TXRF - CiteSeerX

Spectrochimica Acta Part B 61 (2006) 1098 – 1104 www.elsevier.com/locate/sab

A new technique for the deposition of standard solutions in total reflection X-ray fluorescence spectrometry (TXRF) using pico-droplets generated by inkjet printers and its applicability for aerosol analysis with SR-TXRF☆ U.E.A. Fittschen a,⁎, S. Hauschild a , M.A. Amberger a , G. Lammel b , C. Streli c , S. Förster a , P. Wobrauschek c , C. Jokubonis c , G. Pepponi d , G. Falkenberg e , J.A.C. Broekaert a a

University of Hamburg, Department of Chemistry, University of Hamburg, Martin-Luther-King-Platz 6, 20146 Hamburg, Germany b Max Planck Institute for Meteorology, Bundesstraβe 53, 20146 Hamburg, Germany c Atominstitut, Vienna University of Technology, Stadionallee 2, 1020 Wien, Austria d ITC-irst, Via Sommarive 18, 38050 Povo (Trento), Italy e Hamburger Synchrotronstrahlungslabor at Deutsches Elektronen-Synchrotron DESY, Notkestr. 85, 22603 Hamburg, Germany Received 14 December 2005; accepted 15 September 2006 Available online 3 November 2006

Abstract A new technique for the deposition of standard solutions on particulate aerosol samples using pico-droplets for elemental determinations with total reflection X-ray fluorescence spectrometry (TXRF) is described. It enables short analysis times without influencing the sample structure and avoids time consuming scanning of the sample with the exciting beam in SR-TXRF analysis. Droplets of picoliter volume (∼5–130 pL) were generated with commercially available and slightly modified inkjet printers operated with popular image processing software. The size of the dried droplets on surfaces of different polarity namely silicone coated and untreated quartz reflectors, was determined for five different printer types and ten different cartridge types. The results show that droplets generated by inkjet printers are between 50 and 200 μm in diameter (corresponding to volumes of 5 to 130 pL) depending on the cartridge type, which is smaller than the width of the synchrotron beam used in the experiments (b 1 mm at an energy of 17 keV at the beamline L at HASYLAB, Hamburg). The precision of the printing of a certain amount of a single element standard solution was found to be comparable to aliquoting with micropipettes in TXRF, where for 2.5 ng of cobalt relative standard deviations of 12% are found. However, it could be shown that the printing of simple patterns is possible, which is important when structured samples have to be analysed. © 2006 Elsevier B.V. All rights reserved. Keywords: Calibration; Pico-droplet; Inkjet printer; SR-TXRF; Aerosol analysis

1. Introduction The impact of the chemical composition of airborne particles and its particle size dependence on air quality and on climate

☆ This paper was presented at the 11th International Conference on Total Reflection X-ray Fluorescence Spectrometry and Related Methods (TXRF2005), held in Budapest, Hungary, 18–22 September 2005, and is published in the special issue of Spectrochimica Acta Part B, dedicated to that conference. ⁎ Corresponding author. Tel.: +49 40428386099; fax: +49 49428384381. E-mail address: [email protected] (U.E.A. Fittschen).

0584-8547/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.sab.2006.09.009

has become apparent in recent years [1–3]. Particle sources and sinks are strongly size dependent and so is the chemical composition of most aerosols e.g. automobile traffic aerosols with regard to Pb, Pd and Pt [4–6]. Multiphase processes in the aerosol are highly variable both in time and space. Hence, the dynamics of aerosol physics and chemistry needs to be addressed in order to understand the effects of aerosols on atmospheric and cloud chemistry. A number of metals like Fe, Pb and also non-metals like P, S, Cl and Br undergo chemical transformations typically on the time scale of hours. Conventional bulk analytical techniques typically fall short with time resolutions of 6–24 h for size-resolved determination of major

