1728
Anal. Chem. 1985, 57, 1728-1733
added DMF, Me2S0, or HMPA. The temperature rise occurring after the end point AT1 is related to the heat liberated, Q, by the equation
where M,,Ma,M,,and M,are the masses of water, acrylonitrile, propan-2-01, and dipolar aprotic solvent, respectively, and C,, D,, C,, and C, are the corresponding specific heats. AT,, the “corrected” rise in temperature, defined as the temperature rise that would occur if the same reaction proceeded in the absence of the dipolar aprotic solvent, is related to Q by eq 2
The third component of eq 4 is used as the correction factor in obtaining the data for the construction of Figure 4. The reaction rates (avalues) are corrected for changes in the concentrations of the reactants, as the total volume of the titration solution changes, by multiplying a by V (Figure 4a) and VL (Figure 4b) where V is the ratio of the total volume to the initial, i.e., smallest volume used. Figure 4a gives the corrections assuming that the reaction is first order, while Figure 4b assumes a second order reaction. V is given by the equation
V=
V , + 2 mL,
+ 1 mL, + 1 mL, + 1 mL, + 1 mL,
2 mL V, + 2 mLa
where a, w, and p are acrylonitrile, water, and propan-2-01, respectively,and V, is the volume, in mL, of the dipolar aprotic solvent. Registry No. Me2S0,67-68-5;DMF, 68-12-2;HMPA, 680-31-9; water, 7732-18-5; acrylonitrile, 107-13-1.
LITERATURE CITED (1) Szmant, H. H. Ann. N.Y.Acad. Sci. 1975, 243, 20-23. (2) Cowie, J M. G.; Toporowski, P M. Can. J . Chem. 1961, 3 9 , 2240-2243. (3) Craver, J. K. J. Appi. Polym. Sci. 1970, 14, 1755-1765. (4) Holmes, J. R.; Kivelson, D.; Drlnkard, W. C. J. Am. Chem. Soc. 1962, 8 4 , 4677-4686. (5) Tokuhiro, T.; Menafra, L.; Szmant, H. H. J . Chem. Phys. 1974, 6 1 . 2275-2282. (6) Wu, N. M.; Mallnin, T. L. Anal. Chem. 1980, 52, 186-189. (7) Tommila, E.;Murto, M. L. Acta Chem. Scand. 1963, 17, 1947-1956. (8) Rammler, D. H.; Zafforini, A. Ann. N.Y. Acad. Sci. 1967, 141, 13-23. (9) Henderson, T. R.; Henderson, R. F.; York, J. L. Ann. N.Y. Acad. Sci. 1975, 243,3a-53. (10) Tatsumi, C.; Kotani, R. Bull. Univ. Osaka Prefect, Ser. B 1969, 21, 123-131. (11) Jezorek, J. R.; Mark, H. B., Jr. J. Phys. Chem. 1970, 7 4 , 1627-1633. (12) Greenhow, E. J. J. Chem. Sac., Perkin Trans. 2 1978, 1248-1255. (13) Greenhow, E. J.; Spencer, L. E. Analyst (London) 1973, 9 8 , 90-91. (14) Greenhow, E. J.; Dajer de Torrljos, L. A. Analyst (London) 1979, 104, 801-811. (15) Greenhow, E. J.; Spencer, L. E. Analyst (London) 1973, 9 8 , 98-102. (16) Shaw, R. J . Chem. Eng. Data 1969, 74, 461-465. (17) “Dimethyl Sulfoxide Technical Builltin”; Crown Zellerbach Corp: Camas, WA, 1966; p 2. (18) Geller, B. E. Zh. Fiz. Khim. 1961, 35, 2210-2216. (19) Chagas, A. P.;Alroldi, C., personal communication, University of Campinas, 1983. (20) Gutmann, V. Chem. Brit. 1971, 7 , 102-107.
RECEIVED for review December 10,1984. Accepted February 25, 1985.
