Continuous-flow real-time and flow injection determination of

(26) Pella, E.; Colombo, B. Mlkrochlm. Acta 1973, 697-719. (27) DesMarais, D. J.; Hayes, J. M. Anal. Chem. 1976, 48, 1651-1652. (28) Smiley, W. G. Nuc...
0 downloads 0 Views 517KB Size
Anal, Chem. 1987, 5 9 , 127-130 (26) Peiia. E.; Colombo, B. Mlkrochlm. Acta 1973, 697-719. (27) DesMarais, D. J.; Hayes, J. M. Anal. Chem. 1976, 48, 1651-1652. (28) Smiley, W. 0.Nucl. Sci. Abstr. 1949, 3 , 391-392. (29) Hatch, G.; Slade, R., private communication. (30) Beicher, R.; Ingram, 0.Mlcrochem. J . 1966, 7 7 , 350-357. (31) Gonflantinl, R. Nature (London) 1978, 277, 534-536. (32) Baertschl. P. Earth Plenet. Scl. Left. 1976, 37, 341-344. s. Left. 1974, 7 , (33) Hardcastle, K. G.; Friedman, I. G e o ~ h ~ Res. 165-167. (34) Ferhi, A. M.; Letolle, R. R.; Lerman, J. C. Proc. I n t . Conf. Sfable hot. 2nd 1976, 716-724.

(35) Wedeking, K. W.;

127

Hayes, J. M.; unpublished results.

RECEIVED for review January 21, 1986. Accepted August 8, 1986. This work was supported by a cooperative agreement with the United States Environmental Protection Agency (CR-807322). It has not undergone an internal review by the agency* It is not intended to be, nor should it be misconstrued as, a statement or a reflection of agency policy.

Continuous-Flow Real-Time and Flow Injection Determination of Rainwater pH with a Poly(viny1 chloride)-Based Tubular, Membrane Electrode Brooks C. Madsen* and Donald W. Doller Department of Chemistry, University of Central Florida, Orlando, Florida 32816

A continuous-flow method that utilizes a poiy(vinyi chioride)-based tubular membrane electrode is used to determine the pH of ralnwater in real time. A flow inlectlon method based on the same tubular membrane electrode is applied to measurement of Individual sample pH. Results agree favorably when compared with measurements made with a commercial combination electrode. Prominent features tnclude linear calibration curves from pH 2.7 to 5.0 and a sampling rate of 60 h-’ for the flow injection method.

The measurement of rainwater pH and acidity has been the subject of recent studies that describe many sources of error in the measurements, provide specific recommendations for measurement methodology, and describe the overall reliability of the measured results (1-5). The focus of other selected studies is directed toward the role of both weak and strong acids and the influence that ionized weak acids have on pH measurements and on strong acidity as determined by titration (6-8). Dramatic changes in rainwater pH have been documented for samples collected as part of a t least three national/regional monitoring programs (cf. ref 8 and 9) where pH is first measured in a field laboratory shortly after the end of a specified sampling period and later in a central laboratory. The temporal resolution of pH within individual rainstorms has been measured after collection of small composite samples which represent fixed time intervals ( 1 0 , l l ) or fixed sample amounts (12, 13). A previous study continuously measured rainwater pH below 3.0 in several instances using a prototype flow cell which accommodated a commercial pH electrode; however, the corresponding measured specific conductance values did not substantiate the extemely low pH measurement values (14). Another study (15) used a trap that required a minimum of 0.3 mm of rain to accommodate a combination electrode, but few details about system calibration or response were described. Utilization of a tubular flow-through electrode (16)which includes a pH-sensing membrane (17,18) when incorporated into a continuous flow (CF) or flow injection analysis (FIA) system can provide a viable approach to near-real-time rainwater pH measurement with minimal sample requirements. 0003-2700/87/0359-0 127$01.50/0

