Thick-target PIGE analysis of plant materials preconcentrated by dry ashing

Talanta 51 (2000) 717 – 725 www.elsevier.com/locate/talanta

Thick-target PIGE analysis of plant materials preconcentrated by dry ashing K.-E. Saarela a,*, L. Harju a, J.-O. Lill b,c, J. Rajander a, A. Lindroos d, S.-J. Heselius c Laboratory of Analytical Chemistry, A, bo Akademi Uni6ersity, FIN-20500 Turku, Finland b Department of Physics, A, bo Akademi Uni6ersity, FIN-20500 Turku, Finland c Turku PET Centre, Accelerator Laboratory, A, bo Akademi Uni6ersity, FIN-20500 Turku, Finland d Department of Geology and Mineralogy, A, bo Akademi Uni6ersity, FIN-20500 Turku, Finland a

Received 5 July 1999; received in revised form 14 October 1999; accepted 8 November 1999

Abstract Plant materials were dry ashed at 550°C and analysed using particle-induced prompt gamma-ray emission (PIGE). The analyses were performed with an external beam of 3 MeV protons incident on the target. Seven biological certified reference materials were analysed and used for the evaluation of the method for Na, Mg, Al, P and Mn. The elemental concentration to detection limit ratios were greatly enhanced by dry ashing of the biological materials. The concentrations of the elements in ashes were clearly above the values at which reliable analyses can be made. The method was applied to samples of spruce and pine. Due to the low ash content of the wood samples, the sensitivity of the method was radically improved. The detection limits for the five elements studied in spruce wood were in the range 0.014–2.5 mg g − 1. The set-up and the beam current used enabled simultaneous particle-induced X-ray emission spectrometry (PIXE) analyses, with the sensitivity optimised for heavier trace elements. © 2000 Elsevier Science B.V. All rights reserved. Keywords: PIGE; Dry ashing; CRMs; Plant materials

1. Introduction Particle-induced g-ray emission (PIGE) is an analytical technique based upon the measurement of g-ray emission from a sample irradiated with ions from a particle accelerator [1,2]. In PIGE an * Corresponding author. Tel.: +358-2-2154-140; fax: + 358-2-2154-479. E-mail address: [email protected] (K.-E. Saarela)

excited nucleus is produced by a nuclear reaction, normally a (p,p’ g), (p, g), (p,n g) or (p,a g) reaction. The g-ray emitted in the de-excitation is detected and the nucleus can be identified from the g-ray energy. However, this nucleus can be a product of different nuclear reactions. It is therefore difficult to identify the target nucleus and furthermore to relate the detected g-yield to the concentration of a nuclide or an element in the sample. The g-ray yield changes rapidly with the

0039-9140/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 9 1 4 0 ( 9 9 ) 0 0 3 2 6 - 4

718

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

incident particle energy. Therefore thick-target gray yields have been measured and tabulated for different particle energies [3 – 6]. Collected data can be found in the book by Bird and Williams [1]. A flexible experimental set-up for externalbeam particle-induced X-ray emission (PIXE) analyses of thick solid samples has been developed at the Accelerator Laboratory of A, bo Akademi University [7]. The PIXE technique has been employed mainly in the analysis of biological materials. In recent years the sensitivity has been optimised for the analyses of heavier trace elements [8–11]. Biological samples were also dry ashed prior to the analyses in order to increase the sensitivity for these elements [8]. The gain for different biological materials was related to the ash content of the material studied [11]. In the PIXE analyses the sensitivity was hampered especially by the high content of potassium. The spectral background in PIGE is, however, not expected to be negatively affected. This is due to a low g-ray yield for elements like K and Ca. The gain in PIGE can therefore be greater. Light elements that can be determined with PIGE are often difficult to analyse with other methods. The instrumental methods generally used, demand the often laborious step of sample dissolution. Dissolution of ashes is often troublesome and is, e.g. in the case of aluminium, a principal source of error in quantitative analyses [12]. Dry ashing with modern programmable furnaces does not require constant operator attention and is therefore not a laborious step in the analytical procedure. If the ash can be analysed without dissolution the enrichment can be effectively benefited. The use of external-beam PIGE spectrometry has been rather limited [13] and biological applications are limited to a few studies of mainly biomedical samples. In the present work PIGE is evaluated as a complementary technique to thicktarget PIXE performed with 3 MeV protons. The elements Na, Mg, Al, P and Mn are analysed in dry ashed plant materials with PIGE. The method is evaluated by seven biological certified reference materials (CRMs) and applied to the analyses of wood materials. Data on the elemental concentra-

tions in wood are of great industrial importance. Results obtained by PIXE-analyses of trunk wood have been earlier reported by our group [9,10].

