Environ. Sci. Technol. 2008, 42, 6092–6099
Evaluation of Indicator-Based pH Measurements for Freshwater over a Wide Range of Buffer Intensities SHIGUI YUAN AND MICHAEL D. DEGRANDPRE* Department of Chemistry, The University of Montana, 32 Campus Dr., Missoula, Montana 59812
Received March 24, 2008. Revised manuscript received May 19, 2008. Accepted May 26, 2008.
Two different sulfonephthalein indicators, cresol red (CR) with a pKa of ∼8.3 and bromothymol blue (BTB) with pKa of ∼7.4, were tested for an analysis of freshwater over a broad range of pH and total alkalinity values. Measurements from an autonomous sensor system using a 1 cm optical path length were compared to those using a 10 cm path length on a benchtop spectrophotometer. The indicator pH perturbation was quantified with a thermodynamic model and nonlinear leastsquares analysis. The laboratory study found that the perturbationcorrected pH differed between the 1 cm (large indicator perturbation) and 10 cm (small indicator perturbation) optical path length measurements from -0.017 to +0.15 with a median of +0.0041 pH units for CR and from -0.015 to +0.026 with a median of -0.0008 pH units for BTB. Precision was (0.0005-0.013 and (0.0001-0.0027 pH units for the 1 and 10-cm-pathlength measurements, respectively. The autonomous sensor was deployed for 14 days in a local creek. Simultaneous glass pH electrode measurements had a large negative and drifting offset (-0.15 to -0.40 pH units) compared to the indicatorbased measurements. This study is the first in situ comparison between potentiometric and spectrophotometric pH methods in a freshwater system.
Introduction Many contemporary freshwater research and monitoring problems require improved pH measurement technology. Glass pH electrodes, the most common tool for pH analysis, have a number of well-known problems, particularly for freshwater applications (1–4). Liquid junction potential differences between standards and low ionic strength freshwater samples can lead to 0.1-1.0 pH unit errors (5–7). Because pH data are commonly used for geochemical modeling, small errors in pH lead to large errors in geochemical concentrations, especially for predicting inorganic carbon speciation. For example, the error in the calculated partial pressure of CO2 (pCO2) is >25% for a 0.1 pH unit error, potentially leading to incorrect conclusions regarding whether a natural water is a source or sink of atmospheric CO2. Furthermore, reproducible pH measurements are needed to resolve pH trends in aquatic ecosystems as human impacts such as climate change and CO2 acidification become increasingly significant. Indicator-based pH measurements are an appealing alternative to potentiometric pH measurements. Marine * Corresponding umontana.edu. 6092
9
author
e-mail:
michael.degrandpre@
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 16, 2008
scientists have used pH indicators since the 1980s to obtain reproducible long-term seawater pH data (8–10). Seawater pH accuracy and precision of (0.002 and (0.0005 pH units, respectively, have been achieved with shipboard spectrophotometric systems (9, 11). Spectrophotometric pH measurements have only recently experienced renewed interest for freshwater applications (12–14). Several autonomous indicator-based pH systems have been developed, primarily for seawater applications (13, 15–17). Our indicator-based pH sensor, the Submersible Autonomous Moored Instrument for pH or SAMI-pH, was developed for freshwater (12, 13) and seawater (17) applications. An accuracy, relative to laboratory spectrophotometric measurements, of -0.003 ( 0.004 pH units was achieved by SAMI-pH during deployment in a buffered alkaline river (13). Liu et al. (16) deployed their spectrophotometric pH instrument in a river for 2 days, but the accuracy and precision were not reported. While laboratory and field studies have demonstrated the utility of the technique for freshwater analysis, the performance over a broad range of freshwater conditions (e.g., buffer intensity) has not been characterized. Our goal for this study is to explore the use of indicators over the range of pH (6–9) and total alkalinity (AT) (100-1500 µmol kg-1) typically found in freshwater and to further test the methodology in the SAMIpH sensor previously developed and optimized in our laboratory. One concern regarding indicator-based pH measurements, as noted by others (1, 12–14, 18), is that the addition of a weak acid indicator can significantly change the pH of weakly buffered water. The magnitude of the pH perturbation is determined by several factors, such as the indicator pKa, pH and concentration of the indicator solution, and the sample pH and buffer intensity (i.e., the concentration of acid or base required to change the sample pH by one unit; the units are mol kg-1 pH-1). In seawater studies, 10-cmpath-length cuvets are used to minimize the amount of indicator added (9, 10, 19). However, for some applications, a 10 cm path length may be impractical, for example, in hand-held field monitors or low-power autonomous instruments. Conventional long path length cells require a more complex design, with collimating optics that are alignmentsensitive, and low-light throughput. The low-light throughput requires a high-power light source, a serious limitation for long-term autonomous in situ measurements. The large size and volume of 10 cm cuvets are also disadvantages. One alternative, the long path length liquid core waveguide, has low-light throughput and has been found to interact with the indicator (16). Short path length optical flow cells and cuvets avoid many of these limitations. However, because the perturbation may be large for short path length cells, a simple and reliable method for minimizing and correcting for the pH perturbation is needed. In SAMI-pH, which uses a 1-cm-path-length cell, a simple methodology to correct for the pH perturbation was previously suggested (13) and subsequently used for seawater analysis (17). In this study, we propose an improved indicator perturbation correction technique for weakly buffered freshwater and compare the performance of 1 cm (large perturbation) and 10 cm (small perturbation) path length measurements.
