Applicability of a hollow cathode emission source for determining trace elements in electrically non-conducting powders

OW-8547/83/0406254?7 NJ 1983. Perymon

Applicability

of a hollow cathode emission source for determining elements in electrically non-conducting powders

SO3 W/O Press Ltd

trace

S. CAROLI,A. ALIMO~TI Istituto Superiore di Sanita. Laboratorio di Tossicologia. Viaie Regina Elena 299. OOkX-Roma,Italy

and K. ZIMMER E&v& Lorind TudomBnyegyetem. Szervetlen is Analitikai Ktmiai Tans&k, Miueum krt. 4/b, 1088 Budapest, Hungary. (Received

12 January 1982; in

reoisedfirm

13 August 1982)

Abstmc+-The appli~biiity of the hollow cathode emission source for the determination of traoe elements in mineral residues from ashing of biological materials is investigated. The method uses samples previously diluted with a 1:4 graphite~p~r powder mixture to render them electrically conductive and subsequently pressed into the form of hollow cylinders. The procedure was tested with synthetic specimens; the latter contain Ag, Al, B, Cr, Mn, Ni, Pb, Sn, Ti and Zn, and a matrix similar to that of organic materials after mineralization. The use of Ga as an internal standard greatly improved the linearity of log I, vs log c, relatlonship. The precision of the measurements and the detectlon limits were also found to be entirely satisfactory.

1. INTR~OUCTI~N DIRECT SPECTROSCOPIC determinations of trace elements and minor constituents as well as main components of powders (e.g. from ashing of organic materials or through grinding of minerals) can greatly contribute to simplifying a variety of otherwise complicated analytical problems. Therefore, it is understandable that spectroscopists have devoted effort to studying new procedures in this field. Traditional arc and spark methods for powder analysis do not offer sufficient reliability. Loading a mixture of sample and graphite powder into electrode craters suffers from uneven distribution of the specimen and ejection of particles from the excitation zone. Discharges on briquettes prepared from sample-graphite mixtures increase precision noticeably, but do not allow satisfactory detection limits to be reached. An interesting method proposed to improve the feasability and reliability of powder analysis was described by PAPP [ 1, 23; it is based on the preparation of a phenolformaldehyde plastic binding electrode and was successfully applied to the determination of main components and trace elements in minerals, biological samples, plastics, etc. Beside arc and spark methods, discharges under reduced pressure seem suited for the analysis of powders, since the material is volatilized by sputtering, a process which can be well controlled. The method developed by EL ALFYef al. [3] and DC&AN et al. [4J permitted the analysis of the main constituents of powdered samples made conductive after mixing with copper powder. In this second case the glow discharge emission source (GDES), devised by GRIMM[S], was employed. The method can be applied for precise determinations at trace, minor and major constituent concentrations. In former work, the present authors described the use of another discharge at reduced pressure, namely the hollow cathode emission source

[ 1J [2] [3] [4] [SJ

L. PAPP, ~~~r~~jm. Acza, Wien I, 375 (1978). L. PAPP, Appl. S~frose. 32, 247 (1978). S. EL ALFY, K. LAQUAand H. MASSMANN,Z. Anal.Chem.263, 1 (1973) M. D&AN, K. LAQUAand H. MASSMANN,Specfroclrim. Acru 27B, 65 (1972). W. GRIMM, Speetrochim. Acta 23B, 443 (1968). 625

S. CAROLI er al.

626

Table 1.

I-ICES Instrumentation and working conditions

Equipment (supplied by RSV, FRG) vacuum spectrograph 1m/800 electric source HVG 2 Grimm’s glow discharge lamp (modified for use as a demountable HCES)

Srgnal detection and evaluation emulsion, Kodak SA-I films developer, Kodak D-19 (4 mm at 2OkO.l “C) microdensitometer, MD-100 (supplied by Jenopttk, GDR)

(HCES) [6, 7, Sf. This source is characterized by a high analyte residence time and consequently lower detection limits as compared to the GDES. The purpose of the present study was to ascertain whether the HCES could be successfully used for determining trace elements and minor constituents in powdered samples, with particular reference to the determination of the elemental composition of organic material after sample combustion. 2. EXPERIMENTAL 2.1. Spectroscopic instrumentation Details of

the HCES used and the emission spectroscopic equipment are given in Table I.

