WO2019102173A1 - Detector and method for detection of airborne beta particles - Google Patents

Abstract

A detector (1, 2) and associated method for the detection of man-made airborne beta particles in the presence of background beta radiation. The detector detects the energy of particles at a detector head (8), and comprises a low level discriminator (LLD) (20) for distinguishing between noise and beta particles (once alpha particles and optionally gamma particles have been stripped from the count of particles). The LLD threshold value (22) is variable over time according to the electronic noise of the detector head (8), such that the LLD threshold value (22) tracks just above the electronic noise thereby maximising the number of detected beta particles. Electronic noise can be measured directly or derived from a measurement of the temperature of the detector head (8).

Description

Detector and Method for Detection of Airborne Beta Particles

The present invention relates to detectors of airborne radioactive particles and in particular to detectors with improved sensitivity to radioactive particles, including for example beta particles, indeed potentially enabling the improved detection of man-made beta particles (herein also referred to as 'target particles'), which can be damaging to health. These target particles must be detected in the presence of natural beta particles emanating from the ever- present radon decay chains (commonly referred to as background radiation). Such target particles may be emitted from airborne dust, for example as a result of mining, processing, or transportation activities which result in dust, adjacent or in which target particles may be found and carried.

The range of alpha and beta particles in air is very small (typically around a maximum of 20mm and 100mm respectively). As alpha and beta particles are commonly emitted from dust particles, a detector for detection of airborne beta particles can be configured so that dust particles are collected near to a detector head, enabling detection of alpha and beta particles emanating from collected radioactive dust particles.

Detectors in accordance with the present invention are commonly also known as Continuous Air Monitors (CAMs) and are commonly deployed in environments where, for example radioactive materials have been or are being processed. This is because generally such a detector will be deployed to continuously monitor the air in a localised environment for harmful airborne particles, thereby protecting workers from the effects of radioactive dust inhalation. Whilst such detectors are generally deployed for continuous air monitoring, it will be appreciated that continual (i.e. periods of monitoring with breaks in-between), occasional (infrequent, e.g. hourly or daily), or one-off monitoring modes are possible.

The detector such as the one of the subject invention is commonly referred to as a Continuous Air Monitor (CAM), which is common parlance to the skilled person. However, despite the

l word continuous in the expression, it should be understood that the apparatus and method of the subject invention does not need to operate continuously.

Where the words CAM or Continuous Air Monitor is used in this specification, it is to be understood by the skilled person as equivalent to the detector of the present invention.

CAMs have become more sensitive over time as technology has improved. Due to the potentially damaging nature of radiation, it is desirable for CAMs to be as sensitive as possible and this trend of improving sensitivity is likely to continue. It is a simple matter to increase the sensitivity at the expense of more false alarms (i.e. background particles erroneously detected as target particles). However, each false alarm is costly as it may result in evacuation of the area showing an alarm while the incident is investigated. As such, whilst it is desirable to increase the sensitivity this should be achieved without increasing the false alarm rate.

It is desirable to detect beta particle radiation as well as alpha particle radiation because radioactive isotopes typically decay towards stability by emitting both. If an isotope decays directly to a stable daughter via a beta particle (such as Csl37), an alpha particle detector would not see it - as such, beta particle (and/or gamma photon) detection is needed to detect, for example, Csl37.

Alpha particles are typically emitted with more energy than beta particles, and (unlike beta particles) are mono-energetic, having energies characteristic of the emitting isotope. Energy spectrometry therefore makes identification of man-made alpha particles relatively straightforward in the presence of background alpha particle radiation. Patent no GB2482047B discloses such a method of background alpha particle discrimination.

The detection and identification of man-made beta particles is more difficult compared to the detection of alpha particles for two main reasons. Firstly, with beta particles, energy is not concentrated in one or more narrow peaks as is common with alpha particles, but instead energy is spread as a broad spectrum, making the target particles difficult to detect amongst background particles which are also spread over a broad spectrum.

Secondly, the deposited energy of a beta particle as it passes through a CAM's detector is very low relative to alpha particles, and further limited if the beta particle is released with low energy. A thicker detector would increase deposited energy, but increase trapping (a process whereby an electron produced by a traversing radioactive particle is trapped and released later, contributing not to the wanted signal, but to unwanted electronic noise), so there is a trade-off between deposited energy and electronic noise due to trapping.

