15.06.2015

Oven Coulometric Karl Fischer Method

Deliverable 1.2.4

1. Experimental setup

For method development and validation purposes the measurement system was divided into two parts: the coulometric titrator and the vaporization (oven) system (Annexes 1 and 2). The coulometric titrator used was a 831 KF by Metrohm and the vaporization system used was 874 Oven Sample Processor, both devices were connected to a computer equipped with Tiamo – device and data management program supplied also by Metrohm.

The following parameters of the coulometric titrator were studied: polarization current between indicator electrodes, titration speed, time interval between measurement points and two end point criterion – titration speed and indicator electrode potential. For the oven system carrier gas flow speed and oven temperature were studied.

The experiments used to study the effect of different parameters on the performance of the coulometric titrator, were carried out using a gravimetric reference solution composed of methanol and water to minimize matrix effects and side reactions. Furthermore, in order to exclude some of the possible sources of error present when using the oven system, direct injection into the measurement cell was used.

The oven system parameters were studied with the same reference solution as well. However, solid samples were included as well: alpha-D-lactose monohydrate, calcium oxalate monohydrate and a certified reference material with a moisture content of 1 g/100g. In addition, a large number of measurements were made with a variety of real samples for the purpose of developing methods for their moisture content determination and characterization of measurement uncertainty. Dried nitrogen was used as the carrier gas in all of the experiments.

2. Experimental parameters of the O-cKF method

The methods used for different samples is based on a common framework. The main parameters of the coulometric titrator are kept the same as the supplier of the instruments recommends (titration speed: “optimal”, indicator electrode polarization current of 10 µA). For the oven system, the carrier gas flow speed (50 ml/min for the instrument used) is left unchanged as previous experiments did not show large differences between results obtained at different speeds.

Two parameters are usually varied for the measurement system: oven temperature and extraction time. The former must be sufficient for the water to be released from the sample in a reasonable time (under 15 minutes) and low enough to prevent sample decomposition. The extraction time dictates how long the instrument waits for the water to be released from the sample, if this time is too short, the analysis will be stopped before the water is released from the sample. Both of these parameters depend on the sample and therefore must be optimized experimentally. Typical values of oven temperature are between 100 and 200 °C. Typical values of extraction time vary from 200 to 300s.

Rarely, in the case of very small absolute water contents, it is recommended to set titration speed to “slow” as it will keep the titration process more stable. In addition, the relative drift for determining the end-point of the analysis is adjusted according to the measurement cell condition.

For finding the optimal values of the parameters, especially the oven temperature, a procedure, based on series of measurements with temperature gradients, has been developed. It is described in detail in Annex 3.

Annex 4 describes preparation of a reference solution and step by step sequence for measurement of its water content.

3. Validation and measurement uncertainty

3.1 Most influential parameters

Of the parameters listed previously, three had a significant effect on the measurement results.

The most important parameter is the oven temperature. Too low temperature will not release all of the moisture within the sample, a temperature too high can lead to sample decomposition. As real samples can be vastly different, this parameter should be experimentally optimized for each sample (see Annex 3). In addition to oven temperature, the current between indicator electrodes and titration speed had a significant effect on results. Variations in both result in changes in iodine generation stability over time and titration time, with the current between indicator electrodes having a significantly larger impact on titration time.

Details will be given in the Validation and Measurement Uncertainty Report (deliverable 1.2.5)

3.2 Measurement uncertainty estimation

As mentioned before, in addition to validation experiments, a large number of measurements were made with various real samples, their moisture contents were from 0.01 to 23 g/100g. To study the effects, different environmental conditions have, some samples were kept under adverse environmental conditions with relative humidities ranging from 0 to 100%. To give a better representation of real situations, separate measurement uncertainty estimations were made for different measurement situations. In addition, the samples were divided according to their observed moisture content into two groups: under 1 g/100g and more than 1 g/100g.