Critical parameters that affect the measurement of pH by flow injection analysis using tubular poly(viny1chloride) (PVC) flow cells have been evaluated (18),and the application to measurement of soil extract pH has been demonstrated. A single reagent line system into which sample was introduced was used to evaluate the response of several membrane materials over the pH range 2-12 in that study. The electroactive materials evaluated produce calibration curves that showed marked nonlinearity below pH 4 except for tris(2-ethylhexyl)amine, which was reported to yield nearly ideal Nernstian response over the pH range 3.0-8.0. Buffers and samples were prepared in a fashion that maintained an ionic strength of at least 0.01. It was concluded that the dispersion coefficient should approach 1.0 and buffer capacity of the injected sample must exceed that of the carrier to minimize error. Soils generally exhibit some buffer capacity, and pH is measured on an appropriate ionic extract which allows control of ionic strength during sample preparation. By contrast, rainwater pH is typically determined without sample pretreatment and rainwater buffer capacity is extremely small. The latter property represents a fundamental difference which makes it difficult to use a buffered carrier solution that will stabilize the base line and also allow buffer capacities of individual rainwater samples to exceed that of the carrier. The general approach that was described for soil extract pH determination (18) has been adapted with appropriate modifications to the measurement of rainwater pH using CF and FIA. Results compare favorably with results obtained from static pH measurements made with a conventional glass/ reference electrode combination. The CF and FIA methods provide rapid and accurate approaches with the potential for automation and allow rainwater pH to be determined within minutes of sample collection.

EXPERIMENTAL SECTION The FIA manifold that was used is illustrated in Figure 1A. A Technicon proportioning pump was used and appropriate flow established using short sections of Tygon tubing. Variable-speed studies were performed by using a Buchler 2-6100 polystaltic pump. Samples were injected manually for FIA using an Altex rotary injection valve with 0.62-mL loop. The manifold was constructed with 0.5-mm4.d. Tygon tubing, and the flow cell was constructed in a fashion that retained the features of a previously 0 1986 American Chemical Society

128

ANALYTICAL CHEMISTRY, VOL. 59, NO. 1, JANUARY 1987

1 ;:I 8 W

m z n

Rain Standard

7'

R

to 8.0 using constant ionic strength buffers (19). The deionized water carrier solutionwas obtained from a central deionizing system, which consistently provides water of greater than 1 MQspecific resistance. Further purification of the water with a Culligan SR system produced water of greater than 7 MQ specific resistance, which was used for sample and standard preparation. The 1.4 M NaCl and KC1 ionic strength adjustment solutions were prepared from reagent grade chemicals. The rainwater samples used to evaluate the performance of the FIA system were collected on the rooftop of the University of Central Florida (UCF) Chemistry Building as previously described (20). Samples were collected during November 1985 through June 1986. The CF determination of rainwater pH was performed with a system illustrated in Figure 1B. The sampling system (Figure IC) used for the CF determination of rainwater pH during individual storms included a polypropylene funnel mounted 2 m above roof level which was deployed immediatelybefore the onset of rain.

RESULTS AND DISCUSSION F l o w I n j e c t i o n and Continuous-Flow System Variables. Utilization of the FIA technique was accomplished after

C FM t

T -T4

1 ~

locm

?

b

Flgure 1. Flow injection manifold (A), continuousflow manifold (B), and continuous-flow sampling system (C) for real-time pH determination: C, Tygon mixing coil, 0.5 mm i.d. X 0.5 m; CFM, to CF manifold; CI-, 1.4 M NaCI; DB, &bubbler; F, 540 cm2 funnel; FC, to fraction co8ector; G, electrical ground M, pH meter; R, reference electrode, Orion 90-01 or SCE; T I , PTFE tubing, 0.5 mm i.d.; T2, PTFE tubing, 1.5 mm Ld.; T3, tygon tubing, 3 mm i.d. X 2 m; T4, PTFE tubing, 3 mm i.d. X 9.5 m; TME, tubular membrane electrode; V I , rotary injection valve; V2, rotary four-way valve; W, waste.

