The calibration procedure is to obtain a way to estimate the Narrow band absorption ration RN from the broadband absorbption ratio RB we get using LEDs . Linear regression is used to derive the equation to estimate RN (for absorption ratio, narrowband spectrometer) from RB (for absorption ratio, broadband photometer).
The authors state that this relationship is specific to the photometer they built and that other systems will likely differ due electrical and optical variation in the individual components. Presumably this is variation inherent in the manufacturing process.
The authors state that only one calibration is needed.
They authors state that purified indicator is needed for calibration and all subsequent pH measurements. We’ll get to the purified indicator separately, but for now assume that we’ve got some.
To calibrate we need an available narrow band spectrophotometer.
- Obtain purified indicator mCP adjusted to an RN value of 1.6. We calculate that this is a pH of 7.82 … roughly that of seawater
- Calibration seawater.
- The authors used synthetic tris buffered sea-water with an initial pH of of 8.113 (determined with mCP and the narrow band spectrometer), apparently for calibration.
- create 10 or more samples of tris buffered sea water with range of pH
- S =35
- manipulation of pH is done by adding 1 N HCl or 1 N NaOH
- pH errors caused by indicator dye are estimated at ~0.002 at pH 7.9 and significantly below the target accuracy of the broadband photometer of ± 0.01
- N and B comparisons. They used surface seawater collected from the Gulf of Mexico. The seawater was acidified with HCl. It was also watered down literally to manipulate salinity. These trials were used to produce Figure 4 showing the differences in measurement of pH between N and B, and the effects of Salinity and Temperature on N and B measurements.
- salinity manipulated with deionized Hs0
- acidity manipulated with 1 N HCl.
- measure T and S
- Follow protocol to obtain RN but include measure blank on the broadband device before adding mCP
- obtain RN and RB readings
- for each sample
- measure blank for both N and B
- measure absorption for N and B
- calculate RN and RB
- take RN and RB as ordered pairs and run a linear regression to obtain the regression line equation and correlation coefficient, variance and so forth – this is the calibration.
- compare calibration with Yang et.al.
Purified indicator mCP
B.Yang et.al. reference a procedure to purify indicator mCP from sodium salt according to the procedure of Patsavas et al. (2013).
B.Yang reports they employed 10 mmol.L-1 mCP stock solution in 0.7 mol/kg NaCl for all measurements. The R-ratio of the stock mCP solutions was adjusted to 1.6 by addition of 1N HCL or 1N NaOH.
Tris acidimetric SMR 723e was used to prepare tris-buffered synthetic seawater.
From Fig.3 of B.Yang et.al. it’s evident the authors used at least 30 samples of well-buffered solutions with RNs ranging from about 0.5 to 2.5 to obtain their linear regression coefficients.
In Figure 4., differences in the pH readings from broadband and narrowband devices are plotted. Here they used 136 samples over a pH range of about 7.55 to 8.22 or so. Such a run could also be used to confirm the calibration as well.