A typical CAM for the detection of radioactive particles (alpha or beta) comprises two components:

- a radioactivity detector head;

- and a means of data collection and processing.

It may also be that a CAM further comprises a dust-collecting filter with a collecting surface, and a means for drawing or moving air through or towards the filter, such as a vacuum pump or a fan. In a typical configuration, a vacuum pump draws air through the filter, which collects airborne dust. If these additional components are present, the detector head will typically be configured so that it faces the filter's collecting surface. Regardless of configuration, the detector head detects radioactivity and the energy of each detection is measured. Supporting hardware may also be present, particularly to provide a power source, connectivity between the two components, and additional means to enhance their collection/ processing functions, such as processors, memory, and amplification.

Signals at the detector head need to be discriminated between those due to alpha particles, beta particles and gamma photons. This is conventionally done by discriminating between deposited energy, with high energies being classified as alphas and low energies as a combination of gamma photons and beta particles (it is not yet known which). To then discriminate between the combination of gamma photos and beta particles, a gamma detector can be positioned behind a sheet of material with properties such that beta particles are blocked but such that gamma photons pass through easily. This allows the gamma photon count to be determined, which can be subtracted from the combined beta particle and gamma photon count to leave the gross beta particle count.

There is then a background compensation process to strip counts determined to be due to background (i.e. naturally occurring) radiation and finally an assessment of whether the residual count (i.e. that once background has been stripped from the gross count) is likely to be due to man-made radiation. The residual count, i.e. that following the background compensation, will tend towards zero when no man-made radiation is present and could be positive or negative. The residual count is tested in some way to determine if it indicates the presence of man-made radiation. The CAM will be configured to produce an alarm (aural, visual, electronic signal etc) in the event of pre-determined limits being exceeded. For example, a CAM could be configured such that if a residual count greater than 10% of the total number of particles is found, the CAM triggers an alarm. For example, if 100 beta particles are detected (once alpha particles and gamma photons have been subtracted from the count), and the expected count of beta particles due to background radiation in the same period is 86, the residual count is 14. This number exceeds 10% of the gross beta particle count (14 is more than 10% of 100), so an alarm would be triggered if the pre-set limit is set as 10%.

A CAM's detector head typically comprises a 0.2mm thick silicon wafer facing the filter (when present) across a typical 4-5mm air gap. When an alpha particle or beta particle enters the wafer, it produces an electrical charge which is collected and measured externally. At an atomic level, the silicon wafer is an empty space sparsely populated by silicon atoms. When a radioactive particle (alpha or beta) enters, if it grazes a silicon atom, it is likely to dislodge an electron from the silicon atom's outer orbit, producing a free electron (negatively charged) and an ion (the original atom, positively charged because it has lost an electron). The radioactive particle loses only a little energy at each grazing, so multiple electron/ion pairs are produced, and these are detected by applying an electric field across the thickness of the wafer that sweeps electrons/ions to the wafer terminals to produce the external charge.

The energy of a radioactive particle is typically expressed in electron-volts (eV), and silicon at room temperature requires = 3.62eV of energy to produce an electron/ion pair. Alpha particles consist of two protons and two neutrons bound together to form a particle identical to a helium nucleus, making them much larger and heavier than beta particles, so they interact more with matter. Thus, a lOMeV alpha particle can be stopped by 0.2mm of silicon, giving up all its energy, but beta particles typically lose only 200keV of energy in the same distance, resulting in an electrical charge one fiftieth that of an alpha particle. (A reference for this figure of 3.62eV is pp368 "Radiation Detection and Measurement" 4th Ed. Glenn F Knoll. Wiley (2010) ISBN 978-0-470-13148-0). Thus a measurement of the energy incident at or upon the detector head can be used to determine if a beta particle is present. Accordingly, for the reader's ease, the detector head will be assumed to measure energy.

When a beta particle loses 200keV of energy in silicon, it produces 200,000/3.62 = 55,000 electron/ion pairs = 9fC, and this charge is converted to a voltage in the associated charge amplifier by a charge conversion capacitor, whose smallest practical value is typically lpF. Since V = Q/C, this results in only 9mV for 200keV deposited energy, and it must be detected in the presence of the equipment's self-generated random electronic noise (as distinct from external interference which is also sometimes referred to as 'noise'). All subsequent references to noise within this document are to self-generated random electronic noise (commonly referred to just as electronic noise) unless otherwise stated.