designed cell (18). The mixing coil contributes about 0.2-mL internal volume between the mixing tee and the flow cell and caused a delay of less than 10 s between sample injection and FIA peak maximum. Cell potentials were measured with a Corning 130 pH meter and were recorded on strip chart recorders. The tubular flow-through membrane electrode (TME) was constructed from 0.5-mm4.d. Tygon tubing as previously described (16, 18). The membrane was prepared to contain (w/w) 1% tris(2-ethylhexy1)amine(Fluka), 65% bis(2-ethylhexyl)sebacate (Aldrich), 0.6% sodium tetraphenylborate (Aldrich), and 33% poly(viny1 chloride) Aldrich. These reagents were used without additional purification. The internal filling solution was prepared to contain pH 5 buffer and 0.1 M chloride. The flow-through calibration was performed by using dilute solutions of nitric acid and sulfuric acid which were prepared by appropriate dilution of previously standardized 1 N nitric acid and sulfuricacid stock solutions. The pH values for these solutions were calculated from the known acid concentrationsand assumed activity coefficient of 1.00. Mean activity coefficients will vary N H2S04. The from 0.97 to 0.92 for 1.0 X lo-* N to 1.0 X calculated pH values were verified by static measurement ( 4 , 5 ) of the pH using a Corning No. 223447 combination electrode and Coming 130 pH meter. Calibration for the static measurements was achieved with commercially availablepH 4.01 and 7.00 buffers traceable to National Bureau of Standards (NJ3S)standards. The TME response in the CF mode was also evaluated from pH 2.0

optimization of injector sample loop size, carrier solution ionic strength, carrier solution flow rate, and manifold/flow cell inner tubing diameter. Variation of injector loop size from 0.12 to 1.2 mL resulted in an increased dispersion of the injected sample plug. At volumes greater than 0.3 mL dispersion of less than 1.2 was observed. At volumes greater than 1.0 mL, a marked flattening of the peak profile was observed. Calibration curve linearity was adversely affected at higher pH values by smaller loop volumes. At loop volumes greater than 0.6 mL the FIA calibration curve was linear from pH 2.7 to 5.3 based upon measurements on dilute HN03 and H2S04 solutions. A typical calibration curve can be described as mV = (344.0 f 1.7) - (58.1 f 0.4)pH, with a 1.5 standard error of estimate anc = 0.9996 for n = 17. Standard deviations observed for injection of pH 4.0 and 5.0 standards (n = 11) over 2-h time intervals when alternated with sample injections were 0.9 mV and 0.5 mV, respectively. Calibration curve slopes from different days varied from 55 mV to 60.5 mV with mean and standard deviation of 57.7 f 1.5 (n = 15). Use of the FIA manifold in the CF mode where standard acids were substituted for the water carrier stream resulted in a calibration curve mV = (340.3 f 1.4) - (56.2 f 0.4)pH, with a 0.9 standard error of estimate and r = 0.9995 for n = 15. Nernstian behavior of the electrode is observed from pH 3 to 7 when constant ionic strength buffers are introduced in the CF mode. Extremely noisy base lines were observed when water was used as the carrier stream. Base-line stability was improved by addition of either NaCl or KC1 prior to passage through the TME. Approximately 12 cm of tubing was required to connect the TME to the reservoir that housed the reference electrode. It is believed that the low conductivity of solution between the TME and reference electrode causes an effect similar to the appearance of a streaming potential associated with the reference electrode, which contributes to response unstability. A decrease in tubing length between the TME and reference electrode reservoir resulted in a minimal improvement in base-line stability. Base-line noise of about 2 mV caused principally by pump pulsation is observed when 0.1 M C1- is present in the carrier system. A decrease in carrier flow rate or increase in the diameter of the TME and connecting tubing results in increased dispersion and peak broadening. Pump pulsation effects are noticeably reduced when increased diameter tubing is utilized. Sufficient mixing of the injected sample with C1- solution was ensured by use of a 0.5-m mixing coil positioned immediately before the TME. Decrease of mixing coil length causes minimal influence on peak shape. The system described in

ANALYTICAL CHEMISTRY, VOL. 59, NO. 1, JANUARY 1987

129

Table I. Determination of Rainwater pH PH

sample

storage temp, O C 4

033183

combination Electrode 4.52 f 0.03, n = 25 (37

TME FIA 4.47 f 0.02, n = 6 (5 months)

4.3'

A

1o:oo

930

months)

11:10

Time, a.m.

