- Email: [email protected]
Determination of As, Se, Cr, Co and Ni in geochemical samples using the Stabilized Temperature Platform Furnace and Zeeman background correction
Determination of As, Se, Cr, Co and Ni in geochemical samples using the Stabilized Temperature Platform Furnace and Zeeman background correction E. PRUSZKOWSKA and P. BARRETT Perkin-Elmer Corporation.901 Ethan Alien HIghway. Ridgefield,CT 06877. USA (Receiwd
28 July
1983)
Abstract-The Stabilized Temperature Platform Furnace technique with Zeeman background correction was used for the determination of As, Se. Cr. Co and Ni in five Geochemicai Exploration Reference (GXR) samples from the U.S. Geological Survey. After dissolving the samples in a mixture ofaclds, aqueouscalibration curves aere used The method ofaddlGons was not necessary.The resultsare in good agreement with recommended values and recoveries of added analyze were close to IOO”,. The method proved to be simple and accurate.
INTRODKTION ATOMIC absorption spectrometry using the graphite furnace has provided very low detection limits but it is sometimes impossible to obtain these levels in the presence of complex matrices. Matrix dependent errors are often reported. A combination of several improvements has made the analysis of complex samples much less difficult. All of these improvements, including the L’vov Platform, pyrolytically coated tubes, rapid heating of furnace (2OOO”C/s),fast electronics, integrated absorbance signals and matrix modifier, make up a new furnace concept which is called the Stabilized Temperature Platform Furnace [ 1,2]. The platform is heated primarily by radiation from the graphite tube wall, while the tube is heated by an electric current passing through it. There is therefore a time lag between the heating of the tube and the platform. The atomization of a sample is delayed until the atmosphere within the graphite tube is at a nearly stable temperature. This reduces problems related to residence time of the free atoms in the tube and decreases the probability of analyte losses in the form of molecular compounds. Another advantage is that the platform sometimes extends the life of the graphite tube, especially when corrosive samples such as those high in acid concentration are used. We use pyrolytically coated tubes exclusively because they offer remarkable improvement in performance, especially in situations where the analyte or matrix reacts strongly with carbon. The platform is heated primarily from the tube, so it is important that the tube reach final steady state temperature as quickly as possible. For this purpose the fast heating feature of the furnace called “‘max power” is used. A heating rate of approximately 2OOO”C/sis provided. Fast electronics are needed here because peaks are generated very quickly and inaccurate quantitation of peaks can cause errors. In situations where the analyte is vaporized into the gas phase when it is nearly stable in temperature, peak area should be used instead of peak height. The idea of a matrix modifier is associated with the platform concept. Matrix modifiers are solutions that are specific for each analyte and are dispensed on the platform together with the sample. With the use of matrix modifiers, the volatility of the analyte may be reduced and/or the volatility of the matrix increased, so that a larger portion of the matrix may be removed during the char step. Loss of volatile analyte metals during the char step is reduced and the background from the matrix is also reduced. A combination of all these factors will decrease or even eliminate interferences in many cases. The problem of background correction still exists when a background signal appears at the same time as the analyte signal. When the background level is very high, accurate correction may be difficult using conventional continuum source background correction systems. In addition, structured background can cause inaccurate results in continuum [I] D. C. MANNING and W. SLAVIN, Appl. Spectrosc. 37, 1 (1983). [2] W. SLA~IN, G. R. CARNRICK, D. C. MANNING and E. PRUSZKOWSKA,Ar. Specrrusc. 4.69 (1983). U(B) 39:2,3-U
485
486
E. PRUSZKOWSKA and P. BARRETT
systems. High background levels can be compensated by tbe use of the Zeeman technique and spectral interferences do nor occur in Zeeman AA. Therefore, combining the STPF technique with Zeeman background correction has made it possible to develop simple methods for several elements in difficult matrices, such as sea water [3, 4) or biological materials [S, 61. To test the STPF technique with Zeeman background correction we determined several difficult elements in silicate rock samples. We used five Geochemical Exploration Reference (GXR) samples from the U.S. Geological Survey. Data and recommended values for 42 elements in these samples were reported by GLADNEY er al. [7]. We chose to determine five elements (As, Se, Cr, Co, and Ni) because round robin furnace AAS results quoted in the GLADNEYpaper demonstrated poor accuracy and AAS has not been used for the Se determination. We have compared our results against the recommended values, established in the GLADNEY paper. source
EXPERIMENTAL
Equipmem Most of the tests were performed on a Perkin-Elmer Zeeman/5000 equipped with an HGA R-500 and with an AS-40 autosampler. To compare Zeeman and continuum soura: background correction, we conducted some m~urements on a Model 4000 s~rophotometer equipped with an HGA R-500 graphite furnace and an AS-40 autosampler. Peak absorbances and integrated absorbances were calculated on a Data System IO and graphics were plotted on a Hewlett-Packard 7225A plotter. Pyrolytically coated graphite tubes (Perkin-Elmer part No. B0109-322) with solid pyrolytic graphite platforms (P-E part No. BOltl-091) were used. Graphs ofabsorbance profiles have been used to optimize the analytical conditions and some of these graphs have been included in this paper to illustrate the effect of changes in analytical parameters. Ail analytical results have utilized integrated absorbance, A. s, data collected on the Data System 10.
