Direct multi-element analysis of a fluorocarbon polymer via solid sampling-electrothermal vaporization-inductively coupled plasma mass spectrometry

Abstract

This paper reports on the performance of solid sampling-electrothermal vaporization-inductively coupled plasma mass spectrometry for the direct multi-element analysis of two different perfluorosulfonic acid/TFE copolymer samples, which were selected in order to test the potential of this technique for routine control of fluorocarbon polymers. Careful selection of the most suitable isotopes permits the reliable monitoring of the analytes of interest: Cr, Cu, Fe, K, Mn, Pb and Zn. The use of Pd as chemical modifier allows stabilization of all of these analytes during the pyrolysis step (at 800 °C), enabling adequate matrix removal, while the use of a high vaporization temperature (2700 °C) is required for the efficient simultaneous vaporization of these elements. Moreover, the ¹⁰⁵Pd⁺ signal can be used as internal standard, correcting for possible sensitivity drifts. Under these conditions, straightforward calibration with aqueous standard solutions was feasible for all of the elements investigated. The method thus developed exhibits interesting features, such as a low detection limit (ng g⁻¹ range) for most elements, a high sample throughput (15 min per determination), a low sample consumption (a few milligrams only), precision values usually in the 7–12% RSD range and the absence of any sample pretreatment, with the subsequent lower risk of analyte losses or contamination. Therefore, it seems to offer a promising alternative for the laborious procedures currently in use for analysis of these complex samples.

---

1. Introduction

Fluorocarbon polymers have numerous industrial applications owing to their excellent characteristics: inertness, high resistance towards corrosive reagents (e.g., acids and oxidizing agents), thermal stability, ageing resistance and low coefficient of friction.^1–3 Moreover, these materials can be produced with a high purity.⁴ As a result of these qualities, they are extensively used in the semiconductor industry.⁵

For their use in this industry, it is necessary to control the level of metallic impurities in these materials according to the SEMI (Semiconductor Equipment and Materials International) guidelines and, precisely due to the characteristics previously mentioned, this kind of analysis can be complicated. As pointed out before, these materials are difficult to dissolve using acids. The protocol recommended by SEMI (SEMI F48-0600)⁶ comprises several steps preceding the final measurement, which is preferably carried out with inductively coupled plasma mass spectrometry (ICPMS) owing to its high sensitivity and multi-element capabilities. These steps include: (i) calcination of the materials during several hours (up to 18 h, depending on the polymer type) in a platinum crucible; (ii) acid digestion of the ashes with 12 mol L⁻¹ HCl; (iii) evaporation of the digest thus obtained to dryness and addition of 14 mol L⁻¹ HNO₃ for complete dissolution. This procedure is not only time-consuming, but can result in analyte losses for the more volatile elements during the calcination step (500 °C–650 °C) and/or the evaporation step. In addition to this, contamination may be an issue for some elements, which may be present at very low levels (ng g⁻¹). The use of a high purity quartz combustion system has been proposed as an alternative.⁷ One of the problems of this procedure is the release of significant amounts of HF during the combustion of these materials, leading to possible damage to the quartz furnace lining.

Taking into account all these issues, the development of direct solid sampling methods for these materials seems to be a desirable goal. In this way, it would be possible to improve the sensitivity (by avoiding the dilution that accompanies sample digestion), considerably increase the sample throughput (by eliminating the digestion steps) and reduce the risk of contamination/losses. Among the different solid sampling methods that could be suited for this task, solid sampling-electrothermal vaporization-ICPMS (SS-ETV-ICPMS) seems especially appropriate, combining excellent limits of detection with multi-element possibilities.^8,9 Moreover, in spite of the problems described previously for its digestion, this kind of fluoropolymeric matrix should be relatively easy to remove during the pyrolysis step in a graphite furnace, opening the possibilities for the selective vaporization of the analytes.^10,11 If this latter condition is fulfilled, it has been shown that straightforward calibration using aqueous standard solutions may be feasible,^9,11,12 although sometimes the use of standard addition have been recommended.¹³

This paper aims to investigate the potential of solid sampling-ETV-ICPMS for fast and reliable analysis of fluorocarbon polymers. In order to achieve this goal, and due to the lack of certified reference materials, two different perfluorosulfonic acid/tetrafluoroethylene (TFE) copolymer samples were selected for this study. These samples were chosen because they were known to contain measurable quantities of the target metallic impurities. The samples were also analyzed at Solvay’s lab using pneumatic nebulization-sector field-ICPMS, following sample digestion according to the SEMI F48 guideline described earlier.⁶ Therefore, reliable reference values were available for the different analytes of interest, namely Cr, Cu, Fe, K, Mn, Pb and Zn, making it possible to evaluate the performance of SS-ETV-ICPMS for these kind of samples.