Theory Most spectrophotometric pH methods use the diprotic sulfonephthalein indicators such as bromocresol green (20), phenol red (8, 16), or cresol red (CR) (9, 13). The method relies upon the second dissociation of the sulfonephthalein indicator: 10.1021/es800829x CCC: $40.75
2008 American Chemical Society
Published on Web 07/03/2008
Ka
HL- 798 H+ + L2-
(1)
where HL- is the monoprotic form and L2- is the fully deprotonated form. In this study, CR with a pKa of ∼8.3 and bromothymol blue (BTB) with a pKa of ∼7.4 were used. The freshwater pH is calculated from (14):
(
pH ) -log[H+] ) pKa ′ + log
R - e1 e2 - e3R
)
pKa′ ) pKa° - [0.5092 + (T - 298.15) × 0.00085] × 4 ×
(
õ 1 + õ
(2)
)
- 0.3 × µ (3)
where pH is on the free hydrogen ion scale, R is the absorbance ratio A2/A1, A1 and A2 are the absorbances at corresponding peak wavelengths of HL-(λ1) and L2-(λ2), respectively, pKa° is the infinite dilution dissociation constant, and pKa′ is the apparent dissociation constant calculated using the Davies equation at ionic strength (µ) and temperature (T in K). The e’s are ratios of the molar absorptivities (ε) of HL- and L2- at λ1 and λ2: e1 )
εHL2 εL2 εL1 , e ) , e ) εHL1 2 εHL1 3 εHL1
(4)
Note that all activity coefficients, including that for H+, are included in pKa′ because pH is on the free hydrogen ion scale (14, 16). pKa′ accounts for the ionic conditions of the water, and ionic strength must be known to accurately calculate pKa′. The error is largest at low ionic strength due to the nonlinearity of eq 3. Breaking the eq 3 relationship into two linear regions, the sensitivity to ionic strength ∆pKa′/ ∆µ is 0.041/mmol kg-1 and 0.015/mmol kg-1, between µ ) 0.2-1.0 mmol kg-1 and 1.0-5.0 mmol kg-1, respectively. The latter range is more characteristic of freshwater; therefore, a 1.0 mmol kg-1 uncertainty in µ would result in a 0.015 uncertainty in pKa′.
Methods Test Solution Preparation. Synthetic freshwater test solutions were prepared from high-purity Na2CO3 (99.997%,
#10861, Alfa Aesar) with AT adjusted by standardized HCl (0.5165 N, LC15290-4, Fisher Scientific Co.), resulting in solutions comprised of the ionic species H+, OH-, HCO3-, CO32-, Na+, and Cl-. CO2 exchange was carefully avoided to reduce the uncertainties of ionic composition and pH of sample solutions, and to keep the sample pH constant during the measurements. Samples were between pH 7 and 9 for the CR tests and between 6 and 8 for the BTB tests, bracketing their respective pKa’s. The AT of the solutions was between 100 and 1700 µmol kg-1, ionic strengths between 2 × 10-4 and 6 × 10-3 mol kg-1, and buffer intensities between 1 × 10-5 and 2 × 10-3 mol kg-1 pH-1. Although the solutions were carefully prepared, it proved difficult to minimize the uncertainty in the HCl addition while also eliminating CO2 exchange, which subsequently led to relatively large uncertainties in the final pH. For example, for a sample with an AT of 618 µmol kg-1 and a total dissolved inorganic carbon CT of 608 µmol kg-1 (pH ) 8.6091 at 20 °C), a 0.004 g HCl uncertainty causes a 0.01 pH unit error. As a consequence, the solution pH could not be accurately calculated and used as a primary standard for pH. See pH intercomparison below. Indicator Characterization. The molar absorptivities of HL- were determined at a pH of ∼4 for CR (114080-25G, Sigma-Aldrich, Inc.) and ∼3 for BTB (114421-25G, SigmaAldrich, Inc.). A pH of ∼12 was used to determine ε’s for L2for both indicators. Roughly 1:1000 H2L/HL- is present for the low-pH measurements, which leads to a pH error < 0.0010 pH units. The ε’s were measured at four temperatures, ∼10, 15, 20, and 25 °C. Due to instrument bandpass differences, ε’s were determined both on the SAMI-pH and on the benchtop spectrophotometer (Cary 300, Varian, Inc.). Our previous work has shown that, by measuring ε’s for the SAMIpH and benchtop spectrophotometer, the two system pH’s closely match for highly buffered phosphate solutions (