U.E.A. Fittschen et al. / Spectrochimica Acta Part B 61 (2006) 1098–1104

particulate matter components. Fast bulk on-line methods are so far limited to non-metal components, mostly ions. Metals can be addressed by on-line or off-line single particle mass spectrometric methods [7,8]. Since only recently, these methods do allow for quantification on a mass per sampled air volume [9]. TXRF is a very sensitive method for element determinations which has been successfully used for size-resolved aerosol analysis e.g. direct analysis of airborne dust collected with an impactor on glassy carbon and Si-wafer carriers using conventional X-ray tube excitation [10–13]. Even more sensitive is the excitation with synchrotron radiation. Its combination with size-resolved aerosol sampling is promising, as already shown in Ref. [14] using synchrotron radiation in conventional XRF geometry (incident beam 45°). In elemental determinations in aerosol particle samples collected with the aid of impactors and analyzed with SRTXRF, an accurate and reliable calibration up to now, however, is often problematic. The addition of standard solution with micropipettes for aerosol analyses with TXRF may suffer from destroying the sample spots or lines as a result of the relatively high amount of liquid. The alternative to pipetting a standard solution directly onto the aerosol, could be an external calibration. To obtain a good precision with an external calibration, several reflectors prepared with standard solutions ought to be analyzed, which is a time consuming procedure. An additional problem arises from the relatively large diameter from 1 mm and more of the dried spot resulting from 1 μL standard solution. In SR-TXRF (vertical geometry) the

1099

maximum height of the exciting beam is small (∼ 1 mm at 17 keV at the HASYLAB beamline L). While the dried residue is completely illuminated in the propagation direction because of the footprint of the exciting beam, it isn't completely illuminated in the up–down direction. Furthermore the intensity distribution of the beam in up–down direction is not homogeneous. In the middle the intensity is high and declines to the upper and lower end. To be sure that the whole amount of standard is detected, the sample has to be scanned through the exciting beam in up–down direction. Of course this has to be done with any sample that exceeds the 1 mm range. But that is not the case for the aerosols of the PM2.5 (i.e., atmospheric particulate matter smaller than 2.5 μm by aerodynamic size) size range collected by the Berner impactor (0.25–0.06 μm, 0.06–0.015 μm aerodynamic particle diameter). These are collected in small spots of diameters below 500 μm. If it were possible to give a small droplet of a standard solution on these aerosols, the elements could be determined with SR-TXRF in one step without scanning the sample. Calibration with nano-droplets of 10 to 50 nL was already found to give excellent results in the determination of trace impurities with TXRF in semiconductor material [15,16]. Therefore, it was decided to investigate the suitability of a calibration in aerosol sample analysis by SR-TXRF under the use of pico-droplets, generated by inkjet printers and having a volume ranging from 5 to 130 pL. With this technique a standard solution can be spotted directly on the aerosol sample. The procedure for this should be performed as follows. The aerosol particles are collected in spots of

Table 1 Homogeneity and diameter of dried pico-droplets generated by different printers and different types of cartridges Nr.

Printer

Cartridge

Surface glass

1

HP 520

HP 26A-35

x

Surface silicone

Printjob

Temperature [°C]

Quality

Carrier

Droplet diameter (μm)

Image

6.25% grey

21

“normal”

12.5% grey

21

“normal” “draft” “custom” “normal”

Glass Quartz Glass

200/120/40 200/50 200

Inhomogeneous Inhomogeneous Homogeneous

Quartz Glass

100/50/20 200 200 200 100/50 100/30/20 100 100 100 100 150 100 100 40 100 100 100 200 200 200 60

Inhomogeneous Homogeneous Inhomogeneous Homogeneous Inhomogeneous Inhomogeneous Inhomogeneous Homogeneous Homogeneous Homogeneous Homogeneous Homogeneous Inhomogeneous Homogeneous Homogeneous Homogeneous Homogeneous Inhomogeneous Inhomogeneous Inhomogeneous Homogeneous

x 2

HP 500C

HP 26A-35

x

x

HP 3325

HP 25-1 HP 26A-36 HP 26A-37 HP 28A-10

HP 28A-1

x

HP 10 black HP 10 mag HP 78

x

x

3

4

HP 2000c

5

HP PSP1000

x

12.5% grey 12.5% grey tricolour 12.5% grey 10% cyan 10% magenta 10% gelb 10% cyan 10% yellow 10% magenta 12.5% grey 10% magenta 10% magenta

x x x

HP 45

5% yellow 5% magenta 5% cyan 6.25% grey 6.25% cyan 12.5% grey

x

x

21

“best”

Glass

22

“best”

Glass

22

“normal”

Glass Quartz Glass

Glass

Quartz

1100


different sizes on Si-wafer collectors on the impaction stages of a Berner impactor. The latter is known to exhibit excellent characteristics for separations in the size range from 16 to 0.015 μm (aerodynamic particle diameter). The Si-wafer collectors are placed under the nozzle plates of the impactor in such a way, that the aerosols were collected in the middle of the collectors. With common imaging software it is then easily realized to spot one droplet with an inkjet printer in the middle of the same collector, which can be tested beforehand with ink droplets from ink cartridges. Various printer types and cartridge types were tested, in order to get an overview on the performance of the technique in comparison to droplet aliquoting using micropipettes. Further, the possibility to print simple patterns on a given area was tested.

