DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best avaiiable original document.

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of-the United States Government. Neither the United States Govenunent nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or' usefdness of any information, apparatus, product, or ptooess disclo@.or represents that its use would not infiringe privately owned rights. Reference hereh to any specific commercial product, process, or service by trade name, mdemaric, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy. -

Available to DOE and DOE iontractors from the Office of Scientific and TechAcal Information, P.O. .Box 62, Oak Ridge, TN 37831; prices available from (615) 576-8401.

Available to the public from the National Technical-Information Service, U.S. Department of Commerce, 5285 Port Royal Road, Springfield, VA 22161.

WSRC-MS-96-0406

Measurement Assurance Proyram for Deuterium Oxide by Infrared Spectroscopv (u)

Sherold R. Johnson and John P. Clark, Westinghouse Savannah River Company, Laboratories Operations, Dept., 772-D, Aiken, SC 29802, USA 803-557-6820 & 803-952-3345

INTRODUCTION

Analytical chemistry measurements require an installed criterion based assessment pl;ogram to identify and control sources of error. This program should also gauge the uncertainty-about the data. A self-assessment was performed of long established quality control practices against the characteristics of a comprehensive measurement assurance program. Opportunities for improvement were identified. This paper discusses the efforts to transform quality control practices into a complete measurement assurance program. The i-esulting program heightened the laboratory's confidence in the data it generated, by providing real-time statistical information to control and determine measurement quality.

BACKGROUND

The Water Quality Laboratory of the Savannah River Site has performed deuterium measurements in support of the Heavy Water Production - Purification Facility for over forty years. As the Cold War ended, production needs diminished. Reprocessing of heavy water is on-going for use at other Department of Energy complex facilities or for commercial sale. In 1996, nearly 10,OOO deuterium determinations were performed in support of current mission activities.

The analysis of deuterium or heavy water is performed using a Fourier Transform Infrared (FTIR) Spectrometer. The instrument incorporates the Michelson interferometer design with sealed optics. The FTIR analyses are computer controlled, which greatly reduces the subjectivity in chart recorded readings. The manufacturer's software, WinFirst, interacts with Microsoft Excel. The WinFirst software houses the operating, data generation and manipulation commands. The manipulation functions allow placement of the data in designated cells of an Excel spreadsheet. The spreadsheet shown below, details the mole% concentration with respective absorbance readings. Correlation coefficient as the calibration acceptance criterion, graphical calibration curve, linear regression statistics, control standard and sample data are also included on the spreadsheet.

Excel Stweadsheet

0.2 0.4 0.6 0.8

The instrument readability range is within 0.1 through >99 mole% deuterium. Because the operating range is so vast, absorbance responses are non-linear. However, linearity can be established within shorter range subsets. Therefore, eight such subset ranges are in use encompassing the entire range of readability.

QUALITY CONTROL PRACTICES

It has long been said that an analytical measurement uncertainty is no better than the standards used to produce the data. In the analysis of deuterium, there are no standards commercially available that are traceable to a national measurement base, such as the National Institute of Standards and Technology (NIST). The Water Quality Laboratory sought compensation for this shortcoming through participation in a sample exchange program, external to the Site. The laboratory has participated in the exchange program for many years. The material involved in the exchange became well characterized. The Water Quality Laboratory used the remaining quantity as a pgmary .- standard. However, only short-term traceability was achieved.

In later years, much of this characterized material became available for purchase. The Water Quality Laboratory purchased three small lots of material from the Chalk River Facility, Canada. This material provided an internal means of long-term primary standard traceability. A %-gallon drum, of product from the on-site Heavy Water Facility was obtained for the preparation of working standards. This material was internally characterized against the purchased Chalk River material.

In an effort to obtain independent characterization of the 55-gallon drum, aliquots are sent to the Brookhaven National Laboratory. The 55-gallon quantity would support working standard preparations for about two years. Before consumption of the drum quantity, characterization of a

. new drum would begin.

From the 55-gallon drum, working standards are gravimetrically prepared as dilutions of the stock material. The stock material is analyzed at each withdrawal from the drum. This practice serves as an integrity verification measure. Deuterium is hygroscopic. Interaction with atmospheric moisture compromises standard integrity. For this reason, stock material and working standards are stored with installed in-line desiccant traps. Therefore, in the withdrawal of standard from the storage vessel, dry air displaces the liquid.

Working standard preparation utilizes the two-balance , weight-by-difference method as shown below. Using a tared stock delivery and receiving vessel, the weight of stock received in the larger vessel is verified against the weight delivered from the smaller vessel. The diluent is dispensed to a preestablished weight. Balances are calibrated prior to this work, using class-S tolerance mass standards.