Standard solutions were prepared from 1000 mg I - l standards (Aifa Products, Ventron) by dilution with deionized water. The Ni matrix modifier solution was prepared by dissolving pure Ni cups (Perkin-Elmer part No. 303-0813) in 5.6 7; heated HNO, (Uhrex). Magnesium nitrate solution was prepared from reagent grade MgfNO& .6Hz0 (Baker). Water from a Continental Water System was purified in a Mill&ore deionizer consisting of Miiii-RO, and Miiii-Q systems. The geological samples (GXR) were obtained from the U.S. Geological Survey. Procedure
To prepare each sample, 0.5 g of the powdered GXR sample was dissolved in a 64
mixture (5 + 1) of
HF and HCIO*. Each sample was then evaporated to dryness, and the process was repeated two more times. Each sample was then treated with 3 ml of HCIO, only and again evaporated to dryness. The residue was then dissolved in lOOmi of 1% HNO>. Part of the Ca and AI fluorides remained undissolved in the final solution, but that did not affect the determination. Co-precipitation did not occur and solutions with the sediments stored in plastic bottles were found to be stable for at least seven months. AR anafyses reported here used dilutions in water of the original 0.5 g in lOO-ml rock sohttions that varied up to lOO-fold to keep the resulting absorbances less than about 0.3 A. We did not establish the extent of the linear ranges. Ten microliters of the diluted sample or standard solutions and 5 ~1 of the matrix modifier, when used, were automatically dispensed onto the platform using the AS-40 autosampler. The temperature conditions and modifiers were similar to those reported for the Stabilized Temperature Platform Furnace 123 and are summarized in Table 1. After 100 to 150 firings, the tube and platform were replaced on a routine basis; however, our experience does not indicate that this is essential. The contact rings and windows were cleaned to remove accumulated salt approximately once a week.
[3] G. R. CARNRICK, W. SUVIN and D. C. MANNING,Anal. Chem. S3, 1866(1981). [4] E. PRUWKOWSKA, G. R. CARNRICK and W. SLAVIN,Anal. Chem. 55, 182 (1983). [S] E. PRU~LZKOWSKA, G. R. CARNRICK and W. SLAVIN,Clin. Chem. 29,477 (1983). [6] E. PRUSWOWSKA, G. R. CARNRKKand W. SLAVIN,At. Spectrosc. 4, 59 (1983). [7] E. S. GLADNEY,D. R. PERRIN,I. W. OWENSand D. KNAB,Anal. Chem. 51, 1557 (1979).
487
Atomtc absorptton spectrometry Table 1. Zeeman graphite furnace condltlons Element
Modifier
As se Cr co
20 pg NI 100 pg NI 50 c(g Mg (NO,)z.%O 50 pg Mg (NO,),.6H,O
Ni
none
Char temp. (~C)
Atom temp. ( Cl
1400 1400 1300 1300 I200
2500 2100 2500 2400 2500
RESULTS
In order to obtain accurate results for this determination we developed optimum operating conditions for each of five elements: As, Se, Cr, Co and Ni. This involved finding the best charring and atomization temperature as well as the optimum amount of matrix modifier. The choice of the appropriate matrix modifier was based on the previously published work [2]. We found that, with the conditions we used, accurate results were obtained. Our work with As demonstrates how optimum conditions are chosen. Previously Ni was found to be the best modifier for As [ 13. First, a char study for a standard and sample solution was performed at 2500 “C atomization temperature. The results of this study are shown in Fig. 1 as the absorbance profiles for an As standard. The actual char temperatures and integrated absorbance values are indicated on the profiles. Nickel, used as a matrix modifier, converted As to a less volatile compound enabling a higher char temperature to be used. The integrated absorbance did not vary much up to about 1500 “C. Above 1600 “C the integrated absorbance values dropped off sharply, indicating that As was beginning to volatilize. For this work, we selected 1400°C as the optimum char temperature. We also did the char study for one of the GXR samples and found that it can be charred at the same temperature as the standard solution. The results of an As atomization study are presented in Fig. 2. The integrated absorbance decreased very slowly with the increasing atomization temperature. At the same time, the peak became less broad and was shifted to an earlier time in the atomization step. We selected 2500 “C for the atomization temperature. The amount of Ni used as a modifier was also studied, and the results are shown in Fig. 3. Using a fixed amount of As, we found that the integrated absorbance signal began to drop off when less than 5 c(g of Ni was used on the platform. When the amount of Ni was between 5 0. 5
w
1
CHAR 1200
:
0.26
1400
zl
1500
i
1700
0.24 0.24
0. 10
0I 4
0
TIME
Fig. I. Char study for I ng of As. The modifier was 20 pg of Ni and the atomization temperature was 2500°C. The wavelength was 193.7 nm. The temperatures and A.s signals are shown.