2.2. Sample preparation

The following operative scheme was set up: (a) powdered samples (named hereafter “A”) containing the elements to be determined, generally in the form of their oxides, were mixed with an artificial matrix “B” resembling a typical mineral residue of organic material, LiF is added, the mixture is spiked with Ga, 0, as an internal standard, and finally diluted with a 1: 4 graphite-copper powder mixture up to a total weight of ten times that of samples A (exact compositions and weights are set out in tables); (b) the mixtures “F” thus obtained were pressed into the form of hollow cylinders with an external diameter of 5.0 mm, an internal diameter of 3.0 mm, and a length of 7.4 mm; (c) other aliquots of the same mixtures F were stratified over a slightly pressed portion of copper powder so that a pellet embodying a thin layer of sample is obtained after complete pressing (the final disks had a diameter of 8.0 mm and a thickness of 3.00 mm); (d) the pressed cylindrical samples were then placed into supporting copper cylinders (external diameter 7.8 mm, internal diameter 5.0 mm)and closed at the lower end by the pressed disksas shown m Fig. 1. In this way it was possible to make up a hollow cathode whose internal surface (walls and bottom) consisted entirely of sample to be analyzed. To obtain stable pellets the mean size of sample particles must be below 40pm. All chemicals employed for the preparation of pellets were accurately ground to 30pm using a Pulverisette mill (Fritsch, FRG), 15 mm being as a rule sufficient. Standard samples A confining IO different elements at various ratios were prepared starting from a single stock mixture diluted with graphite powder in proportions of 1: 4,l: 5,l: 7.1 : 1I, 1 : 24 and 1: 99. Six standards (Al to A6) were thus obtained. Matrix B was prepared according to the data reported m Table 2. Equal amounts of matrix B and LiF (Merck Suprapur) are brought together in order to favour the formation of neutral species from analytes during discharge. Standards A and matrix B + LiF were subsequently mixed at a ratio 1: 1, doped with Ga, 0, as the internal standard, and finally diluted with a 4: 1 copper (Vaskut, 99.99997; purity )-graphite (Ringsdorff RW A, 99.9998 “/;,purity) mixture in order to render final samples electrically conductive. The percentage composition is given in Table 3 below. Table 4 presents the concentrations of the various elements in the linaf standards F obtained by dilution of the corresponding standards A. For each sample F blanks were prepared following the same procedure detailed above, but substituting an equivalent amount ofgraphite for component A. 0.8 g ofmaterial in total were necessary for preparing a complete hollow cathode charge (i.e. two hollow cylinders and one disk). 161 S. CAROLI and 0. SENOFONTE,Curt. J. Spt~rox. 25, 73 (1980). [73 S. CAROLI, A. ALIMONT~ and P. DELLE FEMMINE, Specrrose. l&t. 12. 871 (1979). [a] S. CAROLI. A. ALIMONTI and 0. SENOFONTE,Spectrose. Left. 13, 307 (1980).

Determination of trace elements in non-conductrng powders

621

-B

Fig. I. Hollow cathode arrangement for the analysis of pressed powders; A, sample-containing pellets; B, copper supporting cylinder; C, sealing O-ring; D, copper supporting plate; E. copper pellet incorporating sample.

Table 3. Percentage composition of samples F for discharge with HCES Table 2. Composition of matrix B Component

Percent

Weight (7;)

Salt

Sample A (I through 6) Matrix B Lithium fluoride Gallium oxide Copper Graphite

43.8 47.1 5.3 2.05 1.75

KCI (Carlo Erba RP) NaCl (Carlo Erba CGS) CaSO,.ZH,O (Merck p.a.) MgO (Merck p.a.) Fe,O> (Merck p.a.)

IO 5 5 0.07 63.95 15.98

A sketch of the die used for preparation of the sample cylinders is shown in Fig. 2. As deduced from exploratory testing, the height of sample containing cylinders should not exceed 10 mm, otherwise the pressure necessary to effectively compress the sample particles (4 tons cm- *, applied for 1 min) can damage the thin-walled hollow piston of the die (part 1 in Fig. 2).

Table 4. Analyte composition of standards F Element concentration

(ppm)

Sample No.

Aga

Alb

BC

Crd

MnC

Nir

Pbs

Snh

Ti’

ZnJ

Fl F2 F3 F4 F5 F6

5 20 40 60 80 loo

5 20 40 60 80 loo

50 200 400 600 800 loo0

5 20 40 60 80 loo

5 20 40 60 80 loo

10 40 80 120 160 200

50 200 400 600 800 loo0

50 200 400 600 800 loo0

5 20 40 60 80 loo

IO0 400 800 I200 1600 zoo0

Elementscontained in the mixturesas:a,Ag,O; b,AI,O,;c, HsBO,;d, Cr,O,;e, MnO,;f,NiO;g, PbO;h, SnO,; 1. TiO,; j, ZnO. All these compounds are of spectral grade purity (Spex Elements Kit 1010).

628

S. CAROLI

et al.