In a well-designed system, the two dominant sources of electronic noise are the silicon detector head and its charge amplifier. With good design, it is possible to reduce amplifier noise such that detector head noise dominates. However, beta particle energies are spread over a wide range, starting from zero energy (i.e. many beta particles are released with low energies), so there comes a point when the voltage developed across the charge conversion capacitor becomes comparable with electronic noise and the particle can no longer be detected. Thus it is essential to minimise electronic noise if low energy beta particles are to be detected, since the energy of some beta particles may be less than the electronic noise of the system and accordingly impossible to detect with confidence. (A reference for further information on electronic noise is "Low Noise Electronic System Design" C D Motchenbacher and J A Connelly. Wiley-lnterscience (1993) ISBN 978-0-473-57742-3).

Noise has many sources, and even an ideal resistor produces noise. Noise is also produced by imperfections. Temperature variations cause different stresses and strains on connections and joints on internal surfaces or on external connections, which may affect noise. Therefore, all electronic devices implicitly generate electronic noise with the passage or path of electricity between components.

To determine whether a signal is a beta particle or just electronic noise, a Low Level Discriminator (LLD) is common in a detector for the detection of beta particles. The purpose of the LLD is to change output state when a signal input exceeds a user-defined value, following which the particle count is then increased by one. From the above discussion, it will be clear that if the user-defined value were set to zero electronic noise would continually be detected (erroneously) as beta particles - any signal, no matter how small, would be counted as a beta particle. There needs to be a threshold below which the signal is deemed to be noise (or a low energy beta particle which cannot be discriminated from the electronic noise due to its low energy), and above which the signal is determined to be genuine beta particle and not due to electronic noise.

In practice an LLD is an electronic comparator with two inputs and one output; the inputs are a threshold value, and a signal value to be tested against the threshold value. The output changes state if the LLD triggers (i.e. the signal to be tested crosses (exceeds) the threshold). The LLD is 'set' by defining the threshold voltage (the 'LLD threshold value'), so 'setting' an LLD means setting the threshold voltage value to an express value. In a typical configuration, the LLD threshold value is therefore set just sufficiently above the noise to acceptably limit the number of counts due to noise without missing too many counts of low energy beta particles. The significance of this is that a correct beta particle count can only be obtained if the LLD is set to OkeV (such that all beta particles are counted regardless of how low energy they are), but this can never be achieved because the LLD threshold value must always be set above the electronic noise. If it weren't set above the noise, the detector would constantly be triggering due to electronic noise rather detection of actual beta particles, which is clearly undesirable and would give a falsely high beta particle count.

CAMs are often positioned in buildings where the temperature can vary significantly through the year, perhaps between -10^°C to +45^°C, which causes the electronic noise of the CAM to vary considerably. Since the LLD threshold value of a CAM must be set above the noise (to avoid counting noise as beta particles), it is typically set such that it sits above the electronic noise throughout the entire operating temperature range. This has the advantage that the LLD threshold value of the specific equipment does not need to be adjusted according to its environment and yet will still detect a broadly acceptable number of target particles. This also makes it easy to re-deploy a CAM to a different environment without re-calibration.

Accordingly, in a CAM configured to detect beta particles, typically the prior art approach is to set the LLD threshold value to approximately lOOkeV. This ensures it sits above the noise for a wide range of operating environments that a CAM can be expected to endure. Such a setting prevents system noise triggering the LLD and increasing the count when there are actually no beta particles present, and thus reduces the possibility of false positives.

A consequence of setting the LLD threshold value at such a level is that some low energy beta particles are missed, as low energy signals are necessarily presumed to be due to electronic noise and not genuine beta particles. If, for example, ³⁶CI were being monitored, this would result in only 80% of ³⁶CI beta particles being detected (i.e. with the other 20% having energy below the lOOkeV LLD threshold value and therefore disregarded). In other words, when a plot of counts against emitted energy is made, 36CI shows an average energy of 235keV, and numerical integration of the distribution >100keV captures only 80% of the emitted counts (as 20% have an energy below lOOkeV). Unfortunately, not all beta particle spectra are the same shape, so existing methods are not able to reliably compensate for failure to detect all of the beta particles (due to the proportion with energy falling below the LLD threshold value), since this loss (the number undetected) depends on the spectrum in question. This problem is typically overcome by calibrating the CAM for known likely emissions.

