'Track etch' or 'plastic' detectors use a small strip of transparent plastic material to record exposure to radon. This material has a regular crystal structure and upon manufacture has few 'defects' in its atomic lattice.

A property of some materials is that they are resistant to chemical attack by selected reagents, but if attack does occur it may start preferentially at an imperfection. This can be a grain boundary (in a metal), a small region of abnormal composition (in an alloy) or simply somewhere where the regular lattice has been compromised.

In common with many dielectric materials, the plastic strips used for radon measurement can be damaged by alpha particles emitted by radon and its daughters. The materials are selected to have a high sensitivity, but they can exhibit high background counts, and even marked differences between one side of a strip and another.

Damage from radon is invisible when it occurs but if material that has been exposed is treated with a caustic solution it is attacked preferentially at the sites of the alpha particle damage. Each imperfection then shows up under a microscope as a small spot. Counting the spots can give a reliable indication of the average radon level to which the strip has been exposed.

At very high radon exposures, the spots may be so numerous that they start to overlap, but statistics and other ingenuity may be applied to determine the correct result. Thus, plastic detectors may be used over a wide range of radon concentration. Identical strips may be used in rooms at 20 Bq/m³ as in those at 20,000 Bq/m³.

However in real life, room air contains both radon and radon daughters and because the relation between their activity concentrations is not fixed, detector material exposed to free room air cannot be used to determine the average radon concentration. This problem is overcome by enclosing the plastic strip in a container (often made of plastic also) and that is designed to admit only radon gas - and not radon daughters. Within the container, radon daughters will form, but their number will bear a fixed relation to the radon concentration in the container, and that in the room. The characteristics of these containers may depend on their shape, size and other factors, but once the relation is known between the average radon exposure and the number of spots that appear on the plastic strips, the device may be said to be calibrated.

These devices are widely used because they are cheap, robust and safe, and because they have such a good 'memory'. Each site of radon damage is preserved within the plastic, and the strip may be processed (chemically etched) months after exposure.

Sometimes, the plastic strips alone are 'calibrated' in test chambers, but different container designs may behave differently in a stable fixed environment than in a real room where there are thousands of fluctuations in pressure caused by wind, movement, and opening and closing of doors. If equilibrium by diffusion alone is assumed, the time constant of some types may be several hours.

Thus, even if only because of the different conditions of calibration and use, it would be wise for scientists to compare results from different types of detectors when they are used in buildings.

In order to confirm similar results from different makes of track etch detectors, the author undertook comparisons of many types within the BRE field trials, and in dozens of houses. Many householders were intrigued to know why three or four different 'radon pots' were variously placed on bookshelves or strung up in cellars. The reason for the work was explained to them quite freely: it was expected that all types would give broadly the same answers.

Thousands of track-etch detectors were used by the author during the winters of 1989/90 and 1990/91. Some were obtained from NRPB (and of several different types), some from TASL in Bristol (again different types), and some from the USA. Early in the work, one type of detector from the USA was rejected for further use after several wholly wrong results were proffered by the laboratory.

At this point, it should be made clear that problems with radon detectors have been widely reported in the USA for many years. No secrecy was found necessary outside of (in the very early days) not identifying the laboratories under scrutiny, so that initial problems of procedure and calibration would not damage reputations unnecessarily.

Also, and to its credit, the US EPA has been open about the problems that were found, as indeed they were bound to be within the Freedom of Information culture. One example of work in which comparisons were made between seven different types of charcoal detectors is that published in mid 1990 in Pennsylvania.

The aim was to test for accuracy and random errors, and with one exception the average result from each set of 15 detectors was within 20%, and with most being closer than 10% to the known average radon level in the houses.

However, one type was in error by over 300% and another showed variations between detectors in the same batch of over 25%. The work followed several other studies in which the accuracy of both charcoal canisters and track etch detectors had been questioned.

Amongst the reasons cited for undertaking the work were that most calibration tests were performed under laboratory conditions, which might not reflect conditions in a home. Similar concerns have been voiced in other published work.

In the UK, there are broadly two reasons why work undertaken for Government Departments is occasionally formally classified under the Official Secrets Act. These are:

1. that the work is or is connected with matters that could harm the security of the State were information to fall into enemy hands.

2. that disclosure of the information could be an embarrassment either to a Minister or other member of the Government, or to a senior Civil Servant. In this context, it would usually be viewed as the Government Department that could be embarrassed.

The contemporary history of radon measurement using track etch detectors in houses may appear to be fully documented. In the USA and in the UK (at NRPB) sophisticated radon test chambers can be used for calibration of detectors. Secondary devices can be used also. For some years, NRPB have operated a device called FRED - Fast Radon Exposure Device, the aim of which is to be able to calibrate batches of radon detectors in a few hours by exposing them to around 100,000 Bq/m³ of radon. The device is described in NRPB report R190. Other publications are also available and show the excellence of comparisons between track etch detectors and 'active' measurements (those taken using electronic equipment) and in some cases, between different types of track etch detectors.