Fig. 2. Die utilized for the preparation of cylindrical pellets. 1, piston; 2, die mam body; 3, sealing Oring; 4, base; 5, central peg. In order to form cylindrical pellets with this die, powders are simply poured into the 1.0 mm hollow space formed between the main body and the central peg after removing the piston. The piston is then inserted and subjected to a pressure of 4 tons cm- ‘. In order to obtain hollow cylinders of the required height (7.4 mm) 600 mg of powdered samples Fare necessary. The pellets thus formed are finally drawn out easily from the die by demounting the base and removing the peg. Two pellets are then placed into the copper cathode as shown in Fig. 1. The same pressure was applied for preparing flat pellets as described above, in this case obviously with a conventional 8.0 mm die. The pressing conditions adopted are a good compromise to achieve good mechanical resistance of the samples, reproducible sputtering and a discharge of high stability.

2.3. Working conditions The working conditions are listed in Table 5. The analytical lines used are given in Table 6. In all cases the first is preferable,

with alternative

lines given in brackets.

Table 5. Operating conditions for hollow cathode discharge analysis of powders Spectrograph entrance slit: 30 pm Concave grating: I m radius, 1200 grooves mm-‘, blazed at 5” 52’. Paschen-Runge mounting Spectral range: 120-600 nm Dispersion: 0.78 nm mm- ’ at 300 nm Resolving power: 30 000 at 300 nm

Transport gas: Argon at 230 Pa Anode-cathode distance: 7.0 mm Applied current intensity: 400 mA Resulting voltage: 370-410 V (depending on sample composition) Predischarge time: 30 s Exposure times: 30 and 60s

Table 6. Spectral lines for analysis of powders Element

Ag Al B Cr Mn Ni Pb Sn TI Zn

Wavelength (nm) 338.289 (328.068) 396.153 (308.215) 249.678 (249.773) 425.435 (284.325) 403.075 (279.827) 352.454 (305.082) 283.307 (405.782) 283.999 (303.412, 286.333) 365.350 307.590 (334.502)

Determinationof trace elementsin non-conducting powders

629

After a preburning time of 30 s, very constant voltage values with less than 0.2 y0 oscillations are attained. For each F sample two consecutive exposures of 30 and 60 s, respectively,were taken on the same film. Each exposure series was then reproduced twice and the values were averaged.

3. RESULTSAND DISCUSSION 3.1. Power 0s detection Insofar as detection limits are concerned, these were calculated using the procedure recommended by KAISERand SPECKER[9]. The following equation was applied: CL=

C”

[

3 J2a,U,+ 1,)

( >I 1,+1, I

(1)

u

where cL = analyte concentration corresponding to the limit of detection, cU = background equivalent concentration, a, = relative standard deviation of background plus blank contributions, and I, and I, are the background and blank intensities, respectively. In the calculation of the detection limit for each element, quantities appearing in Eqn (1) were estimated on the basis of sixteen individual measurements. Results are summarized in Table 7. Values reported therein cannot be considered completely satisfactory. In fact, two of the main advantages of the hollow cathode emission source, i.e. the relatively long time of residence of analyte atoms in the discharge zone and the almost negligible background intensity level, are essential prerequisites for obtaining very favourable detection limits [lo]. The limitation in the case of powder analysis was not the capability of the hollow cathode to reach lower levels, but the fact that chemicals used for preparing samples contained traces of the elements under study, this holdjng particularly for the copper and graphite powders. The use of high purity chemicals thus becomes mandatory to take full advantage of the HCES’s analytical potential. 3.2. Calibration Satisfactory linear dependences of analytical signal on concentration were obtained for all elements with the exception of boron. This may be related to an irreproducible contribution of the emission from boron in the insulating glass cylinder separating the anodic and cathodic blocks of the lamp. In all instances four types of plots were taken into account, i.e. the emission intensity as well as the emission intensity ratio between analyte and gallium were plotted vs. logarithm of concentration (in pg) at two different exposure times (30 and 60 s). The gallium lines chosen were Ga I 403.298 nm and Ga I 417.206 nm. Though in some cases they were far from the spectral range of the analyte lines, they proved to be better than other gallium lines. Figures 3 and 4 illustrate as an example the plots obtained for Ag and Ni. Table 7. Detection limits obtained in the analysis of powders with HCES Element

Detection limit (pg g- ’ )

A8 Al B Cr Mn Ni Pb Sn Ti Zn

0.8 0.6 l

0.1 0.3 1.2 5.0 0.5 2.0 1.0

* Not determined [9] H. KAISER and H. SPECKER,Fresenius’ Z. Anal. Chem. 149,46 (1956). [lo] J. A. C. BROEKAERT,Spectrochim. Acfa 35B, 225 (1980).

630

S.

CAROLI et

al.