The very different behavior of the analytes in a graphite furnace together with their significantly different levels of concentration (from a few ng g⁻¹ for Pb to hundreds of μg g⁻¹ for K) present a challenge. Optimization of the instrumental parameters and the use of Pd, both as chemical modifier and as internal standard, were investigated in order to permit the simultaneous determination of these elements, the goal always being to develop a procedure that allows straightforward calibration with aqueous standard solutions.

2. Experimental

2.1. Instrumentation

All of the ETV-ICPMS measurements were carried out at Ghent University. A Perkin-Elmer (Boston, USA) HGA-600MS electrothermal vaporizer coupled to a Perkin-Elmer Sciex Elan 5000 ICP-mass spectrometer (Concord, Canada) was used for these measurements. The characteristics of the ETV device and its connection to the ICPMS were described in a previous work.¹³ Pyrolytic graphite-coated tubes and cups (cup-in-tube technique for solid sampling) were used in combination with this device. A microbalance (Sartorius M3P, Göttingen, Germany) with a readability of 1 μg was used for weighing the samples.

On the other hand, the dissolution of the samples and the analysis of the digests subsequently obtained were carried out in a class 10 000 clean room at Solvay. An Element 2 (Thermo Electron, Bremen, Germany) sector field-ICPMS instrument was used for this analysis.

2.2. Samples and standards

2.2.1. Standards and reagents. Water was doubly distilled and further purified using a Milli-Q water purification system (Millipore, Billerica, USA). Analyte and modifier (Pd) solutions were prepared daily by appropriate dilution of 1 g L⁻¹ single-element standard solutions (Merck, Darmstadt, Germany) with 0.14 mol L⁻¹ HNO₃. 14 mol L⁻¹ HNO₃ and 12 mol L⁻¹ HCl were purified by sub-boiling distillation in quartz equipment.

2.2.2. Samples. The two samples investigated are perfluorosulfonic acid/TFE copolymers and were provided by the Analytical Technologies Division of Solvay Research & Technology (Brussels, Belgium). They were available as granules, each weighing approximately 15 mg. Both samples are practically identical except for their level of metallic impurities, as will be shown in Section 3.4. They will be referred to as Sample 1 and Sample 2.

2.3. Procedure for the analysis of the samples by SS-ETV-ICPMS

A ceramic knife was used to cut the plastic granules into smaller pieces (of about 1–2 mg). After weighing, the material was loaded into a sample cup using a pair of tweezers. The sample cup was then inserted into the furnace for subsequent analysis using an insertion tool available from Perkin-Elmer. Finally, 10 μL of 20 mg L⁻¹ palladium solution (0.2 μg Pd) were added into the graphite cup using the autosampler. The operating conditions are summarized in Table 1.

Table 1 Instrumental operating conditions and data acquisition parameters for the ETV-ICP-MS measurements

a Added as Pd(NO3)2. b Signal used as internal standard. c Monitored for diagnostic purposes.
ICP-mass spectrometer
RF power	1200 W
Plasma flow rate	12 L Ar min^−1
Auxiliary flow rate	1.2 L Ar min^−1
Carrier flow rate	1.4 L Ar min^−1
HGA-600MS electrothermal vaporizer
Sample mass	1–2 mg
Chemical modifier	Pd (0.2 μg)^a
Temperature program	Temperature	Ramp	Hold time
Drying step	150 °C	10 s	20 s
Pyrolysis step	800 °C	10 s	30 s
Vaporization step	2700 °C	0 s	5 s
Cleaning step	2700 °C	1 s	3 s
Data acquisition
Scanning mode	Peak hop transient
Dwell time per acquisition point	20 ms
Isotopes monitored	^53Cr^+, ^63Cu^+, ^57Fe^+, ^39K^+, ^55Mn^+, ^208Pb^+, ^68Zn^+, ^105Pd^+,^b (^80Ar2^+, ^13C^+)^c

---

In all cases, five replicates (every replicate represents the measurement of 1 solid sample) were carried out for each determination, and the median of the five was taken as the representative value in order to minimize the possible influence of outliers.¹⁴ Calibration was carried out against aqueous standards solutions using the ¹⁰⁵Pd⁺ signal as internal standard. The data used to construct the calibration curve (based on 3 standard solutions and a blank) were gathered in the beginning of every working session and, afterwards, the measurements of all of the solid sample replicates were carried out. Integrated peak area (counts) was selected as the measurement mode in all situations.

