Overcoming the challenges associated with Proficiency Testing (PT) schemes for online chlorine analysis

Speaker / Author: P.J. Lems

Rand Water

PO Box 3526, Vereeniging, 1930,South Africa

e-mail:

Phone: 083566 2176Fax: 016455 2055

Abstract

Proficiency Testing (PT) and Inter Laboratory Comparison (ILC) schemes are Quality Assurance (QA) tools used by laboratories to demonstrate that they are competent to execute analytical methods. Establishment of Rand Water’s online testing capacity required the development of suitable PT/ILC schemes to ensure the accuracy of the data. Over the past four years Rand Water has implemented and managed a PT scheme for online pH and conductivity analyses. The scheme has now been extended to online chlorine analyses.

Establishment of the PT/ILC schemes for online analyses presented a number of challenges. These included: 1) difficulties associated with the introduction of external control samples into online systems; 2) the absence of on-site analysts to receive and process control samples; 3) the distribution of analysers over an extensive reticulation system and 4) the need to measure volatile parameters such as disinfectant residuals.

These challenges were addressed by using portable instruments, rather than reference samples to compare the performance of online analysers. The principle was that a highly trained analyst would travel from site to site and compare measurements made with a single portable instrument against data recorded by online analysers. This presentation will consider the challenges encountered in the establishment of this system and present examples of typical data sets generated by the scheme. It will also attempt to highlight and discuss deviations from more conventional PT schemes.

  1. Introduction

In the last five years Rand Water has invested a great deal of resources in upgrading and improving the online analytical analysis of its water quality. Twenty new online Laboratories had been constructed and the project should reach completion by end 2013. Upon completion Rand Water will have one of the most advanced online water quality monitoring processes in Africa. A total of 350Quality Assuranceanalysers will analyse key parameters such as pH, conductivity, turbidity, ammonia and the chlorine species which include free and total chlorine and monochloramine on 71 pipelines delivering 3800Ml of water every day.

With the investment and improvement in Rand Water’s analytical online measurements capabilities there is a need for better quality assurance of analysis produced. Thus the decision was made that the new online facilities need to be accredited under ISO 17025:2005.

One of the requirements for Accreditation is the participation in a PT/ILC.1

For conventional laboratories PT/ILC are QA tools that are used to show competency in the analysis of analytical methods. ISO 17043:2010 specifies general requirements for the competence of providers of proficiency testing schemes and for the development and operationofproficiencytestingschemes2.Implementation of the standard towards the online measurement applications has proved to be problematic and even ILAC P9:11/2010 clause 4.6 recognises that not all areas are equally suited for PT/ILC and thus recommends that the laboratories discuss and agree on suitable alternative means by which the performance can be assessed and monitored.3

For this reason Rand Water is developing in-house PT/ILC schemes that are modified to fulfil the needs of the online analysers.Rand Water has successfully implemented a PT/ILC for online analysis of pH and Conductivity with only minor deviations from the requirements

With the need to expand the PT/ILC to include the chlorine analysis the preliminary assessments highlighted numerous challenges and it was found that the conventional preparation and distribution of control samples would not yield a workable solution. An alternative method of determining the assigned values used for comparison had to be determined.

The most plausible solution is to use a portable instrument as a reference. A single analyst would make use of a validated instrument to perform analysis on water samples as it is simultaneously analysed by the online analysers. Paired analysis would be used to determine the deviation from the assigned values as calculated with the portable analyser.

  1. Identification of challenges that prevents PT/ILC for chlorine analysis

The challenges that were highlighted by the initial assessment are primarily linked to the use of an external sample as a reference. The first limitation that was identified deals with the design of the online analysers and its inability to analyse external samples. Next two items identified was the volatility of chorine and coupled with that the distribution of samples over an extensive reticulation network. Finally the absence of an onsite analyst to receive and process the control samples was identified.

2.1Challenge of the online analyser design

The online analysers used are not designed for manual or batch operations and as such do necessarily not provide for the introduction of an external sample. WithinRand Water three different types of analytical methods are used for online chlorine determination and each has its unique challenges.

