop (100 µL maximum).The separations were performed on a PRP-X100 (Hamilton, Reno, NV, USA) anion-exchange column (250 x 4.1 mm, 10 µm).An additional Agilent 1100 Series binary pump was used to continuously add the internal standard solution.
The outlet of the chromatographic column was connected through a T piece to a Meinhard concentric nebulizer.An Agilent 7500cx inductively coupled plasma mass spectrometer (Agilent Technologies, Tokyo, Japan) equipped with an octopole reaction cell using helium as a reaction gas to reduce polyatomic interferences on arsenic was used in this work.For HPLC-ICP-MS data acquisition, the "time resolved analysis" mode was used with 1 second of integration time per mass.
For tuning of ICP-MS, a solution containing 10 µg L -1 of As made up in double deionized water filtered through 0.45 µm was monitored at m/z 75 and 72; the ion intensity, resolution and mass axis were optimized.

Standards and reagents
Arsenite (As III ), arsenate (As V ), dimethylarsinic acid (DMA) and arsenobetaine form (AsB) were delivered by Fluka (Buchs, Switzerland), while monomethylarsonic acid (MMA) was from Carlo Erba (Milano, Italy).The stock solutions of arsenic species containing about 1000 mg L -1 of As were prepared in water and maintained at 4 ºC after standardization against an atomic absorption arsenic standard solution (J.T. Baker, Phillipsburg, USA). 13Appropriate dilutions of the stock solution were prepared daily, by weight, using double deionized water to obtain the required concentration.Germanium internal standard and nitric acid was purchased from Fluka.
A previously developed method 5 based on HPLC-ICP-MS coupling system was optimized in order to separate the five arsenic species in human urine.The mobile phase consisted of 4 mM ammonium phosphate (Merk, Darmstad, Germany), 4 mM ammonium hydrogen carbonate (Fluka) and 4 mM ammonium sulfate (Sigma, St. Louis, MO, USA).The pH value was adjusted to 8.9 by the addition of ammonium hydroxide (Trace Select, Fluka).These solutions were filtered through a 0.45 µm membrane before use.

Analytical procedure
Urine samples were diluted 5 fold with 0.1 % nitric acid before injection.The flow coming from the column (0.95 mL min -1 ) was mixed with the internal standard solution containing around 15 ng g -1 of Ge and 4 ng g -1 of As (0.15 mL min -1 ).The signals for m/z 75 and 72 were monitored over time.After smoothing of the data using moving average (n = 5) in order to reduce noise level the isotope ratio 75 As/ 72 Ge was calculated.Then, the on-line pseudoisotope dilution equation was applied to each point of the chromatogram to obtain the mass flow chromatogram.The amount of arsenic in each fraction was determined by integration of the chromatographic peaks using the Origin 5.0 software (Microcal Software Inc., Northampton, MA, USA).Finally, the concentration of arsenic was computed by dividing the As amount found by the injection volume.