E. PRUSZKOWSKAand P BARRETI
0
4 TIME
Fig. 2. Atomization study for
1 ng of As. The modifier was 20 pg of Ni and the char temperature was 1400°C. The wavelength was 193.7 nm.
1-1.2
UC NI
0. 23 0. 26
0_
4
0 TIME
of varying amounts of matrix moddier (Ni) on I ng of As. The char temperature was 1400°C and the atomization temperature was 2500 ‘C. The wavelength was 193 7 nm.
Fig. 3. E&t
and 4Opg, the integrated absorbance signal remained constant, but the absorbance profile became broadened. For all As determinations, we used 2Opg of Ni on the platform. After establishing the proper conditions for the As determination, we investigated the degree to which large amounts of Al in the sample matrix would affect the As determination, since Riley [8] had reported that Al interferes with As at 193.7 nm. Performing measurements at 193.7 nm with continuum source background correction and using standards with a fixed amount of As and varying amounts of Al, we collected data that showed increasing distortion of the As profiles with increasing Al. Two examples are shown in the left-hand panel of Fig. 4. Under these conditions even 25 mg 1- 1 of Al significantly increased the signal and disturbed the absorbance profile. This is in agreement with RILEY [S]. Using Zeeman background correction at the same wavelength, there is very little interference from up to 100 mgl-’ of Al as illustrated in the right-hand panels of Fig. 4. By 100 mgl-’ of Al the b&ground signal was higher than 2.0 A, and the profile was slightly disturbed. [8] K. W. RILEY, At. Spectrosc. 3, 120 (1982).
489
Atomic absorption spectrometry
c____~_
c--
0
c_-_A
4
0
--
4
4
CScr> --
‘IWE
Ftg. 4. The Al interference on 2 ng of As comparing continuum source background correctton with Zeeman effect background correction. The mod&r was 2Opg of Ni and the wavelength was 193.7 nm. The left-hand panels are from the contmuum source and the rtght-hand panels from the Zeeman-elTect correctors
An alternative wavelength for Asat 197.2 nm has half the sensitivity of that at 193.7 nm, but the interference from Al is not as severe. Using continuum source background correction, we found that up to 100 mg I- 1 of Al could be accurately compensated by the background corrector. With 200 mg l- 1 of Al there was a significant decrease in the integrated absorbance value. However, with the Zeeman system at 197.2 nm, we found that as much as 900 mg l- ’ of Al caused only a small decrease in the integrated absorbance value. This means that even in cases of very high Al content and low As concentrations, we could expect accurate results. Our samples were rich in Al (3-21 y; in the rock), so we decided to use the 197.2~nm wavelength even with Zeeman correction. The final solution of the geochemical samples contained less than 0.5 y; HNOJ and traces of HC104. We determined that as much as 5 “/;;HNO, and 2 “/;,HClO, could be used without changing the integrated absorbance signal. Similar experiments were performed for Se, Cr, Co and Ni to arrive at the conditions listed in Table 1. For each element we did char studies with standard and sample solutions. Using the appropriate matrix modifier, we found that sample and standard solutions can be charred up to the same temperature. We also performed atomization studies for all four elements. The amount of matrix modifier was not critical but different levels were tested and the amounts reported in Table 1 were found optimum. During tests with Se we confirmed again that spectral interference from Fe at 196.0 nm [9] does not occur using Zeeman background correction. Figure 5 compares results obtained with continuum source background correction and Zeeman background correction when running one of the geological samples (GXR- 1) with an Fe content of about 25 7; in the rock, 1.2 gl” in the solution. The actual background level from this sample was relatively low, only about 0.1 A. As seen in Fig. 5, the continuum background correction is subject to a spectral interference, resulting in a large negative signal from overcompensation. The plot using the Zeeman corrector has no visible disturbance of the absorbance profile and the integrated absorbance is higher, indicating that we are able to obtain accurate and reliable analytical results. Except for the problem associated with the effect of Al on As, which forced us to use the less sensitive As wavelength, all the conditions for the five elements determined in these rock samples used conditions that were very similar to those suggested earlier [2] for general [9] F. J. FERNANDEZand R. GIDDINGS, AI. Specrrosc.