2. Experimental

2.1. Instrumental set-up Most of the set-up was earlier used in thicktarget PIXE analyses and is described in detail elsewhere [11]. The set-up includes a photo-multiplier (PM) tube monitoring light from the beam path. The current from the PM-tube is integrated over the acquisition time and used for normalising peak areas in the obtained spectra [14]. The particle beam of the A, bo Akademi MGC-20 cyclotron was collimated to a diameter of 1 mm. An Intrinsic Germanium Planar (IGP) detector was used for detecting X-ray emission from the irradiated sample and a High-Purity Germanium detector (Model: GMX-25195-S) for measuring the g-ray emission. The resolution (FWHM) for the HPGe detector was 2.6 keV at 1332 keV (60Co) at measuring conditions. The g-detector was placed 35 mm behind the target at an angle of 45 degrees with respect to the beam path (Fig. 1). Every sample was irradiated for about 11 min with a beam of 3 MeV protons. The beam spot was changed every 90 s. The recorded spectra were analysed using the SAMPO90 computer program [15].

2.2. Sampling of trunk wood Samples of wood were taken in 1994 from a polluted area, Harjavalta (x= 22° 10% 12¦ E, y= 61° 21% 27¦ N), where a Cu–Ni smelter has been active for about 50 years. For Harjavalta, PIXE results reported earlier by our group [10] showed elevated concentrations of heavy metal ions. The other sampling site, Merimasku (x= 21° 48% 49¦ E, y=60° 29% 48¦ N), represents a relatively unpolluted area in the Turku archipelago. The samples from Merimasku were taken in 1997. Mainly Norway spruce (Picea abies) was studied in this work. Discs were cut of the stem within 0.5 m above the ground. In the laboratory, the sampling

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

was made by cutting or drilling the disc in a sectorial pattern in order to get a representative sample of the trunk [9,10].

2.3. Dry ashing and sample preparation The CRMs and the wood samples were dried at 105°C in order to obtain the concentrations on dry matter basis. The dry ashing of the biological materials was performed with a Nabertherm Program Controller S 27 oven to 550°C. Amounts of 1–3 g of the CRMs and 10 – 20 g of the wood samples were weighed for the dry ashing. The temperature was increased very slowly, especially in the range 200 – 350°C, to ensure a complete charring. The pyrolysis is an important step of the procedure and a prerequisite for an ashing temperature not exceeding the adjusted furnace temperature. Both the ashes and the original biological materials were pressed to pellets on a backing of

719

spectroscopically pure graphite. The obtained pellets were 1.5 mm thick with a polycarbonate backing of 0.5 mm. A pure graphite pellet was also prepared and irradiated for an equally long time as the samples to determine the spectral background for the set-up.

2.4. Calibration Experimentally obtained calibration constants KZ for the elements are used for quantification of both the nonashed and the ashed samples. These constants reflect the experimental conditions, including, e.g. geometry and the stopping power of the two types of samples analysed. The stopping power in a sample is related to the atomic numbers of the matrix elements. No stopping-power corrections are performed in this work. The calibration constants KZ are calculated from the peak areas YZ according to KZ = C ZY·QZPM

Fig. 1. Set-up for external-beam PIXE/PIGE analyses of thick samples. The particle beam enters from the right and the g-ray detector is down to the left.

