pH probe drifting and measured values are different than expected

[laurenzane]

Last week on 9/7/22, the bulb of the titrator was left out accidentally for about 2 hours. I attempted the pH calibration which measured values within the accepted range but measured about 9mV as opposed to 0 for the pH 7 buffer. I also ran a CRM but this deviated from the accepted value by over +1%. I chose not to continue with titrations and let the electrode sit in KCl for one week.

Today, Jill and I attempted the pH calibration and noticed that the values for pH 4, 7, and 10 were drifting rather than oscillating between two points and settling on a value.

We flushed the internal and external KCl reservoirs twice and refilled the reservoirs with new solution. We let the probe soak in pH 4 buffer for about 4 hours to reactivate the electrode. After this, we calibrated the pH probe again with new buffers and all values fell within the accepted range according to the manufacturers reference guide. We ran a CRM with a different bottle of CRM that had been previously tested, leading us to believe that the CRM may have not been accurate.

While we resolved that we do not need to replace the probe, we should look out for values drifting during the calibration or TA measurements. The probe is 1-2 years old, so we may need a replacement in the near future (within 6 months- 1 year).

On 9/30/22, Jill, Amy and I replaced the acid titrant and completed a run with two new bottles of CRM. Here are the values we recorded that indicate probe deprecation. For the pH calibration and the CRM values, deviance from the expected was first observed the day the probe was left out (8/25/22). Since 8/25/22, pH calibration values have trended higher and higher for ph 4 and 7 and less negative for pH 10. The deviance is progressively more dramatic with each sample ie: deviance from expected at pH 10 (Sample #3) is more than the deviance at pH 4.

We also tested two CRM bottles, one I had tested (Batch # 176, Bottle A) and a new bottle from the Putnam batch bottled on March 18, 2022 (Batch #202) and also ran a junk sample. The CRM values were within range today, but have deviated from the known value by over 1% for the past few weeks.

slope: -57.06
zero pt: 7.289
pH 4: 187.2
pH 7: 16.4
pH 10: -157.2

CRM176: -0.299%
CRM202: -.223 %

We were not able to calculate TA values because the pH calibrations were so far from the expected values that the R script failed, so we used the pH calibration from last week.

"Error in nls(y ~ ((At + (St/(1+ Ks * Z/(f * Hprime))) + (Ft/(1 + Kf/f * : number of iterations exceeded maximum of 50". We researched this error and it typically occurs when the model does not fit your data.

[hputnam]

I am not clear on the data you are presenting and if they show that we need a new probe now. Please do this more quantitatively in the github issues with more than one data point.

For example, take a picture of or list the criteria shown on the probe manual for the indicators of probe degradation:

zero point
values the probe mV signal can move within so many seconds
slope of the calibration

Also include the plots of the pH 4,7,10 values and slope through time from the data you have all be saving from each calibration since we started using this probe.

Also include plots of the deviation from the CRM through time since we started using this probe with the CRM batch indicated on the plot

These are the data and tests we need to diagnose sensor degradation given that they are ~$630 per probe, we can’t order new ones without these data. Also please post a large sign or generate a checklist for shutting down the instrument that each user needs to complete to ensure the probe is not left dry.

In addition to the quantitative information, there are also several other steps we can try for reactivation, such a regeneration solution i purchased previously, as well as 2-3 of these electrodes we returned from Hawaii on the last trip.

Thanks for your work on this!

[amyzyck]

Zyck_Titrator_Electrode_Diagnostics.pdf

I have attached a report which summarizes the probe health data. The first page of the report is a screenshot of the adjustment and calibration section from the probe manual. Based on section 3.4 in the manual, the zero point for the used electrode has remained within the 6-8 pH range and the slope has not been less than -52 mV/pH in the time period which we've had this probe.

The second page of the report graphs the outputs from the pH calibrations for the 4 (top), 7 (middle), and 10 (bottom) calibration buffers over time (captures the full age of the electrode). I've also attached the pHCalibration Excel document which includes all calibrations from the Puritz and Putnam group from April 2021 to present: pHCalibration.xlsx.

The last page of the report graphs the % deviation from the CRM batch over the age of the electrode with Batch numbers included. The 1% accuracy threshold is denoted with an orange dotted line. The final 2 datapoints are highlighted because we were unable to calculate the CRM values for both CRM batches using the pH Calibration completed that day (R error message included in Lauren's original post). Instead, the CRM values were calculated using the pH Calibration from 9/16/2022. Again, I've attached the CRM Accuracy Data collected from both lab groups from April 2021 to present: Copy of CRMAccuracyData.xlsx