In a CAM there is a discrepancy between alpha particle and beta particle transport across the air gap because although beta particles have negligible interaction with the thin layer of air, alpha particles give up some or all of their energy. The result is that most emitted beta particles reach the detector head, but not all alpha particles.

The significance of the two preceding paragraphs is that the background beta particle count is invariably predicted as being a multiple of background alpha particle count, with the exact multiple being derived directly from theoretical considerations of natural decay chains without regard to practical considerations. Fortuitously, one error mechanism tends to reduce the number of detected background beta particles (those with energy below the LLD threshold value), whereas the other reduces the number of detected background alpha particles (those lost over the air gap), and when =100keV is chosen for threshold value of the LLD, the theoretical multiple approximately predicts beta particle background from alpha particle background.

Adjusting the LLD threshold value significantly away from lOOkeV will result in the two errors no longer conveniently cancelling out. Raising the LLD threshold value will result in more genuine beta particles being missed making it more likely an incident will be missed or under- read, and lowering the LLD threshold value will result in an increased probability of false positives from system noise being erroneously recorded as beta particles. Clearly neither is desirable.

According to a first aspect of the invention there is provided a detector for detecting airborne beta particles which generates an electronic noise in use, said detector comprising a detector head for measuring energy, means of determining an electronic noise value of the detector, a low level discriminator threshold, and a means of adjusting the low level discriminator threshold, wherein the low level discriminator threshold is configured to an initial value, the means of determining an electronic noise value of the detector detects the electronic noise value of the detector, the electronic noise value is compared against the low level discriminator threshold value, and if the electronic noise value differs from the low level discriminator threshold value, the low level discriminator threshold value is varied, and energy measured by the detector head below the low level discriminator threshold value is disregarded.

This allows the LLD threshold value to be lowered when electrical noise permits, potentially reducing the number of disregarded deposited beta particles, and thus improving accuracy. Minimising disregarded beta particles is significant because it is typical to calibrate a detector for ³⁶CI and incorporate a compensating factor such that the detector reads correctly on ³⁶CI, but other nuclides will have a different spectrum shape. If the compensating factor is small, differently shaped spectra introduce small counting errors, but as it rises, so too does the error when counting different nuclides. It also permits the LLD threshold value to be raised when the temperature is increased, ensuring that as equipment noise increases such that it extends above the existing LLD threshold value, the threshold value is moved to compensate such that the noise increase does not increase false positives. The LLD threshold value can be adjusted continually or continuously in response to the electronic noise of the apparatus, being adjusted continuously or continually such that it tracks just above the noise. The LLD threshold value may be set or determined using a means for setting such a threshold, and may be variable over time. The initial value of the LLD threshold value may be pre-set at a common value (such as lOOkeV), or alternatively may be another value such as 0 (zero) or set according to the ambient environment and / or expected electronic noise of the detector in use.

Optionally, the detector comprises a temperature sensor for measuring at least one temperature of the detector, and a means for deriving the electronic noise from the at least one temperature measured by the temperature sensor. This is advantageous as temperature is very simple to measure and monitor, and modification of pre-existing detector equipment is thus relatively simple. The temperature may be measured in more than one place should the user so desire, for example on more than one part of the detector head, and/or one or more external surfaces of the detector.

Optionally, the low level discriminator threshold value is configured to be variable according to a multiple of the standard deviation of the electronic noise value. This provides a constant and known probability of false beta particle counting due to noise. Here the detector's electronic noise value should be understood to mean its standard deviation about zero.

Optionally the detector is configured to determine the electronic noise value continually thereby permitting the low level discriminator threshold value to be varied continually over time. This permits the temperature to be continually sampled and the LLD threshold value to be adjusted continually as the apparatus is operated. This resampling may take place at any appropriate period as dictated by circumstances, e.g. every 30s, every 10 minutes, or hourly.

Optionally the detector comprises a means for directly measuring the electronic noise value of the detector, and the low level discriminator threshold value is variable according to the electronic noise value. This permits an accurate measurement of system noise by measuring it directly, rather than relying on an estimation of noise from a measurement of temperature.

Optionally the low level discriminator threshold value is configured to be variable according to a multiple of the standard deviation of the electronic noise value. This provides a constant and known probability of false beta particle counting due to noise.

