Impact of Instrument Response Variations on Health Physics Measurements

IEEE Transactions on Nuclear Science, Vol. NS-32, No. 1, February 1985

IMPACT OF INSTRUMENT RESPONSE VARIATIONSON HEALTH PHYSICS MEASUREMENTS

Guy A. ArmantroutLawrence Livermore National Laboratory

P. 0. Box 808Livermore, CA 94550

Abst ract

Uncertainties in estimating the potentialhealth impact of a given radiation exposure includeinstrument measurement error in determining exposureand difficulty in relating this exposure to aneffective dose value. Instrument error can be due todesign or manufacturing deficiencies, limitations ofthe sensing element used, and calibration andmaintenance of the instrument. This paper evaluatesthe errors which can be introduced by designdeficiencies and limitations of the sensing element fora wide variety of commonly used survey instruments.The results indicate little difference among sensingelement choice for general survey work, with variationsamong specific instrument designs being the majorfactor. Ion chamber instruments tend to be the bestfor all around use, while scintillator-based unitsshould not be used where accurate measurements arerequired. The need to properly calibrate and maintainan instrument appears to be the most important factorin instrument accuracy.

Problems of Real World Measurements

Actual dosimetric applications of healthphysics survey instruments involve making estimates ofrelative health risks under conditions oftensignificantly different than that found in theoreticalmodels and in the calibration laboratory. An earlierpublication [1] dealt with this problem andconsideration was given to different exposuregeometries and the difference in radiosensitivity ofdifferent body organs.

In actual practice, the radiation fields arealmost never monoenergetic. Even in those rare caseswhere a monoenergetic source may be used, shielding isalso employed which creates a lower energy scatterspectrum. In reactors and nuclear processing plants,multiple radionuclides are usually involved. As aresult, radiation fields usually involve a spectrum ofenergies rather than a discreet energy. Radiationinstrumentation, on the other hand, is usuallycalibrated with monoenergetic sources so that itsresponse is known as a function of energy rather thanas a response to multiple spectral inputs. It shouldbe noted, however, that, since there are an almostinfinite variety of various spectral combinations, ithas not been practical to try to calibrate instrumentsin terms of such a series of "standard" spectra.

A second complication is that, in practice,radiation fields which do not approximate a pointsource or a parallel beam flux are often encountered.These omnidirectional fields can result from multiplesources and from scatter from massive structures andshielding within the vicinity of the source or exposedindividual. Even for those cases involving parallelbeam exposure, the net effect to the individual may bebetter approximated by a distributed source geometrysince the individual will likely move and rotate in theradiation field.

Realistic Dose Estimates

The two principal factors to consider inestimating real world instrument performance are thoseof energy spectrum distribution and incident direction

of the radiation. In the case of distributed energy,it is possible to consider the response of theinstrument to different types of spectra which arerepresentative of the range of spectra which arenormally encountered. This can be readily donecomputationally provided the monoenergetic response ofthe instrument and the relative energy distribution ofthe spectrum being considered are known.

The problem of the effect of differentexposure geometries is best considered by using thedifferent conversion factors discussed elsewhere [1,2]which relate the effective dose under these differentexposure conditions to the exposure from a parallelbeam source as measured at a point in space by theinstrument. Under these conditions, the measuredinstrument response for a parallel source can be usedin this evaluation.

Comparison Approach for Instruments

Measurement of the instrument response to apoint source (which approximates a parallel beamradiation field) has been made at Lawrence LivermoreNational Laboratory (LLNL) for representativeinstruments from the major classes of instruments ascategorized by the type of detection element used inthose instruments. In addition to the LLNL data,additional data from the published literature have beenreviewed and included in this analysis [3-7]. Thisdata has been converted to a form convenient forcomputational comparison of the different instrumentsfor different incident spectra and dose conversionfactors.

In order to evaluate the response of theseinstruments for different energy spectraldistributions, several "standard" energy spectra wereselected. The principal source of these spectra isfrom a study recently completed by P. L. Roberson atBattelle Pacific Northwest Laboratories for the NRC[8]. These spectra were taken for numerous locationswithin operating and shut down nuclear power plants andrepresent the types of spectra to which individuals maybe exposed.

The specific spectra considered are as shownin Table 1.