2.4. Procedure for dissolution and analysis of the samples for comparison purposes

Analysis of the samples under investigation was also carried out following the SEMI F48-0600 guideline⁶ in order to obtain reference values. Three samples of each material were weighed (∼0.5 g) into Pt crucibles and subsequently heated on a Bunsen flame for approximately 30 min, in order to calcine the polymer matrix without allowing the crucibles to become red hot. The crucibles were afterwards introduced into a muffle furnace and kept at 585 °C for 12 h in order to eliminate any residual carbon. After cooling down, the residues were taken up in 2 mL of 12 mol L⁻¹ HCl. This solution was evaporated to dryness on a hotplate and the remaining residue dissolved in 1 mL of hot 14 mol L⁻¹ HNO₃. At this point, Sc, In and Bi were added to this solution as internal standards. Finally the solution was diluted to 50 mL with Milli-Q water. Analysis of the resulting solution was carried out using sector field-ICPMS.

3. Results and discussion

3.1. Preliminary experiments

The first experiments aimed at assessing the potential of ETV for the vaporization of the polymeric materials subject to study. The two samples investigated are practically identical except for their content of metallic impurities (Sample 1 shows lower levels for most of the target elements) and, therefore, they showed the same furnace behavior. By monitoring the ¹³C⁺ signal, further confirmed by the smoke coming out the tube, it was observed that the matrix begins to vaporize at 650–700 °C. A temperature of 800 °C is sufficient for efficient matrix removal, leaving no residue in the cup. Therefore, in theory, the use of this pyrolysis temperature should minimize any matrix effect.

The selection of the isotopes most suited for the determination of the analytes of interest is a very relevant issue, owing to possible spectral overlap. The samples show a high level of S, F and C, in addition to the high amount of C that is always vaporized in an electrothermal vaporizer, especially if a high vaporization temperature is used. Although most of the matrix components should be removed during the pyrolysis, it would be preferable to avoid any overlap between the analyte signals and those of polyatomic species containing these elements whenever possible, considering the low levels at which many of the analytes may be present. A list of potential interferences is shown in Table 2 and the isotopes finally selected are listed in Table 1. For Mn, the only isotope available, ⁵⁵Mn⁺, should not be affected by serious interferences. For Pb, the most abundant isotope (²⁰⁸Pb⁺) should also be suitable. On the other hand, for Cr, it has already been reported that monitoring of ⁵³Cr⁺ results in the best detection limits when using ETV-ICPMS, owing to the occurrence of ⁴⁰Ar¹²C⁺, which affects the use of the most abundant isotope (⁵²Cr⁺).^15,16 In a similar way, monitoring of ⁵⁷Fe⁺ is preferred since it is well-known that the ⁵⁶Fe⁺ signal suffers from spectral overlap from ⁴⁰Ar¹⁶O⁺. For Cu, the most abundant isotope (⁶³Cu⁺) was selected since the Na level in this sample (and hence also the ⁴⁰Ar²³Na⁺ signal intensity) is expected to be low. Moreover, the use of the ⁶⁵Cu⁺ signal is hampered by the overlap with those of ³²S³³S⁺ and ³²S³²O₂¹H⁺, among others (see Table 2). These interferences were indeed observed to be relevant for these samples, leading to high baselines and irreproducible data. In the case of Zn, all of the isotopes are affected by the same problem, overlap with sulfur-containing polyatomic ions. Taking into account the low abundance of ³⁴S⁺ (4.2%) compared to that of ³²S⁺ (95%), the monitoring of ⁶⁸Zn⁺ should be preferable. Finally, K is the analyte most affected by spectral interferences. The signal intensity for the most abundant isotope (³⁹K⁺) overlaps with the signal for the hydride of a minor isotope of Ar (³⁸Ar¹H⁺) and is also affected by the tailing of ⁴⁰Ar⁺ and ³⁸Ar⁺ signals. These interferences are severe and govern the limit of detection that can be obtained by means of this technique for K, especially considering that the situation for the other K isotopes is even worse. Fortunately, this is the element that is present at the highest levels in the samples (tens or even hundreds of μg g⁻¹) and it should be feasible to carry out a reliable determination at these contents. Fig. 1 shows the transient signal obtained for an aliquot of K standard solution containing 50 ng of K. The baseline signal demonstrates the magnitude of the interference. However, it is also clear that the isotope ³⁹K⁺ should be suitable for monitoring tens of nanograms of K.

Fig. 1 Influence of Ar-based interferences on the ³⁹K⁺ signal (50 ng K) under the conditions used in this work (see Table 1).