When analysing a simple sample with the oven system under reasonable conditions, a relative combined standard uncertainty of 0.7% was obtained. Ifa real sample with a low moisture content is analysed and improper sample handling techniques are applied, a relative combined standard uncertainty of 10% was obtained, for samples with a high moisture content this is 3%. The results are handled in depth in the Validation and Measurement Uncertainty Report (deliverable 1.2.5).

4. Annexes

Annex 1

The measurement system used

Figure 1. Measurement cell with a generator electrode with a diaphragm.

Figure 2. Oven system, 874 Oven Sample Processor

Figure 3. Opened database in the control software Tiamo.

Figure 4. Method development in Tiamo

Annex 2

Oven (vaporization) system scheme

Annex 3

O-cKF method development for a new sample type

  1. First gradient
  2. The sample amount should be minimal, however no less than 10 mg when using an analytical balance.
  3. The gradient should cover a wide temperature range, for example 50-250°C and the speed of the gradient can be high: 10°C/min.
  4. Plot the results on a graph and determine the temperature, at which the water is released(Figure 5). This analysis will also provide a rough indication of the moisture content of the sample.
  5. If no water is seemingly released from the sample, the experiment should be repeated with a larger sample amount.

Figure 5. An example water release profile.

  1. Second gradient
  2. Adjust the sample amount, so the absolute water content in the measured sample is around 1000 µg.
  3. The temperature range should be narrower and must cover the area where water was released. The gradient should be slower.
  4. Fit for purpose method
  5. Better repeatability is achieved with methods using a constant temperature.
  6. With a constant temperature, there is no need to cool the oven system and sample throughput is higher.
  7. The used temperature should correspond to the maximum speed of water release, if no sample decomposition occurs.
  8. Validation
  9. Analysis at different temperatures.Measurements should be made at least at three different temperatures. One of these would be the previously selected temperature, the second 10°C above and the third 10°C below the selected temperature. At each temperature at least three replicatemeasurements should be made.
  10. To estimate reproducibility, measurement at the selected temperature should be made on two additional days, each day three replicate measurements.

Annex 4

Preparation of reference solutions and analysing it with cKF

  1. Sampling methanol.
  2. Allow the container that you want to use, to equilibrate for at least 24 hours.
  3. Weigh the container with its septum (using a balance with the highest possible resolution in that region).
  4. Pour methanol into the container and seal it immediately to limit contamination by water vapour.
  1. Determination of methanol water content.
  2. Carry out a measurement series of pure methanol to determine its own water content. Take the sample through the septum with a syringe, to avoid pressure differences, puncture the septa with another syringe that is filled with molecular sieves (“drier”).
  3. If measurements are also planned with the oven system, measure the methanol water content with the oven system as well.
  1. Spiking with water.
  2. Weigh the container with the added methanol and calculate the mass of methanol in it.
  3. From the mass and moisture content of methanol, you can calculate the amount of water that must be added to achieve the desired moisture content.
  4. Draw the needed amount of water into a syringe.
  5. Add the water through the septum.
  6. Weigh the container again to determine the exact amount of water added, from that the reference water content can be calculated.
  1. Moisture content determination:
  2. Direct injection – a number of replicate measurements with good repeatability (first measurement results are usually higher and are excluded), a total of around 8 replicate measurements are required.
  3. Take enough sample for all measurements in one syringe, to avoid pressure differences use the “drier”.
  4. Calculate the sample mass for each injection from the weight differences of the syringe.
  5. Oven system – three blank vial measurements at the start of the series, followed by about six replicate measurements with the sample.Sample amounts should be as uniform as possible.
  6. Vials and caps need to equilibrate at least 24 hours prior to their use in anhydrous atmosphere. If possible, blank vials should be analysed at different points during the process.
  7. Add samples into vials though septa and use the “drier” to avoid pressure differences.
  8. If possible avoid preparing the samples many hours before the analysis.

For a better comparison, carry out measurements with both direct injection and the oven system on the same day.