Table 1.

#1 Low Energy

#2 High Energy Spectrum

#3 Medium Energy Spectrum

133Xe Plume.

Turbine floor 272 nearviewing gallery.Operating BWR, Location A,Site 0.

Reactor Vessel SamplingStation. Operating RWR,Location J, Site 0.

#4 Instrument Comparison "Unity" spectrum.

In addition to the reactor facility spectra,a low-energy spectrum characteristic of a 13 Xe plumefrom a reactor was considered as being representativeof the type of low energy spectrum which may beencountered in normal gamma ray survey work. Finally,

0018-9499/85/0002-0918$01.00 ©1985 IEEE

918

919

a 'unity" spectrum was used to evaluate instrumenteffects independent of spectral bias.

The response of the different instruments tothe different spectra for the different operationalquantities can be performed in a straightforwardfashion using the expression:

Relatve Repons Instrument ReadingRelative Response True Dose Rate

ni AEi Si Ri di

ni

AEi Si CXi Kiiw=1

where

AE. = energy increment

Si = spectral weighting over AE

RicxiJi

relative instrument response at ERemis conversion factor at ERoentgen 1

Measured units per unit exposurerate at E

Ki = Radiation exposure rate inRoentgens at E.Unit Flux I

These calculations are done by firstintegrating the spectrally weighted instrumentresponse. Then the actual spectrally weightedoperational quantity of interest (such as effectivedose at 1 mg/cm2) is calculated. The ratio of thesetwo quantities then gives the accuracy with which agiven instrument is able to measure a given operationalquantity when used in a given radiation fieldcorresponding to one of the "standard" energyspectra. Values greater than unity would indicate thatthe instrument is over responding, while values lessthan unity would indicate that the instrument is underestimating the operational quantity of interest.

Instrument Comparisons

Estimates of instrument measurement accuracyand an assessment of the errors in measuring thedifferent operational quantities have been performedfor the following situations:

Case a. Instrument variations only for evaluatingexposure rate without spectral weighting orcorrection for operational quantities.

Case b. Instrument variations only for evaluatingexposure rate but weighted by the instrumentresponse to various spectra.

Case c. Errors introduced when measurements are madefor the different spectra if the assumption ismade that dose exposure (rems - R). Theoperational quantity desired is dose at 1000mg/cm2.

Case d. Instrument variations for evaluating dose at1000 mg/cm2 for the different spectralconversion factors are applied to the readingsand this is the operational quantity that isdesired.

Case e. Errors introduced in evaluating the dose rateunder different spectral conditions by usingthe conversion factors for effective dose at1000 mg/cm2 when the radiation isomnidirectional and the effective dose rate iscalculated based on the ICRP26 weighting ofthe different organ doses.

Case f. Errors introduced in evaluating the dose rateunder different spectral conditions if theassumption is made that dose - exposure (rems

R) when the radiation is omnidirectional andthe effective dose rate is calculated based onthe ICRP26 weighting of the different organdoses.

Case g. Errors introduced in evaluating the dose rateunder different spectral conditions by usingthe conversion factors for peak dose at 7mg/cm2 (which is close to the dose equivalentindex) when, in fact, the proper operationalquantity is dose at 1000 mg/cm2.

Based on these cases, 99 instruments werecompared. Instrument types included those based onscintillators, ion chambers, proportional counters,scintillators, and semiconductor detectors. Inaddition, three thermoluminescent dosimeters (TLD-LiF)were evaluated for comparison.

Instrumental differences were first comparedbased on the "Case a" criteria above and the resultsare shown in Table 2.

Table 2.

Summary of Instrument Response Accuracy for DifferentSurvey Instruments Assuming a Unity Energy Spectrum andRem R (Case a).

NuInstrument Type Ins

GM CountersIon ChambersProportional CountersScintillatorsSemi conductorsTLDs (LiF)

Relativeimber of Response Accuracytruments Low Hign Average

44452323

0.312 1.46 1.03 ± 0.190.689 1.58 1.07 i 0.25-- -- 1.050.3020.980.99

0.4000.991.35

As can be seen, all of the survey metersexcept those based on scintillators had an averageresponse which ranged between 0.99 and 1.07. In thecase of GM counters and ion chambers, sufficient dataexists to determine averages and standard deviations asindicated. On the basis of the existing data, theredoes not appear to be a statistically significantdifference in the accuracy of making dose ratemeasurements by using either an instrument based on aGM counter or one based on an ion chamber. Inpractice, there is a greater difference betweendifferent instruments within a given category thanbetween instrument categories.