2.1.1Amperometric analyser

The amperometric analysers manufactured by Capital Controls, the 1870Eperforms continuous measurements whereby an acetic buffer is continuously added to the sample as it enters the analyser. It was considered that the installation be modified to allow for an external sample to be fed into the analyser. However taking into consideration the volume of samples required for this type of analysis, the number of analysers and the modifications that had to be made to introduce these samples it was found to be unfeasible.

2.1.2Colorimetric analyser

The next type of analyser is a Hach CL17 colorimetric analyser. These are new analysers that have been installed in the newly upgraded online laboratories. The analyser performs an analysis every two minutes. A sample flows into the analyser where buffer and DPDindicator reagents are added.4 Colour reaction is allowed to take place and a measurement is made. This analyser could be capable of measuring an external sample, however during the commissioning of the analysers in the online facilitates it was found that the variations in the flow rate had a negative impact on the analysers performance. To rectify the problem the suppliers recommended that a constant head device be fitted to the installation to minimise the flow variations. This addition however removed the possibility of having external samples analysed by the analysers.

2.1.3Polarographic analyser

The final type of chlorine analyser used in Rand Water is primarily found in the distribution network. It is an ATI Q45 analysers that utilizes a membrane-covered polarographic sensor that does not require the addition of chemical reagents.As this analyser has a probe sensor that could be removed it is the only analyser that could allow for a manual sample to be analysed.

2.2Volatility of chlorine

The volatile nature of the chlorine means that the concentration would decrease until there were no residual left.This makes it impractical to prepare a control sample in advanced and send it to the participant. The volatility also affects the availability of Certified Reference Materials(CRM).The CRM’s that was identified includes a sample matrix and concentrated chlorine sample that has to be prepared prior to analysis in accordance with the suppliers specifications. This is again not a process that is accurately performed in the field environment.

2.3An extensive reticulation network

The third challenge links to the volatile nature of the chlorine samples and that a control samples would have to be transport over an extensive reticulation system. Some of the key analysers are installed on the ends of the distribution which covers 3000km of pipelines. With such great distances between sites and the time taken to travel it does indicate that a PT/ILC would take a couple of days to be completed, as is the case with the pH and conductivity PT/ILC.With chlorines low stability in natural waters it would be impractical to store control samples for extended periods.

2.4Absence of an on-site analyst

The last challenge identified was the absence of an on-site analyst to receive and process the control samples. The online Analysers are installed and operated in such a manner that very little human interaction is required. Dedicated maintenance personnel are responsible only for the maintenance of the analyser. As per the different requirements the maintenance teams will clean any deposits from the analysers and replenish the chemicals required. Since these maintenance staff do not have any experience with the actual preparation and analysis of samples it was thought that to have them analyse samples would introduce an error with is normally not associated with the online analysis.

  1. Development of the PT/ILC method for online chlorine analysers

With the development of the PT/ILC for online chlorine analysis two key areas had to be addressed. The first deals with the validation of the analysers that would be used as the reference and the second are the actual process followed to conduct the PT/ILC. The processes followed for the two activities will be discussed and the results are included.

3.1Instrument validation.

The key area that was identified with regards to a good PT/ILC scheme was the accurate determination of the assigned value. This approach makes useof an instrument that was validated and is check using NIST traceable standards, to determine the assigned value.

Standard methods list eight different methods for the determination of chlorine. Analysis of each method revealed that the DPD colorimetric method appears the most feasible for field analysis. The main reason for selecting this method is the simplicity thereof. The more sensitive methods such as the amperometric titrations cannot be performed with the required accuracy in a field environment. Thus the DPD colorimetric method was selected for the determination of freeand total chlorine residuals. The instrument selected for this method is a Hach DR890 colorimeter.

3.1.1Principal of method.

N,N-diethyl-p-phenyldiamine (DPD) is the reagent that is added to the sample. Chlorine oxidizes this DPD to form two possible oxidation products, a magenta coloured compound known as WursterDye and a colorless imine compound:

Figure 3: DPD-chlorine reaction products

The intensity of the colour is directly proportional to the amount of chlorine present in the sample. The sample are analysed photometrically at wavelengths ranging from 490 to 555nm. 5The supplier’s method for the DR890 does provide separate DPD SwifTest reagentsthat are used for free and total chlorine determination.