Table 2 Available isotopes for the target elements and list of the most relevant potential interferences considering the matrix composition

Isotopes (abundance)	Possible interferences
^50Cr (4.3%)	^32,33,34S^18,17,16O^+, ^36Ar^14N^+
^52Cr (83.8%)	^40Ar^12C, ^36Ar^16O^+, ^34,36S^18,16O^+, ^38Ar^14N^+
^53Cr (9.6%)	^40Ar^13C^+
^54Cr (2.4%)	^54Fe^+, ^38Ar^16O^+, ^40Ar^14N^+
^63Cu (69.1%)	^40Ar^23Na^+
^65Cu (30.9%)	^33S^32S^+, ^32S^32S^1H^+, ^32S^32O2^1H^+, ^40Ar^12C^13C^+
^54Fe (5.8%)	^54Cr^+88Ar^16O^+, ^40Ar^14N^+
^56Fe (91.7%)	^40Ar^16O^+40Ca^16O^+
^57Fe (2.2%)	^40Ar^17O^+40Ca^17O^+, ^38Ar^19F^+
^58Fe (0.3%)	^58Ni^+40Ar^18O^+, ^40Ca^18O^+
^39K (93.1%)	^38Ar^1H^+40Ar^+ and ^38Ar^+ (tailing)
^40K (0.001%)	^40Ar^+40Ca^+, ^24Mg^16O^+
^41K (6.9%)	^40Ar^1H^+40Ca^1H^+, ^24,25Mg^17,16O^+
^55Mn (100%)	^40Ar^14NH^+, ^36Ar^19F^+
^204Pb (1.5%)	^204Hg^+
^206Pb (23.6%)
^207Pb (22.6%)
^208Pb (52.3%)
^64Zn (48.9%)	^32S^32S^+32, S^32O2^+, ^64Ni^+, ^46,48Ca^18,16O^+
^66Zn (27.8%)	^32S^34S^+, ^34S^32O2^+, ^32S^16O^18O^+,
^67Zn (4.1%)	^34S^33S^+, ^34S^32S^1H^+, ^33S^16O^18O^+,^34S^16O^17O^+
^68Zn (18.6%)	^34S^34S^+, ^34S^16O^18O^+

---

Another difficulty to be considered when optimizing the working conditions is the large difference in the concentration of the elements to be monitored: many of these are present at the ng g⁻¹ level, but some are present at the μg g⁻¹ level. The high sensitivity of ETV-ICPMS is suitable for direct monitoring of those analytes present below the μg g⁻¹ level. For those elements present at the μg g⁻¹ level or higher, on the other hand, the OmniRange option (which can reproducibly reduce the sensitivity of the mass spectrometer for every isotope monitored) was used to diminish the sensitivity, ensuring working within the linear range for all signals of interest and protecting the detector from overloading. Therefore, OmniRange settings for those elements present at higher levels (K and Fe in both samples, in addition to Cr, Cu and Zn in Sample 2) were briefly optimized at the beginning of every working session in order to obtain signal intensities for all of the elements in the range 25 000–250 000 counts.

3.2. Optimization of the temperature program and evaluation of the performance of Pd as chemical modifier

Since applying the optimum temperature program can be critical, especially for a multi-element determination, the program needs to be carefully optimized. In fact, recent papers have shown that the potential of ETV-ICPMS for multi-element determination may have been underestimated.^17–19 In this regard, it has been demonstrated that a quadrupole-based ICPMS instrument enables the simultaneous monitoring of at least 20 isotopes, without significant degradation in terms of precision and sensitivity.¹⁷ However, in spite of this encouraging conclusion, the limiting factor in many cases may be finding a temperature program suitable for all of the elements of interest while maintaining the capability of separating in time the vaporization of the matrix from that of the analytes, which is very important for direct solid sampling.¹¹ Appropriate conditions are not always found and the separate monitoring of different groups of analytes is sometimes necessary,^12,20,21 thus compromising the sample throughput. The aim of this work was to find suitable conditions for the simultaneous monitoring of all of the elements of interest in one single firing, regardless of their different volatilities, since a high sample throughput is considered of the utmost importance for the routine control of the samples under investigation.

Fig. 2 shows the maximum pyrolysis temperatures that can be allowed without loss of analytes, in the absence of any modifier, for aqueous standard solutions with analyte contents similar to those found in Sample 1. It can be seen that, if a pyrolysis temperature of 800 °C is used, as required for efficient matrix removal, losses of Pb and Zn can occur. This point was confirmed by vaporization of the solid samples and monitoring of the signals for all of the analytes during the pyrolysis. In order to solve this problem, the use of a chemical modifier is needed.

Fig. 2 Maximum pyrolysis temperature and minimum vaporization temperature required for ensuring maximum sensitivity for the different target analytes in the presence and in the absence of Pd (200 ng Pd, added as Pd(NO₃)₂).