The lowest instrument response in the GMcounter category was a Phillips X-Ray Monitor which hasa maximum advertised operating energy of 80 keV andperhaps should not have been considered in thiscomparison. The next lowest instrument response was aWallac Automatic Alarm Dosimeter--RAD 21 which had anaccuracy of 0.797. This compares favorably with thelowest responding ion chamber instrument.

920

Many of the GM counter and ion chamberinstruments have a movable shield which is used fordiscrimination against beta rays. An analysis has beenmade of the effect of the movable shield or cap on theoverall response of instruments for which measurementshave been made. Table 3 shows the results of thesedifferences. As can be seen, there is littledifference between instrument response with or withoutthe shield for the unity spectrum, though the GMcounter-based instrument response was about 5 percenthigher without the shield.

Table 3.

Effect of Beta Shield on Instrument Accuracy forDifferent Spectral Inputs Assuming Rem R (Case a).

Tabl e 4.

Instrument Accuracy for the Measurement of DifferentRanges of Spectral Energy for the Different InstrumentCategories Assuming Rem R (Case b).

Instrument

GM Counter

Low Energy(Spectrum 1)

High Energy(Spectrum 2)

Relative Response AccuracyLow High Average

0.339

0.301

8.05 1.99 ± 1.67

1.49 1.06 ± 0.19

Instrument

GM Counters

Shield OpenShield Closed

Ion Chambers

Shield OpenShield Closed

Relative Response AccuracyUnity Low Energy High Energy

Spect rum Spectrum #1 Spectrum #2

1.09 t 0.21 2.53 i 1.23 1.14 ± 0.221.04 ± 0.14 1.07 ± 0.37 1.05 ± 0.14

1.06 i 0.29 1.62 i 0.97 1.09 ± 0.381.07 ± 0.20 1.16 ± 0.60 1.12 ± 0.23

Medium Energy(Spectrum 3)

Ion Chamber


High Energy(Spect rum 2)


0.298

0.409

0.720

0.740

1.42

3.69 1.19 ± 0.66

1.82 1.12 ± 0.39

1.82

The effect of the shield becomes morepronounced when the energy spectrum being monitoredfavors either the low- or high-energy spectrum. As canbe seen, GM counters with the shield open have asignificantly higher response (2.53 vs. 1.07) at lowerenergies. Ion chambers also have a somewhat enhancedresponse (1.62 vs. 1.16) but less than the GMcounters. At high energies, the difference is muchless (1.14 vs. 1.05) for the GM counters and isactually reversed (1.09 vs. 1.12) for the ion chambers,presumably due to reduced build-up in the ion chamberwithout the shield in place.

As can be seen in Table 3, spectral weightingcan have a significant influence on the response of thedifferent instruments. For this reason, instrumentaccuracies were, evaluated for the three differentspectra described earlier in this report. Table 4shows the summary of the response variations for thedifferent instrument types, again assuming thatrem R.

As expected, the most significant variationswere for the low-energy spectrum where window andenergy cutoff effects are most pronounced. Byinstrument type, the ion chambers showed the mostuniform overall response under different spectralconditions. The GM counters showed the characteristicover response at low energies which might be expected.

The scintillator-based instruments showed thelargest range of response, especially at lower energieswhere one instrument over responds by 68.4X for the133Xe plume spectrum. UJnder response at high energies,which might be expected to be significant for thescintillators based on their monoenergetic response athigher energies, was not significant. This is due tothe fact that the high energy spectrum used hassignificant low-energy components which largelybalances out the instrument response.

