Radon in the Living Environment, 98 RETROSPECTIVE ASSESSMENT OF HISTORIC RADON CONCENTRATIONS IN NORWEGIAN DWELLINGS BY MEASURING GLASS IMPLANTED Po-21 AN INTERNATIONAL FIELD INTERCOMPARISON Aleksandar Birovljev 1, Rolf Falk 2, Ciara Walsh 3, Francesca Bissolo 4, Flavio Trotti 4, James P. McLaughlin 3, Johan Paridaens 5, Hans Vanmarcke 5 and Anikken Heiberg 1 1 Norwegian Radiation Protection Authority, (NRPA) Østerås, Norway 2 Swedish Radiation Protection Institute, (SSI) Stockholm, Sweden 3 Physics Dept, University College Dublin, (UCD) Ireland 4 CRR-ULSS 2 Veneto region, (CRR) Verona, Italy 5 SCK-CEN, Boeretang 2, B-24 Mol, Belgium The first Norwegian study of historic radon concentrations in 17 dwellings in the high radon areas in Norway has been conducted as part of an international field intercomparison during 1998. The investigation is part of SINI (an acronym for Sweden, Italy, Norway and Ireland) international collaboration on retrospective radon measurements in several European countries having different climates and living conditions. The retrospective radon concentration is estimated via measurements of Po-21, the long-lived decay product of Rn-222 implanted in glass surfaces of objects like pictures, mirrors, cabinet-glass etc, the method called surface trap. Three different surface trap techniques to assess the implanted Po-21 activity and two different procedures to estimate retro radon from Po-21 data were used. The Po-21 and the retrospectively estimated radon results agree reasonably well over a wide range of concentrations. Historic radon concentrations were also estimated from analysis of a smaller number of volume trap samples (pieces of spongy materials), and the results compared to those from surface traps. The retro radon results correlate with contemporary radon results with correlation coefficient of.877. To evaluate uncertainty in Po-21 measurements due to varying position on the glass a study of spatial homogeneity of three sample glasses was conducted and variations between 12% and 18% were found. Key words: retrospective radon estimates, Po-21, surface traps, Norway, intercomparison INTRODUCTION Recent scientific efforts to reduce uncertainties in epidemiological studies which investigate relationship between radon and lung cancer, have focused on techniques that better estimate peoples historic exposure to radon and radon progeny. Although significant methodological advancements (McLaughlin, 1998) have been made during most recent years, a widespread field tests of these, so called retro detectors are still scarce. Two dominant retrospective techniques that have been developed during recent years are known as surface traps (Samuelsson et al. 1992, Falk et al. 1996) and volume traps (Oberstedt et al. 1996). Both techniques are based on measurements of Po-21, a long-lived Rn-222 progeny that accumulates in certain objects found in dwellings with radon exposure. In the surface trap technique the measurements of alpha activity from Po-21 atoms implanted by recoil in glass surfaces of objects like pictures or mirrors are conducted, whereas the volume trap technique is based on measurements of Po-21 deposited on inner surfaces of porous materials, such as spongy mattresses, cushions etc. The Po-21 in volume traps originates from radon, which has diffused into the porosity of the material leading to a simple relation: one atom of Rn-222 that decays in the volume trap produces one Po-21 atom. For the surface traps the relationship of Po-21 817
98 Radon in the Living Environment, concentration to the time-integrated airborne concentration of radon gas or its short-lived progenies is more complicated and usually a model such as Jacobi model (Jacobi 1972) is needed to obtain a meaningful estimate. In the present paper the results of measurements using different retrospective techniques in houses in two high radon rural areas in Norway are presented. The measurements have also functioned as a field intercomparison of surface trap retro detectors belonging to the four participating laboratories. For the houses where volume trap samples were available the retrospectively estimated radon concentration was compared with volume trap results for the same houses. The investigation is part of SINI international collaboration on retrospective radon measurements in several European countries having different climates and living conditions. MATERIALS AND METHODS A surface trap retro-detector should fulfill two main criteria: (a) be able to measure Po-21 activity (with alpha energy of 5.3 MeV) which is implanted in the up to 1 nm thick layer of the glass surface, (b) the alpha background activity in the glass, which may vary from sample to sample, should be discriminated. Several different principles may be used to achieve these two requirements. The participating laboratories use three different surface trap techniques to measure Po-21 surface activity. An overview of the techniques used is shown in table 1. All the different detector types have the same