The performance of Pd (introduced as Pd(NO₃)₂) was tested for this purpose. Pd was observed to successfully stabilize the analytes, resulting in an increase of the maximum pyrolysis temperature that can be used for most of the target elements, but also in an increment in the temperature that is needed for their efficient vaporization. The only exception to this behavior is K. In this regard, it can be pointed out that there is little information available about the performance of Pd as modifier for this element, since it is seldom determined by means of GFAAS or ETV-ICP-based techniques, and in the few cases in which it was actually determined no modifier was needed.^22–25 For this element no sign of interaction with Pd was observed: the signal profile and the sensitivity were practically independent of the presence or absence of modifier.

In conclusion, it can be said that the use of Pd as modifier is beneficial for our purpose since it prevents any losses of volatile analytes at the pyrolysis temperature needed (800 °C). This aspect was confirmed by monitoring the pyrolysis step during ETV analysis of the actual solid samples. However, as a result of the further stabilization achieved, the vaporation temperature should also be increased. In this regard, the temperature finally selected was 2700 °C, in order to obtain the highest sensitivity for the most refractory analyte: Cr in this case. Additional experiments were carried out evaluating Rh and Pt as alternative modifiers, but the results obtained did not significantly differ from those observed with Pd, and thus the latter was used as modifier in all further experiments.

The amount of Pd used may also have an impact on the results obtained in a multi-element determination (compromise conditions may be needed) and was therefore optimized.^26,27Fig. 3 shows the comparison of the sensitivity obtained for all of the analytes as a function of the amount of Pd introduced. Different trends could be observed for the different elements, mainly depending on their volatility. On the one hand, sensitivity increased with the amount of Pd used for the most volatile elements (Pb and Zn). This is likely the result of a physical effect (Pd acting as carrier) rather than a chemical interaction, since the optimization was carried out using a low pyrolysis temperature (600 °C), at which no analyte losses should occur. This carrier effect has already been well documented to be especially relevant for volatile elements, for which the transport efficiency in the absence of any modifier is poor.^28–30 For the rest of the elements, the sensitivity does not change significantly up to 200 ng of Pd. For higher amounts, the sensitivity seems to decrease, possibly as a result of overstabilization (the signals clearly show more tailing). In addition to this, when the amount of Pd reaches 0.5 μg some signs of plasma overloading (resulting in signal suppression) can be observed by monitoring of the ⁸⁰Ar₂⁺ signal.^31,32 It can also be pointed out that K is the analyte least affected by the presence of Pd, which is logical since it does not seem to interact with it as discussed before. K is only very mildly affected by any signal suppression produced by Pd since it is vaporized well before the modifier, as can be seen in Fig. 4. Taking all these factors into consideration, 200 ng of Pd were selected as the optimum for further experiments.

Fig. 3 Influence of the amount of Pd on the sensitivity for the different target analytes under the conditions used in this work (see Table 1).

Fig. 4 Comparison of the signal profiles obtained for: (A) an aqueous standard solution containing 0.40 ng Cu (peak area = 125 400 counts), 0.45 ng Cr (peak area = 114 400 counts), 10 ng Fe (peak area = 84 870 counts), 65 ng K (peak area = 82 880 counts), 0.13 ng Mn (peak area = 71 800 counts), 0.011 ng Pb (peak area = 26 830 counts), 0.14 ng Zn (peak area = 33 290 counts) and 200 ng Pd (peak area = 27 420 counts); and (B) 0.983 mg of a solid sample (Sample 1), containing approximately 0.41 ng Cu (peak area = 138 180 counts), 0.44 ng Cr (peak area = 111 060 counts), 9.8 ng Fe (peak area = 84 560 counts), 64 ng K (peak area = 82 070 counts), 0.13 ng Mn (peak area = 76 960 counts), 0.011 ng Pb (peak area = 25 800 counts), 0.14 ng Zn (peak area = 31 960 counts) and 200 ng Pd (peak area = 27 210 counts).

3.3. Direct vaporization of the solid samples

Once the operating conditions were optimized, further experiments were carried out comparing the signals obtained on vaporizing solid samples and aqueous standard solutions containing similar amounts of the analytes. Fig. 4 shows an example of this comparison. As can be seen, under the conditions used, direct vaporization of the solid samples leads to well-defined unimodal signal profiles, with moderate tailing (excluding the high baseline for ³⁹K⁺, the origin of which was already discussed in Section 3.1.), very similar to those obtained for aqueous standard solutions both in shape and, most important, in peak areas. Moreover, the Ar dimer signal shows no evidence of suppression, confirming the successful removal of most matrix components. Therefore, external calibration with aqueous standard solutions seems possible for the samples studied in this work.