Up until now, we have been evaluatinginstruments on the basis that rems - R in estimatingdose (rems) based on instrument readings which areusually in units of exposure (R). 10CFR20 [2] definesconversion factors which are energy dependent for

Proportional Counters




Sci nti 1 1 ators




Semi conductors



TLD



1.29

1.06

1.06

8.79

0.650

0.854

4.94

1.11

1.10

68.4

1.52

1.58

0.899 27.18

1.02

0.99

0.99

1.41

1.34

1.34

making corrections to instrument readings for thedetermination of dose, and we have evaluated responseusing those factors. In addition, we have consideredadditional conversion factors discussed in an earlierpublication [1] which may give an effective dose whichis more directly proportional to health risk effects.These different conversion factors were considered as

921

Cases c through g above, and the calculated instrumentresponses for each of these cases has been performed.

This data has been condensed into Table 5 forthe GM counters and ion chambers. The values indicatedshow the range between the lowest and the highestresponses for each instrument type and energyspectrum. Of these, Cases e and f show the largestvariations with the main discrepancy occurring at lowerenergies. This occurs since the conversion factor hasthe greatest difference, and instrument effects aremost pronounced in this energy range.

The actual variation between the differentmeasurement assumptions is somewhat clouded in Table 5,however, since a large number of instruments wasconsidered resulting in a significant range invalues. Table 6 shows the variations in response forthe Victoreen 471 whose energy response as measured atLLNL was found to be relatively flat with energy. Alsoshown in Table 6 are the response variations forrepresentative instruments from the other instrumentcategories. Note the particularly good response of thediamond detector and the relatively nonuniform responseof the scintillator.

Case c and Case d are the two most commonmeasurement modes of assuming that rem = R (Case c) orcorrecting using Cx for dose at 1 mg/cm2 (Case d) whenthe proper operational quantity is dose at 1 mg/cm2.As expected, the only significant difference occurs atlow energies, with the exception of the scintillatorwhich also shows a significant change at higherenergies due to its strong weighting toward a low-energy response. In general, instrument variations aremuch larger than the 3-4 percent change seen for mostspectra by making the corrections for dose at 1mg/cm2. Thus, if corrections are going to be used withany meaning, great care needs to be exercised ininstrument selection and calibration.

Case c considers the possibility that healthrisks may be proportional to the ICRP26 weightedconversion factors. If this is the case, then errorson the order of 3x at low energy and 1.6x at medium andhigh energy are being introduced by using the 10CFR20conversion factors for dose at 1 gm/cm2, and higherstill if the assumption that rem R is made. This isclearly a more significant factor than making the Cxcorrections for dose at 1 gm/cm2, but involves thedetermination of whether the ICRP26 weighted Cx factoris, in fact, the correct health risk factor.

It should be noted that the observedvariations in instrument response noted in Table 4 forthe different categories of instruments and even withininstrument categories significantly exceed the factorsinvolved in the different measurement assumptions.This further emphasizes that if care is not taken inthe selection and calibration of an instrument, thereis no point in worrying about which conversion factorto use.

Conclusion

The results of this study indicate severalfactors of importance to health physics measurements.These are:

b) More demanding applications, such as lowenergy survey work, are better served with an ionchamber instrument.

c) Instrument variations tend to faroutweigh correction factor effects for dose at 1gm/cm2 unless great care is taken in instrumentselection and calibration.

d) Selection of an appropriate conversionfactor to represent possible health risk effects canresult in variations in readings from 3x at lowenergies to 1.6x at higher energies. These alternatecorrection factors are not included in 10CFR20 atpresent.

e) The diamond detector had a responsecharacteristic comparable to a good ion chamber and mayhave some significant operational advantages.

f) GM counters operated without their betashield at low energies may show a significant over-response.

g) Scintillator-based instruments should notbe used in applications where accurate measurements areneeded due to their extremely strong variation inresponse with energy.

References

tl] Armantrout, G. A., "Health Physics Instruments:What are the current needs?" IEEE Trans. on Nucl.Sci., Vol. NS-30, No. 1, p. 521 (1983).

[2] Code of Federal Regulations 10, Part 20, NationalArchives of the United States (1983).

[3] Portable Radiation Survey Meters, LLNL Health andSafety Manual M-010, Supplement 33.22, LawrenceLivermore National Laboratory, Livermore, CA(1976).

[4] Instrument Evaluations No. 1 through No. 30, NRPB-lEl through NRPB-1E30, National RadiologicalProtection Board, Harwell, Didcot, U.K. (1974through 1983).

