does increase the sensitivity of the collection step, it does so a t the expense of selectivity. Equation 9 predicts a 14-mV anodic stripping peak potential shift as the scan rate is increased from l to 3 V/min. A 13-mV anodic shift was observed experimentally. The stripping electrode potential a t which the collection peak is observed shifts 29 mV anodically over the same scan rate range, resulting in the increase in hE with scan rate indicated in Figure 6. Collection Peak Current vs. Concentration. The collection peak current was linear in the nanomolar and subnanomolar range with deposition times ranging from 6 to 12 min as indicated in Figure 7. (The sharply rising charging current baseline associated with the stripping step precluded setting the recorder at a sufficiently high sensitivity to make accurate stripping peak current measurements feasible for subnanomolar lead concentrations.) By least squares analysis of a standard addition procedure, the concentration of lead in the 0.1 M HCl blank was found to be 3.9 X M with a standard deviation of 0.3 X M. Linearity out to a t least M lead was also observed.
of the induced current and to construct the appropriate apparatus to minimize it.
CONCLUSIONS
RECEIVEDfor review August 2, 1976. Accepted October 4, 1976. This research was supported in part by a grant from the Environmental Protection Agency (No. R-804179-01-O),and in part by a grant from the Office of Water Research and Technology (No. A-053-WIS).
These preliminary experiments indicate the improved sensitivity and precision obtained with ASVWC a t tubular electrodes as compared to conventional ASV. To fully realize these advantages, work is in progress to investigate the nature
LITERATURE CITED (1) T. R . Copeiand and R . K. Skogerboe. Anal. Chem., 46, 1257A (1974). (2) D. C. Johnson and R. E. Allen, Talanta, 20,305 (1973). (3) D. Laser and M. Ariel, J. Electroanal. Chem. lnterfacial€lectrochem., 49, 123 (1974). (4) W. R. Seitz, R. Jones, L. N. Klatt, and W. D. Mason, Anal. Chem., 45, 840 (1973). (5) S. H.Lieberman and A. Zirino, Anal. Chem., 46, 20 (1974). (6) D. K, Roe and J. E. A. Toni, Anal. Chem., 37, 1503 (1965). (7) M. Stulikova, J. Electroanal. Chem. lnterfacial Electrochem., 48, 33 (1973). (8) G. E. Batley and T. M. Florence, J. Electroanal. Chem. lnterfacialElecfrochem., 5 5 , 23 (1974). (9) W. J. Blaedel and L. N. Klatt, AnaLChem., 36, 879 (1966). (IO) W. J. Albery and M. L. Hitchman, "Ring-Disc Electrodes", Oxford University Press, London, 1971. (11) W. J. Blaedel and G. W. Schieffer, Anal. Chem., 46, 1564 (1974). (12) W. J. Blaedel and G. W. Schieffer, J. Electroanal. Chem. lnterfacialElectrochem., submitted April, 1976. (13) T. M. Florence, J. Electroanal. Chem. lnterfacial Electrochem., 27, 273 (1970). (14) W. J. Biaedel and R. A. Jenkins, Anal. Chem., 47, 1337 (1975). (15) W. T. de Vries and E. Van Dalen, J. Electroanal. Chem. lnterfacial Electrochem., 14, 315 (1967).
Determination of Total Fluoride in Soil and Vegetation Using an Alkali Fusion-Selective Ion Electrode Technique Neil R. McQuaker' and Mary Gurney Environmental Laboratory, Water Resources Service, Department of Environment, 3650 Westbrook Crescent, Vancouver, B.C. V 6 S 2L2, Canada
A rapid NaOH fusion-selective ion electrode (SIE) technique for the determination of total fluoride In both soil and vegetation is described. Prior to the SIE determination, the high alkaline content of the fused sample is reduced by the addition of HCI. When a pH of 8-9 is reached, the sample is filtered and subsequently buffered to pH 5.2. It is found that when the sample is filtered under slightly alkaline conditions that interfering cations, such as aluminum and iron, are removed as insoluble oxides. This makes the time consuming distillation step recommended in the classical Willard and Winter procedure unnecessary. In the case of vegetation samples, a precision (as a mean relative standard deviation) of 4.3% is obtalned In the range 50-600 ppm F. A corresponding value of 4.1 % Is obtained for soil samples in the range 30-700 ppm F. When authentic samples are spiked with NaF quantitative recoveries are obtained. If a 0.5-g sample is used, the detection limit Is 3 ppm F.