Additional experiments were conducted to investigate the possible effect of the sample mass on the final result. As shown in Fig. 5 for Fe and Mn (similar results were obtained for the rest of the analytes), no significant effect was found for masses up to 5 mg (the use of higher masses is not feasible considering the dimensions of the cup used for solid sampling). These data prove the suitability of the ETV for efficient vaporization of this type of polymeric material. Using high masses could be beneficial for the monitoring of elements present at very low levels. In this work, also considering the total amount of material available for the complete study, samples masses between 1 and 2 mg were selected for the determinations.

Fig. 5 Influence of the sample mass on the results obtained with solid sampling-ETV-ICPMS. For every value, 5 solid sample replicates were measured and the ¹⁰⁵Pd⁺ signal was used as internal standard. Uncertainties are expressed as ±standard deviation. The dashed line represents the reference value.

Finally, the use of the ¹⁰⁵Pd⁺ signal as internal standard was considered before final analyses of the samples were carried out. The use of this internal standard seemed promising since: (i) Pd is already added as chemical modifier and therefore its use does not complicate the analytical procedure; (ii) it is seldom present at high levels in these kinds of samples; (iii) it is an element of medium mass and, thus, should exhibit an “average” behavior in the plasma. Using Pd as both chemical modifier and internal standard has been evaluated with satisfactory results in two previous works aiming at single-element determination of two very different elements (Cr and I),^16,33 but no information on its performance for multi-element analysis was available.

Experiments proved that the use of ¹⁰⁵Pd⁺ as internal standard is also useful in this case. Basically an internal standard can help to correct for some matrix effects (e.g., matrix-induced signal suppression),^16,33 but this is not a significant advantage for this work since these effects were not observed. However, it can also help to correct for any variations in the experimental conditions during the measurements, helping to improve both the precision and the accuracy. In this regard, it has already been reported by Martin-Esteban et al. that a downward drift in sensitivity during vaporization of many solid samples (owing to deposits formed on the sampling and skimmer cones) occurs frequently.³⁴Fig. 6 shows the results obtained during an entire working session (monitoring of 15 solid samples after calibration with aqueous standard solutions). There is a clear drift in sensitivity affecting all of the analytes (for the sake of simplicity only Fe and Mn results are shown), resulting in lower results later on in the session. This drift not only results in a decreased precision, but could eventually also lead to a significant bias. Of course, it can be corrected by frequent recalibration, but this would negatively affect the sample throughput. As can be seen in Fig. 6, this effect can be corrected for by using the ¹⁰⁵Pd⁺ signal as internal standard, ensuring results that do not deviate more than 15% from the expected value for every solid sample. Repeatability improves from 12–13% to 6–7% RSD, a level that is difficult to further improve in direct solid sampling-ETV-techniques simply as a result of the degree of inhomogeneity of the samples analyzed.³⁵ Therefore, the ¹⁰⁵Pd⁺ signal was used as internal standard in the final analysis of the solid samples.

Fig. 6 Effect of the use of the ¹⁰⁵Pd⁺ signal as internal standard during a working session on the repeatability of the results. Every value represents the result for an individual solid sample. Calibration was carried out using aqueous standard solutions, before the monitoring of all of the solid samples. The dashed line represents the reference value.

3.4. Performance of solid sampling-ETV-ICPMS for sample analysis

The samples investigated were finally analyzed by means of SS-ETV-ICPMS, as described in Section 2.3., using the optimum conditions summarized in Table 1. The results obtained for these samples are shown in Table 3. Eight (Sample 2) and ten (Sample 1) different determinations, performed on at least four different days, were carried out for each sample. The samples were also analyzed according to the SEMI 48 procedure as detailed in Section 2.4., using pneumatic nebulization-sector field-ICPMS after sample dissolution in order to obtain reliable reference values for method validation.

Table 3 Analysis of the fluorocarbon polymer samples by means of solid sampling-ETV-ICPMS: quantitative results and limits of detection obtained (3s definition).^a The uncertainties are provided as 95% confidence intervals. Reference values obtained by means of solution-sector field-ICPMS are provided for comparison purposes

	Cu	Cr	Fe	K	Mn	Pb	Zn
a Every determination consisted of the measurement of 5 replicate solid samples (≈15 min of work).
LODs (μg g^−1) ETV-ICPMS	0.002	0.020	0.030	0.500	0.001	0.001	0.010
Sample 1 (μg g^−1)
ETV-ICPMS	0.44 ± 0.03	0.50 ± 0.03	10.0 ± 0.5	67 ± 4	0.13 ± 0.01	0.012 ± 0.002	0.13 ± 0.02
Reference value	0.42 ± 0.07	0.44 ± 0.07	10.0 ± 0.8	65 ± 1	0.13 ± 0.01	0.011 ± 0.001	0.14 ± 0.02
Sample 2 (μg g^−1)
ETV-ICPMS	1.9 ± 0.2	2.4 ± 0.2	4.0 ± 0.4	270 ± 10	0.17 ± 0.02	0.088 ± 0.013	2.4 ± 0.2
Reference value	1.6 ± 0.3	2.5 ± 0.1	3.5 ± 0.1	280 ± 40	0.15 ± 0.01	0.078 ± 0.006	2.7 ± 0.1

