Underway pH, fCO₂, and TCO₂ Measurements on P16N_2006

The USF group (Dr. Robert Byrne, PI) was responsible for underway surface pH, fCO₂ and TCO₂ measurements during the CLIVAR/GO-SHIP Repeat Hydrography Section P16S_2005 using the automated Multi-Parameter Inorganic Carbon Analyzer (MICA) system. Drs. Xuewu Sherwood Liu and Renate Bernstein operated the system during Leg 1 and Drs. Robert Byrne, Zhaohui 'Aleck' Wang, and Johan Schijf and Mr. Ryan Bell operated the system during Leg 2 of the cruise.

Technical details and performance evaluation of the MICA system can be found in Wang et al. (2007). The system consists of three seawater channels (surface seawater fCO₂, TCO₂, and pH). All measurements (three channels) were made at constant temperature (25°C) and were based on similar spectrophotometric principles. The system operates autonomously with a sampling frequency of ~7/hour. For each sample, all three parameters are measured and recorded simultaneously.

Spectrophotometric pH measurements were based on the method described in Clayton and Byrne (1993), but used thymol blue as the pH indicator (Zhang and Byrne 1996; Wang et al. 2007). Thymol blue was directly injected into a stream of underway seawater and absorbance was monitored spectrophotometrically. Sample pH is linked to the absorbance ratio (R), dissociation constant (K_I), and molar absorbance ratios (e₁, e₂, and e₃) of the indicator with the following equation:

pH_T = -log_TK_I + log[(R - e₁)/(e₂ - Re₃)] (1)

where the subscript T denotes parameters expressed on the total hydrogen ion scale. The calibrated constants of thymol blue are given by Zhang and Byrne (1996) as:

-log_TK_I = 4.706S/T + 26.3300 - 7.17218 logT - 0.01736S (2)

e₁ = -0.00132 + 1.6*10^-5T, (3)

e₂ = 7.2326 - 0.0299717T + 4.6*10^-5T², (4) and

e₃ = 0.0223 + 0.0003917T (5)

where T is absolute temperature in K, and S is salinity.

For seawater fCO₂ measurements, Teflon AF 2400 (DuPont) is used as both a CO₂ permeable membrane and a liquid-core waveguide (LCW) (Wang et al. 2007). Phenol red is used as the indicator (Yao and Byrne 2001). During each CO₂ measurement, the indicator solution, composed of Na₂CO₃ with constant total alkalinity (TALK), is motionless inside the LCW. The seawater samples were directed to flow outside the LCW. After CO₂ molecules equilibrate by diffusion with the LCW's internal solution, the equilibrium pH was determined by measurements of absorbance ratios. fCO₂ was then derived from the equilibrium pH with the following equation:

fCO₂ = TA/(2K_oK^'₁K^'₂[H⁺]^-2 + K_oK^'₁[H⁺]^-1) = a/L + b (6)

where K_o is the Henry's Law constant, K^'₁ and K^'₂ are carbonic acid dissociation constants, and L = 2K_oK^'₁K^'₂[H⁺]^-2+ K_oK^'₁[H⁺]^-1. Parameters "a" and "b" are derived through a calibration procedure using CO₂ gases at known concentrations. The pH of the indicator solution is determined spectrophotometrically, and the L term in Eq. 6 is then calculated. Since the TALK of the internal solution is constant, sample fCO₂ has a linear dependence on 1/L with a slope of "a" and an intercept of "b." The calibration constants, "a" and "b," account for all uncertainties in Eq. 6 including the absolute value of TALK.

Spectrophotometric measurements of TCO₂ using Teflon 2400 AF LCWs have been described by Byrne et al. (2002). Water samples are acidified before measurements (pH ~ 2.7), whereupon the total CO^*₂ concentration equals the TCO₂ concentration. After the internal Na₂CO₃ -indicator (bromcresol purple) solution attains CO₂ equilibrium with the acidified water samples across the LCW CO2-permeable walls, the TCO2 of the outer solution can be written as:

log DIC = log[(K_o)_ex/(K_o)_i] + B - log[(R - e₁)/(1 - Re₃/e₂], (7) and

log[(K_o)_ex/(K_o)_i] = [50.20(µ_ex - µ_i)/2.303][0.023517 - 0.023655(T/100) + 0.0047036(T/100)²] (8)

where B is an experimentally derived constant determined via calibration of the TCO₂ channel against CRM. The subscript "ex" refers to the acidified outer solution, "i" indicates the internal solution, and µ is ionic strength.

For each of the three indicators used, three wavelengths are chosen for measurement of absorbance. Two wavelengths assess the absorbance peaks of acid and base forms of the indicator, while a third wavelength serves as a reference wavelength. Absorbance varies at the acid and base wavelengths in response to pH changes, but not at the reference wavelength. Absorbance measurements at acid and base indicator maxima are used for determination of all CO₂ system parameters. The wavelengths chosen for the three channels are listed in Table 7.

The indicator solution for CO₂ measurements consists of a solution of 2 µM phenol red, 225 µmol/kg total alkalinity (Na₂CO₃), and 0.2 µM sodium lauryl sulfate. For TCO₂ measurements, the indicator solution is made of 2 µM bromcresol purple in 1000 µmol/kg total alkalinity (Na₂CO₃) and 0.2 µM sodium lauryl sulfate. The reference solutions of the CO₂ and TCO₂ measurements are similar solutions that contain no indicator. For pH measurements, the thymol blue stock solution is made in Milli-Q water with a concentration of 1.5 mM. The R ratio of thymol blue solution is adjusted (R~0.77) to minimize the magnitude of indicator-induced pH perturbations. All indicator and reference solutions are stored in gas-impermeable laminated bags. 3N HCl in Milli-Q water is stored in a 250 mL glass bottle and used to acidify TCO₂ samples.

