Analytica Chimica Acta

Transcription

1 Analytica Chimica Acta 676 (2010) Contents lists available at ScienceDirect Analytica Chimica Acta journal homepage: Quantitative analysis of heavy metals in automotive brake linings: A comparison between wet-chemistry based analysis and in-situ screening with a handheld X-ray fluorescence spectrometer R. Figi a,, O. Nagel a, M. Tuchschmid a, P. Lienemann a,b, U. Gfeller a, N. Bukowiecki a,c a Empa, Swiss Federal Laboratories for Materials Testing and Research, CH-8600 Duebendorf, Switzerland b Zurich University of Applied Sciences ZHAW, CH-8820 Wädenswil, Switzerland c Paul Scherrer Institut, Laboratory of Atmospheric Chemistry, CH-5232 Villigen PSI, Switzerland article info abstract Article history: Received 3 June 2010 Received in revised form 21 July 2010 Accepted 21 July 2010 Available online 30 July 2010 Keywords: Brake pads Extraction High pressure asher Inductively coupled plasma optical emission spectrometry X-ray fluorescence Two extraction procedures for ecologically relevant elements present in automotive brake linings (Sb, Bi, Pb, Cd, Cr (total), Co, Cu, Mo, Ni, Sr, V, Zn, Sn) were developed and validated, applying a high pressure asher (HPA-S) and microwave extraction, respectively. Both of these methods allowed for the quantitative analysis of the extracted elements by inductively coupled plasma optical emission spectrometry (ICP-OES). The results were compared to measurements using a handheld energy-dispersive X-ray fluorescence spectrometer (ED-XRF), being in discussion by regulating agencies as in-situ screening tool for brake pads. The comparison indicates that the handheld ED-XRF analysis is basically an efficient screening tool for a reliable assessment of trace metal contents in automotive brake pads with respect to legal standards. While a quantitative determination of elements like Cd, Co, Cr, Mn, Mo, Ni, Pb and Sb was achievable, other elements (V, Cu, Bi, Zn, Sn and Sr) could only be determined qualitatively due to the special matrix characteristics of brake pads Elsevier B.V. All rights reserved. 1. Introduction Brake wear from automotive vehicles contains trace metals which are subject to pollution of roadside soils, drain water from roads and the ambient air [1]. Brake linings can be divided into four main categories. Semimetallic brake linings consist of 30 65% metals, mixed with graphite, filler and binder materials. This is the most frequent category used in European vehicles. Organic brake linings are based on fibers gained from glass, rubber, carbon (Keflar, Twaron) and also contain filler materials as well as temperature resistant synthetic or natural resins. Low-metallic brake linings contain the same organic species, with additional 10 30% of metals (mostly steel or copper). Finally, ceramic brake linings consist of ceramic fibers, filler and binder materials and low amounts of metals. Table 1 lists trace elements which are often used in brake linings and are subject to adverse health effects in their prevailing chemical speciation. To date, certified reference materials (CRM) for brake pads do not exist. While material fluxes and emissions can be reasonably estimated and modeled for a real-world traffic fleet, reliable analytical methods for the compositional analysis of individual brake linings are still rather scarce [2 4]. Quantitative knowledge on the Corresponding author. address: renato.figi@empa.ch (R. Figi). compositions of brake linings is, however, important also for legal agencies, to comply with environmental policies. In the European Union, the guideline 2000/53/EC sets limits for the use of heavy metals to be used in the manufacturing process of light duty vehicles [5]. This paper presents both a high pressure asher (HPA-S) and a microwave (MW) extraction procedure for ecologically relevant elements present in brake linings, allowing for the quantitative analysis of the extracted elements by inductively coupled plasma optical emission spectrometry (ICP-OES). This method is subsequently used as independent reference method for comparison with in-situ measurements on solid brake pad samples using a handheld energy-dispersive X-ray fluorescence spectrometer (ED- XRF). The latter method is currently in discussion to be used as screening method by numerous regulating agencies to check a large number of brake pad samples. To investigate the influence of sample homogeneity, the ED-XRF screening was performed both in situ on selected brake pads as well as on pulverized brake pad samples. 2. Methods and experiments 2.1. Sample selection and preparation Brake lining samples from 14 different vehicle types were available for this study (Table 2). While 10 of the samples represented /$ see front matter 2010 Elsevier B.V. All rights reserved. doi: /j.aca

