r”lanra Vol. 28, pp. 37 to 41 0 Pergamon Press Ltd 1981. Printed m Great Britam
AN INSTRUMENTAL METHOD OF SOLUTION PRECONCENTRATION P. A. MICHALIK and R. STEPHENS Trace Analysis
Research
Centre, Department of Chemistry, Halifax, Nova Scotia, Canada
(Receiued 28 Maq’ 1980. Accepted
9 July
Dalhousie
University,
1980)
Summary-Aerosol concentration trostatic trapping technique. The those of voltage, trap dimensions, the technique offers an inherently
before atomic-absorption measurement is achieved by use of an eleceffects of various experimental parameters are examined, including gas flow-rate through the trap and trapping time. It is concluded that clean method of solution preconcentration which is capable of enhancing the signal by an order of magnitude or more.
Various
methods
samples
before
are chemical
available analysis.
for
turi surfaces is minimal, while droplets which touch the walls of the spray chamber tend to condense out and therefore cease to be important in the analysis. Similar reasoning can be applied to the heating chamber and condenser which are used to effect desolvation (although such units can introduce an added risk of interference effects*). Provided that the temperature of the heating chamber is not so high as to induce boiling, and that the diameter of the condenser is not so small as to force gas to bubble through the condensed solvent, wall contamination ceases to be important because any solution affected by it is not used in the analysis. Suitable desolvation systems have been described by a number of workers;9-15 they include devices capable of handling high concentrations of matrix salts,9*‘3-‘5 which is clearly desirable for any system which is ultimately intended for practical application.
preconcentrating
A number
of excel-
lent texts have been written which give the appropriate details.’ All such methods involve some degree of sample handling, with associated risks of contamination, adsorption losses. etc. These risks increase as the analyte concentration in the sample falls, which leads to correspondingly increased demands on the skill of the analyst and eventually gives rise to a limiting concentration at which the analysis ceases to be feasible. These problems have been discussed by Chakrabarti et al.’ among others.3-6 The present paper examines an attempt to develop a method of instrumental preconcentration suitable for use with a conventional flame atomic-absorption spectrometer. The method is intended to reduce sample handling to near the irreducible minimum, defined in this case by the design of the sample-introduction system used with such spectrometers. The method adopted is electrostatic dust-precipitation. The aerosol from a conventional pneumatic nebulizer is dried in a heating chamber, desolvated, and collected on a wire electrode which is connected to a high-voltage supply. When sufficient sample has been collected the wire is heated to volatilize the collected material. The flame support-gas (oxidant only) is used to transport the initial aerosol to the wire collector, and later to carry the volatilized material into the flame for atomization and measurement. The first part of the procedure resembles the electrostatic dust-precipitators used on the large scale industrially; the second part is analogous to the furnace-flame combinations which have been described recently.’ It is felt that the risks associated with handling very dilute solutions are minimal when pneumatic nebulizers are used. These devices have evolved to the point where it is difficult to conceive of a method which will involve less sample handling, and yet will still serve to introduce a sample solution into a flame. The effectiveness of these devices arises because sample contact-time with the inlet capillary and ven-
EXPERIMENTAL The present work is concerned primarily with the development of an efficient aerosol trap, rather than of a complete analytical system. The latter is a more complicated problem, which will be addressed subsequently and in appropriate stages. Cadmium, lead and zinc were selected for testing the apparatus, because of their ease of volatilization, and the experiments described are confined to examination of the behaviour of these three elements, but there is no reason for supposing that their trapping behaviour is likely to differ from that of other elements. Apparatus
Figure I shows a block diagram of the apparatus. The heating chamber and condenser are similar in design to those described by Veillon and Margoshes.‘” Their insertion causes about a 25% loss of the aerosol delivered by the nebuhzer. The flow-splitter was used only to determine the effect of flow-rate on aerosol trapping efficiency; otherwise the entire flow of support gas passed through the aerosol trap. The trap itself is shown in Fig. 2. The body of the device consists of a length of glass tube carrying a few turns of copper wire on the outside to provide an earth (ground).
37
P. A.
38
MICHALIK
and R.