3.61
(1982).
490
E. PRUSZKOWSKAand P. BARRETT
AA-EC
ZAA
0.091
0. 146
6
0
TIME
Fig. 5. Comparison profiles for Se in GXR-I wrth continuum source background correctton (upper) and Zceman eflbct background correctton (lower). The modifier was 100~1g of NI. the char temperature was 1100°C. theatomization temperature was 2IOO”Cand the wavelength was 196.0 nm.
furnace determinations. The slopes of the working curves for the rock matrices were the same as had been reported for simple standards. Specifically, the L’vov Platform was used for all determinations including Ni, Co and Cr. Our optimization tests for Ni indicated no advantage from the use of the Mg(NO& matrix modifier that we had previously suggested, so no modifier was used. The Mg(NO& modifier permitted use of slightly higher char temperatures for Co and Cr, as we previously had found. For Se, the somewhat larger amount of Ni as a matrix modifier seemed preferable to our earlier recommendation. The characteristic amounts found for the metals studied in the paper were: 17and 33 pg/O.O044A-s for As at 193.7 nm and 197.2 nm, respectively;and 253.0, 9.5 and 13.5 for Se, Cr, Co and Ni, respectively. Only in the case of Ni did the characteristic amounts differ significantly from our previous publication [2], probably reflecting an error in the standards that had been used for that paper. Using these optimum conditions we determined all five elements in the five geological samples. Two separate aliquots of the powdered rock were weighed and dissolved on different days. Two separate dilutions were made of each original rock aliquot. Each of these four solutions were analysed five times and the statistical calculations were performed on the 20 separate determinations for each rock and each analyte. Thus the statistical values should reflect the aggregate errors of the full procedure. The recovery experiment used a single dilution of each of the two rock aliquots. To each of these solutions three separate additions of standard solution were made providing six recovery solutions for each rock and each analyte. All determinations were made against the same working curves and the six recoveries for each rock and analyte were averaged to produce the recovery data in Table 2. Our results along with the recommended values and AA results from the work of GLADNEY et41.[7] are given in Table 2. Agreement between our results using the Stabilized Temperature Platform Furnace technique with Zeeman effect background correction and the recommended values is generally very good.
CONCLUSION
The results obtained for the geological samples using the STPF technique with Zeeman background correction are in good agreement with recommended values and have good
491
Atomic absorption spectrometry Table 2. AS, Se, Cr, Co and Ni m geological samples
Sample
Element
GXR-1
As
GXR-2
!se Cr co Ni As
se Cr
co Ni GXR-3
As
se Cr co Ni GXR-5
As
se Cr co
25 320 k 26 22*1a 21*13 46*20 36&24 < 10 38&20 16*14 11*1 26* 14 27*3 40 6500*400 21 f 10 S&l9 40*3 67*23 47*3 40 la*5 -
se
al+36 38kl6 36+1 79&18 96+9 60 3OOiso -
Cr co Ni
7Ok32 24+ 18 34*17
Ni GXR-6
Results by AA from [l] olg/g)
As
Recom. values from [l] (rl3/k%)
Results from this study GueM
Recoveries this study
( 7”)
460+30
454*11
108
18.6 f 0.8 lo+2 9.3 * 1.1 42klO 31*5
17.3 f 0.8 15+1 8.4 f 0.7 45i3 27+2
98 97 87 104 98
0.74f0.11 37+10 9k2
0.74kO.12 37 + 1 9.5 + I.0
95 100 95
18*3
21+2
103
4000*450
3460+90
98
< 0.25 16+1 42+3
104 a7
55 + 5
60*4
109
12k3
13k2
96
1.1*0.1 100*5 30&5
0.95 f 0.20 109*4 31*2
95 101 95
63il
78k3
106
34Oi30
310*30
97
1.07*0.13 96k 10 14*3 22+4
1.09*0.13 9a*3 14 f 1 26i3
89 96 94 100
0.22 f 0.02 19*1 48+5
precision. The technique is simple and reliable. After dissolving the sample in a mixture of acids, aqueous calibration curves are used. The method of additions was unnecessary. Recoveries of analyte added to the samples were about 100 % for each element. We believe that the technique is suitable for the determination of other elements and can help solve many problems occurring during analyses of geochemical materials.