basic construction. They consist of pieces of plastic track-etch material (SSNTD) which are fixed to the glass surface using contact paper. The differences among the detectors are in the techniques for control of the low energy background alpha activity from the glass. The technique used by UCD and SSI is based on the combination of a CR-39 and a LR115 detector, while CRR uses alpha spectroscopy on CR-39, and these methods have been described elsewhere (Falk et al. 1996; McLaughlin, 1998; Trotti et al., 1997). NRPA uses two 4 cm 2 pieces of CR-39, one with 23 µm Mylar absorber sandwiched between CR-39 and the glass. The absorber reduces the energy of the alpha particle by 3.5 MeV and is used to discriminate most of the low energy background. The signal from the CR-39 with absorber is used as a measure of Po-21 concentration, whereas the second CR-39 piece is used to assess possible anomalies with respect to background. The assembling of the detector requires very low radon environment and special handling when detector is mounted to the glass surface to avoid electrostatic attraction of radon progeny. The retro detectors from all four laboratories have been calibrated at SSI in Sweden by exposure on glasses of known activity of implanted Po-21. The calibration glasses have been measured earlier using high precision active devices at the University of Lund (Samuelsson et al. 1999) and at BfS in Germany. The calibration plot for NRPA relating CR-39 track density to implanted Po-21 activity is shown in figure 1. In table 1 it is also shown which methods each participating laboratory uses to estimate radon concentration in retrospective from the Po-21 data. To estimate the radon concentration in the past from the measured 21 Po, a model developed by UCD is used. This algorithm is based on room models describing the behaviour of radon and its progeny in rooms (Jacobi 1972), implantation 818
Radon in the Living Environment, 98 studies (Cornelis et al 199) and a recursive algorithm (Nyblom et al. 1992). This model is sensitive to the values of a number of key room parameters. These principally are the characteristics of the room aerosols, the room surface to volume ratio and the ventilation rate which are chosen on the basis of a completed questionnaire. The uncertainties involved in this approach are recognised but by using the likely full range of room parameters, it is possible to estimate the range of values within which the mean historical radon level would lie. A sensitivity analysis of the model used here has been reported, and the aerosol concentration was found to be the most significant parameter (Walsh et al. 1999). FIELD WORK LOCATIONS AND PROCEDURES In total 17 detached single family houses in two rural areas have been chosen, 14 at Kinsarvik, Western Norway, and 3 near the town of Fredrikstad, SouthEastern Norway. Both locations have earlier been identified as ellevated radon areas. The fieldwork was conducted by NRPA during spring summer 1998. The houses were selected to provide a wide variation of radon concentrations known from earlier radon gas surveys. A standardized procedure for selection of glass objects in the houses based on earlier SSI and UCD experience was applied (McLaughlin, 1998). First choices are usually glasses found on photos or pictures that are easier to date. Also glasses of other types of objects such as mirrors, doors between rooms and clocks etc. have been used. Window glasses are usually avoided as the airflow pattern and the deposition of progenies near windows is not typical, and also due to UV influence on the detector materials. Only one glass per house was used in the SINI intercomparison. Area of the glass was washed using water and washing up liquid detergent, before the four detectors were attached to the glass surface. Detectors were attached in the same region of the glass, very close to each other. A photograph was taken of the part of the room where detectors were placed. The measurement period was between 2 and 3 months, from April to June 1998. After the exposure the detectors were secured from further exposure by attachment to clean paper or plastic surfaces and sent to the participating laboratories for etching and analysis. Pieces of spongy mattresses (volume traps) were collected in 7 houses. RESULTS AND DISCUSSION The Po-21 results of SINI field-intercomparison are shown in figure 2a and 2b, where the measured implanted Po-21 activity (y-axis) is plotted Vs the identification code of the houses (xaxis). The results have similar overall trend. The mean relative variation (half difference between maximum and minimum implanted Po activity divided by the mean for each house) is approximately 3 %. The mean Po-21 concentration calculated for all the houses but separately for each laboratory is shown in table 2. It can be noticed that in spite of the fact that all laboratories participated in the same intercalibration, there appear to be systematic differences. NRPA and UCD measured on average higher results than SSI and CRR. The reason for the elevated Po-21 values from UCD was essentially a "signal to noise" problem arising from the fact that the detectors, prior to field use, had been stored for approximately seven months at NRPA. During this time the background alpha track densities on the plastic detectors arising from the intrinsic alpha activity of cover paper in contact 819