Influence of Laboratory Environment on the Precision and Accuracy of Trace Element Analysis S. B. Adeloju and A. M. Bond* Division of Chemical and Physical Sciences, Deakin University, Waurn Ponds, Victoria 321 7, Australia The Influence of the laboratory air quality on the precklon and accuracy of elemental analysls at the lower trace/ultratrace levels has been lnvestlgated. Conalderable varlatlons were observed In a conventional laboratory for the perlodlc voltammetrlc measurement of standard solutlons of selenlum, copper, lead, zlnc, nlckel, and cobalt, owlng to the varlabllity of alrborne contamlnatlon under thls condltlon. However, the results obtained for cadmlum were identical under clean room and conventlonal laboratory condlons. The sensltlvlty of the cathodlc strlpplng voltammetry of selenlum to other metal Ions and the absence of analytlcal blank for thls element provlde a novel approach for monltorlng the varlablllty of alrborne contamlnatlon In conventional laboratorles. The effect of such varlablllty on the preclslon and accuracy of trace element analysis Is demonstrated for some blologlcal and environmental samples.
Despite the wide range of instrumental techniques that are now available for trace and ultratrace analysis, the key to
successful and reliable determination of inorganic elements at these levels lies in the ability to control the analytical blank. The four main sources of contamination that can cause variability in blank levels are (i) the laboratory environment, (ii) reagents, (iii) the apparatus, and (iv) the analyst (I). Evidently, careful control of the last three sources requires operation in a laboratory environment of an acceptable standard. Although the control of temperature and humidity of analytical laboratories with sophisticated instrumentation has now become a common practice, the need for regulation of airborne contamination is still not regarded as mandatory by most. These contaminants, usually in the form of dusts, mists, and fumes, circulate in the atmosphere and can enter laboratories through any vent (2). Consequently, the composition of the air in the laboratory will be about the same as that of the surrounding atmosphere, but even more seriously,this will fluctuate with the prevailing atmospheric conditions. The common constituents in ambient air are aerosols composed mainly of solid and liquid particulate matter which can cause serious contamination problems at the trace/ultratrace levels. Murphy (3),Patterson and Settle ( 4 ) ,Zief and Mitchell (5), Zief and Nesher (6),and several other workers have demon-
@ 1985 American Chemical Society 0003-2700/85/0357-1728$01.50/0
ANALYTICAL CHEMISTRY, VOL. 57,
strated that airborne particulate can contribute significantly to trace element contamination from the environment. In a conventional laboratory in Washington DC, Murphy (3) found a lead content of 0.77 bg/m3 while Boutron (7) and Zief and Nesher (6)observed an increase in concentration by a factor of up to 2 or more when solutions were analyzed for iron. A rather unusual example of airborne contamination was demonstrated by Patterson and Settle (4) for the determination of lead in High Sierra lakes where water samples were collected by two different sampling methods. In one of these methods, the water sample was collected and raised in a bucket to a helicopter while the other was simply taken aboard a small boat. The sample air-freighted in the helicopter gave lead value of 0.3 ng/g which was 20 times greater than the value of 0.015 ng/g obtained for the sample collected aboard the boat. Evidently, the former sample was more exposed to airborne contamination (in addition to the exhaust gases from the helicopter) than the latter sample. In another example, the analysis of hydrofluoric, hydrochloric, and nitric acids for aluminum, iron, calcium, magnesium, lead, titanium, and boron in a closed system containing inert gas gave concentrations an order of magnitude lower than values obtained in a laboratory atmosphere (8). It is convincing from these examples that analysis at the lower trace/ultratrace levels must be performed in controlled laboratory environment to prevent the sample from being contaminated by artifacts contributed by the laboratory atmosphere or unclean containers. Despite the considerable developments in the design and construction of clean rooms, comprehensive investigations of the practical significance of these types of facilities for elemental analysis at the trace/ultratrace levels have only scarcely been made ( 4 , 5 ) . Over the years, several different constructions of clean rooms have been described (1,9), but often with insufficient data to demonstrate the need for such facilities. In this paper, a comprehensive investigation of the influence of laboratory air quality on the precision and accuracy of elemental analysis at the lower trace/ultratrace level is reported. The periodic monitoring of the analytical blank under controlled and uncontrolled laboratory atmosphere by voltammetric techniques was useful for observing the variability of airborne contamination under these conditions. The effect of this variability is demonstrated for the determination of trace elements in biological and environmental samples. The typical blank levels of elements such as