4.21 i 0.01, n=2

010986

25

4.20 f 0.02, n=7(5

012786

25

EPA 1157

25

4.21 f 0.06, n=2 4.05, n = 1

4.17 f 0.02, n = 6 (5 months) 4.05, n = 1

(4.03 pH) EPA 2467 (3.94 pH)

25

3.96, n = 1

3.96, n = 1

months)

7:20

7:OO

Time. p.m.

Figure 1A allows 60 sample injections/h. When sample loop size is reduced to 0.3 mL sample throughput can be increased to 80f h. The CF manifold (Figure 1B) was constructed for use in the real-time measurement of rainwater pH. Flow rates were reduced in the CF manifold vs. the FIA manifold to allow for processing of reduced quantities of sample. At the 1.4 d / m i n sample flow rate a minimum of 0.03 mm/min rainfall must be continuously collected by the funnel to ensure an uninterrupted supply of sample to the TME. Calibration of the CF system can be accomplished during periods of no rain or when insufficient rainfall occurs. Standard solution selection is accomplished with the four-way valve (Figure 1B). Base-line noise of less than 0.5 mV is observed in the CF manifold. A typical calibration curve based on introduction of pH 2.7 to pH 5.0 dilute HN03 and H2S04standards was mV = (343.2 f 1.4)- (57.0 f 0.4)pH, with a 1.0 standard error of estimate and r = 0.9997 for n = 16. The debubbler used in the CF system was constructed from a disposable 5-mL pipet tip with the sample inlet positioned at the 0.5-mL level with excess sample/air withdrawn at the 1-mL level. A single TME electrode has been used for the majority of reported FIA and CF measurements. The electrode continues to provide a nearly Nernstian slope when calibration in the CF mode is performed. Some peak broadening and less than Nernstein slope has been observed in the FIA mode after 4 months of use and may suggest a finite lifetime of use for the described TME. Several TME's have been prepared over an &month period from a single solution that contains membrane material described previously. Reproducibility of response from electrode to electrode is excellent. pH Determinations by FIA and Conventional Electrode Methods. During late 1985 and early 1986 several rainwater samples were collected. Upon collection pH was determined by the proposed FIA method and the combination electrode method. Sample pH was determined by both methods on selected rainwater samples stored at room temperature and under refrigeration over a 6-month period. Two NBS traceable reference rainwater samples obtained from the U S . EPA were also included. Results are summarized in

5.0

ITw 130

a 00

3 20

Time, p m

Flgure 2. Real-tima pH and individual fraction pH for three storms: (A) SQ58, March 14, 1986; (B) SQ61, June 9, 1986; (C) SQ62, June 15, 1986. DW is deionized water; S1,S2, and S3 are HNO, standards of pH 3.99, 4.14, and 4.65, respectively.

Table I. These results demonstrate the utility of the TME FIA method. Agreement within 0.05 pH units is achieved when these results are compared with results of combination electrode measurements. Precision of measurementsover time from both methods is no larger than *0.06 pH units. A potential source of error can be produced by the high ionic strength of solution that occurs when salt is added. The degree of ionization of any weak acid component that may be present will be influenced in an unpredictable fashion and cause a modest change in pH which depends upon weak acid composition of the sample. The consistency demonstrated for pH measurements between TME results with salt added and combination electrode results without salt suggests that the changed ionic strength caused minimal error for samples used in this study. Real-TimeRainwater pH Measurement. During early 1986 several rainstorms which occurred on the UCF campus were utilized to evaluate the reliability of the CF system. Figure 2 illustrates the pH behavior that was observed as three different storms progressed. Rainwater in excess of the 1.4 mL/min required to operate the CF system was diverted from the debubbler to a fraction collector. Several fractions of rainwater, typically representing 0.5-1.0 mm of rain collected over 1-20-min intervals were collected. Sample pH for each