3.3. Analytical precision

Statistical treatment of the data obtained for each element clearly demonstrated that the best results are achieved with exposure times of 30 s and using gallium as internal standard. The only exception was Ti: in fact, no reliable graph could be plotted for this element under the conditions mentioned above owing to the very low levels of blackening, and an exposure time of 60s was therefore required. This conclusion was reached using the procedure described by NALIMOV [l l] as applied by AZIZ et al. [12]. Regression equations and regression coefficients s(log I) calculated under these conditions for each element are reported in Table 8. The reproducibility of the determinations was deduced by repeating exposures ten times for ail standards F under a given set of strictly defined conditions of gas

0

3.0

-2 3

.

0

*

z

2

x

4

2.5

.o 5

2.0,-’

A.0

/ I

o/

3.5

0.5

I

1.5

1.0

109 c(wmi

Fig. 3. Calibration graph of Ag in powders. 0, log I,, 30s; 0 , log f,, 60s;

2.0

l,

-1.5

log (I,/[,), 30% A ,

log U,lU, 60s.

-1.5 2.0

2.5 lop CtPPmt

Rg. 4. Calibration graph of Ni in powders. o, log f,, 30s; 0, log I,, 60s; l, log (l,,/&), 30 %,A , log (b/Lb

i-1t1 V. V. NALIMOV, The PI

Application

OJ Mathematical

60 s. Statistics

( 1969). A. Azrz, J. A. C. BROEKAERT and F. LEES,Speetr~him.

to Chemical

Analysis, Pergamon Press, Oxford

Rcta 360, 251 (1981).

Determinationof trace elements in non-conducting powders

631

Table 8. Regression equations and regression coefficients for the analysis of powders by means of HCES using gallium as internal standard Elements wavelength (nm)

Regression equation log cx = log CL+ 9 long(~x/~,)

Regression coeihcient s(log 1)

log c = 1.929-k 1.251 log(l/f,) log c = 2.787 + 2.307 log(f/f,) log c = 2.146+ 1.295 logI/!,) log c = 2.221 +2.212 iog(f/f,) lof c = 2.467+ 1.154 ~0g(~/f,) log c = 3.240 i- 1.190 log(lfl,) log c = 3.450-k 1.299 log(l/f,) log c = 3.367 + 1.077 log(l/l,) log c = 3.842 + 1.273 log(l/l,)

Ag 338.289 (a) Al 396.153 (a) Cr 425.435 (a) Mn 403.075 (a) Ni 352.454 (a) Pb 283.307 (a) Sn 283.999 (a) Ti 365.350 (b) Zn 307.590 (a)

0.0622 0.0347 0.0423 0.0348 0.0480 0.052s 0.0227 0.0671 0.0238

(a) exposure time = 30 s; (b) exposure time = 60 s

pressure, current intensity and discharge time (230 Pa, 400 mA and 30 s, respectively). The range of the relative standard deviation of the estimate as defined by NALIMOV[l I] is given for each element in Table 9. 4. CONCLUSXON On the basis of these results the fotlowing condusions can be drawn: (i) the method proposed is suitable for reliable analysis of trace elements in powdered samples; (ii) it is possible to determine a unique set of experimental parameters which allows the simultaneous determination of numerous elements; (iii) the linearity of the plots extends over at least two orders of magnitude; (iv) an internal standard seems to yield better results than plotting emission intensities only. The method can be advantageously used for the direct analysis of trace elements in biological materials after ashing. An inspection of the cathodes after discharge revealed that the attack on the inner surface was more concentrated on the lower half and bottom of the cathode itself. This is in accordance with results of previous studies by CAROLI~~al. [13,14], and in practice permits the use of only one cylinder of pressed powder instead of two. The second cylinder, which practically does not contribute to the discharge, can simply be a blank placed over that containing the sample within the cathode. This limits the amount of sample F required for the discharge to only 650-700 mg. This amount corresponds to about l-l 5 mg of ashed material, which is indeed a small quantity. Dilution of the original material to be analyzed is not a serious drawback since the final ~on~ntrations are still adequate to attain sufficiently intense signals. Table 9. Reproducibility of elements determination in powders by means of HCES Element

Ag Al Cr Mn Ni Pb Sri Ti Zn

Range of relative standard deviation (“/,) 1.3-2.0 0.7-1.9 1D-3.2 1.1-2.3 l.tYkl.5 1.2-1.9 0X-2.2 1.4-3.5 0.62.7

[t3] S. CAROLI,A. ALIMONTIand F. PETRUCCI, Anal. C&m. Acta 136, 225 (1982). SENOFONTE and K. ZIMMER,Spcrrox. Left. 14,575(1981).

[I43 S. CAROLI,A. ALIMONTI,0.