98 Radon in the Living Environment, with the detectors has grown to levels comparable to or even greater than the subsequent signal track density from exposure of the detectors to glass objects in the field. This is not normally a problem as the time interval between detector assembly and use in the field is usually only a couple of weeks and the low background acquired during this period is easily corrected for by means of control detectors. A positive outcome of this experience has been an improvement in UCD background control by the choice of new cover paper material of very low alpha activity which is now being used in this work. Due to this background problem there is no entry made for UCD values in Table 2. The average value for the UCD background affected Po-21 values was 57.55 Bq/m 2. The background problem for UCD data should be borne in mind when reading figures 2a and 2b. The Po-21 results were used to calculate the historic average radon concentrations in the houses, which are estimated over the number of years the glass object was exposed to radon. The laboratories used two different approaches as depicted in table 1. The other input parameters of the modified Jacobi model were roughly estimated by observation of the room where measurements were conducted and classifying in one of the three categories shown in table 3. Other parameters such as attachment coefficient (.47 cm 3 /h), attached deposition velocity (.5 m/s) and unattached deposition velocity (5 m/s) were taken from the literature. The results of the calculations are shown in figure 3. The mean relative variation (half difference between maximum and minimum estimated Rn concentration divided by the mean for each house) is approximately 6 %. This very high variation is due to different calculations of radon estimates from Po-21 data. The circles in figure 3 are the results of volume traps (pieces of spongy mattresses) found in some of the houses. The volume traps were analyzed by SCK-CEN. The volume trap samples were not always found in the same rooms as the surface trap samples, and they were sometimes of different age compared to the glass samples. Thus the differences in estimated radon concentration from surface and volume traps were expected. The sample from house no. 5 was a piece of mattress that was stored in the attic of the house for a number of years, and hence the much lower volume trap value for this house. Other six volume trap results compare reasonably well with the surface trap results. Contemporary radon measurements in the same houses where retro measurements were done have been conducted during last 2-3 years. These results are compared with the retrospectively estimated radon results in figure 4. The Pearson s correlation coefficient of.877 was obtained for the statistically significant result at.1 level (one tailed probability distribution). Comparing results with the Y = X line in figure 4 it can be seen that most contemporary radon results are higher than the retro results for the same houses. The arithmetic mean of contemporary results is 373 Bq/m 3 and of retro results is 2417 Bq/m 3. This means that radon concentration in most of the houses has increased compared to the historic average levels. There are, however a few houses where opposite trend is observed. It should be noted that considerable uncertainties are associated with both retro and contemporary measurements. The houses in one of the locations (Kinsarvik) where 14 houses were measured are built on an endmoraine characterized by extreme porosity and permeability. Daily and weekly changes of radon concentrations indoors are highly dependent on climatic conditions and can be extreme. In one of the houses 6 Bq/m 3 radon concentration was measured on one day while the next day the concentration was only 2 Bq/m 3. As the concentrations in many of the houses were in periods 82
Radon in the Living Environment, 98 extremely high the track-etch detectors used in earlier surveys were easily saturated with tracks, and this increased the uncertainty of the measured radon gas concentration. Due to atypical seasonal variations of radon concentrations indoors it was difficult to predict year average radon concentrations from a few months measurements with any certainty. Year to year variations of average radon concentrations are also expected to be considerable. It is plausible that the uncertainty in predicting indoor radon year average in such moraine or other uncompacted earth geological structures is comparable or even grater than the uncertainties associated with retro measurements or at least volume trap measurements. If future and more detailed studies