STEPHENS
Aerosol Trap
FLAME Nebulizer Heating Condenser Chamber
2 Flow Splitter
Fig. 1. Block diagram
In all cases a winding of about one turn per inch was found to be quite sufficient. Unless otherwise stated, the tube was 45 cm long and 2.5 cm in diameter. The ends of the tube were closed with rubber stoppers, each with a 5 cm length of 0.5-mm bore glass tube inserted through its centre. A length of 22-gauge chrome1 wire (the collector) formed the axis of the system. For trapping a sample the collector wire was connected to a S30 kV d.c. supply (Precise Measurements Co., Flemington, N.J.). No difference in behaviour was found for positive or negative polarity of this supply; the experiments described here were all done with a positive collector potential. To volatilize the collected material the chrome1 wire was disconnected from the high-voltage supply, and connected to a variable transformer to produce the usual electrothermal volatilization. During this heating cycle slight tension was maintained in the wire to counteract its thermal expansion. This arrangement is only applicable for relatively volatile materials, but was extremely convenient for the present investigation because factors such as trap length, outer tube diameter and diameter of the collector electrode could be readily altered. The system therefore permitted the effects of such variables to be established; this information is needed before a trap can be designed which is capable of handling more intractable materials. To operate the trap, the high-voltage supply was connected and a solution aspirated as usual. Atomic-absorp-
,+_
Chrome1
Wire
of the apparatus
tion readings could be taken during this time if required. At the end of the trapping time aspiration was stopped and the collector wire was heated to drive off the sample. This produced an absorption signal as a single pulse (examples are given in Fig. 8 below). No problem was encountered with background absorption, since all the solutions used were made from the pure salts dissolved in demineralized distilled water. Washing by aspirating a few ml of distilled water proved to be adequate for cleaning the unit between readings. Mechanism of aerosol collection The aerosol particles generated by pneumatic nebulization can acquire quite a high electrostatic charge. However, most of this charge appeared to be dissipated in the heating chamber and condenser (some quite spectacular electrical discharges were observed in the former). Thus the collection process appears to operate on electrically neutral particles. This view is confirmed by the observation that the polarity of the high voltage supply can equally well be positive or negative, virtually lOOy/, trapping efficiency being attainable in either case. Therefore the force responsible for aerosol collection is considered to arise from the interaction of the electrostatic field in the trap with the electric dipole which that field induces in each particle. For such a force to cause physical transport of the particle it is necessary for the field to be highly inhomogeneous. The production of such a field is accomplished here by use of a wire electrode of diameter very small in comparison with that of the coaxial earthed winding. Measurement of trapping ejiciency The trapping efficiency, n, of the device is defined as the fraction of the aerosol mass entering the trap which is actually retained there. Values of n were determined as follows. Calibration curves were plotted for each element by use of the sample introduction system shown in Fig. 1. but with the collector voltage set at zero. The atomicabsorption signal was then measured with the collector voltage set at the required value. The signal was converted into sample concentration by use of the calibration curve. The ratio, r, of this concentration to the corresponding value at zero collector voltage gives the mass fraction of aerosol leaving the trap. Hence n = (1 - r) x lOOa<. RESULTS AND DISCUSSION
___________ -----------:
of trapping parameters
Injuence
From, Condenser
11
Figures length kV,
:II
and
tration that
3 and on
for a trap
trap
length
Supply
Fig. 2. The aerosol
voltage trap.
and
used
voltage
length and
vice versa.
of n on the
voltages and
of 14 and
solution
of 45 cm. These
trapping
effects then
the dependence
for collector
collector
complementary Power
4 show
of the trap
on
voltage n. That
the smaller In practice
results
tend
a useful
show
to produce
is, the
the trap
18
concen-
can upper
higher
the
be made, limit
of
Solution preconcentration
about 18 kV for the voltage was found. At higher voltages flashover became a problem. Therefore, a trap length of 45 cm was used for subsequent work, which gave trapping efficiencies of !W-lOO~O for voltages in the range 14-18 kV. The effect of solution concentration on n (Fig. 4) shows a similar pattern for each element examined. In each case there is a slight reduction in n as the concentration, and hence the total mass of the particles entering the trap, is increased. The differences in concentration at which n falls below 100°~ for different elements are presumed to arise from differences in the physical properties, such as density and dielectric constant, of the particles involved. It is also possible that the gas-flow patterns in the particular apparatus used exert some influence. This matter has not yet been investigated in detail. However, the results clearly suggest that no fall-off of n is likely to occur at very low concentrations, where high efficiency is most necessary (this conclusion is also supported by the results in Fig. 7 below). The effects of support-gas flow-rate and of trap diameter on n are interdependent. Examination of’n for various trap diameters and a fixed volumetric gas
n 100% .-------------------
/p/
#/
150
300 Trap Length (mm
39
450
1
Fig. 3. Effect of trap length on n: 0 = 18 kV; X = 14 kV. Values shown were measured for Pb at 283.3 nm; similar results are observed for Cd and Zn.
I 5
10
15
20
30
25
35
40 cont.
1ppm)
SO
m t
I 100
Q2
Q5
B-_o--0-A-A
#----o_
n SO
x-x_
Q8
-0
X-
cont. ( ppm)
‘.O
1.5
20
Zn -0 cl_ X-x-
x-x
04
06
4kV
SO
0.1
Q2
a3
concyppm)
Fig. 4. Effect of solution concentration
0.7
and collector voltage on n.
P. A. MICHALIKand R. STEPHENS
40
is probably
a result of the correspondingly reduced residence time of aerosol particles in the trap. In this situation less time is available for particles to cover the distance to the collector from their particular line of flow through the trap. Thus fast-moving particles which enter the trap far from the collector are unable to reach it before they leave the system.
loo-n
;:: ‘.
a__
.A....
““‘.“..
50 --
“‘......
.._...