720

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

3. Results and discussion

3.1. Spectrum interpretation and background contributions

Fig. 2. PIGE-spectra of nonashed and dry ashed Wood Fuel (NJV 94-5).

where CZ is the certified concentration of the element Z and QPM the integrated charge on the target. Elemental concentrations in the ashed samples are calculated from the ash content. Assuming constant isotopic ratios for the elements the g-ray yield from a certain nuclide can be considered representative for the element. The following CRMs are used for calibration and evaluation of the method: Pine Needles SRM 1575 and Tomato Leaves SRM 1573 from National Institute of Standards and Technology (NIST), Energy Grass (Phalaris arundinace L.) reference material NJV 94-4, Energy Forest (Salix) NJV 94-3, Wood Fuel NJV 94-5 and Energy Peat (Carex) NJV 94-1 from the Swedish University of Agricultural Sciences and Wheat Flour ARC/CL-3 from the Agricultural Research Centre of Finland.

In Fig. 2 the PIGE spectra for both nonashed and ashed Wood Fuel are given. For simplicity only the peaks of Mn-55 (126 keV), Mg-25 (585 keV), P-31 (1266 keV), Al-27 (171, 844 and 1014 keV) and Na-23 (440 and 1636 keV) are marked. These peaks are a result of nuclear excitation by inelastic scattering and correspond to transitions to the ground state except for the 171 keV peak of Al and the 1636 keV peak of Na, which feed the transitions to the ground states. The peaks from Mg and Mn are lost in the background of the spectrum of the biological sample. The aluminium line at 171 keV (more precisely 170.68 keV) corresponds to a transition between the 1014 keV (1014.43 keV) and the 844 keV (843.76 keV) energy levels and has the lowest yield of these three lines. The lower yield is partly compensated by the higher detector efficiency in this spectral region. No background peak was observed at 171 keV. At 1014 and 844 keV there were contributions from the aluminium in the surrounding constructions, like the beam line, as the set-up was originally designed for thick-target PIXE. The background contributions to the Al lines in Wood Fuel were 40% of the peak areas for the biological material but only 1% of the peak areas for the ashes. The background contribution was subtracted from the peak areas in the calculations of the concentrations. The yield at 440 keV from Na-23 was higher than at 1636 keV. However the measurement of background radiation revealed a peak at 440 keV. This background contribution was 0.6–4.0% for the ashes compared to 8–60% for the biological materials.

3.2. Determined calibration constants Calibration constants KZ determined and used in this work are presented in Table 1. The constants were determined both for ashed material (KZ,ASH) and nonashed material (KZ,BIO). In the calculation of the elemental calibration constants,

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

CRMs with high certified concentration and small confidence intervals were used. The elemental concentrations in the ashed CRMs were calculated from the ash content obtained in the dry ashing process and used in the calculation of calibration constants for the ashed material KZ,ASH. The ratios of KZ,BIO to KZ,ASH are also included in the table. This ratio will be independent of the concentration of the analyte element in the original sample. The KZ,BIO/KZ,ASH ratios, mostly in the range 1.1 – 1.3, reflect the poor counting statistics for some of the nonashed samples, the interference from the background radiation and the matrix effects. The type of calibration standard should be similar to that of the sample. KZ,BIO should be used when analysing biological samples and KZ,ASH for analyses of ashed samples.

3.3. Validation of the method for quantitati6e analyses The results of the PIGE determinations of Na, Mg, Al, P and Mn in the seven biological CRMs are shown in Table 2, where quantification was done by using the g-lines with the highest KZ-values in Table 1. The results were obtained in single PIGE determinations. The certified or recom-

721

mended concentrations and their intervals are included in Table 2. Consensus values compiled by Roelandts and Gladney for Na and Mg in Pine Needles and Na in Tomato Leaves are included as well [16]. All five analyte elements were certified only in one of the seven CRMs analysed. For most elements and lines, results obtained with combined dry ashing and PIGE analyses agreed with the certified values. Manganese and also in some cases, magnesium and aluminium, were impossible to detect without preconcentration. Results obtained for all five elements without preceding dry ashing differed greatly from those reported. The discrepancies can be attributed to the low concentrations in nonashed materials and to the background contributions discussed in chapter 3.1. From the results in Table 2 it can be concluded that no major losses of any of the above mentioned elements occur during the dry ashing. According to Bock [12] no Mg or Al will be lost during dry ashing at 550°C. When the sample results in an alkaline ash it can safely be ashed up to 800°C without loss of phosphorous. According to Gorsuch [17] upper limits of 500–550°C seem to be advisable for manganese. Losses of sodium will be, to some extent, a function of both the chemical form in which the element occurs and