Selection of internal standard and development of the equation for on-line pseudo-isotope dilution analysis
In order to appropriately correct for matrix-induced signal enhancement or suppression as well as for drift instability of the instrument, the analyte and the internal standard should undergo an equal relative signal intensity drift.To this end, mass-to-charge ratio and ionization potential of both elements should be as close as possible, being especially critical the first factor.In this regard, selenium seems the best candidate to use as internal standard for arsenic speciation. 12,14However, this element is often present in urine samples, which can produce errors in the normalization.By contrast, germanium is rarely present in urine samples and has been satisfactory used for matrix effects correction. 15As a consequence, the isotope 72 Ge was selected as internal standard.
The proposed procedure is based on post-column isotope dilution analysis. 16Briefly, this technique consists in the on-line addition of an isotopically enriched solution of the sought element after the chromatography separation to modify the original isotope abundances in the sample.The resulting isotope ratio (in the mixture) of the most abundance isotope in the sample and the spike permits to calculate the endogenous concentration contained in each chromatographic peak.In the case of arsenic, a germanium internal standard is used instead of a spike, owing to its monoisotopic character.Since both elements have different ionization efficiencies, the experimental isotope ratio in the mixture R m ( 75 As/ 72 Ge) will not provide the As/Ge molar ratio.The instrumental response of Ge present in the mixture must be previously normalized to As in order to correlate R m with the analyte concentration.For this purpose, a known amount of As was added within the internal standard solution and the corresponding isotope ratio R IS ( 75 As/ 72 Ge) was measured.Consequently, R m and R IS can be expressed as follows: were As s N (mol g -1 ) shows the amount of As in the sample with density d s (g mL -1 ) pumped at a flow rate f s (mL min -1 ), which is mixed with As IS N (mol g -1 ) of As arising from the internal standard solution pumped at a flow rate f IS (mL min -1 ) and density d IS (g mL -1 ).The term is the concentration of Ge normalized to As, which must not be confused with the true amount of Ge.Indeed, ' Ge IS N would represent the mol g -1 of 72 Ge contained in the internal standard solution if the ionization efficiency were the same as As.When we combine eqns.( 1) and ( 2) the following expression is obtained: Please note that the true amount of Ge is not needed in the calculation.Rearranging eqn.
(3) for As s N , we obtain:   has the units of ng min -1 and it is the mass flow of the sample eluting from the column, MF s .Then, the final pseudo-isotope dilution equation has the form: If the analyte concentration changes with time, e.g., during the chromatographic peak, MF s will also change with time.The integration of the chromatographic peak in the mass flow chromatogram will give the amount of As in that fraction.The concentration is then easily calculated knowing the sample volume injected.Eqn. ( 6) was thus used for calculations in the present work.

Analytical results
In a preliminary study, the concentration of the As primary standard in the internal standard solution was optimized.On the one hand, it has to be taken into account that a high enough amount of exogenous As is required to minimize the m/z 75 background influence.
On the other hand, the higher amount of 75 As coming from the post-column solution the higher baseline noise, leading to poorer detection limits of the endogenous species.Thereby, a concentration of ca. 4 ng g -1 was selected as a compromise.The case of Ge internal standard concentration is much less critical because it does not contribute to the signal of the analyte eluting from the column and does not influence the final results.In addition, the chromatographic and post-column flow rates were tested.The final values used for the mobile phase and the internal standard solution (0.95 and 0.15 mL min -1 , respectively) allowed the elution of all the species in a proper time without sacrificing the accuracy of the post-column flow rate.
To a better understanding of the procedure, Fig. 1 illustrates the conversion from original ICP-MS intensities to mass flow.The chromatograms corresponding to m/z 75 and 72 (a) are first transformed into the isotope ratio chromatogram (b).It is worth stressing that 75 As and of the isotope ratio chromatogram provides R IS (see Fig. 1b).Next, equation ( 6) is applied to the whole chromatogram.Finally, the mass flow peaks of Fig. 1c are integrated and divided by the injection volume (50 µL).It should be remarked that the present strategy permits to correct for errors derived from instrumental instabilities and matrix effects in the whole chromatogram, since the Ge internal standard is continuously added to the effluent from the column.
The proposed procedure was applied to the analysis of a blank urine sample spiked at 1, 5 and 10 µg As L -1 (concentrations referred to the diluted urine injected).It is worth noting that the blank urine used corresponds to an unexposed human and no As species were found when it was analyzed by the conventional calibration method.Ten replicates for each of the three fortification levels were carried out.Additionally, in order to check the suitability of 72 Ge as internal standard the results were compared with those obtained using internal standard calibration.To this end, calibration standards containing 0-20 µg L -1 of As for each compound were injected by triplicate within the post-column solution.Then, the isotope ratio chromatogram was plotted (as exemplify in Fig. 1b).Satisfactory recoveries, between 90 to 105%, were obtained both for the medium and highest fortification levels when pseudounspecific IDA was used.At the lowest fortification level, which was closed to the detection limit, recoveries were in the range of 96-115% and coefficients of variation were below 10% (Table 1).No significant differences were noticed between both methods, thus it seems that calibration-free measurements based on eqn.( 6) can be performed for the quantification of As species in urine.
Intermediate precision (n = 9) was also estimated by analyzing replicates of the medium fortification level on 3 different days.The coefficient of variation was found to be <8% in all cases.Detection limits, defined as three times the signal-to-noise ratio in the mass flow chromatogram were determined for the blank urine sample spiked at 1 µg L -1 of each As species.As can be seen in Table 1, LODs were <0.7 µg L -1 in the diluted urine.In fact, the continuous addition of arsenic post-column to normalize de germanium response increase notably the detection limits.However, these values are satisfactory to evaluate the potential risk of people exposed to inorganic arsenic.Actually, the American Conference of Governmental Industrial Hygienists (ACGIH) and Deutsche Forschungsgemeinschaft (DFG) set the BEI and BAT values for occupational arsenic exposure as 35 µg As L -1 and 50 µg L -1 , using the sum of inorganic arsenic, MMA and DMA. 17 The validity of the proposed method for correcting instrument signal drift was tested with the injection of a 15-h batch run.Fig. 2a compares the intensity chromatograms corresponding to the first and last injections of the experiment.Appreciable signal drift was observed between them, resulting in lower peak areas for the second injection.The application of eqn.( 6) lead to the mass flow chromatograms presented in Fig. 2b.As expected, no significant differences were noticed between injections, which confirm that appropriate correction of signal drift is achieved.It is worth mentioning the severe signal suppression at the dead volume caused by the high salt content of urine sample (Fig. 2a).In this case, instrumental instability was not totally overcame (Fig. 2b), most probably because the Ge internal standard suffers more signal depression than As in the presence of high concentration of Na.Anyway, such anomalous behavior does not affect any chromatographic peak Finally, the quantification of a human urine sample containing different species of arsenic was performed by the present methodology, internal standard calibration and standard additions (Table 2).The concentrations calculated by pseudo-unspecific IDA were in very good agreement with those obtained using the other quantification strategies.These data confirm the suitability of the developed procedure for arsenic speciation studies in human urine.