Fig. 2. Scan across an aerosol sample (0.25–0.06 μm aerodynamic diameter, 48 h sampling time) referring the fluorescence intensity of iron obtained with the SR-beam: 50 μm wide SR-beam in steps of 50 μm.

2. Experimental 2.1. Chemicals Standard solutions containing 1 g/L of cobalt and scandium (Merck, Darmstadt, Germany), subboiled HNO3 prepared from HNO3 63% from Merck and bi-distilled water were used in this work. Spotting with pico-droplets was performed on quartz carriers having a roughness below 5 nm and a hydrophobic surface. To generate a hydrophobic surface on the reflector, it was treated with 1 μL of 10% silicone solution in Isopropanol from Merck. 2.2. Instruments For aerosol sampling a 12-stage low-pressure Berner impactor covering the size range from 16 to 0.015 μm (aerodynamic particle diameter, Haucke KG, Gmunden, Austria) was used. Silicon wafers were chosen as material for the impaction

plates, because they show very good performance in TXRF in comparison with quartz and glassy carbon carriers as a result of their low surface roughness [13]. The silicon carriers were cut in pieces of 24.7 × 15.0 mm from a silicon wafer (diameter 200 mm) with the aid of a diamond saw. To make sure that the aerosols were collected exclusively in the middle of the silicone carriers, the other parts were protected by cellulose nitrate filters from Sartorius (pore diameter: 3 μm, filter diameter: 40 mm) that were removed after the sampling. All printers that were used for pico-droplet generation were commercially available from Hewlett Packard (Models: see Table 1). They were slightly modified, in such a way that the form feed was bypassed and the printing could be triggered on demand. Patterns were generated with Corel PHOTO-PAINT 12 software. A transmitting light microscope (Axiovert 100, Zeiss, Oberkochen, Germany) was used for the sizing of the generated ink droplets. 2.3. Measurements The TXRF measurements were performed with an 8030C instrument (Atomika, now Cameca, Oberschleissheim, Germany).

Fig. 1. Atmospheric aerosols collected on silicon wafers placed at the different impaction stages of a Berner impactor (without using a cellulose nitrate filter mask).

Fig. 3. Spectrum of ambient air aerosol (2.00–1.00 μm aerodynamic diameter, 1 h sampling time) sampled in Hamburg at the top of an 80 m high university building obtained with conventional TXRF using X-ray tube excitation (a) and spectrum obtained with SR-TXRF at HASYLAB beamline L, for the same sample (b).


Fig. 4. Slightly modified inkjet printer HP 500 C with sample holder and cartridge.

All samples were analyzed by exciting with the Mo Kα-line (17.4 keV) under the use of 200–1000 s integration times. The SR-TXRF measurements were performed at the set-up for TXRF at HASYLAB beamline L, being described in detail in Ref. [17]. A peltier-cooled silicon drift detector (VORTEX, SII NanoTechnology, Northridge, USA) was used. Aerosol samples were excited with SR-radiation monochromized to 17.4 keV with the aid of a multilayer monochromator and an integration time of 100 s was used. The Kα-lines were used for the determination of all elements investigated, excepted for lead, for which L-lines were used. To compare the precision of the preparation of standard samples by spotting with micropipettes and with printing, a calibration curve was generated once by successively adding 1 μL with an micropipette and once by successively adding

1101

droplets with the printer on a quartz reflector (single dose of 16 droplets in the case of TXRF with the Atomika 8030C and with 1 droplet when using SR-TXRF). To study the diameter of dried pico-droplets and the distance between droplets in a printed pattern, a line of droplets was printed with a 1 g/L cobalt standard solution on silicon substrates and scanned with a SR-beam of 50 μm and 250 μm in height in steps of 50 μm and 250 μm respectively. The X-ray optics used were a remote controlled cross slit system with 3-mm thick tungsten jaws. It was stated that the diameter of a spot didn't exceed the distance that is given by the full width at half maximum of the intensity profile resulting from the SR-TXRF line scan. To demonstrate the excellent potential of SR-TXRF for studies on atmospheric aerosols, particles were collected with a Berner impactor on silicon carriers for 1 h on the top of a university building in 80 m height in Hamburg. The particles on the silicon carriers from stages containing particles with aerodynamic particle diameter from 2.00 to 1.00 μm, 1–0.25 μm, 0.25–0.06 μm and 0.06–0.015 μm were analyzed with both conventional TXRF and SR-TXRF. 3. Results and discussion To clarify the demands of a calibration in the case of real aerosol samples, aerosols were collected with a Berner impactor on silicon carriers. The size of the sampled aerosol spots from the stages with particles of an aerodynamic diameter from 0.25 to 0.06 μm and from 0.06 to 0.015 μm is small. An image of the sample pattern on three different impaction stages is shown in Fig. 1. A scan over the aerosol samples with a narrow SR-beam