selenium, copper, lead, cadmium, zinc, nickel, and cobalt are given and the levels at which they can be adequately determined under the different laboratory conditions are indicated. EXPERIMENTAL SECTION Reagents and Standard Solutions. All acids and ammonia solution used were Aristar grade (B.D.H. Chemicals), while other reagents were of analytical grade purity. Ammonia-ammonium chloride buffer solution, dimethylglyoxime (dmgH2),standards, and distilled deionized water were prepared as previously described ( I 0-1 3). Instrumentation. All experiments were performed on EG&G Princeton Applied Research microprocessor-controlledinstrumentation as previously described (10-13). A medium-size mercury drop with a surface area of 0.015 cm2was used in all cases. Glassware. All glassware (borosilicate) and polyethylene bottles were soaked in 2 M nitric acid for at least 7 days, washed three times with distilled deionized water, soaked in distilled deionized water, and finally soaked in 0.1 M hydrochloric acid until ready for use. Biological and Environmental Materials. Bovine liver sample was obtained from the U.S. National Bureau of Standards, Washington, DC. The veal sample was prepared by slicing a portion into small sections and then homogenizing in a Waring Blendor. Sufficient water was added to form a liquid paste and
NO. 8, JULY 1985
1729
the sample was blended for another 5 min at high speed. The homogenized veal sample was then freeze-dried,ground to a fine powder, and finally sieved to ensure uniform particle size. The seawater sample was collected from 100 m depth at Queenscliff, Victoria, Australia. Sample Decomposition. The bovine liver sample was decomposed by dry ashing using sulfuric acid as the ashing aid, as previously described (13),while the veal sample was wet digested with nitric and sulfuric acid mixture, as recently reported (12, 14). In both cases, a 0.5-g sample was used. Working Area. The results reported for clean laboratory in this study were obtained under clean air conditions controlled at a temperature of 22.5 i 0.5 "C. All sample and solution preparations were made in a class-100 clean room while the analytical measurements were performed in a class-1000 clean room. Both of these laboratories form part of Deakin University Trace Analysis Unit. Comparative results were obtained in a conventional laboratory maintained regularly at a temperature of 18 & 1 "C. RESULTS AND DISCUSSION Voltammetric Approach for Monitoring Airborne Contamination. The periodic measurement of the voltammetric responses of some elements in standard (or blank) solution should, ideally, provide useful information on the variability of the analytical blank level resulting mainly from airborne contamination under different laboratory environments. While such approach may be useful in assessing the reliability of the results obtained for some elements at the trace/ultratrace levels, every precaution must be taken to ensure that the data obtained in this way are truly indicative of the variability in the analytical blank. The concept of working in a closed system, maintained with all voltammetric techniques, provides a unique approach for making such comparative measurements, but other factors that may result in variability in the measurement must also be carefully controlled. In particular, possible variations that may result from chemicals and instrumental malfunction must be eliminated. The avoidance of these problems in this study requires preparation of all the standards and other chemicals in the clean room (class 100) and storage in precleaned polyethylene bottles over the period of measurement. Subsequently the solutions were left in the respective laboratory for at least 24 h prior to making the first measurement. In addition, the same batches of chemicals and standards were used during this period. The problems associated with instrumental variation were easily circumvented by use of a microprocessor-controlled polarographic analyzer which is based on an autoranging system, requiring no manual instrumental setting except for the usual input of desired voltammetric parameters, and timing of the preconcentration step of stripping, and adsorption voltammetry is made automatically by the built-in timer in the instrument. Under the clean laboratory conditions maintained regularly at 22.5 f 0.5 "C,no significant variation was observed for repeated peak current measurements of several aliquots of standard solutions at different time intervals. However, in the conventional laboratory with no clean air facilities and temperature maintained always at 18 f 1"C,the measured peak current value varied at different times for the several aliquots of the standard solutions. It appears from all indications that these variations resulted from the variability in the airborne contamination in the uncontrolled laboratory environment. Nevertheless to ensure that there was no contribution from the instrument in these variations, all measurements were made in both laboratory conditions after 1h from the time the instrument was switched on. The observed variations resulting, under these conditions, from the variability of the airborne contamination in controlled and uncontrolled laboratory environment are considered below for a number of trace elements.