prove this to be true all the earlier problems of assessing the year average radon concentrations in houses built on highly permeable ground could be avoided by a single measurement using retro detectors. With improvements of the retro techniques, it is possible that retro detectors will in the future find new use as a quicker, if not more reliable method to assess year average exposures to radon and progenies in houses built on uncompacted earth (moraines, eskers,) where extreme and atypical variations of radon concentrations are expected. The uncertainties involved in the surface trap retro techniques are related not only to the precision of Po-21 measurements, but also to many physical processes like the deposition of airborne progenies, implantation and removal mechanisms of the implanted Po-21 atoms. One of the questions that might be posed is an error that results by placing the retro detector at one position on the glass compared to some other position. We have therefore empirically studied the spatial homogeneity of the implanted Po-21 in the glass, to evaluate this contribution to the overall uncertainty. Three large picture glasses were selected from three different houses. All three glasses hung on the wall without any objects in front of them. The Po-21 activity implanted in the three glasses was mapped by placing several hundred 4 cm 2 CR-39 track-etch detectors at a distance of 3 cm from each others center in two perpendicular dimensions. The track-etch detectors were analyzed and three-dimensional graphs produced. The results from two of the glasses are shown in figure 5. The relative variation of track density is equal to the relative variation of implanted Po-21, and was quantified by one standard deviation divided by the mean. For the three glasses studied this fraction was 12% (figure 5cd), 14% and 18% (figure 5ab). However, it can be seen from figure 5a that a maximal error of up to 5% can result by measuring at positions on the glass with minimum and maximum activity. The Po-21 activity tends to be higher at edges and corners while the part in the middle of the glass is usually more stable. The origin of this spatial variation of the implanted Po- 21 activity is not quite clear. The uneven deposition of the short lived radon progeny due to the turbulent airflow in the vicinity of the glasses seems to be the most likely effect. CONCLUSIONS The SINI Norway field intercomparison of surface trap retrospective techniques has produced reasonably good performances of the participating laboratories. Average half relative total variation among the laboratories for Po-21 measurements is 3 %, whereas for radon measurements it is around 6%. There is clearly room for improvements, especially in the procedure to estimate radon concentration from the Po-21 data. The contemporary radon concentrations in the dwellings correlates with retrospectively estimated values with the correlation coefficient of.877 (figure 4). The results indicate that retro techniques may be suitable for quicker determination of a year 821
98 Radon in the Living Environment, average exposure to radon and its progenies in houses built on uncompacted earth (moraines, eskers etc) where large variations of radon concentration is expected. Homogeneity tests on three sample glasses show that on average differences of between 12% and 18% in the measured implanted Po- 21 may be expected as the position of the detectors on the glass is varied. It appears that the middle part of the glass is more homogeneous than areas close to edges and corners. ACKNOWLEDGEMENTS The authors wish to acknowledge the generous cooperation of the people of the communities of Kinsarvik and Kraakeroey in the fieldwork described here. The assistance of the European Communities through research contract FI4P-CT95-25 is also gratefully acknowledged. REFERENCES [1] Cornelis, J., Landsheere, C., Poffijn, A and Vanmarcke, H. Experimental and Theoretical Study of the Fraction of Po-21 adsorbed in Glass. In:: Fredrick T. Cross, editor. 29 th Hanford Symposium on Health and the Environment, Indoor Radon and Lung Cancer: Reality or Myth? 15-19 October 199. Richland, Washington, US; pp11-113. [2] Falk R, Mellander H, Nyblom L, Østergren I. Retrospective assessment of radon exposure by measurements of Po-21 embedded in surfaces using alpha track detector technique. Environ. Int 1996; 22(S1) [3] Jacobi W. Activity and potential alpha energy of Rn-222 and Rn-22 daughters in different air atmospheres. Health Phys. 1972; 22:441-45 [4] Lively RS, Steck DJ. Long term radon concentrations estimated from 21 Po embedded in glass. Health Physics 1993; 64(5):485-49. [5] McLaughlin JP. The application of techniques to assess radon exposure retrospectively. Radiat. Prot. Dosim. 1998; 78(1): 1-6. [6] Nyblom, L. and Samuelsson, C. The determination of the activity of serially transforming radionuclides by a recursive technique. Rad. Prot. Dosim. 