Several conditions may be essential for full comparison of passive radon detectors in a field trial situation. These are:

1. that a number of detectors of each type must be kept unexposed in order that the background count can be determined,

2. that a group of each type must be exposed side by side to allow the mean and standard deviation of each type to be estimated and

3. that continuous measurement of the radon concentration in each room must be conducted in order to provide an absolute reference.

Other requirements may be considered essential also, including that the age of detectors should be taken into account by processing laboratories.

It is of course inherent within the basic assumptions for long term averaging of radon concentration by track etch detectors that they should respond reasonably linearly to airborne radon concentration - because the variation of radon concentration in buildings can vary over orders of magnitude.

This is quite distinct from concern about any non-linear response to integrated concentration, because this can be more a matter of assessment of overlapping tracks by the scanning equipment.

To take into account known aging behaviour of track etch detectors laboratories need to know their age and time of exposure. Many track etch detectors are exposed for one month only whereas others are exposed for six or seven months.

NRPB have recorded degradation of sensitivity of 18% over six months in test houses, but (apparently) not outdoors. Temperature may be a determining factor, and with warmer houses perhaps yielding lower results. An overall correction factor of 9% would be the 'best guess' for 18% degradation over the monitoring period. It is usual to assume a linear decay.

However, there are no uniquely correct conditions for a successful field comparison. Conditions may be chosen to meet the trial objectives.

Rigorous calibration procedures.

Full comparison of detectors would include assessment of background, calibration and linearity, as well as reproducibility. In these cases, active measurements might be used and background counts could be taken from non-exposed plastic. Each side of each individual sheet of plastic from which strips were cut could be assessed for background count. This is usually in the range 0.1 to 0.5 tracks/mm².

The calibration of detector strips is also a simple matter and related to the number of tracks that are detected per unit area per unit of radon exposure. Plastic strip detectors are inherently linear in response over many orders of magnitude as the latent tracks do not normally interact with each other.

Assessment of the standard deviation would be essential for some studies. Standard deviation is a basic statistical concept and may be derived for any group of detectors of the same type by exposing many of them side by side under laboratory or real life conditions and determining the spread of results. The origin of any scatter would depend upon the integrated radon exposure.

At low values it might be determined by the number of tracks per unit area. At high exposures, problems of overlapping tracks and interaction may occur. However, any apparent errors might be as much due to the scanning equipment as to the detectors themselves.

At very low exposures, say below 20 Bq/m³ for three months, there will only be a small number of tracks on the plastic, and uncertainty due to Poisson statistics may be 10% or greater. The sensitivity is generally lower than 10 tracks per mm- for a years' exposure at 20 Bq/m³. Thus detectors used for 3 months, as is usual, may exhibit fewer than 2 tracks per mm- at typical radon levels, against a background of perhaps 0.5 tracks per mm². This is why track etch detectors cannot be used reliably for short period measurements at low radon levels. Other types do not suffer these problems (see Section 8).

However, knowing the background and calibration characteristics from laboratory data, a detector used in a building may be processed to determine the average radon level. For each individual measurement there will be a random but unknown error. This is distinct from systematic error between different batches or types of detectors. Systematic error can be investigated by simple comparisons of detectors.

Simple comparison tests.

In these tests, building owners (usually householders) would be asked to place radon detectors in batches within rooms. Typically, three or four different detectors would be placed on a book case, bedside table or kitchen cabinet, and (as is usual) well away from direct sunlight or draughts from doors or windows.

If it were desired to reduce the effect of random error then multiple units of each type could be supplied to each householder. This is not done as a matter of course, and it is inescapable that no number of detectors of a single type can possibly be used to detect or remove systematic error if this is present within one or more batches of detectors, or within some aspect of the processing procedures.

The appeal of track etch detectors is their simplicity of use and low cost. Indeed, it is essential for a domestic measurement programme that householders should not have to concern themselves with any aspect of calibration.

All the householder (or research scientist) need do is to place the detectors in a room, leave them for the specified period, and return them by post to the processing laboratory. The received result should be corrected for the known background and age of the plastic strip, and for its calibration.

Correction of data to yield an estimate of the annual average radon concentration in a house introduces an entirely separate approximation, and one that may be in serious error for some houses. This is not considered further here, but see Section 8.

A reasonable analogy to simple comparisons of radon detectors would be comparison of three or four different voltmeters.

If a scientist purchased ten batteries of unknown voltage and measured their characteristics using the different voltmeters, and if each gave different answers then it would be necessary to question which one or more was reading incorrectly.

It would not be necessary for the scientist to know all about the development and calibration of volt meters, merely that they gave markedly different answers when used for the purpose for which they were intended and under appropriate conditions.