2 R. Figi et al. / Analytica Chimica Acta 676 (2010) Table 1 Common trace elements contained by automotive brake linings. In addition to the listed trace elements, Si, Ba, Fe and Zr are further elements often found in brake linings in percentage-wise amounts. Element Compound Sb Sb 2S 3 Bi Pb PbO/PbS Cd Cr (total) Cr 2O 3 Co Cu Cu/Cu 2S Mo MoS 2/Mo 2O 3 Ni Ni Sr Sr V Zn ZnO/ZnS Sn SnS/SnS 2 Table 3 Optimized procedure for extraction by HPA-S. Time (min) End temperature ( C) Delay time (min) Step Table 4 Optimized procedure for extraction by microwave-assisted extraction. Time (min) Power (W) Temperature inside ( C) 2: : : : Temperature outside ( C) brake linings from new cars, three samples originated from used vehicles. Pulverized samples were gained from the considered brake linings by grinding off the lining layer of the brake pads using a tempered tungsten carbide molding cutter (Starmax Prealpina Suisse ). The gained material was subsequently homogenized and sieved to a grit size below 0.5 mm, using a chromium steel sieve. For an in-depth comparison between the presented wetchemistry analysis and the measurements with a handheld ED-XRF spectrometer, a brake lining containing a significant amount of most of the elements listed in Table 1 was selected. This preselection was achieved by a qualitative scan of all brake pads by wavelength dispersive XRF (WD-XRF, Philips/PANalytical PW2400, [10]) Wet-chemical analysis Chemicals HNO 3 (E. Merck, 65% p.a.), HCl (E. Merck, 37% p.a.) and HClO 4 (E. Merck, 70% p.a.) or mixtures were used in the extraction of automotive brake samples. The water used in the experiments was purified in an Milli-Q reagent grade 18 M water system (Millipore). All standard reference solutions for the ICP-OES calibration procedures are from Alfa Aesar specpure Sample extraction using a HPA-S High Pressure Asher The wet-chemical extraction of the elements listed in Table 1 was performed using a HPA-S High Pressure Asher (Anton Paar Inc.) [6 8]. Five duplicates of 0.1 g sample material were extracted applying an acid mixture consisting of 5 ml HCl (27% p.a.), 1.6 ml HNO 3 (65%, p.a. and 0.25 ml HClO 4 (70%, p.a.) in a 70 ml quartz digestion vessel, at an end temperature of 320 C and a pressure of 135 bar. This mixture was optimized for a complete extraction of Sn and Sb. Since Zr and Si were not within the scope of analysis, HF was not included. For a quantitative extraction of Cr, both the use of HClO 4 and the temperature of 320 C were necessary to quantitatively dissolve chromium carbides formed during the extraction. The detailed temperature profile for the extraction is shown in Table 3. The extracted solutions were filled up to a defined volume with ultrapure water Sample extraction using microwave extraction For additional comparison, another five duplicates of 0.1 g sample material were extracted in 50 ml Teflon vessel by medium pressure microwave-assisted extraction (MLS Start microwave oven), using the identical acid mixture as described above. The extracted solutions were filled up to a defined volume with ultrapure water. The extraction parameters are listed in Table Quantitative analysis The determination of the elements of interest in the extracts was performed by inductively coupled plasma optical emission spectrometry (ICP-OES, Varian Vista Pro Radial, Varian Inc.). The considered emission lines are listed in Table 5. The Ar line was used as internal standard. The sampling inlet system consisted of a twister cyclonic spray chamber (VA/60), a quartz radial torch (Varian) and a Gemcone (Inc.) high nebulizer. The power was 1.10 kw, flows were 10.5 L min 1 (plasma), 1.5 L min 1 (auxiliary) and 0.70 L min 1 (nebulizer). A replication time of 40 s and a stabilization time of 20 s were applied. The view height was 12 mm, and the number of replicates was 3. Table 2 List of brake pads. Table 5 Detection limits for the investigated elements (determined according DIN 32645). Manufacturer Model Element ICP-OES emission line Detection limit in ppm ( g/g) Golf Bora Golf T4 Transporter Opel Corsa Renault Megane Ford Focus Audi A4 Mercedes A BMW 3-er Peugeot 205 Peugeot 207 Fiat Punto Toyota Yaris Bi Cd Co Cr Cu Mn Mo Ni Pb Sb Sn Sr V Zn