Three Ocean Optic 2000 spectrophotometers were used to determine indicator absorbance for each of the three measurement channels. The light assemblies, spectrophotometers, and optical cells are connected with optic fibers. The light assembly of each channel consists of a high-temperature tungsten lamp with blue and short-pass filters to achieve an improved balance of spectral intensity between 430 and 700 nm.

The optical cells of the CO₂ and TCO₂ channels are custom-machined from PolyEtherEtherKetone (PEEK) rods. The PEEK optical cell has a 27 mm OD and a 2 mm ID with a length of 15 cm. A Teflon AF 2400 LCW is located inside this optical cell. The sample inlet and outlet, and two optical fibers that connect the optical cell with the light source and spectrophotometer, are inserted into the ends of the LCW with two custom-made PEEK connectors. The ends of the LCW are sealed with two O-rings that are housed inside the connectors. The PEEK connectors allow both reagent and light to pass through the LCW. The pH optical cell is machined from a PEEK rod, but does not require special connectors since no LCW is used.

Indicator and reference solutions are pumped through separate lines into their respective channels by digital peristaltic pumps. Surface seawater is obtained continuously with a shipboard pumping system. It first flows through a SBE 49 CTD that records salinity and temperature, and is then pumped through three seawater channels (fCO₂, TCO₂, and pH). Before entering the TCO₂ channel, seawater samples are acidified with ~3 N HCl using a digital peristaltic pump. The mixing ratio is approximately ~700 (seawater to HCl). An in-line coil is used to facilitate mixing. For pH measurement, thymol blue is mixed with seawater sample at a mixing ratio of ~700 (seawater to thymol blue), whereupon the final thymol blue concentration in the mixed sample is ~ 2 µM. This low indicator concentration results in very small indicator-induced pH perturbations (< 0.001 pH units). An in-line mixing coil is also used in this case.

All channels are thermostated in a Lauda E100 water bath that is set to 25 ± 0.1°C. All samples, reference and indicator solutions are temperature pre-equilibrated to 25°C in the water bath using PEEK and glass coils. All measurements, as well as calibrations, are taken at this temperature.

All components of the system are connected to a custom-made electronic motherboard that is controlled by a PC. The interface program cycles to operate the MICA autonomously. The time required for each measurement cycle depends on the equilibration time (7 min for the seawater fCO₂ and TCO₂ channels) and flushing time selected for the indicator/reference solution and samples (~2 min). The equilibration time required for pH measurements is very short. The following sequence was used during a measurement cycle:

1. Flush pH reference (seawater samples without indicator solution).
2. Flush reference for seawater fCO₂ and TCO₂.
3. Read and store reference readings.
4. Flush indicator solutions for seawater fCO₂ and TCO₂; mix thymol blue with seawater samples (pH measurements); acidify TCO₂ samples.
5. Seawater fCO₂ and TCO₂ equilibration (7 min).
6. Read and store measurements.
7. Repeat steps 4-6 five times.
8. End of one measurement cycle and repeat from the beginning.

Prior to the cruise, the CO₂ channel was calibrated against five standard CO₂ gases ranging from 280 to 550 ppm (xCO₂). TCO2 was also calibrated before the cruise using a CRM. Thymol blue has been calibrated previously for seawater pH measurements (Zhang and Byrne, 1996). During the cruise, CO₂ gas standards and CRM were used periodically to evaluate calibration consistency for CO₂ and TCO₂ measurements. Re-calibration was performed if necessary.

The calibration for the CO₂ channel is related to xCO₂ readings of the standard CO₂ gas tanks, even though it is fCO₂ that drives equilibrium across the LCW membrane. Based on DOE (1994), xCO₂, pCO₂ and fCO₂ can be calculated from each other at various measurement conditions (temperature, salinity, and pressure). The MICA seawater fCO2 measurements reflect fCO2 at 25°C with 100% water vapor content. The fCO2 at 25°C was corrected to in-situ temperature to compare with PMEL LICOR underway pCO₂ measurements. The temperature correction used in this case is given by Millero (2007):

d(lnfCO₂)/dT(deg^-1) = 0.044 - (8*10^-5)S (9)

No correction was required for the MICA TCO₂ measurements since TCO₂ is expressed in µmol/kg. The MICA pH measurements gave pH values on the total scale at 25°C.

The auxiliary salinity, temperature, and pressure data used for these corrections and calculations were obtained from underway measurements using shipboard instruments.

Quality control analysis was performed in three steps. First, suspect data were removed, as necessary, following reported cruise log malfunctions (e.g., possible sample contamination and observed malfunction of the thermostat) and spectrophotometric anomalies (e.g., sudden decrease of baseline light signal indicating entrapment of air bubbles). Secondly, data points with deviations > 3 standard deviations (p < 0.01) from the running average of the data were removed. Finally, two of the three measured carbonate parameters (fCO₂, TCO₂, pH) were used to calculate the third parameter, which was compared with direct MICA measurements of the parameter. Suspect data were eliminated based on this internal consistency check.