3.1.2Validation process

The Hach DR890 colorimeter had been validated in accordance with the work instruction used by the Rand Water inorganic Laboratory.6 The validation was conducted on both chlorinated and chloraminated water sources. The validation was conducted over three non-consecutive days and each day the sample analysis was repeated seven times.

The validation included the following analysis:

  1. Hach Free Chlorine Dispenser- reagents weighed
  2. Hach Total Chlorine Dispenser-reagents weighed
  3. DR/Check Absorbance standard, Lot A3085, Exp. Mar 2015 - absorbance standard analysis at different wavelengths (420λ, 520λ, 560λ, 610λ)
  4. DPD chlorine LR standards kit, lot A3086A, exp. March 2015- analysis of blank, std1,std2 and std3
  5. Blank Millipore water with Free DPD reagent
  6. Blank Millipore water with Total DPD reagent
  7. Blank Millipore water with chlorine spike
  8. Free chlorine water sample from VereenigingA19 pipeline water -
  9. Free chlorine A19 sample with spike
  10. Free chlorine A19 sample diluted 10 times
  11. Low level free chlorine sample from Palmiet A6 pipeline (pre-chloramination)
  12. Total chlorine Palmiet O2 pipeline water(chloraminated)-
  13. Total chlorine Palmiet O2 sample diluted 10 times

3.1.2Validation results

3.1.2.1Detection limits (LOD) and limits of quantification (LOQ)

The LOD and LOQ were calculated by combining the three days results for the blank analysis performed on reagent grade water with the addition of free and total reagents.

Table 1. Summary of the calculated average LOD and LOQ

Blank Millipore water with free DPD / Blank Millipore water with total DPD
Mean / 0.011 / Mean / 0.014
Standard deviation / 0.005 / Standard deviation / 0.005
LOD / 0.026 / LOD / 0.029
LOQ / 0.059 / LOQ / 0.064

From table 1 it can be seen that the detection limits are well below the minimum residual concentration of 0.2 mg/l free chlorine at are required at the point of delivery.7

3.1.2.2Precision

The precision was determined by utilising the % RSD of the various samples analysed. The table below contains the average value for each sample analysis performed over the three days as well as the spike precision

Table 2: Determination of the precision

Free chlorine VG site water A19 / Low level free chlorine Palmiet A6
Day 1 / Day 2 / Day3 / Day 1 / Day 2 / Day3
Mean / 0.95 / 1.30 / 1.12 / Mean / 0.65 / 0.85 / 0.89
Sigma / 0.01 / 0.02 / 0.01 / Sigma / 0.01 / 0.01 / 0.01
% RSD / 1.22 / 1.29 / 0.85 / % RSD / 1.06 / 1.33 / 0.65
Free chlorine VG sample with spike / Total chlorine Palmiet suction kiosk O2
Day 1 / Day 2 / Day3 / Day 1 / Day 2 / Day3
Mean / 1.69 / 1.95 / 1.80 / Mean / 1.74 / 1.78 / 1.73
Sigma / 0.02 / 0.03 / 0.01 / Sigma / 0.03 / 0.02 / 0.02
%RSD / 1.41 / 1.34 / 0.68 / % RSD / 1.47 / 1.10 / 1.15

The %RSD values as shown in table 2 are all less that 2% with a mean of 1.12%

3.1.2.3Accuracy

The method adopted in the validation for the determination of accuracy was the use of NIST traceable standards, Absorbance standards and % spiked recoveries. The NIST traceable standard is a chlorine standard from Lot Nr. A3086A and expires March 2015. The absorbance standard,CAT No. 27639-00, is a set of three standards that are analysed at four different wavelengths. The spiked samples were Vereeniging A19 water sample that was spiked with a chlorine stock solution.