28 July
1983)
Abstract-The Stabilized Temperature Platform Furnace technique with Zeeman background correction was used for the determination of As, Se. Cr. Co and Ni in five Geochemicai Exploration Reference (GXR) samples from the U.S. Geological Survey. After dissolving the samples in a mixture ofaclds, aqueouscalibration curves aere used The method ofaddlGons was not necessary.The resultsare in good agreement with recommended values and recoveries of added analyze were close to IOO”,. The method proved to be simple and accurate.
INTRODKTION ATOMIC absorption spectrometry using the graphite furnace has provided very low detection limits but it is sometimes impossible to obtain these levels in the presence of complex matrices. Matrix dependent errors are often reported. A combination of several improvements has made the analysis of complex samples much less difficult. All of these improvements, including the L’vov Platform, pyrolytically coated tubes, rapid heating of furnace (2OOO”C/s),fast electronics, integrated absorbance signals and matrix modifier, make up a new furnace concept which is called the Stabilized Temperature Platform Furnace [ 1,2]. The platform is heated primarily by radiation from the graphite tube wall, while the tube is heated by an electric current passing through it. There is therefore a time lag between the heating of the tube and the platform. The atomization of a sample is delayed until the atmosphere within the graphite tube is at a nearly stable temperature. This reduces problems related to residence time of the free atoms in the tube and decreases the probability of analyte losses in the form of molecular compounds. Another advantage is that the platform sometimes extends the life of the graphite tube, especially when corrosive samples such as those high in acid concentration are used. We use pyrolytically coated tubes exclusively because they offer remarkable improvement in performance, especially in situations where the analyte or matrix reacts strongly with carbon. The platform is heated primarily from the tube, so it is important that the tube reach final steady state temperature as quickly as possible. For this purpose the fast heating feature of the furnace called “‘max power” is used. A heating rate of approximately 2OOO”C/sis provided. Fast electronics are needed here because peaks are generated very quickly and inaccurate quantitation of peaks can cause errors. In situations where the analyte is vaporized into the gas phase when it is nearly stable in temperature, peak area should be used instead of peak height. The idea of a matrix modifier is associated with the platform concept. Matrix modifiers are solutions that are specific for each analyte and are dispensed on the platform together with the sample. With the use of matrix modifiers, the volatility of the analyte may be reduced and/or the volatility of the matrix increased, so that a larger portion of the matrix may be removed during the char step. Loss of volatile analyte metals during the char step is reduced and the background from the matrix is also reduced. A combination of all these factors will decrease or even eliminate interferences in many cases. The problem of background correction still exists when a background signal appears at the same time as the analyte signal. When the background level is very high, accurate correction may be difficult using conventional continuum source background correction systems. In addition, structured background can cause inaccurate results in continuum [I] D. C. MANNING and W. SLAVIN, Appl. Spectrosc. 37, 1 (1983). [2] W. SLA~IN, G. R. CARNRICK, D. C. MANNING and E. PRUSZKOWSKA,Ar. Specrrusc. 4.69 (1983). U(B) 39:2,3-U
485
486
E. PRUSZKOWSKA and P. BARRETT
systems. High background levels can be compensated by tbe use of the Zeeman technique and spectral interferences do nor occur in Zeeman AA. Therefore, combining the STPF technique with Zeeman background correction has made it possible to develop simple methods for several elements in difficult matrices, such as sea water [3, 4) or biological materials [S, 61. To test the STPF technique with Zeeman background correction we determined several difficult elements in silicate rock samples. We used five Geochemical Exploration Reference (GXR) samples from the U.S. Geological Survey. Data and recommended values for 42 elements in these samples were reported by GLADNEY er al. [7]. We chose to determine five elements (As, Se, Cr, Co, and Ni) because round robin furnace AAS results quoted in the GLADNEYpaper demonstrated poor accuracy and AAS has not been used for the Se determination. We have compared our results against the recommended values, established in the GLADNEY paper. source
EXPERIMENTAL
Equipmem Most of the tests were performed on a Perkin-Elmer Zeeman/5000 equipped with an HGA R-500 and with an AS-40 autosampler. To compare Zeeman and continuum soura: background correction, we conducted some m~urements on a Model 4000 s~rophotometer equipped with an HGA R-500 graphite furnace and an AS-40 autosampler. Peak absorbances and integrated absorbances were calculated on a Data System IO and graphics were plotted on a Hewlett-Packard 7225A plotter. Pyrolytically coated graphite tubes (Perkin-Elmer part No. B0109-322) with solid pyrolytic graphite platforms (P-E part No. BOltl-091) were used. Graphs ofabsorbance profiles have been used to optimize the analytical conditions and some of these graphs have been included in this paper to illustrate the effect of changes in analytical parameters. Ail analytical results have utilized integrated absorbance, A. s, data collected on the Data System 10.