[5] Wolf, M. A., et al., "Use of Cadmium TellurideDetector in a New Tiny Personal RadiationChirper," IEEE Trans. on Nucl. Sci., Vol. NS-26,No. 1 (1979).

[6] Planskay, B., "Evaluation of Diamond RadiationDosimeters," Phys. Med. Biol., Vol. 25, No. 3, pp.519-532 (1980).

[7] Krohn, B. J., "The Photon Energy Responses ofSeveral Commercial Ionization Chambers, GeigerCounters, and Thermolumi nescent Detectors," LANLReport LA-4052, Los Alamos National Laboratory(1969).

[8] Roberson, P. L., "Measurement of Low- and High-Energy Photon Spectrum at Commercial NuclearSites," 4th Quarterly Progress Report, NRC FIN No.B2419-1, Battelle Pacific Northwest Laboratory,Richland, WA (1983).

a) There is not a significant differencebetween GM counter-based instruments and ion chamber-based instruments for general purpose survey work whenapplied to real world spectra, provided the instrumentsare properly calibrated and maintained.

922

Tabl e 5.

Measurement Error Ranges for GM Counters and Ion Chambers in Response to Three Different SpectraMeasurement Assumptions (Cases c-g).

for Different

Instrument

Relative Response Accuracy - Range of ValuesCase c Case d Case e

Meas. + rem - R Cx @ 1 gm/cm2 Cx @ 1 gm/cm2Cx #ICRP26

Dose + Cx @ 1 gm/cm2 Cx @ 1 gm/cm2 Weighted

GM CountersLow Energy (Spectrum 1)High Energy (Spectrum 2)Medium Energy (Spectrum 3)

Ion Chambers

Low Energy (Spectrum 1)High Energy (Spectrum 2)Medium Energy (Spectrum 3)

0.2700.2900.290

0.3300.6930.710

- 6.43- 1.43- 1.36

- 2.95- 1.75- 1.75

0.381 - 8.980.308 - 1.490.296 - 1.43

0.461 - 4.050.726 - 1.830.746 - 1.83

1.140 -0.487 -0.480 -

27.902.362.31

1.38 - 12.161.15 - 2.881.21 - 2.96

Case frem - R

Cx #ICRP26Wei ghted

0.812 -0.468 -0.461 -

Case g2

Cx @ 0.007 gm/cm2

Cx @ 1 gm/cm2

19.302.272.20

0.983 - 8.861.09 - 2.761.15 - 2.82

0.3910.3090.298

0.4730.7300.750

Tabl e 6.

Measurement Errors for SpecificAssumptions (Case c-g).

Instruments in Response to Three Different Spectra for Different Measurement

Instrument

Meas. +

Dose + C,

Victoreen 471 Ion ChamberLow-Energy (Spectrum 1)

High Energy (Spectrum 2)Medium Energy (Spectrum 3)

Eberline PRM-7 ScintillatorLow Energy (Spectrum 1)High Energy (Spectrum 2)Medium Energy (Spectrum 3)

Di amond DetectorLow Energy (Spectrum 1)High Energy (Spectrum 2)Medium Energy (Spectrum 3)

TLD (LiF)Low Energy (Spectrum 1)High Energy (Spectrum 2)Medium Energy (Spectrum 3)

Relative Response Accuracy - RangeCase c Case d Case erem = R Cx @ 1 gm/cm2 Cx @ 1 gm/cm2

Cx #ICRP26

@ 1 gm/cm2 Cx @ 1 gm/cm2 Weighted

0.8360.9630.960

7.020.6240.818

0.7180.9610.971

0.7870.9560.951

1.0651.0031.005

9.820.7370.963

0.9791.001.02

1.0360.9960.994

3.1981.5861.625

29.501.161.56

2.941.591.65

3.111.571.61

! of ValuesCa'se frem R

Cx #ICRP26Weighted

2.5091.5221.554

21.090.9861.32

0.2161.521.57

2.361.511.54

Case gmCx @ 0.007 gm/cm2

Cx @ 1 gm/cm2

1.1491.0051.006

10.140.7430.968

1.021.011.02

1.090.9970.995

- 9.29- 1.49- 1.43

- 4.21- 1.83- 1.83

Documents

Impact of Instrument Response Variations on Health Physics Measurements