Table 1 Calibration constants (Kz) for biological materials and corresponding ashes (Energy Forest, Na, Wheat Flour, Mg and P, Pine Needles, Mn and Al)a Nuclide

Na-23 Na-23 Mg-25 Mg-25 Mg-25 Mg-26 Al-27 Al-27 Al-27 P-31 Mn-55 Mn-55 a

Energy (keV)

440 1636 390 585 975 1809 171 844 1014 1266 126 931

Biological sample

Ashed sample

Ratio KZ,BIO/KZ,ASH

Concentration

KZ

Concentration

KZ

0.029 0.029 0.562 0.562 0.562 0.562 0.545 0.545 0.545 2.09 0.675 0.675

3.14 0.637 n.d. 0.0477 n.d. n.d. 0.0880 0.251 0.314 0.0699 n.d. n.d.

1.81 1.81 61.8 61.8 61.8 61.8 22.1 22.1 22.1 230 27.4 27.4

3.69 0.561 0.0222 0.0381 0.0101 0.00111 0.0457 0.190 0.265 0.0553 0.0445 0.0325

Elemental concentrations are in mg g−1 by dry weight. n.d., not determined.

0.85 1.13 1.25

1.92 1.32 1.19 1.26

722

Table 2 Elemental concentrations in CRMs obtained by PIGE analysesa Element /Energy (keV)

Na

Mg 585 585 Al 171 844 1014 171 844 1014 P 1266 1266 Mn 126 931

Energy Forest NJV-94-3

Wood Fuel NJV 94-5

Energy Peat NJV 94-1

Energy Grass NJV 94-4

Pine Needles SRM 1575

Tomato Leaves SRM 1573

Wheat Flour ARC/CL-3

Ash (%) certified by preconc. by preconc. direct analysis direct analysis certified by preconc. direct analysis certified by preconc. by preconc. by preconc. direct analysis direct analysis direct analysis certified by preconc. direct analysis certified by preconc. by preconc.

1.603 29 94.3

1.221 40 912 39.7 35.9 37.5 31.8 300 9 18 294

3.963 (80 9 56) 32.2 32.5 23.1 11.1 770 9 89 844 357 900 979 992 997 957 419 664 703 450 979 406 642 36 9 3.8 48.3 44.8

6.762 (300 9150) 240 239 243 182 880 974 963 719 1800 9 170 1740 1490 1490 978 1230 1440 1090 9 90 1235 1310 160 9 16 159 123

2.466 36 9 16b 31.6 28.6 31.4 23.7 1180 9120b 1090 742 545 9 30

19.410 4809 100b 495 516 311 248 (7000) 5730 4092 (1200) 1380 1220 1180 642 893 979 34009 200 3600 2760 238970 242 306

0.909

340 9 36 339 361 (20 9 15) 32.8 18.0 17.7

500 9 100 639 813 60 911 86.4 60.2

260 9 32 273 269 269 198 296 345 210 9 28 203 571 210 9 38 203 190

1200 9 200 1050 1270 675 9 15

4.4 4.0 17.0 24.9 562 9 10

11.6 8.3 7.8

2090 9 140 12.87 90.34

a Certified concentrations are in bold face and recommended within brackets, the underlined concentrations were used for calculating the calibration constants (Table 1) and for quantification, all concentrations are in mg g−1 by dry weight. b Consensus values according to Roelandts and Gladney [16].

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

440 1636 440 1636

Obtained concentration

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

723

Table 3 Concentration to detection limit ratios (C/DL) in the biological CRMs, Energy Forest (Na, Mg, P, Mn) and Wood Fuel (Al) and in their corresponding ashesa Element

Energy (keV)

Nonashed material C

DL

C/DL

C

DL

C/DL

20.4 2.1 5.3

1800 21 000 31 000 3740

4.1 405 237 723

439 52 131 5

30

21 000

89

237

Energy Forest Na Mg P Mn

440 585 1266 126

29 340 500 60

1.4 165 94

Wood Fuel Al

1014

260

21

a

Dry-ashed material

Concentrations are in mg g−1.

the nature of the material in which it is found [17]. Ashing temperatures of both 450 and 550°C have been recommended for biological material [12].

method is radically improved, due to the high preconcentration factor obtained.