Until recently the total fluoride analysis of many substances has relied upon the classical Willard-Winter procedure ( I ) . This well known procedure depends upon (i) the release of fluoride from the sample by a suitable degradation procedure such as alkali fusion, (ii) separation from interfering ions by distillation from perchloric acid, and (iii) determination of the
fluoride by a suitable analytical step. The reliability of the Willard-Winter procedure is well established (2). I t is, however, very time consuming. Recently, more rapid methods, which omit the distillation step and use a selective ion electrode (SIE) determination, preceded by either an alkali fusion ( 3 , 4 )or an acid/alkali leach (4-6) procedure, have been proposed. These methods have been restricted to vegetation samples where cations, such as aluminum and iron, are not usually present in sufficient concentration to interfere with the SIE method. It is to be noted that even though these methods have been found to be generally acceptable for vegetation samples, it is possible that some of the bound fluorides may not be recovered in the case of the leach procedures (6). This contrasts with the alkali fusion techniques which should allow for the complete release of bound fluorides. In present work we have explored the feasibility of extending a modification of the rapid NaOH fusion-SIE method proposed by Baker ( 3 ) to include soil as well as vegetation samples. Baker, prior to the SIE determination, filtered his samples under strongly alkaline conditions. However, preliminary work done in this laboratory indicated that both aluminum hydroxides and silicates are appreciably soluble under these conditions (7). Therefore, unlike the insoluble iron oxides, they are not retained by the filter paper but rather they enter the filtrate where they are subsequently able to complex the fluoride when the sample is buffered to p H 5.2. ANALYTICAL CHEMISTRY, VOL. 49, NO. 1, JANUARY 1977
53
Table I. Concentrations of Selected Elements in the Soil Samples Concentration, mg/g Sample No.
s1
s2 s3 s4 s5 S6
A1 6.8 8.6 32.6 9.4 9.7 8.2
Fe 12.3 63.3 344.0 19.8 32.4 78.6
Ca 130.0 18.4 99.0 4.8 0.9 5.4
Mg
R E S U L T S AND DISCUSSION
40.0 3.7 9.7 7.0 5.5 10.6
Calibration. The procedure just outlined ensures that both samples and standards will not only have the same blank value but also be 0.5 N in NaC1. The importance of equivalent ionic backgrounds for both samples and standards becomes apparent when we consider the relationship governing SIE measurements. If we assume that the volume change during the measurements is negligible, then measurements using the SIE are expected to conform to
In this work we have attempted to render both the silicates and amphoteric aluminum hydroxides insoluble by reducing the high alkaline content of the sample prior to filtration. This was done by adding concd HCl until a pH of 8-9 was reached. The sample was then filtered and subsequently buffered to pH 5.2. It was hoped that by filtering the sample under these slightly alkaline conditions that interfering cations such as aluminum, iron, calcium, magnesium, and silicon would be preferentially removed as insoluble oxides, thus making the time consuming Willard-Winter distillation step unnecessary for both vegetation and soil samples. EXPERIMENTAL Apparatus. The potential measurements were made using a Fisher Accumet Model 520 Digital pH/ion Meter equipped with an Orion Model 94-09A fluoride electrode and a calomel reference electrode. Calibration was achieved, in the range 0-55 pg F, by adding known aliquots of NaF standards (see below) to a representative blank solution. Reagents. The following reagents and solutions were prepared according to the following instructions: Stock Fluoride Solution (1000 p p m F ) . Dissolve 2.210 g pre-dried reagent grade NaF in distilled water and dilute to 1 1. Total Ionic Strength Adjustment Buffer (TZSAB). To 300 ml distilled water add 58 ml glacial acetic acid and 12 g sodium citrate dihydrate. Stir to dissolve and then adjust the pH to 5.2 using 6 N NaOH. Cool and dilute to 1 1. Sodium Hydroxide Solution. Dissolve 670 g NaOH pellets in distilled water and dilute to 11. Store in a polyethylene container. Procedure. The samples as received by the laboratory were dried at 105 O C and then homogenized; the soil and vegetation samples were ground so as to pass 100 and 40 mesh sieves, respectively. Approximately 0.5 g of the prepared sample was then weighed accurately and transferred to a 130-ml nickel crucible. Next, the sample was moistened slightly with distilled water. This was followed by the addition of 6.0 ml of the sodium hydroxide solution. The crucible was then tapped slightly so as to uniformly disperse the sample in the sodium hydroxide