1730
ANALYTICAL CHEMISTRY, VOL. 57, NO. 8, JULY 1985
,J - ,l -
w5
035
065
I
065
1
4
0
-E , V
Cathodic stripping voltammetric (CSV) determination of selenium under (a) clean room and (b) conventional laboratory conditions: (1) 0.1 M HCI, (2) 10 lg/L Se(IV) In 0.1 M HCI, DPCSV, t = 300 s (including 15-s equilibration time), scan rate = 2 mV s-1" Figure 1.
8
DAY
Periodic monitoring of airborne contamination using selenium stripping peak as an indicator under (0)clean room and (H) conventional laboratory conditions: 5 Ng/L Se(IV)in 0.1 M HCI; other conditions as in Figure 1. Flgure 2.
Table I. Typical Blank Levelsnof Some Trace Elements under Different Laboratory Conditions
element Cd
cu
Pb Zn Se co Ni
blank level, pg/L conventional clean labb labb 0.010 f 0.001 0.050 0.003 0.020 0.001 0.050 0.003
*
30 pg/L. While it may be possible to determine this element at lower levels, repeated scrupulous cleaning of laboratory ware will most certainly be necessary, but despite the additional time often spent on this, adequate reproducibility is not always guaranteed. On the contrary, the results obtained for cadmium under the two different laboratory environments were quite similar. Figure 6 shows that comparable variability in the peak currents were observed in both cases. This may indicate that the analytical blank was similar in both laboratories throughout the period of examination. However, the use of longer deposition time (30 min) revealed, as shown in Table I, that the blank level in the uncontrolled laboratory environment was 5 times higher than that obtained in the clean room. Nevertheless both of the blank levels are low enough to permit precise and accurate determination of cadmium under both laboratory conditions. The variability, as dem-
a
4
-E,V
DAY
Flgure 5. Periodic variation of lead peak current of a standard solution under (0) clean room and (). conventional laboratory conditions: 0.5 pg/L Pb(I1) in 0.1 M HCI; other conditions as in Figure 4. I
I
O.O2t
0
8
4
DAY
Flgure 6. Periodic variation of cadmium peak current of a standard solution under (0)clean room and (D) conventional laboratory conditlons: 0.5 pg/L Cd(I1) in 0.1 M HCI; other conditions as in Figure 4; (a) X = 25.7 nA and (b) R = 26.0 nA.
onstrated by the results in Figure 6 for the standard solution (0.5 pg/L), was often less than *2% under both clean room and conventional laboratory conditions. Other Trace Elements. Figures 7-10 show that, unlike cadmium, the voltammetric responses for copper, zinc, nickel, and cobalt varied considerably from one day to the next in the uncontrolled laboratory environment possibly as a consequence of airborne contamination. In contrast, the variability observed for the measured peak currents of the four elements under clean room condition was between only 1and 3%, which is further in support of this view. Also the typical blank levels, as shown in Table I, clearly indicate that copper and zinc can only be comfortably determined in uncontrolled laboratory environment at concentrations >30 pg/L and for cobalt and nickel, the realistic concentration, in terms of precision and accuracy, under the same condition appears to be >5 pg/L. Levels substantially lower than those can be comfortably determined for the four elements under clean room condition. From the data in Table I, it is quite evident
1732
ANALYTICAL CHEMISTRY, VOL. 57,NO. 8, JULY 1985
n 0.0
f 0
8
4
0
8
4
DAY
DAY
Figure 7. Periodic variation of copper peak current of a standard solution under (0) clean room and (W) conventional laboratory conditions: 5 pg/L Cu(I1) in 0.1 M HCI; other conditions as in Figure 4.
Figure 9. Periodic Variation of nickel peak current of a standard solution under (0)clean room and (W) conventional laboratory conditions: 0.5 pg/L Ni(I1) in 0.1 M NH,/NH,Ci 5X M dmgH,, DPAV, t , = 60 s (including 15-sequilibration time), scan rate = 4 mV
+
S'.
