150 likes | 340 Views
Sources of error in etched track measurements. – or – why aren’t my results as good as yours when we’re using the same detectors? Fero Ibrahimi. Accuracy & Precision Passive radon detectors. Assessing Accuracy Reference value / Calibration - Radon Chamber / Box
E N D
Sources of error in etched track measurements – or – why aren’t my results as good as yours whenwe’re using the same detectors? Fero Ibrahimi
Accuracy & PrecisionPassive radon detectors • Assessing Accuracy Reference value / Calibration - Radon Chamber / Box Chamber / Box Instrumentation – itself calibrated / intercomparison ! Assessing Precision Standard error of the mean (SEM) (s / n) Standard deviation (SD) Assessing Both Internal control - Blind test / dummy customer External control – Certification Assurance / Proficiency Test / Validation Scheme External control - Intercomparison Exercises
IntercomparisonsNRPB / HPA Annual Exercise • Since 1997 • 40 passive detectors • 10 x transits – subtracted from exposures • 10 x ‘low’ radon exposure ~ 0.1 - 0.2 MBq m-3 h • 10 x ‘UK action level’ exposure ~ 0.2 – 1.0 MBq m-3 h • 10 x ‘high’ exposure ~ 1.0 – 2.0 MBq m-3 h • 3 different radon exposures & equilibrium factors (F) • For each exposure set • (Net) Absolute % Difference • % Standard Deviation • For all 3 exposure sets • Mean % Difference • Mean % Standard Deviation • Sum • Rank results • Grade ‘A’ < 10%
Sources of Measurement Uncertainty • Radon calibration reference value • Radon-222 source 3.1% at 2 sigma (95%) Confidence Level - PTB • HPA Radon Chamber - minimum 5.3% • Laboratory • Etching equipment • Counting system - Track Recognition, Focus, Scratches • Track overlap – calibration curve correction • Ageing / Fading effects • Seasonal / temperature corrections (Miles, 2001)
Other sources of error • Passive detectors Diffusion cups / casings – Rn-220 (Tn) Etched track material / polymer: • chemicals – monomer, initiator, plasticiser • cure cycle variation • Etching chemicals • Laboratory • Personnel
Track overlapCalibration curve corrections Counting whole etched tracks Counting foreground pixels (px)
Ageing / Fading EffectsHardcastle & Miles (1996) Combined ageing & fading correction factor = 0.0007M2 + 0.0142M + 0.9528
Why aren’t your results as good as somebody else’s when you’re using the same detectors? • What Quality Assurance checks are you doing? • Personnel – adequate + continued training / support • Radon Chamber / Box instrumentation calibration / intercomparison • Detector calibration - material sensitivity & background – HPA each sheet • Etch System – HPA every time • Count System– HPA every time • Track overlap – calibration linearity • Ageing & Fading effects • Seasonal / temperature effects on annual average concentration
Why aren’t your results as good as somebody else’s when you’re using the same detectors?continued • How often should you assess your measurement system? • Minimum: internal blind test – every 6 months ? • Better: internal blind test – every batch of etch track material • Even better: external proficiency / intercomparison test – 3 diff exposures • Best: all of the above! • Any Questions / Comments ?