This article introduces the use of configuring sample conductivity functionSievers^®M9 TOC analyzerProcess improvement in detecting the first stage conductivity. Many users choose M9 analyzer to detect the first stage conductivity, in order to simplify the cumbersome process of using desktop instruments to detect conductivity and improve detection efficiency.Using M9 analyzer to detect conductivity can greatly reduce the inherent instability of using desktop instruments to detect conductivity, while achieving automation of data transmission.

However, in some cases, significant sources of instability still exist, sometimes causing the test results reported by the analyzer to exceed the ± 2% accuracy limit specified in the pharmacopoeia. This application literature provides suggestions to users to help them improve their detection process capabilities and reduce the instability of out of specification (OOS) results.

The traditional M9 conductivity detection procedure involves first performing a single point calibration of 1.4 mS/cm, followed by confirmation with 25 µ S/cm HCl. According to the requirement of ± 2% accuracy specified by USP, the confirmation result must be within 24.5-25.5 µ S/cm to be considered passed. When conducting a 25 µ S/cm confirmation, small deviations in the 1.4 mS/cm calibration can cause the confirmation result to exceed the specification standard by ± 2%. In addition, when CO in the air₂When entering the standard solution, it will increase the conductivity of the HCl standard.

We evaluate the causes of unstable detection through experimental data and determine the conditions that can significantly reduce instability, in order to improve the efficiency of the detection process and increase the success rate of meeting the ± 2% specification standard. We used 12 M9 analyzers for single point calibration at 1.4 mS/cm and 100 µ S/cm, respectively, and then tested the accuracy of each confirmed concentration point within a certain range.

We have determined the optimal calibration/confirmation combination for stability by calculating the process capability that meets the ± 2% specification standard.

Figure 1 shows the result data detected by multiple analyzers at 5, 10, 25, 50, and 100 µ S/cm. The data set shown at the top of the graph is from a calibration of 1.4 mS/cm, and the data set shown at the bottom is from a calibration of 100 µ S/cm. The dashed line represents the specification limit.

Figure 1: Six different confirmation points

Comparison of conductivity accuracy between two calibration methods

Table 1 lists the process capabilities of each confirmed concentration for two different calibrations, as well as the corresponding probabilities of exceeding specification standards. If the process capability index is lower than 1.3, it indicates that the process cannot meet the specification limit with 99.99% confidence. Table 1 quantifies the performance improvement amplitude when comparing the 100 µ S/cm calibration with the 1.4 mS/cm calibration. For example, the measurement process combining 1.4 mS/cm calibration and 25 µ S/cm HCl confirmation has a maximum probability of exceeding specifications of 31% due to measurement deviation, while the measurement process combining 100 µ S/cm calibration and 25 µ S/cm HCl confirmation reduces the probability of exceeding specifications to below 1%.

Table 1: Comparing different calibration and confirmation combinations

Process capability and frequency exceeding specifications

The calibration tasks and confirmation standards for firmware and software used in the 2.0 upgrade version can provide users with flexible and robust measurement processes, greatly improving process capabilities. Here are our operational suggestions:

Performing 100 µ S/cm calibration instead of 1.4 mS/cm calibration can improve the accuracy of conductivity measurements for various confirmation standards used in pharmacopoeia water testing. This suggestion can significantly solve the problem of unstable measurement results and improve the performance of pharmacopoeia compliance. Users can decide whether to implement this suggestion based on their existing process capabilities and frequency of exceeding specifications. If the instability of the operation is already small, users may not see significant process improvements.