3.5. Analysis of trunk wood 3.4. Limits of detection and quantitati6e determination The enrichment obtained by dry ashing clearly improves the accuracy of the analyses. The concentration to detection limit (C/DL) ratios for some energy lines of elements in both the nonashed Energy Forest and Wood Fuel CRMs and the corresponding ashes are reported in Table 3. The detection limits were determined according to general practice as three times the square root of the background area below the peak divided by the calibration constant [11]. As a limit of quantification (LOQ) 10 times the square root of the background (or ca 3.3× DL) has been given [18]. Many of the concentrations in the biological CRMs were too low to fulfil this criterion. In Energy Forest the C/DL ratio 2.1 for magnesium is too low for reliable quantification. For the ashes the concentrations clearly exceed the limits of quantification. Table 3 clearly shows that the C/DL ratio has to be considered when directly analysing biological materials with thick-target PIGE. The detection limits for the ash matrix are higher than for the corresponding biological material. However, by using dry ashing the over-all sensitivity of the

The generally low ash content of spruce and pine wood makes this type of sample especially suitable for the analytical method described. The results of the analyses of some trunk wood samples by combined dry ashing and thick-target PIGE are given in Table 4. Elemental lines given in Table 3 were used for analytical determinations. The concentrations of Na, Mg, Al, P and Mn are given for the nonashed material on dry matter basis. Without enrichment by dry ashing the concentration levels of the elements observed were too low for quantification. Quite few data have been published on the content of light elements in wood. This holds especially for spruce wood. For the phosphorous concentration of spruce Young et al. [19] and Bailey et al. [20] report 50 mg g − 1, which is close to the values obtained in this work. The data for P and Mn from simultaneous PIXE-analyses are also given in Table 4 for comparison. For manganese good agreement was obtained for the two methods. A systematic error can be seen for the samples from Merimasku as PIXE gave somewhat lower concentrations. Larger deviations were obtained for phosphorous between the two methods. The phosphorous con-

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

724

centrations were close to the limit of quantitative determination for the PIXE method (LOQ = 53 mg g − 1) [11]. The PIXE analyses of phosphorous are affected by escape peaks, low detector efficiency and intense bremsstrahlung [21]. By considering the ash percentage, the detection limits for the whole analytical procedure can be calculated. For the seven spruces analysed the mean ash concentration was 0.347 weight%. The detection limits for Na, Mg, P, Mn and Al in ashes of plant materials can be seen from Table 3. When using thick-target PIGE in combination with dry ashing the following detection limits were obtained for spruce wood: Na 0.014 mg g − 1, Mg 1.4 mg g − 1, Al 0.3 mg g − 1, P 0.8 mg g − 1 and Mn 2.5 mg g − 1. Hall [22] reported working detection limits for tree-rings obtained by thin-target PIXE/PIGE and different sample preparation techniques. The detection limits achieved for Na, Mg, Al and P in the present work with the thick-target technique are clearly lower than those reported by Hall [22].

4. Conclusions Particle-induced g-ray emission (PIGE) analyses of plant materials were performed at conditions

optimised for external-beam thick-target particleinduced X-ray emission (PIXE) analysis. The samples were analysed as thick targets. The elements Na, Mg, Al, P and Mn were determined with the method. The PIGE analyses of low concentrations of elements are problematic due to contributions from background radiation. However, by dry ashing the materials to 550°C prior to the analysis, the contributions from the background became negligible. The method was evaluated using seven certified reference materials. Experimental calibration constants and detection limits were determined for the five elements in the nonashed and ashed sample. Good agreements were obtained between certified values and the results from the PIGE analyses of the ashes. No significant losses of the elements studied were observed. The concentrations in the ashes were clearly above the limits of quantitative determination (LOQs). In plant materials like wood the concentrations in nonashed samples are well below the LOQs. In ashed samples the detection limits for PIGE are somewhat higher than in nonashed samples. However, considering the preconcentration factor, the detection limits for the whole method were greatly

Table 4 Thick-target PIGE analyses of wood samples preconcentrated by dry ashinga Sample

Ash (%)