A
;8k”
*k” 2k”
-----------_________~
. . . . .._....__.___
A
Ej’kt of trapping time
40 -30.. 20-m IO20
* 100
40
Flow Rate (ml/set) Fig. 5. Effect of flow-rate on n for various trap lengths and collector voltages, measured for 15 ppm Pb. Trap lengths: q45cm;o30cm;A15cm. flow-rate showed that no measurable variation occurred. This suggests that any reduction in n due to increased trap diameter is counteracted by a corresponding increase in n due to the decreased linear gas flow-rate through the trap, which leads to an increased residence time for the aerosol particles. The influence of flow-rate alone on n was found by using the flow-splitter (Fig. 1) to control the proportion of the support gas flowing through the trap. Results are shown in Fig. 5. The fall in n with increasing flow-rate
down
”
lOO%-
The most important property of the present device for analytical purposes is probably its ability to maintain almost constant trapping efficiency over a period of time. The effect is shown for lead in Fig. 6. The data show that a slight reduction of trapping efficiency with time occurs at high solution concentrations, presumably because of significant surface coverage of the electrode, and perhaps failure of particles to adhere to areas already covered. At low concentrations, however, it appears to be feasible to extend trapping times almost indefinitely. Times up to 20 min have been used and by no means represent an upper limit. To illustrate the potential capability of the system, five replicate measurements were made, in each of which 100 ml of a lo-ng/ml lead solution were trapped. The corresponding absorption signals observed at 283.3 nm on volatilizing the trapped samples are shown in Fig. 7. The reproducibility of the system is felt to be adequate (and in any case is probably governed by the volatilization step rather than by the preconcentration process). No difficulty
--50
Pam x 100 ppm
X
n
go-80-70.-
60
120 Time
Fig. 6. Effect of trapping
Fig. 7. Repetitive
(set)
time on n. Values measured
for Pb at a collector
iLu_ signals,
each obtained
I 240
180
after trapping 1OOml of a lo-ng/ml voltage = 18 kV.
voltage
solution
of 18 kV.
of Pb. Collector
Solution preconcentration
41
was encountered in obtaining the data in Fig. I despite the fact that the measurements were made at the end of the experiments described above, by which time the system had been exposed to large volumes of solutions containing high concentrations of all three
by the use of a heated chrome1 wire for the volatilization step. Some potential solutions to these problems have been found, and will be discussed in subsequent communications.
elements examined; this provides some confirmation that this preconcentration step is inherently clean. For comparison with the data in Fig. 7, the detection limit of the standard unmodified mstrument for lead was found to be about 0.2 ppm at 283.3 nm.
Acknowledgement-The authors are indebted to the Natural Sciences and Engineering Research Council. Canada for support of this work.
Thus it appears clear that the sensitivity can be enhanced by at least an order of magnitude with this system. CONCLUSIONS
The experiments done show that it is feasible to use electrostatic trapping of aerosol particles as a simple, clean and effective method of sample preconcentration before atomic-absorption measurements. The risks of sample loss or contamination during the process are felt to be minimal because once the sample has left the nebulizer the preconcentration is carried out quite automatically, and effectively in the vapour phase: if any aerosol does touch a potentially contaminating surface it tends to condense and become lost to the analysis. By the time the sample reaches a possible source of contamination (the collector), preconcentration is complete and therefore contamination has ceased to be particularly important. Encouragingly, no limits have been found for the degree of preconcentration obtainable with the technique. However, it is clear that this is only likely to apply to pure solutions such as were used here; it is almost certain that in practice some limit will be set by the composition of the sample matrix. A further problem that arises is the lack of generality imposed
REFERENCES
1. See, e.g., R. Mavrodineanu and H. Boiteux, Flame Spectroscopy, Wiley, New York, 1965; G. F. Kirkbright and M. Sargent, Atomic Absorption and Fluorescence Spectroscopy, Academic Press, New York, 1974. 2. K. S. Subramanian, C. L. Chakrabarti and I. S. Maines, Anal. Chem., 1978, SO, 444. 3. G. C. Eicholz. A. E. Naeel and R. B. Hughes. ibid.. 1965, 37, 863. 4. D. E. Robertson, Anal. Chim. Acta, 1968, 42, 533. 5. R. A. Durst and B. T. Duhart, Anal. Cbem., 1970, 42, 1002. 6. W. G. King, J. M. Rodriguez and C. M. Wai, ibid.. 1974,46, 771. D. J. Koop, M. D. Silvester and J. C. Van Loon, paper presented at the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 1978. 8. P. W. J. M. Boumans and F. J. DeBoer, Spectrochim. Acta, 1976, 31B, 355. 9. A. Venghiattis, Appl. Opt., 1968, 7, 1313. 10. A. Hell, W. F. Ulrich, N. Shifrin and J. RamirezMunoz, ibid., 1968, 7, 1317. 11. C. VeilIon and M. Margoshes, Spectrochim. Acta, 1968. 23B, 553. 12. G. Uny, J. N’Guea Lotton, J. P. Tardiff and J. Spitz, ibid., 1971, 26B, 151. 13. H. J. Issaq and L. P. Morgenthaler, Anal. Chem., 1975. 47, 1661. 14. Idem, ibid., 1975, 47, 1668. 15. Idem, ibid., 1975, 47, 1748.