Optionally the detector is configured to determine the electronic noise value continually thereby permitting the low level discriminator threshold value to be varied continually. This permits the noise to be periodically measured and the LLD threshold value to be adjusted continually as the apparatus is operated. This resampling may take place at any appropriate period as dictated by circumstances, e.g. every 30s, every 10 minutes, or hourly. Optionally the detector comprises a means of compensating for an expected count of beta particles due to naturally occurring radiation. This improves the accuracy of the apparatus by compensating for beta particles already present in the environment due to background (naturally occurring) radiation. It provides for compensation of the gross beta particle count in consideration of the expected number of beta particles due to naturally occurring radiation, resulting in a more accurate determination of whether or not target particles are present.

Optionally the detector comprises a means of modifying the expected count of beta particles due to naturally occurring radiation according to the low level discriminator threshold value. This improves the accuracy of the background compensation by modifying the expected count when the low level discriminator threshold value is adjusted and is necessary because in a constant background, changing the LLD threshold value changes the number of background beta particles detected. If the beta background prediction were not modified according to the LLD threshold value, it would be incorrect, leading to a false assessment of man-made radiation and false alarms or missed release detection.

Optionally the detector comprises a filter for collecting dust. This enables radioactive dust to be trapped near the detector so that its radiation can be measured. By trapping the particles on a filter accurate modelling and calculations of what is collected or is trapped there can be performed. This may particularly be the case when the distance therebetween is known.

Optionally the detector comprises a vacuum pump for forcing air to flow through the filter. This provides a controlled air flow through the filter (useful for later calculation of air activity).

Optionally the detector comprises a means of measuring an air resistance across the filter and a means of adjusting the expected count of beta particles due to naturally occurring radiation according to the measured air resistance across the filter. This is useful because an increase in filter air resistance indicates dust build-up on the filter. As the dust layer thickens, background alpha particles emitted from deep below the dust expend all their energy traversing the dust (self-absorption) without reaching the detector, so they are not counted. Beta particles are less affected by dust, so background beta particles emitted from deep within the dust still reach the detector. Basing expected beta particle background on detected alpha particle background risks false beta particle alarms unless alpha self-absorption is compensated for.

Optionally the detector comprises an air gap between the filter and the detector head, and a means of modifying the expected count of beta particles due to naturally occurring radiation to compensate for losses due to the air gap. This permits accurate modelling and compensation for the particles lost, and the particle energy lost, as result of the air gap. This provides for more accurate background compensation and a thus a more accurate determination of whether or not target particles are present.

Optionally the detector comprises a means of compensating for the age of air. This improves the accuracy of beta particle detection.

According to a second aspect of the invention, there is provided a method of detecting airborne beta particles comprising the steps of: taking a detector according to the first aspect of the invention; determining an electronic noise value of the detector; setting the low level discriminator threshold to an initial value; determining the electronic noise value of the detector;

- comparing the low level discriminator threshold value to the determined electronic noise value;

- setting or varying the low level discriminator threshold value according to the determined electronic noise value.

This allows the LLD threshold value to be lowered when electrical noise permits, enabling more accurate beta particle counting, as a higher proportion of low energy particles are counted. It also permits the LLD threshold value to be raised when the temperature is increased, ensuring that as equipment noise increases such that it extends above the existing LLD threshold value, it is moved to compensate such that the noise increase does not increase false positives. The initial value of the LLD threshold value can be pre-set according to a knowledge of the environment in which the apparatus will be deployed, pre-set at a common value (such as lOOkeV), or can be determined from an initial determination of noise. Setting the LLD threshold value according to noise may cause it to be adjusted from its previous setting, or alternatively mean retaining its previous value where noise has not changed or has only changed by an inconsequential amount. The setting of the LLD threshold value according to noise ensures that the threshold value it is set to the optimum position at that point in time.

Optionally the step of determining the electronic noise value of the detector comprises determining the electronic noise value of the detector by measuring the electronic noise directly. This permits an accurate reading of noise to made and ensures that the LLD threshold value is set according to the actual noise of the system.

Optionally the steps of determining the electronic noise value of the detector by measuring the electronic noise directly, and setting the low level discriminator threshold value according to the electronic noise value are repeated two or more times. This permits the LLD threshold value to be repeatedly (continually) or continuously set or adjusted over time, tracking just above the noise. Where the electronic noise value of the detector is consistent over a time period this may mean that the LLD threshold value is repeatedly set to the same value.