0
8
4
DAY
Flgure 8. Periodic variation of zinc peak current of a standard solution under (0)clean room and (W) conventional laboratory conditions: 5 pg/L Zn(I1) in 0.1 M HCI; other conditions as in Figure 4.
that the use of clean room condition is quite vital for precise and accurate determination of the seven elements considered in this study at the lower trace/ultratrace levels. Effect of Analytical Blank Contribution on the Precision and Accuracy of Trace Element Analysis in Biological and Environmental Samples. The variability of the analytical blank, as demonstrated in this study, can influence the precision and accuracy of trace element analysis in real samples significantly. Even the adoption of the approach of blank subtraction is sometimes no more than a futile exercise owing to periodic variability of analytical blank levels in uncontrolled laboratory environment. On the contrary, the ability to control and maintain the elemental contribution from the blank solution at the lowest possible level at all times is the main fador, demanding the use of clean room condition for trace and ultratrace analysis. The results in Table I1 show that, in spite of the observed variability of the analytical blank in uncontrolled laboratory environment, selenium and cadmium can be adequately determined in bovine liver sample. However, the result obtained for lead in the conventional laboratory was considerably higher
L 0
8
4
DAY
Flgure 10. Periodic variation of cobalt peak current of a standard solution under (0)clean room and (B)conventional laboratory conditions: 0.5 pg/L Co(I1) in 0.1 M NH,/NH,CI 5 X lo-' M dmgH,;
other conditions as in Figure 9.
+
Table 11. Influence of Laboratory Environment on the Determination of Some Elements in Bovine Liver Samplea amt of element, bg/g element
clean labb
conventional labb
certified valueC
selenium cadmium lead
1.13 f 0.03 0.27 f 0.02 0.34 f 0.02
1.03 f 0.05 0.29 f 0.04 0.47 f 0.08
0.27 f 0.04 0.34 f 0.08
1.1 f 0.1
Sample decomposed by wet digestion using nitric and sulfuric acid mixture. Blank values were less than the mean deviation of the element being determined. *Error is mean deviation based on triplicate determination. Error is standard deviation based on results from various instrumental techniques.
ANALYTICAL CHEMISTRY, VOL. 57.
Table 111. Influence of Laboratory Environment on the Determination of Cadmium and Lead in Veal Sample” element cadmium lead
amt of element, uala clean labb conventional labb 0.35 f 0.01 0.19 f 0.02
0.39 f 0.03 0.30 f 0.05
Sample decomposed by dry ashing, using sulfuric acid as ashing aid a t 500 “C. Blank values were less than the mean deviation of the sample being determined. Error is mean deviation based on triplicate determination.
Table IV. Influence of Laboratory Environment on the Determination of Some Trace Elements in Seawater Samplesn
elements cadmium copper lead zinc
amt of element, pg/L clean labb conventional labb acidified nonacidified acidified nonacidified pH 2.7 pH 8.2 pH 2.7 pH 8.2
0.05 f 0.01 0.81 f 0.04 0.63 f 0.03 1.71 f 0.05
0.06 f 0.02 0.56 f 0.05 0.42 f 0.04 0.32 f 0.09
0.06 7.22 1.72 4.07
f 0.02 f 1.95 f 0.82 f 2.59
0.08 4.76 1.27 3.52
f 0.02 f 3.00 f 0.56
f 1.81
Water samples were kept in the clean room but were exposed
to the different laboratories during acidification and determination step. Error is mean deviation based on triplicate determihation.
NO. 8, JULY 1985
1733
of this element at the ultratrace level is therefore considerably less than for the other three elements. The results in Table IV also indicate that the portion of the element accessible for the ASV determination is dependent upon the final pH of the solution. It appears acidification of the sample to pH 2.7 enables adequate accessibility of the four elements in addition to permitting their simultaneous determination by ASV.