1992; 45:1 [7] Oberstedt S, Vanmarcke H. Volume traps A new retrospective radon monitor. Health Phys. 1996; 7(2):223-226 [8] Samuelsson C, Johansson L, Wolff M. 21 Po as a tracer for radon in dwellings. Radiat. Prot. Dosim. 1992;45(1/4): 73-75. [9] Samuelsson C, Falk R, Roos B. Alpha particle emission from reference glass surfaces implanted with 21 Po. Proc. of Radon in the Living Environment Workshop, April 1999; Athens, Greece (in this volume). [1] Trotti F, Lanciai M, Mozzo P, Panepinto V, Poli S, Predicatori F, Righetti F, Tacconi A, Tanferi A. Improvements in radon retrospective assessments based on analysis of 21 Po embedded in glasses. In: Proc. IRPA Symp. on Radiation Protection, Prague 8-12 September 1997. [11] Walsh C, McLaughlin J.P. Correlation of 21 Po implanted in glass with radon gas exposure: Sensitivity analysis of critical parameters using a Monte-Carlo approach. Proc. of Radon in the Living Environment Workshop, April 1999; Athens, Greece (in this volume). 822
Radon in the Living Environment, 98 Table 1: Overview of techniques used to measure surface activity of Po-21, the laboratory technique (etching and counting of tracks), and the method of retrospective estimation of Rn-222. UCD SSI CRR NRPA Po-21 measurement Laboratory technique Estimation of Rn concentration Po-21 signal from Chemical etching and "Jacobi" room model with combination of LR-115 tracks counting, manual modulation of parameter and CR-39 detectors or by image analyzer values Po-21 signal from combination of LR-115 and CR-39 detectors Spectroscopy on tracks from a single CR-39 detector Po-21 signal from combination of two CR- 39 detectors (one with an absorber) Chemical etching and tracks counting by image analyzer Chemical etching and tracks analysis by image analyzer Chemical etching and tracks counting by image analyzer Experimental calibration for Po-21 concentration vs Rn-222 exposure "Jacobi" room model with modulation of parameter values "Jacobi" room model with modulation of parameter values Table 2: Statistics for individual laboratories and the total mean. *For UCD values see text Laboratory Average Po results (Bq/m 2 ) Average Rn results (Bq/m 3 ) NRPA 53.4 2265 SSI 41.6 1863 CRR 41.8 1822 *Total: 45.36 1983 Table 3: Model parameter values were classified in three different room categories: low, medium and high. Parameter Low Medium High Surface/volume (m -1 ) 8 5 3 Ventilation rate (h -1 ).2.5 1.5 Aerosol conc. (m -3 ) 25 25 15 823
98 Radon in the Living Environment, 45 Po-21 conc. exposure(hbq/m 2 ) 4 35 3 25 2 15 1 5-5 Calibration plot Linear fit R=.988-1 -2-1 1 2 3 4 5 6 7 8 9 1 Track density (# tracks/cm 2 ) Figure 1: Calibration plot for NRPA detectors. Detectors were exposed to 5 different cumulative exposures on glasses of known implanted Po-21 activity. Data were fitted by a straight line using least square fit. 824
Radon in the Living Environment, 98 4 35 SINI Norway intercomparison (a) 45 4 (b) Average Po-21 conc. (Bq/m 2 ) 3 25 2 15 1 5 Po-21 (Bq/m2) 35 3 25 2 15 1 NRPA SSI UCD CRR 5-5 2 4 6 8 1 12 14 16 18 2 22 2 5 8 Location ID 11 14 19 Location ID Figure 2: (a) SINI field intercomparison results of glass implanted Po-21 measurements in Norwegian houses. Error bars are the total range of results for each house (b) In this plot the results of each laboratory are denoted with a different symbol. 825
98 Radon in the Living Environment, 14 Retrospectively estimated radon concentration (Bq/m 3 ) 12 1 8 6 4 2 SURFACE TRAPS VOLUME TRAPS -2 2 4 6 8 1 12 14 16 18 2 22 Location ID Figure 3: (a) SINI field intercomparison results of retrospectively estimated radon concentration in 17 Norwegian houses. Error bars are the total range of results for each house. The circles represent retrospective results of volume traps. Contemporary radon conc. (Bq/m 3 ) 25 2 15 1 5 Pearson's correlation coefficient =.877. Significant at.1 level Y = X -2 2 4 6 8 1 12 14 Retrospectively estimated radon conc. (Bq/m 3 ) Figure 4: Correlation between SINI field intercomparison mean retrospectively estimated radon concentration vs contemporary radon concentration in 17 Norwegian houses. The correlation is statistically significant at.1 level (1-tailed probability distribution) with Pearson s coefficient of.877. 826
Radon in the Living Environment, 98 (a) Track density (#track/cm 2 ) 4 3 2 1 1 2 X (cm) 3 4 5 1 2 3 Y (cm) 4 Frequency count 4 3 2 1 (b) 2 25 3 35 4 Track density 7 (c) 6 1 (d) Track density (# tracks/cm 2 ) 5 4 3 2 1 1 2 3 4 5 6 7 8 X (cm) 5 4 3 2 1 Y (cm) Frequency count 8 6 4 2 35 4 45 5 55 6 65 Track density Figure 5: Results of mapping of two large picture glasses. In 3D plots X and Y are spatial coordinates and Z is the track density (a) Glass of average implanted Po-21 activity of 92 Bq/m 2 which was mapped using 216 CR-39 track-etch detectors; (b) Histogram showing variation of track density values measured on glass (a); (c) Glass of average implanted Po-21 activity of 582 Bq/m 2 which was mapped using 368 pieces of CR-39 track-etch detectors; (b) Histogram showing variation of track density values measured on glass (c). 827
98 Radon in the Living Environment, 828