According to a third aspect of the invention, there is provided a method of detecting airborne beta particles comprising the steps of taking a detector according to the first aspect of the invention wherein said detector comprises a temperature sensor for measuring at least one temperature of the detector; the step of determining the electronic noise value of the detector comprises measuring at least one measured temperature of the detector, and determining the electronic noise value of the detector by estimating the noise from the measured temperature of the detector. This permits the temperature of the detector to be used to determine the detector noise and may be useful where it is difficult or impractical to measure the noise directly. This permits the LLD threshold value to be set according to the noise of the system derived from a simple temperature measurement. This also has the advantage that the counting of particles does not need to be suspended whilst the npise is determined. Measurement of temperature is advantageous as temperature is very simple to measure and monitor, and modification of pre-existing detector equipment is thus relatively simple.

Optionally the steps of measuring a measured temperature of the detector, determining the electronic noise value of the detector by estimating the noise from the measured temperature of the detector, and setting the low level discriminator threshold value according to the electronic noise value are repeated two or more times. This permits the LLD threshold value to be repeatedly (continually) or continuously set or adjusted over time, tracking just above the noise.

Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

• Figure 1 is a block diagram of a first embodiment of the invention, showing a detector (up until the output of the LLD) according to a first embodiment of the present invention wherein temperature is measured and this measurement is processed to estimate system noise, which in turn is used to adjust the LLD threshold value.

· Figure 2 is a block diagram of a first embodiment of the invention, showing a detector

(up until the output of the LLD) according to a second embodiment of the present invention wherein noise is measured directly at the input to the LLD, and this measurement of electronic noise is used to adjust the LLD threshold value.

In the drawings, like parts are denoted by like reference numerals.

With reference to Figures 1 and 2, all embodiments of the detector (1, 2) of the present invention are capable of being configured like a conventional detector known in the art. They comprise a filter (3), a detector head (8) (for detecting alpha particles and beta particles), a vacuum pump (5) for drawing air over the filter, and a microcontroller (31) for processing the data from the detector head (8), and other components. The microcontroller (31) comprises an internal Analogue to Digital Converter (ADC). The microcontroller (31) could alternatively be a microprocessor plus external memory or other processing means. Alternatively a fan could be used instead of a vacuum pump (5).

The detector head (8) is the transducer that interacts with radiation to produce an electronic charge. It incorporates a 0.2mm silicon wafer detector (not shown). The detector head (8) faces the filter (3) for dust (7) collection across an air gap of 4 to 5mm. Both the detector head (8) and dust collection filter (3) are situated within a funnel (4). The funnel (4) includes a gate (not shown) that clamps the filter (3) in position. A vacuum pump (5) causes an airflow (6) through the funnel (4) which causes dust (7) to be deposited on the filter (3). Wafers of different thicknesses could be used depending on the purity of the silicon. The air gap could· be outside of this range, but this range is typically used as it yields good results.

Radioactive dust particles (7) caught by the filter (3), may emit alpha particles and / or beta particles, and if these reach the detector head (8), they create an electrical charge within the silicon wafer. The signal (10) from the detector head (8) passes through a charge amplifier (11) a shaping amplifier (12) and coupling network (13) producing a bipolar pulse. The amplitude of this bipolar pulse is approximately 250mV per MeV (therefore lOOkeV at the detector head (8) generates 25mV at the output of the coupling network (13)). The coupling network (13) consisting of a capacitor (14) and resistor (15) is necessary to block DC offsets if present due to the electronics.

This signal from the coupling network (13) is then used as the signal input (21) of the LLD (20). This LLD (20) determines if an energy level detected is due to noise or due to a genuine beta particle, so that noise doesn't skew the beta particle count. This LLD (20) is commonly referred to as a "beta LLD" or "beta particle LLD", to distinguish it from other LLDs which may be present in the detector (1,2), e.g. an LLD to distinguish between alpha particles and beta particles by considering energy against a threshold value. All references to LLD within this document refer to this beta LLD, unless otherwise specified. The LLD (20) is a comparator that compares an amplified and shaped pulse from the detector head (8), which arrives at the signal input (21) with the LLD threshold value (22) and changes output state when the signal input (21) exceeds the beta LLD's threshold value (22). Where the signal input (21) is below the LLD threshold value (22), it is disregarded. "Below" is to be construed by the skilled person as meaning "at or below" as the user may require. The LLD (20) produces a rectangular pulse at the output (23), with the leading edge of this rectangular pulse being produced as the input signal (21) exceeds the LLD threshold value (22) and the trailing edge of this rectangular pulse is produced as the signal input (21) falls below the LLD threshold value (22) as it decays to zero. It is conventional to trigger and count pulses using the leading edge, so that is the approach employed here. It is possible that the LLD threshold value (22) will be set or determined using a means for setting such a threshold value, and it is possible that the LLD threshold value (22) will be variable or varied over time. The user could configure the detector (1,2) to always configure the LLD threshold value (22) to be the electronic noise value that has been measured.