Table 11. AniodCation Ratio Summaries and Comparison of pH Measurements for Samples Collected during Rainstorms SQ58, SQ61, and 8 6 6 2

mean anion/cationo mean anion/cationb linear regression eq for pH, vs. pH, n std error of estimate

SQ58 3/1/86

SQ6l 6/9/86

SQ62 6/15/86

1.00 i 0.13, n = 9 0.99 f 0.12, n = 9 pHt = (0.05 f 0.20) + (0.985 f 0.043)pHC 13 0.02

1.02 i 0.08, n = 31 0.98 f 0.09, n = 31 pHt = (0.56 f 0.12) + (0.866 i 0.028)pHC 34 0.03

pH, = (-0.16 f 0.12) + (1.031 i 0.027)pHC 21 0.02

Uses pH measured by combination electrode, pH,.

Uses pH measured by TME FIA, pH,

130

ANALYTICAL CHEMISTRY, VOL. 59, NO. 1, JANUARY 1987

fraction was determined by the FIA and Combination electrode methods. Reliability of these pH measurements were verified by calculation of anionlcation (equiv/equiv) and conductivity/calculated conductivity ratios (21) after determination of concentrations for major cations and anions in each sample by flame spectroscopy (20) and various FIA methods (22-24). Rainstorms designated SQ58 and SQ61 provided 9 and 31 fractions, respectively, of sufficient volume to allow measurements of major anion and cation concentrations. The resulting pH measurements and anion/cation ratio summaries from SQ58 and SQ61 and pH measurement summary for SQ62 are presented in Table 11. A paired sample t test suggests no significant difference in measured pH between the FIA and combination electrode methods for SQ58, SQ61, and SQ62 data. The pH data summaries that are presented in Table I1 illustrate the consistency of the TME FIA and combination electrode measurements. The linear regression equations presented in Table I1 for SQ58 and SQ62 further demonstrate the consistency of pH measurementsby the two methods. The 0.866 slope of the regression equation for SQ6l may suggest a bias; however, close inspection of the data reveals four samples where differences in excess of 0.1 pH unit occur. Omission of data for these samples would have increased the slope to approximately 0.94, which is much less suggestive of a bias. The anion/cation ratios calculated for each sample suggest that no serious error has been introduced in the measurement of pH by either method. The general reliability of the CF measurement of rainwater pH is illustrated in Figure 2 where the pH measured with the combination electrode for individual fractions is compared on a time scale consistent with fraction collection as each storm progressed. Each storm provided a unique and somewhat complex pH profile as measured in the CF mode. This complexity was verified from measurement of pH on individual rainwater fractions collected simultaneously with that required for operation of the CF manifold. These fractions are generally representative of the rainfall that occurred over the specified time intervals; however, during periods of very gentle rainfall the fraction collected may represent as little as 10% of total rainfall because the major portion is directed to the CF manifold. Individual fraction pH values measured by the combination electrode method are displayed using the bar chart profiles in Figure 2. Satisfactory agreement between CF pH and fraction pH is observed for SQ58 and SQ6l. Fraction pH values are generally higher than CF pH for SQ62. This observation may be important because pH change after sample collection has previously been documented (8, 9). The pH measurements with the combination electrode and by FIA were performed within 5 h of sample collection for SQ58, however, approximately 20 h lapsed between collection and measurement for both SQ61 and SQ62. Fraction pH was measured within 1h of collection for three samples from SQ62. These measurements 4.58 vs. 4.71, 4.59 vs. 4.62, and 4.46 vs. 4.47 suggest that a change in pH had occurred during roomtemperature storage of some SQ62 fractions. These prelim-