---

As can be seen, there is a good agreement (differences within experimental errors) with the reference values for both samples, proving that, under the conditions used, the ETV-ICPMS method provides accurate results when calibrating with aqueous standard solutions. Concerning the reproducibility of the method, the RSD values range between 7 and 12% for most analytes. There are only two discrepancies: Pb in both samples (≈17% RSD) and K in Sample 2 (4% RSD). These results can be explained by considering that Pb is the element present at the lowest levels (tens of ng g⁻¹), and more heterogeneity can be expected at such levels.³⁵ The better precision achieved for K in the sample in which it is present at the highest levels (hundreds of μg g⁻¹) is in good agreement with this explanation. In any case, these uncertainty values can be considered as acceptable for the fast and routine control of these materials.

Detection limits were also calculated (3s definition) for the ETV-ICPMS approach and are also reported in Table 3. It can be appreciated that LODs in the low ng g⁻¹ are achieved for those elements for which the most abundant isotopes can be monitored, while LODs at the tens of ng g⁻¹ level are obtained for those elements for which a less abundant isotope needs to be monitored (Fe, Cr, Zn). The only exception to this rule is K. Owing to the interferences discussed in Section 3.1., a significantly higher LOD is obtained. As mentioned before, this LOD is still low enough, considering the high contents at which this element is usually found in these and many other fluoro carbon polymers. However, for those samples in which the K content would be below the μg g⁻¹ level, ICP-MS instrumentation capable of coping with these interferences should be evaluated (e.g., instrumentation equipped with a collision/reaction cell).

Finally, in order to further evaluate the robustness of the method proposed, additional analyses (200 solid samples) of Sample 1 were carried out on 10 different days. The same graphite parts could be used during these sessions, and the method proved able to provide a stable response as can be seen in Fig. 7. However, signals of deterioration of the graphite cup and tip (used for closing of the tube during vaporization) could be observed. Thus, it would not be recommended to use these pieces for the analysis of more than 200 solid samples, while the graphite tube may still be used for more firings.

Fig. 7 Reproducibility of the results obtained on ten different measuring days. For every value, 10 solid sample replicates were measured using the conditions shown in Table 1. Uncertainties are expressed as ± standard deviation. The dashed line represents the reference value.

4. Conclusions

It has been demonstrated that solid sampling-electrothermal vaporization-inductively coupled plasma mass spectrometry is suitable for the direct multi-element analysis of the perfluorosulfonic acid/TFE copolymer samples under investigation.

It is clear that this technique requires careful optimization of the operating conditions (e.g., isotope selection, temperature program, use of Pd as both chemical modifier and internal standard) in order to obtain the best possible performance. However, once this optimization has been carried out in the proper way, the technique exhibits very attractive features, the most noteworthy ones being the ability to obtain fast and reliable results at trace and ultratrace levels while maintaining the maximum straightforwardness for calibration, using just aqueous standards.

The simplicity of the procedure finally developed contrasts with the complex protocols required for digestion of these samples, suggesting that solid sampling-electrothermal vaporization-inductively coupled plasma mass spectrometry can be an attractive option for the routine control of other kinds of fluoro carbon polymers as well.

Acknowledgements

This study was financially supported by the Spanish Ministry of Science and Technology (BQU 2002-0089), the Diputación General de Aragón (DGA), the University of Zaragoza (Ibercaja Project 230-121) and the Fund for Scientific Research – Flanders (FWO-Vlaanderen, research project nr. G.0037.01). Martín Resano acknowledges the Spanish Ministry of Science and Education (Secretaría de Estado de Universidades e Investigación, Programa Nacional de Ayudas para la Movilidad de Profesores de Universidad) and Maite Aramendía the ‘Europa’ program from CAI and DGA (CONSI + D; Grant reference CD17/05) for financially supporting their stay at Ghent University.