The charge amplifier (11) comprises an op-amp configured for use as a charge amplifier (11). A lpF charge conversion capacitor (not shown) is connected between the inverting input and the output of the charge amplifier (11). The non-inverting input is grounded and the detector head (8) is connected between ground and the inverting input. The charge amplifier (11) action forces detector head (8) current to flow through the charge conversion capacitor, producing a voltage.

The skilled person will appreciate that the apparatus (1, 2) needs be configured to distinguish between signals at the detector head (8) due to alpha particles, gamma photons and beta particles, to ensure the beta particle count is not affected by alpha particles and gamma photons.

A particle with an energy above a specified level (typically 200keV deposited energy) is deemed an alpha particle, and particles with an energy below a specified level (typically 200keV deposited energy) is a beta particle or noise (or alternatively a gamma photon). A similar LLD (not shown), distinct from the LLD (20), having a threshold input set to 200keV deposited energy allows discrimination between beta particles and alpha particles. Alpha particle energies are measured and recorded as an energy spectrum (i.e. the number of particles having an energy falling within an energy range is recorded against a series of adjacent energy ranges, to form a spectrum) as this is necessary for alpha background compensation.

At the output of the coupling network (13) the signal (40) is fed into a peak hold circuit (41) and an 8-bit ADC (42). At the output (43) of the ADC (42) the signal can be used to determine and construct the alpha particle spectrum. The ADC (42) is fired once per pulse.

If required (based upon an understanding of the environment in which the detector (1,2) will be deployed), gamma photons can be counted using a second detector head and associated electronics (charge amplifier, shaping amplifier etc.) sitting in parallel (not shown), with the second detector being shielded from beta particles leaving the dust (7) trapped by the filter (3). The count from the second detector is subtracted from the beta particle count, to prevent gamma photons falsely being detected as beta particles. This is necessary because beta particles and gamma photons produce indistinguishable charges within the detector head (8) of the detector (1,2), potentially causing false beta particle counts in a gamma field. It is known that beta particles are stopped by thin aluminium, so by positioning this second (identical) detector (not shown) behind the first detector head (8), with a shield (not shown) in between that stops beta particles but is substantially transparent to gamma photons, the beta particles and gamma photons can be distinguished. Any count registered by the second detector cannot be beta particles emitted from dust particles (7) trapped on the filter (3), and the count is presumed to be due to gamma photons.

Provided that the gamma count is synchronous with the raw beta particle count, gamma detections can be subtracted from the detector head's (8) beta particle count to give the number of gamma-compensated beta particles, which is also recorded. Although the gamma detector (not shown) is habitually positioned behind the detector head (8), this is not essential as it responds primarily to the ambient gamma field rather than emissions from dust particles (7) trapped the filter (3). The gamma detector is only employed where gamma emissions are likely to be present where the detector (1, 2) is deployed.

Where beta particles are of a low energy they may still develop a charge in the detector (1,2) but may not be counted, as it is not possible to know whether the charge developed is due to an actual beta particle or due to random electronic noise. The threshold below which energy is disregarded due to being considered indistinguishable between representing genuine beta particles and representing noise is, as discussed, determined by the setting of the LLD threshold value (22).

Thus the data from the detector (8) is discriminated between alpha particles and beta particles according to deposited energy, then recorded as a raw count of beta particles, and an energy spectrum for alpha particles. Data is continuously recorded over multiple time frames, for which associated start and stop time stamps are recorded.

As previously discussed, the detector (1, 2) comprises an LLD (20) with two inputs - a signal input (21) represented as a voltage, and a LLD threshold value (22) input represented as a voltage which the signal input (21) is compared to. The detector (1, 2) stores the LLD threshold value (22), which may be adjusted over time. This is stored as a non-volatile value in a Digital to Analogue Convertor (DAC) (32), but may alternatively be stored within the microcontroller (31). In practice, a digital potentiometer could be used to perform the function of the DAC (32).