Table 3. Calculated % Accuracy for each sample

ABS standard at 420 λ wavelength
STD 1 / STD 2 / STD 3
Accuracy % / 98.86 / Accuracy % / 100.64 / Accuracy % / 100.92
ABS standard at 520 λ wavelength
STD 1 / STD 2 / STD 3
Accuracy % / 99.17 / Accuracy % / 100.62 / Accuracy % / 100.91
ABS standard at 560 λ wavelength
STD 1 / STD 2 / STD 3
Accuracy % / 97.95 / Accuracy % / 99.73 / Accuracy % / 100.26
ABS standard at 610 wavelength
STD 1 / STD 2 / STD 3
Accuracy % / 98.67 / Accuracy % / 99.63 / Accuracy % / 100.11
Chlorine standard (Lot Nr. A3086A)
STD 1 / STD 2 / STD 3
Accuracy % / 102.72 / Accuracy % / 102.30 / Accuracy % / 102.69
Free chlorine VG sample with spike
Day 1 / Day 2 / Day3
Spike recovery % / 94.06 / Spike recovery % / 94.24 / Spike recovery % / 94.87

The average accuracy that was determined using the standards was 100.35% and an average spiked recovery of 94.39%.From the analysis of the validation results the method had been found to be fit for purpose.

3.2PT/ILC method

Having established that the portable method for determination of free and total chlorine is fit for purpose the method for comparative analysis had to be developed

3.2.1Apparatus used during the PT/ILC run

  1. Hach DR 890 colorimeter serial nr: 120290C88637
  2. Hach Free Chlorine SwifTest dispenser
  3. Hach Total Chlorine SwifTest dispenser
  4. Glass beakers (100 mL)
  5. 10ml Eppendorf research plus pipette
  6. Eppendorf Tips (10ml)
  7. Hach Sample cell (25ml)

It should be noted that all glassware used arefree of chlorine demand and disinfectant chlorine species.

3.2.2Verifications to be performed in the laboratory prior to testing

Prior to the start of the PT/ILC a set of NIST traceable standards that are kept in the controlled environment of the laboratory are analysed on the colorimeter. The standard confirms that the instrument response is within the tolerance range as listed on the certificate of analysis and that the performance is still the same as during the initial performance evaluation.

Figure 1: Validation of Primary chlorine Standard

Expired NIST traceable standard are also verified against the current principal NIST standards. These expired standards are transported with the analyser and are used to ensure that the instrument is not damaged during transportation. These travel standards are analysed before and after the comparative analysis for a specific site is conducted.

Figure 2: Validation of travel chlorine Standard

3.2.3Performing the measurement

At the participant site the travel standards are analysed prior to any analysis being conducted. The data is collected in a control chart that is used for monitoring of trends.

Figure 3: Control chart of chlorine standards

The glass beaker and sample cells are rinsed with sample to be analysed.A new sample is collected in the 100ml glass beaker. The Eppendorf pipette is used to pipette the required 10 ml of samples into the sample cell. The blank sample is capped, the sample cell wiped clean and placed in the analysers. The instrument is then zeroed to compensate for interferences from oxidized forms of manganese in the water.8Once the instrument zero has been established a new sample is to be collected.

The sampling has to be timed such that it is representative of the sample analysed by the online analyser. For both the Capital Control and ATI the speed of response has to be taken into consideration when sampling. The instrument reading that corresponds to the particular sample will only be available after approximately 1½ to 2 minutes.9This is due to the continuous flow of sample and the time to displace the previous sample. For the Hach analysers the instrument’s waste line is visually observed for the presents of water. Once water is wasted it indicates that the analyser is taking a sample. The analysis is performed in triplicate to compensate for the deviations that may occur during the sampling.

Once a sample is collected it is pipetted into the sample cell. The DPD free or total Chlorine SwifTest is inverted, the reagents are added, and the sample cell is capped and wiped clean. With Free chlorine the sample is analysed immediately while for total chlorine it is recommended that the analysis is performed 2 minutes after the addition of the reagent.8The measured values for both the analysers are recorded on the result sheet.

Prior to departure that travel standards are again measured to ensure the instruments performance during the PT/ILC are within the expected results.