References

1. S. Ebnesajjad and P. R. Khaladkar, in Fluoropolymer applications in chemical processing industries: the definitive user’s guide and databook, William Andrew Publishing, New York, 2004.
2. K. E. Perepelkin, Fibre Chem., 2004, 36, 43–58.
3. Y. Sano, World Pumps, 2001, 421, 32–38.
4. J. R. Moody, J. Cryst. Growth, 1988, 89, 43–48.
5. C. W. Extrand, J. Fluorine Chem., 2003, 122, 121–124.
6. SEMIF-48-0600, Test method for determining trace metals in polymer materials, available from http://www.semi.org, last accessed February 2006.
7. M. Takenaka, Y. Yamada, M. Hayashi and H. Endo, Anal. Chim. Acta, 1996, 336, 151–156.
8. A. Martín-Esteban and B. Slowikowski, Crit. Rev. Anal. Chem., 2003, 33, 43–55.
9. F. Vanhaecke, M. Resano and L. Moens, Anal. Bioanal. Chem., 2002, 374, 188–195.
10. R. E. Sturgeon and J. W. Lam, J. Anal. At. Spectrom., 1999, 14, 785–791.
11. M. A. Belarra, M. Resano, F. Vanhaecke and L. Moens, Trends Anal. Chem., 2002, 21, 828–839.
12. M. Resano, L. Balcaen, F. Vanhaecke, L. Moens and I. Geuens, Spectrochim. Acta, Part B, 2002, 57, 495–511.
13. F. Vanhaecke, M. Resano, M. Verstraete, L. Moens and R. Dams, Anal. Chem., 2000, 72, 4310–4316.
14. M. A. Belarra, M. Resano and J. R. Castillo, J. Anal. At. Spectrom., 1999, 14, 547–552.
15. D. C. Grégoire and R. E. Sturgeon, Spectrochim. Acta, Part B, 1993, 48, 1347–1363.
16. M. Resano, E. Garcia-Ruiz, F. Vanhaecke, C. Crespo and M. A. Belarra, J. Anal. At. Spectrom., 2004, 19, 958–965.
17. M. Resano, M. Verstraete, F. Vanhaecke and L. Moens, J. Anal. At. Spectrom., 2001, 16, 1018–1027.
18. J. D. Venable, D. Langer and J. A. Holcombe, Anal. Chem., 2002, 74, 3744–3753.
19. K. C. Friese, K. H. Grobecker and U. Wätjen, Fresenius’ J. Anal. Chem., 2001, 370, 499–507.
20. L. F. Dias, G. R. Miranda, T. D. Saint’Pierre, S. M. Maia, V. L. A. Frescura and A. J. Curtius, Spectrochim. Acta, Part B, 2005, 60, 117–124.
21. M. Bettinelli, U. Baroni, S. Spezia and C. Terni, At. Spectrosc., 2000, 21, 195–204.
22. K. C. Friese and V. Krivan, Anal. Chem., 1995, 67, 354–359.
23. K. C. Friese and V. Krivan, Fresenius’ J. Anal. Chem., 1999, 364, 72–78.
24. Z. F. Queiroz, F. J. Krug, P. V. Oliveira, M. M. Silva and J. A. Nóbrega, Spectrochim. Acta, Part B, 2002, 57, 49–61.
25. B. Welz and M. Sperling, Atomic Absorption Spectrometry, Wiley VCH, Weinheim, 3rd edn, 1999, p. 541.
26. J. P. Byrne and G. Chapple, At. Spectrosc., 1998, 19, 116–120.
27. A. G. Coedo, T. Dorado, I. Padilla, R. Maibusch and H. M. Kuss, Spectrochim. Acta, Part B, 2000, 55, 185–196.
28. T. Kántor, Spectrochim. Acta, Part B, 1988, 43, 1299–1320.
29. D. C. Grégoire, in Discrete Sample Introduction Techniques for Inductively Coupled Plasma Mass Spectrometry, ed. D. Barceló, Elsevier, Amsterdam, 2000, ch. 3, p. 347.
30. K. Grünke, H. J. Stärk, R. Wennrich, H. M. Ortner and J. A. C. Broekaert, Fresenius’ J. Anal. Chem., 1997, 359, 465–468.
31. D. Beauchemin, J. W. McLaren and S. S. Berman, Spectrochim. Acta, Part B, 1987, 42, 467–490.
32. F. Vanhaecke, G. Galbács, S. Boonen, L. Moens and R. Dams, J. Anal. At. Spectrom., 1995, 10, 1047–1052.
33. M. Resano, E. Garcia-Ruiz, L. Moens and F. Vanhaecke, J. Anal. At. Spectrom., 2005, 20, 81–87.
34. A. Martin-Esteban, B. Slowikowski and K. H. Grobecker, Talanta, 2004, 63, 667–673.
35. M. A. Belarra, M. Resano and J. R. Castillo, J. Anal. At. Spectrom., 1998, 13, 489–494.

---

Footnotes

† Presented at the 2006 Winter Conference on Plasma Spectrochemistry, Tucson, AZ, USA, January 8–14, 2006.

‡ On leave from the Department of Analytical Chemistry, Faculty of Sciences, University of Zaragoza, Pedro Cerbuna 12, E-50009, Zaragoza, Spain.

---