Neither would it be necessary to know the exact voltage of each battery in order to say that the voltmeters gave different answers and that something was clearly wrong.

Similarly, if two different types of radon detectors were to give broadly the same result but another type gave different answers then there could be some suggestion - but no proof - as to which type was misreading.

It would be of some concern if substantial systematic differences between detectors were found when they were used for assessing the low integrated radon concentrations found in most houses. This is because of the possible effects upon national statistics were systematic error to have been present in a large number of detectors used for gathering this information.

Systematic or indeed random errors at higher integrated radon concentrations are of less concern, simply because the number of houses yielding such results is much smaller.

For any set of measurements, two types of error may be present:

1. random error and which could be ameliorated to any desired extent by undertaking multiple readings, and

2. systematic error associated with any one detector type and scanning system and which cannot be detected using multiple measurements.

Validation of detectors, and results.

A validation scheme is now in place in the UK, with the aim of producing confidence in radon measurements. It is interesting nevertheless to consider some of the published data from the United States and work undertaken in the UK.

In the United States, some early measurements using track etch detectors may be open to question. A substantial programme of field trials and test comparisons was undertaken by the EPA but the General Accounting Office has been critical of some commercial and calibration work. Laboratory procedures had to be tightened in order to produce acceptable standards. Consequently, many companies can now achieve within 20% of the correct answer most of the time, and one or two claim within 5%.

There has been much less work undertaken to compare the results from different types of detectors when used in buildings. In the UK over the winters 1989/90 and 1990/91 thousands of detectors were placed in field trial houses in order to determine radon levels. A number of these were used for comparison purposes and some of the results were to be published in a research paper submitted to the EPA. (see Section 44).

One obvious consequence of different results from different types of detectors is uncertainty as to whether many houses are above or below the action level.

Simple comparisons of two or more sets of detectors may be represented graphically, as shown in figure 1 below. Ideally results would all lie about the 1 : 1 line and with random error being indicated by the scatter of the individual data points. Results similar to those shown in figure 2 would indicate systematic error.

Unfortunately, and despite the scale of the comparisons undertaken in the UK and the fact that many of the results are known to dozens of householders already, they cannot be reported here.

In comparing three detector types, data from any one may be plotted along the abscissa. Data from the other two is plotted as ordinate. All axes have the same scale. Divergence of the 'best-fit' lines indicates systematic error between detector types. Results as shown in figure 1 indicate only random error.

These graphs are for illustration only and do not represent actual data.

Only a few general scientific comments can be made which are independent of the actual results that were obtained.

If most or all of the results were as shown in figure 1, then there would be very little justification for withholding publication. Indeed, the results would support the credibility of data already notified to tens of thousands of householders.

However, if many or all of the results were as shown in figure 2, then there could be considerable embarrassment especially for any laboratory of national or international standing that was involved in the calibration or scanning work. This would be especially the case if they were responsible for calibrating or validating more than one of several types of detectors that gave systematically different answers.

In view of the results it is interesting to note that soon after the unexpectedly large response to DOE's spring 1991 mail-drop offering free radon tests to all householders in Devon and Cornwall, a large company with nuclear links became involved in radon testing. This was simply because NRPB alone could not cope (and did not wish to cope) with a huge volume of routine measurement work.

The corporate symbol for this company is a hexagon - and a container of this shape was designed and mass produced. To help ensure acceptance of the new product, a similar yellow plastic was used for the container as had been used for NRPB dome detectors over the preceding years. However, there was obviously no time for lengthy testing side by side of the two different types of 'yellow' detectors, those using the familiar dome container and those using the larger hexagonal unit. It remains an interesting question whether some correction between results obtained by the two types may someday have to be applied.

Some householders whose homes were monitored in detail by the author and using many detectors have since been offered re-monitoring - but using only a single NRPB type of detector. Thus, comparisons seem not to be being repeated, and with only one type of detector there can (of course) be no dispute as to the result.

The three types of detectors used by the author in the 1989/90 BRE field trials were:

NRPB yellow domes, as used in regional survey work

TASL medicine pot designs, as used by IEHO in much of their early survey work

NRPB medicine pots, as used in the 2000-house survey of the UK.

In the 1990/91 comparisons, the types used were:

NRPB yellow domes (again)

A new design from TASL of Bristol

A popular type from the USA distributed by Tech Ops.

The Tech Ops design is different from others used in the UK in that it relies on a filter paper rather than on a small crack or diffusion through the plastic container to keep out radon daughters whilst allowing eventual equilibrium of radon gas. Many types sold in the USA have utilised a similar design.

The results obtained by the author included the first substantial international comparison between these American detectors and types used routinely in the UK. It was expected that all would give broadly the same answers.

Finally it may be recorded that in 1987, at an international symposium on the natural radiation environment, held only a few weeks before the author took charge of BRE's radon work for the UK government, it was stated of radon in a UK keynote address:

next page

top of radon section