Fig. 5. Ink pico-droplets generated by inkjet printers: HP 3325 on glass surface (a), PSP 1000 on silicone coated surface (b) HP 2000 c on glass surface (c) Photosmart P (PSP) 1000 on silicone coated surface (d).

1102


sample shows that the sensitivity and the peak to background ratio in the case of SR-TXRF is considerably higher than in the case of laboratory TXRF instrument using a conventional X-ray tube (Fig. 3). The printing of single pico-droplets with various printer types (one is shown in Fig. 4), and different cartridges was investigated. The diameter and the homogeneity of ink droplets printed on hydrophobic and non-hydrophobic quartz and glass carriers were determined and the results are given in Table 1 and Fig. 5. The printing was defined as homogeneous, when the droplets (20 to 30 droplets were studied) had all the same size

Fig. 6. Two scans obtained with the SR-beam across a line of three droplets of a Co solution (1 g/L) spotted with HP 500 c printer (∼130 pL), diamonds: 250 μm wide SR-beam in steps of 250 μm and triangles: 50 μm wide SR-beam in steps of 50 μm (a); light microscope image of ink pico-droplets printed by HP PSP 1000 printer (∼ 5 pL) on a silicone coated quartz reflector (b).

(∼ 50 μm width) in steps of 50 μm shows that the spreading for iron, which can be used as marker for the aerosol, is about 500 μm (Fig. 2). Therefore, these aerosol spots can be completely illuminated by the synchrotron beam in the vertical direction. To get information about the elemental composition, the samples then were analysed with TXRF and SR-TXRF. A comparison of TXRF and SR-TXRF spectra of the same aerosol Table 2 Precision of printed cobalt standard solutions calibrated with 10 ng of scandium and the apparatus factors of the Atomika 8030C Number

1 2 3 4 5 6 7 8 9 10 11 12 Mean STD RSTD

Hp 3325 16 droplets used cartridge 1 g/L Co

HP 500c 16 droplets unused cartridge 10 g/L Co

ng(Co)

ng(Co)

0.83 0.91 1.39 1.19 1.05 1.12 1.11 1.06 0.83 1.09 0.92 1.20 1.06 0.17 16%

10.2 10.5 10.7 11.2 10.3 11.5 10.8 12.0 11.1 11.9

11.0 0.6 6%

Fig. 7. Calibration curves obtained with mean values from three series for: five times 1 μL of a 1 g/L cobalt standard solution brought successively with a micropipette on a quartz reflector supplied with 10 ng of scandium standard solution residue (a), five times 16 droplets of a 1 g/L cobalt standard solution spotted successively with a HP 500 C printer on a quartz reflector supplied with 2.3 ng scandium standard solution (b); five times 1 droplet of a 1 g/L cobalt standard solution spotted successively with a HP 500 C printer on a quartz reflector (c) (measurements (a) and (b) represent data obtained with the 8030 C Atomika and for (c) measurements were done at HASYLAB, beamline L).

U.E.A. Fittschen et al. / Spectrochimica Acta Part B 61 (2006) 1098–1104 Table 3 Single blank values as estimated from the slope of the calibration curve, using the fundamental parameter approach [18] Element and line

Blank values (pg) from area where a single Blank values (pg) (∼ 130 pL) droplet of Co solution (1 g/L) from area next to the was spotted on a silicone coated quartz printed Co droplet reflector

Cl K 15.70 KK 1.16 Ca K 0.60 Ti K 1.20 Cr K 1.14 Mn K 0.97 Fe K 4.00 Co K ref. 160 Ni K 1.40 Cu K 3.90 Zn K 1.70 Rb K 0.80 Pb L 0.004