solution. Once this had been achieved, the sample was placed in an oven set to 150 "C for 1 h and then removed. After the sodium hydroxide had solidified, the crucible was placed in a muffle furnace set to 300 "C. The temperature was then raised to 600 "C and the sample was fused at this temperature for 30 min. After the sample had been removed from the muffle furnace and allowed to cool, 10 ml of distilled water was added to the sample. The sample was then heated slightly so as to facilitate the dissolution of the sodium hydroxide fusion cake. Next, about 8 ml concd HCl was slowly added, with stirring, so as to adjust the pH to 8-9. During this process Fisher Alkacid Test Ribbon was used to monitor the adjustment of the pH, and care was taken to avoid producing either neutral or acidic conditions in the sample. Once the acid'fied sample had cooled, it was transferred to a 100-mlvolumetric fl,+*;k, diluted to volume, and then filtered through dry Whatman No. 40 filter paper. It is to be noted that a representative blank was prepared by carrying 3 blanks through the above procedure and then combining them. Prior to making the potential measurements, the pH of the prepared blank and the samples was adju ,ted to 5.2 by adding 25.0 ml filtrate and 25.0 ml TISAB to a 150-ml polyethylene beaker. Data for the standard curves were obtained by using a 0.100-ml Eppendorf pipet and 10 and 100 ppm F standard solutions to add 5 successive 1-pg F aliquots followed by 5 successive 10-pg F aliquots to the pre54
pared blank solution. The cumulative amounts of fluoride added were thus in the range 0-55 pg F and the resulting data allowed for calibration over this range (see Results and Discussion). For all potential measurements, the stabilization time was 5 min and the temperature was 22 & 1 "C.
* ANALYTICAL CHEMISTRY, VOL. 49, NO. l , JANUARY 1977
Ei = -S log Ci
+E
(1)
where Ci is the pug F yielding a millivolt potential of Ei; S is the Nernst slope (RTIF)which equals 58.5 mV a t 22 "C,and B is a millivolt potential which is dependent on the ionic background. If the ionic background is constant, E is constant and a t constant temperature Equation 1 yields a linear relationship which is defined by the constants S and B. This means that a t 22 "C, Equation 1can be rearranged to yield
c; =
AC
where AC = C,+] - C, and AE = E, - E,+, w i t h j > i. Equation 2 will allow us to use the data from the standards to compute the blank value in pg F. We can then use Equation 1to construct a linear curve where C, = yg FSTD pg FBLK. Since the ionic backgrounds for the standards and samples are equivalent, we can reference our samples against the calibration curve and then obtain the required result by subtracting the previously computed blank value. Typical blank values, obtained during the course of this work, were of the order 1.5 pg F. We have called the foregoing calibration procedure the single solution technique. When compared with the usual procedure of preparing a series of standard solutions (8),it has the following advantages: (i) it provides for the accurate determination of the blank, (ii) it provides for a linear curve even a t low levels as a result of the blank correction, and (iii) it provides for enhanced sensitivity and accuracy as a result of (i) and (ii). The calibration curves in addition to being extremely linear were also very stable; correlation factors were invariably better than 0.99998 and the fluctuations in the constants S and B , from curve to curve, were minimal. The respective means and their standard deviations, for 7 curves, were -58.4 f 0.3 and 193.2 f 0.5 mV. Range. The standard deviation associated with 12 typical replicate blanks (treated as samples) was 0.2 pg F and, using the criterion of twice the standard deviation, a detection limit of 0.4 pg F is indicated. In terms of ppm fluoride in the original 0.5-g sample, this translates into a detection limit of 3 ppm F. This value may be compared with a value of 7.5 ppm F indicated in Jacobson and Heller's work (6) where a 0.5-g sample and an acid/alkali leach procedure is used. If we use our high standard of 55 pg F, the range becomes 3-440 ppm F. Results from this work, however, indicate that the curve is linear up to a t least 550 pg F and this extends the upper limit of the range to in excess of 4000 ppm F. I t is to be emphasized that if sample dilution is required, the representative blank solution used to produce the standards must also be similarly diluted. Precision a n d Accuracy. Interferences will affect both the precision and accuracy of the results. It is well known that some cations are capable of complexing fluoride and so de-
+
Table 11. Concentrations of Selected Elements in the Soil Sample Filtrates Concentration, ppm Sample No.
A1
Fe
Ca
Mg
Si
s1