3 48 R. Figi et al. / Analytica Chimica Acta 676 (2010) Handheld X-ray fluorescence spectrometry Both the original brake linings and the pulverized samples (Section 2.1) were analyzed without further treatment using a handheld energy-dispersive X-ray spectrometer (Niton XL3t, Niton Inc., USA), applying a counting time of 270 s. Since matrix elements like O and C cannot directly be detected using XRF, matrix effects are empirically corrected by the instrument under the assumption of a specific sample matrix composition. The calibration mode used in this study was originally designed for a semiquantitative determination of economically relevant metals in mineral (silicate-oxide type) matrices (e.g. for use in ore prospecting). This mode was considered to best fit the average matrix of a brake pad sample. This measuring mode was calibrated for silicate-oxides matrices. The influence of not directly detected elements with atomic numbers lower than 16 (S) was estimated via the Compton peak in the fluorescence spectra. While matrix corrections can be accurately performed for a well-known matrix [9], the estimation of the matrix may result in considerable uncertainties in the determined elemental concentrations (see Section 3). 3. Results and discussion Fig. 1 shows the qualitative WD-XRF spectrum of a brake lining originating from a Bora (2000), which showed a manifold presence of the trace elements. The analysis results for this sample will be shown in the following figures. Detailed analysis results for all of the investigated brake linings are listed in the Supplementary Information available for this manuscript. For quality assurance, the residua of the extracts were filtered and analyzed using WD-XRF (Fig. 2). The qualitative spectra showed no significant line intensities resulting from the elements of interest. The elemental recovery rates for both extraction procedures and the ICP-OES analysis are shown in Fig. 3, determined with help of standard solutions. The elemental detection limits were deter- Fig. 1. WD-XRF spectrum of a -Bora brake pad (pulverized sample). LiF200, PE, PX1: WD-XRF crystals.

4 R. Figi et al. / Analytica Chimica Acta 676 (2010) Fig. 2. WD-XRF spectrum of the residuum after HPA-S extraction. LiF200, PE, PX1: WD-XRF crystals. mined based on DIN (indirect procedure) and are shown in Table 5. Fig. 4 compares the HPA-S extraction to the microwaveassisted extraction for a sample with a chromium content clearly below 0.1%. A good agreement was found for virtually all of the considered elements. In contrast, Fig. 5 compares the HPA- S extraction to the microwave-assisted extraction for a sample with a high total chromium content (>1 g/100 g). Again a good agreement was found. The only exception was chromium, which could not be quantitatively extracted using microwave-assisted extraction since the maximum temperature of 290 C was not sufficient to crack high concentrations of new-formed chromium carbide. Fig. 6 shows the comparison between the wet-chemistry based analysis of the selected brake pads and the respective elemental contents determined by handheld ED-XRF in the untreated (solid) brake linings. A good agreement was found for Mo, Pb, Sb, Mn and Sn. This indicates that for these elements the homogeneity of the elements of interest in the solid brake pad samples is sufficiently high for screening via handheld ED-XRF. In contrast, the Cu contents determined via handheld ED-XRF were systematically underestimated by 20 30%, while Sr was systematically overestimated by 40 50%. This points to a calibration problem due to the specific matrix characteristics of brake pads and the applied calibration mode. Due to spectral interferences with other elements, the V and Ni contents obtained from handheld ED-XRF were considered to be incorrect for the given proportions of these elements. Furthermore, the interference of Pb-L lines with Bi-L lines resulted in an incorrect determination of low Pb contents (<0.01%) in samples with increased Bi contents (>0.2%). Cd and Co were not determinable using handheld ED-XRF due to contents in brake pads which usually were below detection limit. Likewise, the low concentrations of Bi were overestimated by ED-XRF in the pulverized sample and not detected at all on the solid sample, due to a higher influence of sampling inhomogeneity in this concentration regime.