23.60 4.56 4.37 0.52 2.7 3.2 4.10 – 1.50 3.30 – 0.013

(diameter ± 2 μm), didn't have satellite droplets and had a nearly round shape. It was found that the quality of the cartridge is important so as to enable a production of droplets of uniform shape and diameter. It was expected, that the droplets printed from the cobalt standard solution have nearly the same size, as the ink droplets. For a verification of this point commercially available unused cartridges were depleted, purged and filled with the 1 g/L cobalt standard solution. The dry solution residues of single droplets (∼130 pL) of 1 g/L cobalt standard solution printed on silicon reflectors were scanned first with a narrow SR-beam (∼ 50 μm width) in steps of 50 μm. The scan shows that the diameter of the droplets was smaller than 200 μm (diamonds in Fig. 6). With the microscope used here, it was not possible to see the droplets of the standard solution. This will be done in the future. The feasibility for spotting according to a simple pattern is demonstrated by printing 3 droplets of cobalt solution in a line and calculating the image with Corel PHOTO-PAINT software. The dried solution residues were scanned with a SR-beam (∼ 250 μm width) in steps of 250 μm. The scan shows equally spaced peaks and demonstrates that pattern spotting is feasible (triangles in Fig. 6). The precision of the spotting of a certain amount of cobalt solution was studied with two cobalt solutions containing 1 g/L and 10 g/L of Co, under the use of a calibration with 1 μLaliquots of a 10 mg/L scandium standard solution and taking into account the apparatus factor for the 8030C TXRF. The results are shown in Table 2. The precision obtained strongly depend on the individual cartridge. The results of the comparison of the precision of the preparation of standard samples by spotting with micropipettes and with printing are given in Fig. 7. The precision obtained when an aliquot of 16 droplets of a 1 g/L cobalt standard solution is spotted with an inkjet printer is not as good as the one obtained in the work with micropipettes as reflected by the correlation coefficients. However, the precision obtained when an aliquot of 1 droplet of a 1 g/L cobalt standard solution is spotted with an inkjet printer is better than that obtained in the work with the mi-

1103

cropipettes. This was found to apply, when the spotting with micropipettes was performed by well-trained workers, otherwise also the printing of aliquots of 16 droplets was found to be more reliable than in the case of working with micropipettes. From the slope of the calibration curve the absolute amount transferred with a single printed droplet from a 1 g/L cobalt solution can be estimated to be 160 pg. That agrees very well with the data given by the printer producer, who stated that a volume of 130 pL is ejected from a single heating of a chamber of this type of cartridge. A further point to be clarified in view of trace and ultra trace analyses when aliquoting with inkjet printers, is the extent to which blank values occur. Analyses of droplets of a blank solution spotted with a printer on a “clean” reflector with SR-TXRF show that the blank values estimated from a calibration with fundamental parameters in SR-TXRF, as described in Ref. [18], are of the same order of magnitude as the blank values for non-spotted areas on the reflectors themselves (Table 3). The question whether the blanks stem from the cleaning procedure of the silicon wafer reflectors needs further investigations. 4. Conclusions The calibration in element determinations under the use of pico-droplets generated by inkjet printers shows some very promising features. The diameter of the dried droplets ranges from 50 to 200 μm for ink droplets printed on hydrophobic surfaces. The reliability of the dosing of standard solution seems to be satisfactory and enables a spotting of well-defined absolute amounts in the picogram range and below for cobalt, for instance. Further, it was shown that a printing according to a given pattern in the micrometer range can easily be done with the aid of common computer software. When applying this calibration procedure to elemental determinations in aerosol samples with SR-TXRF, the dosing of amounts less than 200 fg of an element, as present in many airborne aerosol samples, can be estimated to be well possible. Acknowledgement This work was supported by the European CommunityResearch Infrastructure Action under the FP6 “Structuring the European Research Area” Programme-Integrating Activity on Synchrotron and Free Electron Laser Science (I-03-089EC). References [1] K. Katsouyanni, G. Touloumi, E. Samoli, A. Gryparis, A. Le Tertre, Y. Monopolis, G. Rossi, D. Zmirou, F. Ballester, A. Boumghar, H.R. Anderson, B. Wojtyniak, A. Paldy, R. Braunstein, J. Pekkanen, C. Schindler, J. Schwartz, Confounding and effect modification in the short-term effects of ambient particles on total mortality: results from 29 European cities within the APHEA2 project, Epidemiology 12 (2001) 521–531. [2] J. Cyrys, J. Heinrich, A. Peters, W. Kreyling, H.E. Wichmann, Emission, Immission und Messung feiner und ultrafeiner Partikel, Umweltmed. Forsch. Prax. 7 (2002) 67–77. [3] J.E. Penner, M. Andreae, H. Annegarn, L. Barrie, J. Feichter, D. Hegg, A. Jayaraman, R. Teaitch, D. Murphy, J. Nganga, G. Pitari, Aerosols, their direct and indirect effects, in: J.T. Houghton, et al., (Eds.), Climate Change

