Estimation of uncertainty on pH measurement results: a direct pathway to quality assurance

A technique for estimation of measurement uncertainty of routine pH measurement using pH

meters; Thermo scientific Orion Star and Basic 20 is presented. There are issues associated with

pH measurements, what we actually measure is not usually what we would expect or even intend,

and it is common practice that analysts just take the immediate reading without doing repetitions.

The pH measurements made from one chemical laboratory are not consistent with those made in

a different laboratory and, the pH meter readings generally do not agree closely with the calculated

pH values. The result of first approach, that is single experiment (3 repetitions), uncertainty

evaluation was done according to the guide to Quantifying Uncertainty in Analytical Measurement

(QUAM) and/or the guide to the expression of uncertainty in measurement (GUM) using

uncertainty budget table as a tool. On second approach, we established correspondence between

observed and predicted uncertainties with new derived equations to calculate the total uncertainty.

At acidic pH -2.08 uncertainty was ± 0.02 and basic pH 13.3, the uncertainty was ± 0.01, the

results of calculations similar on both approaches. Results of single experiment over a short period

of time confirm that individual uncertainty values at particular pH values correspond to the

manufacturer’s specification, but not to the expected pH values. Repeatability conditions and

pooled calibration method were used for further assessment. Repeated analysis under similar

measurement conditions and experimental detail were performed to measure pH of numerous

buffers and sample solutions, and then, pooled calibration, the basic statistical calculations, the

Horwitz equation, coefficient of variation (CV%) and the law of propagation of uncertainty (LPU)

in a spreadsheet model were used for the analysis of uncertainty. Pooled calibration created a

satisfactory correspondence between predicted pH values and those observed by experiment. The

Horwitz equation constituted an expert judgment on the performance of the meters, it indicates

poor performance at pH value -2.08. The results correspond to the CV% of [𝐻

+] and CV% of pH

values as well. HorRat ratio showed significant difference between the coefficients of variation at

pH -2.08. At pH 13.3, there was no significant difference between the coefficients of variation

and variances are homogenous.