Standard solutions were prepared from 1000 mg I - l standards (Aifa Products, Ventron) by dilution with deionized water. The Ni matrix modifier solution was prepared by dissolving pure Ni cups (Perkin-Elmer part No. 303-0813) in 5.6 7; heated HNO, (Uhrex). Magnesium nitrate solution was prepared from reagent grade MgfNO& .6Hz0 (Baker). Water from a Continental Water System was purified in a Mill&ore deionizer consisting of Miiii-RO, and Miiii-Q systems. The geological samples (GXR) were obtained from the U.S. Geological Survey. Procedure
To prepare each sample, 0.5 g of the powdered GXR sample was dissolved in a 64
mixture (5 + 1) of
HF and HCIO*. Each sample was then evaporated to dryness, and the process was repeated two more times. Each sample was then treated with 3 ml of HCIO, only and again evaporated to dryness. The residue was then dissolved in lOOmi of 1% HNO>. Part of the Ca and AI fluorides remained undissolved in the final solution, but that did not affect the determination. Co-precipitation did not occur and solutions with the sediments stored in plastic bottles were found to be stable for at least seven months. AR anafyses reported here used dilutions in water of the original 0.5 g in lOO-ml rock sohttions that varied up to lOO-fold to keep the resulting absorbances less than about 0.3 A. We did not establish the extent of the linear ranges. Ten microliters of the diluted sample or standard solutions and 5 ~1 of the matrix modifier, when used, were automatically dispensed onto the platform using the AS-40 autosampler. The temperature conditions and modifiers were similar to those reported for the Stabilized Temperature Platform Furnace 123 and are summarized in Table 1. After 100 to 150 firings, the tube and platform were replaced on a routine basis; however, our experience does not indicate that this is essential. The contact rings and windows were cleaned to remove accumulated salt approximately once a week.
[3] G. R. CARNRICK, W. SUVIN and D. C. MANNING,Anal. Chem. S3, 1866(1981). [4] E. PRUWKOWSKA, G. R. CARNRICK and W. SLAVIN,Anal. Chem. 55, 182 (1983). [S] E. PRU~LZKOWSKA, G. R. CARNRICK and W. SLAVIN,Clin. Chem. 29,477 (1983). [6] E. PRUSWOWSKA, G. R. CARNRKKand W. SLAVIN,At. Spectrosc. 4, 59 (1983). [7] E. S. GLADNEY,D. R. PERRIN,I. W. OWENSand D. KNAB,Anal. Chem. 51, 1557 (1979).
487
Atomtc absorptton spectrometry Table 1. Zeeman graphite furnace condltlons Element
Modifier
As se Cr co
20 pg NI 100 pg NI 50 c(g Mg (NO,)z.%O 50 pg Mg (NO,),.6H,O
Ni
none
Char temp. (~C)
Atom temp. ( Cl
1400 1400 1300 1300 I200
2500 2100 2500 2400 2500
RESULTS
In order to obtain accurate results for this determination we developed optimum operating conditions for each of five elements: As, Se, Cr, Co and Ni. This involved finding the best charring and atomization temperature as well as the optimum amount of matrix modifier. The choice of the appropriate matrix modifier was based on the previously published work [2]. We found that, with the conditions we used, accurate results were obtained. Our work with As demonstrates how optimum conditions are chosen. Previously Ni was found to be the best modifier for As [ 13. First, a char study for a standard and sample solution was performed at 2500 “C atomization temperature. The results of this study are shown in Fig. 1 as the absorbance profiles for an As standard. The actual char temperatures and integrated absorbance values are indicated on the profiles. Nickel, used as a matrix modifier, converted As to a less volatile compound enabling a higher char temperature to be used. The integrated absorbance did not vary much up to about 1500 “C. Above 1600 “C the integrated absorbance values dropped off sharply, indicating that As was beginning to volatilize. For this work, we selected 1400°C as the optimum char temperature. We also did the char study for one of the GXR samples and found that it can be charred at the same temperature as the standard solution. The results of an As atomization study are presented in Fig. 2. The integrated absorbance decreased very slowly with the increasing atomization temperature. At the same time, the peak became less broad and was shifted to an earlier time in the atomization step. We selected 2500 “C for the atomization temperature. The amount of Ni used as a modifier was also studied, and the results are shown in Fig. 3. Using a fixed amount of As, we found that the integrated absorbance signal began to drop off when less than 5 c(g of Ni was used on the platform. When the amount of Ni was between 5 0. 5
w
1
CHAR 1200
:
0.26
1400
zl
1500
i
1700
0.24 0.24
0. 10
0I 4
0
TIME
Fig. I. Char study for I ng of As. The modifier was 20 pg of Ni and the atomization temperature was 2500°C. The wavelength was 193.7 nm. The temperatures and A.s signals are shown.