PIGE

PIXE

Na

Mg

Al

P

Mn

P

Mn

Harja6alta Pine Spruce 1 Spruce 2

0.223 0.322 0.330

28.7 18.5 5.3

86 70 76

3.1 2.5 4.7

53 35 30

43 22 38

71 41 12

61 17 69

Merimasku Spruce 1 Spruce 2 Spruce 3 Spruce 4 Spruce 5

0.320 0.513 0.319 0.313 0.314

3.0 9.9 5.5 9.6 5.4

107 187 95 104 120

6.4 9.8 9.9 7.7 5.8

23 41 21 27 72

23 62 31 16 24

37 30 49 85

17 49 22 12 19

0.347 21

8.2 64

108 36

6.7 40

36 50

31 50

42 57

29 73

Mean spruce RSD (%)

a The samples were collected from Harjavalta and Merimasku in SW Finland. The concentrations in the two last columns were determined with PIXE. All concentrations are given as mg g−1 by dry weight.

K.-E. Saarela et al. / Talanta 51 (2000) 717–725

enhanced by the dry ashing. In spruce samples the average ash content was 0.347% by dry weight and a detection limit of e.g. 0.014 mg g − 1 for sodium can be achieved.

Acknowledgements The financial supports provided by The Foundation of A, bo Akademi and The Foundation for Research of National Resources in Finland are gratefully acknowledged.

References [1] J.R. Bird, J.S. Williams, Ion Beams for Material Analysis, Academic Press, Australia, 1989, p. 719. [2] B. Borderie, Nucl. Instr. Meth. 175 (1980) 465. [3] G. Demortier, J. Radioanal. Chem. 45 (1978) 459. [4] M.J. Kenny, J.R. Bird, E. Clayton, Nucl. Instr. Meth. 168 (1980) 115. [5] A. Anttila, R. Ha¨nninen, J. Ra¨isa¨nen, J. Radioanal. Chem. 62 (1981) 293. [6] J. Ra¨isa¨nen, R. Ha¨nninen, Nucl. Instr. Meth. 205 (1983) 259.

.

725

[7] J.-O. Lill, K.-E. Saarela, F.J. Hernberg, S.-J. Heselius, L. Harju, Nucl. Instr. Meth. B 83 (1993) 387. [8] K.-E. Saarela, J.-O. Lill, F.J. Hernberg, L. Harju, A. Lindroos, S.-J. Heselius, Nucl. Instr. Meth. B 103 (1995) 466. [9] L. Harju, J.-O. Lill, K.-E. Saarela, S.-J. Heselius, F.J. Hernberg, A. Lindroos, Nucl. Instr. Meth. B 109-110 (1996) 536. [10] L. Harju, J.-O. Lill, K.-E. Saarela, S.-J. Heselius, F.J. Hernberg, A. Lindroos, Fresenius J. Anal. Chem. 358 (1997) 523. [11] J.-O. Lill, L. Harju, K.-E. Saarela, A. Lindroos, S.-J. Heselius, Anal. Chim. Acta 378 (1999) 273. [12] R. Bock, A Handbook of Decomposition Methods in Analytical Chemistry, International Textbook Company, Edinburgh, 1979, p. 444. [13] J. Ra¨isa¨nen, Nucl. Instr. Meth. B 40/41 (1989) 638. [14] J-O. Lill, Nucl. Instr. Meth. B 150 (1999) 114. [15] A. Aarnio, J.T. Routti, J.V. Sandberg, J. Radioanal. Nucl. Chem. 124 (1988) 457. [16] I. Roelandts, E.S. Gladney, Fresenius J. Anal. Chem. 360 (1998) 327. [17] T.T. Gorsuch, The Destruction of Organic Matter, Pergamon, Oxford, 1979, p. 491. [18] L.H. Keith, Anal. Chem. 55 (1983) 2210. [19] H.E. Young, V.P. Quinn, Tappi 49 (1966) 190. [20] J.H.E. Bailey, D.W. Reeve, J. Pulp Paper Sci. 20 (1994) 83. [21] O. Valkovic, M. Jaksic, S. Fazinic, V. Valkovic, G. Moschini, E. Menapace, Nucl. Instr. Meth. B 99 (1995) 372. [22] G.S. Hall, Nucl. Instr. Meth. B 49 (1990) 60.