CONCLUSION The results reported in this study clearly demonstrate the significance of clean room conditions for precise and accurate determination at the lower trace/ultratrace levels. Apart from cadmium for which little or no contamination was experienced, most common elements cannot be adequately and accurately determined in uncontrolled laboratory environment even at the lower trace level. Every endeavor should therefore be made when embarking on inorganic elemental analysis at the trace/ultratrace levels to bring part of the laboratory to clean room conditions, even if only possible by installation of relatively inexpensive clean benches. Also, other precautions such as removal of all metallic installations and painting of walls, ceiling, pipes and other metallic parts with nonabrasive metal-free plastic paint must be given high priority. It is now more convincing than ever that reliable trace element data can only be obtained by working under such a clean, highquality environment. Registry No. Cd, 7440-43-9; Cu, 7440-50-8; Pb, 7439-92-1; Zn, 7440-66-6; Se, 7782-49-2; Co, 7440-48-4; Ni, 7440-02-0; HzO, 18-5.
than that obtained in the clean room. The individual results obtained for triplicate determination, 0.38,0.57, and 0.49 Fg/g, suggest that the higher value may due to contamination from the laboratory environment. Similarly higher results obtained for lead in the veal sample, as shown in Table 111,confirmed this view. The results obtained for cadmium in both biological samples agreed favorably under the different laboratory conditions. For environmental samples such as seawater, airborne contamination poses a serious threat for precise and accurate determination because most of the elements are present a t the ultratrace level. This view is well demonstrated by the results in Table IV which show that considerably higher concentrations were obtained for copper, lead, and zinc, despite the fact that the same batch of sample used was stored in precleaned polyethylene bottles and kept under clean room conditions. The separate portions of the sample used were only exposed to the different laboratory environments during the acidification and/or during the voltammetric determinations. The excellent precision of the results obtained under the clean room conditions implies that the undesirable reproducibility of the results obtained in the conventional laboratory was due to airborne contamination, which undoubtedly will also affect the accuracy of the data obtained for the three elements. The results obtained for cadmium, again, agreed favorablyunder both laboratory conditions. The requirement for clean room conditions for the determination
LITERATURE CITED Moody, J. R. Anal. Chem. 1982, 5 4 , 1358A-1376A. Chem. Eng. News 1971, July 19, 29-33. Murphy, T. J. I n “Proceedlngs of the Seventh Materlal Research Symposium”; U S . Government Printing Office: Washlngton, DC, 1976; pp 509-539. Patterson, C. C.; Settle, D. M. I n “Proceedings of the Seventh Material Research Symposium“; US. Government Printing Offlce: Washington, DC, 1976; pp 321-351. Zlef, M.; Mitchell, J. W. “Contamination Control in Trace Element Analysis”; Wiiey: New York, 1976. Zief, M.; Nesher, A. G. Environ. Sci. Techno/. 1974, 8 , 677-678. Boutron, C. Anal. Chim. Acta 1972, 6 1 , 140-143. Allmarin, I. P., Ed. “Analysis of Hlgh Purity Metals”; Israel Program for Scientific Translations: Jerusalem, 1968; pp 1-31. Austin, P. R. “Clean Rooms of the World”, Ann Arbor Science: Ann Arbor. -MI. .._. 1967. Adeloju, S. B.; Bond, A. M.; Brlggs, M. H.; Hughes, H. C. Anal. Chem. 1983, 55, 2076-2082. Adeloju, S. B.; Bond, A. M.; Briggs, M. H. Anal. Chim. Acta 1984, 164. 181-194. Adelbju, S. B.; Bond, A. M.; Brlggs, M. H. Anal. Chem. 1984, 56, 2397-2401. Adeloju, S. B.; Bond, A. M.; Noble, M. L. Anal. Chlm. Acta 1984, 161, 303-314. Adeloju, S. B.; Bond, A. M.; Briggs, M. H. Anal. Chem. 1985, 57, 1386- 1390. Adeloju, S. B.; Bond, A. M.;Hughes, H. C. Anal. Chim. Acta 1983, 148, 59-69. (16) Forbes, S.; Bound, G. P.; West, T. S. Taianta 1979, 26, 473-477. (17) Henze, G. Mikrochim. Acta 1981, 1 1 , 343-349.
RECEIVED for review December 31,1984. Accepted March 25, 1985.