E. PRUSZKOWSKAand P BARRETI
0
4 TIME
Fig. 2. Atomization study for
1 ng of As. The modifier was 20 pg of Ni and the char temperature was 1400°C. The wavelength was 193.7 nm.
1-1.2
UC NI
0. 23 0. 26
0_
4
0 TIME
of varying amounts of matrix moddier (Ni) on I ng of As. The char temperature was 1400°C and the atomization temperature was 2500 ‘C. The wavelength was 193 7 nm.
Fig. 3. E&t
and 4Opg, the integrated absorbance signal remained constant, but the absorbance profile became broadened. For all As determinations, we used 2Opg of Ni on the platform. After establishing the proper conditions for the As determination, we investigated the degree to which large amounts of Al in the sample matrix would affect the As determination, since Riley [8] had reported that Al interferes with As at 193.7 nm. Performing measurements at 193.7 nm with continuum source background correction and using standards with a fixed amount of As and varying amounts of Al, we collected data that showed increasing distortion of the As profiles with increasing Al. Two examples are shown in the left-hand panel of Fig. 4. Under these conditions even 25 mg 1- 1 of Al significantly increased the signal and disturbed the absorbance profile. This is in agreement with RILEY [S]. Using Zeeman background correction at the same wavelength, there is very little interference from up to 100 mgl-’ of Al as illustrated in the right-hand panels of Fig. 4. By 100 mgl-’ of Al the b&ground signal was higher than 2.0 A, and the profile was slightly disturbed. [8] K. W. RILEY, At. Spectrosc. 3, 120 (1982).
489
Atomic absorption spectrometry
c____~_
c--
0
c_-_A
4
0
--
4
4
CScr> --
‘IWE
Ftg. 4. The Al interference on 2 ng of As comparing continuum source background correctton with Zeeman effect background correction. The mod&r was 2Opg of Ni and the wavelength was 193.7 nm. The left-hand panels are from the contmuum source and the rtght-hand panels from the Zeeman-elTect correctors
An alternative wavelength for Asat 197.2 nm has half the sensitivity of that at 193.7 nm, but the interference from Al is not as severe. Using continuum source background correction, we found that up to 100 mg I- 1 of Al could be accurately compensated by the background corrector. With 200 mg l- 1 of Al there was a significant decrease in the integrated absorbance value. However, with the Zeeman system at 197.2 nm, we found that as much as 900 mg l- ’ of Al caused only a small decrease in the integrated absorbance value. This means that even in cases of very high Al content and low As concentrations, we could expect accurate results. Our samples were rich in Al (3-21 y; in the rock), so we decided to use the 197.2~nm wavelength even with Zeeman correction. The final solution of the geochemical samples contained less than 0.5 y; HNOJ and traces of HC104. We determined that as much as 5 “/;;HNO, and 2 “/;,HClO, could be used without changing the integrated absorbance signal. Similar experiments were performed for Se, Cr, Co and Ni to arrive at the conditions listed in Table 1. For each element we did char studies with standard and sample solutions. Using the appropriate matrix modifier, we found that sample and standard solutions can be charred up to the same temperature. We also performed atomization studies for all four elements. The amount of matrix modifier was not critical but different levels were tested and the amounts reported in Table 1 were found optimum. During tests with Se we confirmed again that spectral interference from Fe at 196.0 nm [9] does not occur using Zeeman background correction. Figure 5 compares results obtained with continuum source background correction and Zeeman background correction when running one of the geological samples (GXR- 1) with an Fe content of about 25 7; in the rock, 1.2 gl” in the solution. The actual background level from this sample was relatively low, only about 0.1 A. As seen in Fig. 5, the continuum background correction is subject to a spectral interference, resulting in a large negative signal from overcompensation. The plot using the Zeeman corrector has no visible disturbance of the absorbance profile and the integrated absorbance is higher, indicating that we are able to obtain accurate and reliable analytical results. Except for the problem associated with the effect of Al on As, which forced us to use the less sensitive As wavelength, all the conditions for the five elements determined in these rock samples used conditions that were very similar to those suggested earlier [2] for general [9] F. J. FERNANDEZand R. GIDDINGS, AI. Specrrosc.