Standard Preparation Data

uocEYsIL)DEsIRED FImI Resut! (Uorr)

0 2 0 For a 1 mer dum: 999.48 F a a 2 S'tet 1998.95 For a 4 liier d u n e : 3997.91

99.700 99.691

Waler 0.52 1 .05 2.09

(W Set WL-5551

weight W S J , uncemtainty R.6 10 IO oO>26140 0 OoO17161 0.0006

100 l O O O O 8 2 l f i 0 00026937 0.0200 ZOO' 200 01606379 0 00012937 0.0006

2000- 2000 05706

SELF-ASSESSMENT AND FINDINGS

The Water Quality Laboratory took a proactive look at current good laboratory practices against a comprehensive measurement control program. Several noteworthy practices were currently in use. They included

Use of an internal, primary standard that was characterized using Nuclear Regulatory

Utilization of an external, independent laboratory to assist in the characteriz+ation of bulk

Verification of stock material integrity at each withdrawal from the drum; Standard protection from moisture degradation; Gravimetric standard preparation.

Guide, protocol (NUREG-01 18);

on-site stock material;

Nevertheless, some opportunities for improvement were noted. They included:

Independence between calibration and control standards; Control charting; Statistical software to monitor method and user performance; Determination of total measurement uncertainty.

IMPROVEMENTS

Independent Calibration - Control Standards For each of the eight operation sub-ranges, a five-point calibration is performed. The mid-point of the five standards had been used as the control standard. Since these standards were not prepared from an independent source, systematic drifts in the deuterium analysis would be undetected.

A separate %-gallon drum was obtained from the SRS Heavy Water Facility. Characterization proceeded through a statistically designed experiment, randomizing the analysis order of the Chalk River standard, the previous 55-gallon drum, the latest exchange material and the newly obtained drum. Each analysis sequence began with a full five point calibration. The randomized analysis order included duplicate analysis of each material. An example of the randomized sequences are shown below. (Note that each analysis sequence began and ended with the primary standard.)

I

Designed Experiment

Randomized Analysis Order

Sequence 1 Sequence 2 Sequence 3 Kev CR OD ND EM OD ND EM CR

CR ND EM OD ND EM OD CR

CR EM OD ND EM = Exchange Material EM OD ND CR

CR = Chalk River Primary Standard OD = Old 55-gallon Drum ND = New 55-gallon Drum

c

The exercise served a two-fold purpose: characterization of the supplemental stock material and continued participation in the exchange program. As for characterization of the new drum, statistical comparison against Chalk River revealed acceptability of the material for working standard preparation. Based on the experiment, use began on the nominal mole% concentration. Uncertainty statistics will be finalized after Brookhaven testing is complete.

Data Trending

Another improvement opportunity included the need for statistical software to evaluatg the overall method performance. In previous years, method uncertainty was quoted as that of theChalk River standards. This did not represent the true uncertainty picture. An automated system of data collection was needed to effectively control chart, trend data, and evaluate user performance. Previous data tracking was tolerance based between preset ranges.

Statistical software was installed to track data, calculate overall method uncertainty and evaluate operator performance. The Process Measurement Assurance Program (PMAP) software utilizes the generated data and the random / systematic uncertainties associated with the stock material to calculate the overall method uncertainty. Listed below is the method performance at the upper and lower range of readability.

Deuterium Analysis Method Performance

Range Uncertainty Systematic Random

0.1 - 1.0 mole % > 99 mole %

-0.01 mole % -0.02 mole %

+/- 0.04 mole % +/- 0.05 mole %

A control chart of each sub-set range was established, as well as for the analysis of the stock material before each working standard preparation session. This provided an opportunity to use the data generated as an integrity check. Hence, this information tracks drift over time. Additionally, P W calculates method performance and provides F and t test statistics for determining changes in process capabilities as compared to previous performance. Below is an example of the control charting feature.

Control Charting

FTIR Method Performance 89.00 I

4

It has long been known that atmospheric conditions , namely temperature and humidity were common causes of variation in the data. P W control charting has allowed the laboratory to gauge the extent of such common variation causes. Temperature drifts as much as 6 degrees F, beyond that at calibration can cause data shifts rendering out-of-control situations. While short-term calibration is performed to compensate for operating condition changes, PMAP assists in visualizing the on-set of the problem. Due to this knowledge, steps have been taken to stabilize temperatures reasonably well.

SUMMARY L- - The implementation of measurement assurance attributes for deuterium analyses, proved to be a creditability boost for the Water Quality Laboratory. Establishing independence between calibration and control standards was a leap forward. Automation, to control the process, yielded increased confidence in data released to the customer. Interpreting the data with statistical tools provided a media to defend the data's validity.

REFERENCES

User's Guide, "Using WinFirst - Fourier Infrared Software Tools for Microsoft* Windowsm from ATI Mattison", Analytical Technology, Inc., 1994.

ATI Instrument Specification Guide, "GenesisTM Series F"', Mattison Instruments, Inc., 1992.

* The information contained in this article was developed during work under Contract No. DE- AC09-89SR18035 with the U. S. Department of Energy.