Conclusions
A new procedure for the simultaneous determination of AsB, As III , As V , DMA and MMA in human urine which does not require any methodological calibration graph and allows correcting for instrumental instabilities has been developed.For this purpose, species unspecific isotope dilution analysis has been adapted to As using 72 Ge as an additional isotope of the sought element.
The proposed method has been successfully validated in spiked and unspiked human urinesamples.In addition, the results were in excellent agreement with internal standard calibration and standard additions.
The need for addition a known amount of As to normalized the Ge concentration increases appreciably the LODs.This fact however, do not hampers the correct quantification of toxic inorganic species of arsenic in urine of exposed humans Therefore, the possibility to carry out the quantification of As species in a single run provided by the developed procedure could be very useful to assess workplace, drinking water or food exposure to inorganic arsenic.Table 1.Analytical characteristics of the pseudo-unspecific IDA procedure obtained for a blank urine sample fortified at three different levels.
) Concentrations in mol g -1 can be expressed as concentrations in weight by taking into account the atomic weight of the element.Since the atomic weight of As in the sample ( As s AW ) and in the internal standard solution ( As IS AW ) are the same, eqn.(4) becomes: concentrations (ng g -1 ) of As in the sample and internal standard solution, respectively.

Figure 1 .
Figure 1.-Conversion process from intensities to mass flow using the pseudo-isotope dilution equation.

Figure 2 .
Figure 2.-Use of Ge as a pseudo-isotope of As to correct for instrumental signal drift observed during a 15-h batch run.

Figure 1 a
Figure 1

Figure 2 a
Figure 2

Table 2 .
The uncertainty in the values corresponds to 1 s standard deviation of 10 independent HPLC-ICP-MS injections.Comparison of different methodologies to correct for matrix effects in the analysis of a human urine sample.
a b Recoveries calculated using internal standard calibration, for comparison.c Detection limits referred to diluted urine sample.