3.61
(1982).
490
E. PRUSZKOWSKAand P. BARRETT
AA-EC
ZAA
0.091
0. 146
6
0
TIME
Fig. 5. Comparison profiles for Se in GXR-I wrth continuum source background correctton (upper) and Zceman eflbct background correctton (lower). The modifier was 100~1g of NI. the char temperature was 1100°C. theatomization temperature was 2IOO”Cand the wavelength was 196.0 nm.
furnace determinations. The slopes of the working curves for the rock matrices were the same as had been reported for simple standards. Specifically, the L’vov Platform was used for all determinations including Ni, Co and Cr. Our optimization tests for Ni indicated no advantage from the use of the Mg(NO& matrix modifier that we had previously suggested, so no modifier was used. The Mg(NO& modifier permitted use of slightly higher char temperatures for Co and Cr, as we previously had found. For Se, the somewhat larger amount of Ni as a matrix modifier seemed preferable to our earlier recommendation. The characteristic amounts found for the metals studied in the paper were: 17and 33 pg/O.O044A-s for As at 193.7 nm and 197.2 nm, respectively;and 253.0, 9.5 and 13.5 for Se, Cr, Co and Ni, respectively. Only in the case of Ni did the characteristic amounts differ significantly from our previous publication [2], probably reflecting an error in the standards that had been used for that paper. Using these optimum conditions we determined all five elements in the five geological samples. Two separate aliquots of the powdered rock were weighed and dissolved on different days. Two separate dilutions were made of each original rock aliquot. Each of these four solutions were analysed five times and the statistical calculations were performed on the 20 separate determinations for each rock and each analyte. Thus the statistical values should reflect the aggregate errors of the full procedure. The recovery experiment used a single dilution of each of the two rock aliquots. To each of these solutions three separate additions of standard solution were made providing six recovery solutions for each rock and each analyte. All determinations were made against the same working curves and the six recoveries for each rock and analyte were averaged to produce the recovery data in Table 2. Our results along with the recommended values and AA results from the work of GLADNEY et41.[7] are given in Table 2. Agreement between our results using the Stabilized Temperature Platform Furnace technique with Zeeman effect background correction and the recommended values is generally very good.
CONCLUSION
The results obtained for the geological samples using the STPF technique with Zeeman background correction are in good agreement with recommended values and have good
491
Atomic absorption spectrometry Table 2. AS, Se, Cr, Co and Ni m geological samples
Sample
Element
GXR-1
As
GXR-2
!se Cr co Ni As
se Cr
co Ni GXR-3
As
se Cr co Ni GXR-5
As
se Cr co
25 320 k 26 22*1a 21*13 46*20 36&24 < 10 38&20 16*14 11*1 26* 14 27*3 40 6500*400 21 f 10 S&l9 40*3 67*23 47*3 40 la*5 -
se
al+36 38kl6 36+1 79&18 96+9 60 3OOiso -
Cr co Ni
7Ok32 24+ 18 34*17
Ni GXR-6
Results by AA from [l] olg/g)
As
Recom. values from [l] (rl3/k%)
Results from this study GueM
Recoveries this study
( 7”)
460+30
454*11
108
18.6 f 0.8 lo+2 9.3 * 1.1 42klO 31*5
17.3 f 0.8 15+1 8.4 f 0.7 45i3 27+2
98 97 87 104 98
0.74f0.11 37+10 9k2
0.74kO.12 37 + 1 9.5 + I.0
95 100 95
18*3
21+2
103
4000*450
3460+90
98
< 0.25 16+1 42+3
104 a7
55 + 5
60*4
109
12k3
13k2
96
1.1*0.1 100*5 30&5
0.95 f 0.20 109*4 31*2
95 101 95
63il
78k3
106
34Oi30
310*30
97
1.07*0.13 96k 10 14*3 22+4
1.09*0.13 9a*3 14 f 1 26i3
89 96 94 100
0.22 f 0.02 19*1 48+5
precision. The technique is simple and reliable. After dissolving the sample in a mixture of acids, aqueous calibration curves are used. The method of additions was unnecessary. Recoveries of analyte added to the samples were about 100 % for each element. We believe that the technique is suitable for the determination of other elements and can help solve many problems occurring during analyses of geochemical materials.