1104

[4] [5]

[6]

[7]

[8]

[9]

[10]

U.E.A. Fittschen et al. / Spectrochimica Acta Part B 61 (2006) 1098–1104 2001 — the scientific basis, Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK, 2001, pp. 289–348. K.R. Spurny: Analytical chemistry of aerosols, CRC Press, Boca Raton, USA, 486 pp. G. Lammel, A. Röhrl, H. Schreiber, Levels of atmospheric lead and bromine in Germany: post-abatement levels, variability and trends, Environ. Sci. Pollut. Res. Int. 9 (2002) 397–404. F. Zereini, F. Alt, J. Messerschmidt, A.v. Bohlen, K. Liebl, W. Püttmann, Concentration and distribution of platinum group elements (Pt, Pd, Rh) in airborne particulate matter in Frankfurt am Main, Germany, Environ. Sci. Technol. 38 (2004) 1686–1692. R. Gieray, G. Lammel, G. Metzig, P. Wieser, Size dependent single particle bulk analysis of droplets and interstitial particles in an orographic cloud, Atmos. Res. 30 (1993) 263–293. S.H. Pastor, J.O. Allen, L.S. Hughes, P. Bhave, G.R. Cass, K.A. Prather, Ambient single particle analysis in Riverside, California by aerosol timeof-flight mass spectrometry during the SCOS97-NARSTO, Atmos. Environ. 37 (2003) 239–258. J.L. Jimenez, J.T. Jayne, Q. Shi, C.E. Kolb, D.R. Worsnop, I. Yourshaw, J.H. Seinfeld, R.C. Flagan, X. Zhang, K.A. Smith, J.W. Morris, P. Davidovits, Ambient aerosol sampling using the Aerodyne Aerosol Mass Spectrometer, J. Geophys. Res. 108 (2003) 8425, doi:10.1029/ 2001JD001213. B. Schneider, The determination of atmospheric trace metal concentrations by collection of aerosol particles on sample holders for total-

[11]

[12]

[13]

[14]

[15]

[16] [17]

[18]

reflection X-ray fluorescence, Spectrochim. Acta Part B 44 (1989) 519–524. R. Klockenkämper, H. Bayer, A.v. Bohlen, M. Schmeling, D. Klockow, Collection of airborne particulate matter for a subsequent analysis by total reflection X-ray fluorescence, Anal. Sci. 11 (1995) 495–498. M. Schmeling, D. Klockow, Sample collection and preparation for analysis of airborne particulate matter by total reflection X-ray fluorescence spectrometry, Anal. Chim. Acta 346 (1997) 121–126. F. Esaka, K. Watanabe, T. Onodera, T. Taguchi, M. Magara, S. Usuda, The use of Si carriers for aerosol particle collection and subsequent elemental analysis by total-reflection X-ray fluorescence spectrometry, Spectrochim. Acta Part B 58 (2003) 2145–2155. N. Bukowiecki, M. Hill, R. Gehrig, C.N. Zeicky, P. Lienemann, F. Hedegüs, G. Falkenberg, E. Weingartner, U. Baltensperger, Trace metals in ambient air: hourly size-segregated mass concentrations determined by synchrotron-XRF, Environ. Sci. Technol. 39 (2005) 5754–5762. T.C. Miller, C.M. Sparks, G.J. Havrilla, M.R. Beebe, Semiconductor applications of nanoliter droplet methodology with total reflection X-ray fluorescence analysis, Spectrochim. Acta Part B 59 (2004) 1117–1124. T.C. Miller, G.J. Havrilla, Nanodroplets: a new method for dried spot preparation and analysis, X-Ray Spectrom. 33 (2004) 101–106. C. Streli, G. Pepponi, P. Wobrauschek, C. Jokubonis, G. Falkenberg, G. Zaray, A new SR-TXRF vacuum chamber for ultra trace analysis at HASYLAB, beamline L, X-Ray Spectrom. 34 (2005) 451–455. D.K.G. de Boer, Fundamental parameters for X-ray fluorescence analysis, Spectrochim. Acta Part B 44 (1989) 171–1190.