5 50 R. Figi et al. / Analytica Chimica Acta 676 (2010) Fig. 3. Elemental recovery rate (%) for the two presented extraction methods. In the current legal guideline 2000/53/EC [5], lead is subject to be entirely banned from brake pads. While the lead content related to copper, in the following expressed as Pb(Cu), was limited to 4% in 2003, this limit is currently being decreased to 0.4%. From an analytical perspective, Pb(Cu) is not only sensitive to the Pb measurement itself, but also to the accurate determination of Cu. With the handheld ED-XRF spectrometer, both elements were subject to bias as stated in the previous paragraph. For samples with Cu <1% the Cu(Pb) value cannot be obtained due to the insufficient detection limit for the determination of the respective Pb content. Fig. 7 shows the Pb(Cu) value for a selected brake pad. While the wet-chemistry based value is below the legal limit, the values estimated by handheld ED-XRF exceeded this limit. This example illustrates that although handheld ED-XRF can be suitable for a quick triage of samples with Cu(Pb) values clearly below or above the legal limit, Cu(Pb) values close to the limit Fig. 4. Comparison between extraction procedures (HPA-S vs. microwave) for the considered elements (analyzed by ICP-OES). Sample: -Bora brake pad.

6 R. Figi et al. / Analytica Chimica Acta 676 (2010) Fig. 5. Underestimated Cr determination using microwave-assisted extraction (MW) in a brake lining with high Cr content. Sample: Mercedes A brake pad. Fig. 6. Comparison between wet-chemistry analysis and mobile-xrf analysis of a brake pad sample. value cannot be reliably assessed for legal purposes with this method. 4. Conclusions and outlook Table S1 (see Supplementary Information to this article) shows that the composition of individual brake pads was highly variable. This was also observed in other studies [1]. The presented extraction methods for trace elements in brake linings both allowed for a reliable quantitative analysis using ICP-OES. However, for analysis of Cr, the microwave-assisted analysis did not provide a quantitative extraction. Since the extraction temperature was not sufficient for the complete oxidation of carbon and chromium acts as carbide former, hardly soluble chromium carbide is formed. The use of a tungsten carbide grinder during sampling preparation was essential, to avoid a contamination of the sample by steel. Perchloric acid was necessary to provide enough chlorine to dissolve the strongly matrix-bound Sn and Sb and also acted as an oxidation agent, together with the temperature above 300 C in the HPA-S extraction, for the dissolvement of chromium carbide which was formed during the extraction. The applied brake pad screening procedure using the handheld ED-XRF spectrometer provided a reliable determination of Mo, Pb,

7 52 R. Figi et al. / Analytica Chimica Acta 676 (2010) Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi: /j.aca References Fig. 7. Lead content related to copper, Pb(Cu), determined in a Bora brake pad. Sb, Mn and Sn. In future work the screening procedure will be refined, by the design and validation of a brake pad standard sample intended for use with handheld ED-XRF measurements. Acknowledgements The assistance of Ms. K. Uhlmann in the experimental part is gratefully acknowledged. We thank the Empa machine shop (Erwin Pieper and his team) for support. Financial support was granted by the Swiss Federal Office for the Environment (FOEN/BAFU). [1] N. Bukowiecki, P. Lienemann, M. Hill, R. Figi, A. Richard, M. Furger, K. Rickers, G. Falkenberg, Y.J. Zhao, S.S. Cliff, A.S.H. Prevot, U. Baltensperger, B. Buchmann, R. Gehrig, Environmental Science & Technology 43 (2009) [2] P.J. Blau, Compositions, Functions and Testing of Friction Brake Materials and Their Additives, Oak Ridge National Laboratory for U.S. Department of Energy, [3] D. Chan, G.W. Stachowiak, Proceedings of the Institution of Mechanical Engineers Part D-Journal of Automobile Engineering 218 (2004) 953. [4] P. Kennedy, J. Gadd, Preliminary Examination of Trace Elements in Tyres, Brake Pads and Road Bitumen in New Zealand, Ministry of Transport, Welllington, [5] EU, index.htm, [6] E. Ikavalko, T. Laitinen, H. Revitzer, Fresenius Journal of Analytical Chemistry 363 (1999) 314. [7] R.T. White, Journal of the Association of Official Analytical Chemists 72 (1989) 387. [8] M. Würfels, E. Jackwerth, M. Stoeppler, Fresenius Zeitschrift fur Analytische Chemie 317 (1984). [9] N. Broll, R. Tertian, X-Ray Spectrometry 12 (1983) 30. [10] C.N. Zwicky, P. Lienemann, X-Ray Spectrometry 33 (2004) 294.