inary observations suggest that the CF and/or FIA methods will be useful in the study of pH changes that occur after sample collection. Temperature compensation has not been incorporated during CF pH measurement; however, this influence should be minimal for these three storms because atmospheric temperatures ranged between 2 1 "C and 27 "C. Discontinuitiesin the CF profiles in Figure 2 occur because insufficient rainfall was collected to allow continuous operation of the CF manifold. The CF profile for SQ62 includes regions where HN03standards were introduced for calibration purposes. The dilute HNOBand H2S04standards with pH below 5.0 used for calibration have been found to be stable for at least 4 months when stored in polypropylene bottles. The FIA method has been demonstrated to be rapid, precise, and accurate for measurement of rainwater pH below 5.3. Above this value calibration is impractical using dilute acid solutions. The influence of dissolved carbon dioxide in samples of pH greater than 5.0 also becomes significant. System startup time of less than 10 min is typical. The CF method is comparable to the FIA method, and both methods hold promise for application in situations that dictate rapid or frequent measurement of rainwater pH. Registry No. Water, 7732-18-5; PVC, 9002-86-2.

LITERATURE CITED (1) Galloway, J. N.; Cosby, B. J., Jr.; Likens, G. E. Limnol. Oceanogr. 1979, 24, 1161. (2) Tyree, S. Y., Jr. Atmos. Environ. 1981, 75,57. (3) McQuaker, N. R.; Kluckner, P. D.; Sandberg, D. K. Envlron. Sci. Technol. 1983, 17, 431. (4) Koch, W. F.; Marinenko, G. ASTMSpec. Tech. Publ. 1983, No. 823, 10. (5) Slsterson, D. L.; Warfel, B. E. Int. J. Envlron. Anal. Chem. 1983, 18, 143. (6) Keene, W. C.; Galloway, J. N. Atmos. Environ. 1984, 18, 2491. (7) Lindberg, S. E.; Coe, J. M. Tellus, Ser. B 1984, 368, 166. (6) Keene, W. C.; Galloway, J. N. Atmos. Environ. 1985, 79, 199. (9) Hendrlckson, E. R.; Kosky, K. F.; Schert, J. D. "Florida AcidDeposifion Stody-Monirwing Program Phase I Monitoring Report". Environmental Science and Engineering Contract Report, 1982. 10) Raynor, G. S.; Hayes, J. V. Water, A k , Soil Pollut. 1981, 75,229. 11) DePena. R. G.; Carlson, T. N.; Takacs, J. F.; Holian, J. 0. Atmos. Environ. 1984, 78, 2665. 12) Pellett, G. L.; Bustin, R.; Harris, R. C. Water, Air, Soil Pollut. 1984, 21, 33. 13) Seymour, M. D.; Stout, T. Atmos. Envlron. 1983, 17, 1463. 14) Reddy,.M. M.; Libermann, T. D.; Jelinski, J. C.: Caine, N. Arc. A/p. Res. 1985, 77, 79. (15) Kolton-Shapka, R.; Lakrltz, Y.; Lurla, M. Atmos. Environ. 1984, 18, 1245. (16) Meyerhoff, M. E.; Kovach, P. M. J. Chem. Educ. 1983, 60,766. (17) Schulthess, P.; Shljo, Y.; Pham, H. V.; Pretsch, E.; Amman, D.; Simon, W. Anal. Chim. Acta 1981, 737,111. (18) Hongbo, C.; Hansen, E. H.; Ruzicka, J. Anal. Chim. Acta 1985, 169, 209. (19) Perrin. D. P.; Dempsey, B. Buffers for pH and Metal Ion Control; Chapman and Hall: London, 1974. (20) Madsen, 0. C. Atmos. Environ. 1981, 15, 653. (21) Galloway, J. N.; Likens, G. E. Tellus 1978, 30, 71. (22) Madsen, B. C. Anal. Chlm. Acta 1981, 124, 437. (23) Madsen. B. C.; Murphy, R. J. Anal. Chem. 1981, 53, 1924. (24) Slanina, J.; Bakkar, F.; Bruyn-Hes, A,; Mols, J. J. Anal. Chlm. Acta 1980, 113,331.

RECEIVED for review July 7, 1986. Accepted September 8, 1986.