The LLD threshold value (22) is adjustable using DAC (32) and is set such that is can track just above the noise, assuring maximum beta particle sensitivity. The analogue potential divider (33) scales the typical 0 to +2.5V range of the DAC (32) to the typical 0 to 50mV range needed for the LLD (20). The LLD (20) does not need to be a dedicated physical component; any means which can compare two levels and adjust an output signal accordingly is suitable to perform the function of the LLD (20) and would be recognised by a skilled person as being interchangeable.

The detector (1, 2) may comprise a temperature sensor (30) (as in the first embodiment (1)), or measure the noise directly (as in the second embodiment (2)).

Where the detector (1) is configured to measure temperature (as in the first embodiment (1)), the signal from the temperature sensor (30) is be configured such that it stores and sets the LLD threshold value (22) through a microcontroller (31). Other means could be used than the microcontroller (31). The microcontroller (31) takes the data from the temperature sensor (30) and uses a pre-defined look-up table to vary the digital code fed into the or DAC (32) that adjusts the LLD threshold value (22). Alternatively, an analogue temperature measurement could be taken from the temperature sensor (30), processed in analogue and used for the LLD threshold value (22).

Due to the need to discriminate between alpha particles, beta particles, gamma photons and noise, a detector (1,2) in practice is likely configured with a number of components vyhich function as LLDs in the sense that two things need to be compared. There may be multiple components in a working embodiment of this invention which could be viewed as an LLD as there are likely to be multiple means to discriminate between a signal and a threshold (e.g. to discriminate alpha particles from the remainder). However, this invention relates to the LLD (20) which determines if a detected energy level corresponds to a beta particle or is attributable to noise.

As stated earlier, good design should ensure that system noise is dominated by detector noise. Provided that detector head (8) noise is known at the time (either by direct toe-dipping measurement of the electronic noise, or by prediction from measured temperature), the LLD threshold value (22) can be set to a fixed multiple of detector head (8) noise at that time and this allows a constant and known false alarm probability irrespective of temperature. The LLD threshold value (22) is set to a fixed multiple of the detector head's (8) predicted noise (voltage) at that (measured) temperature, or where the noise is measured directly a fixed multiple of the measured electronic noise value.

Common to all embodiments of the detector (1, 2) described herein is the process of background compensation. This method uses alpha particle spectrometry to allow background radon to be quantified, and in conjunction with a known decay series, the number of beta particles due to radon can be predicted from the number of detected radon alpha particles. For example, the ²³⁸U series (colloquially known as ²²²Rn) series is as follows:

The ²³⁸U series seen by a detector (1,2) starts at ²²²Rn gas (typically emitted from the ground) and ends at ²¹⁰Pb because the quantities on a detector (l,2)filter (3) combined with a 22.3 year half-life mean that no activity is seen. This truncated series is presented as a table, showing energies.

When a detector (1,2) detects a 7.69MeV alpha particle, it must come from ²¹⁴Po decaying to ²¹⁰Pb, but the likelihood of seeing the 5.3MeV alpha particle from ²¹⁰Pb as it decays to ²ⁱ⁰Bi is negligible because the long half-life means that it happens very infrequently. Now consider i the situation of detecting a 6.00MeV alpha from ²¹⁸Po. ²¹⁸Po is on the filter (3) of the detector (1,2) and stays there, decaying through ²¹⁴Pb, ²¹⁴Bi, and ²¹⁴Po, each releasing an alpha particle or beta particle. If we consider from the decay of ²²²Rn to the decay of ²¹⁴Po, we have a total time of 3.1 + 26.8 + 19.9 = 49.8 minutes. Now consider entering the decay chain randomly. The probability of entering the chain before ²¹⁸Po has decayed is 3.1/49.8 = 6.2%, so that is the probability of two alpha particles (6.00MeV, 7.69MeV) and two beta particles (22QkeV, 500keV) being emitted. Moving down the decay chain, the probability of entering after ²¹⁸Po has decayed but before ²¹⁴Pb has decayed is 26.8/49.8 = 54%, and this is the probability of one alpha particles (7.69MeV) and two beta particles (220keV, 500keV) being emitted. Using this methodology, we can calculate the number of alpha particles and beta particles emitted for each entry point. If we multiply the probability of each scenario by the number of alpha particles and beta particles emitted at each point, then sum all beta particle probabilities and divide that by the sum of all alpha particle probabilities this results in a beta particle to alpha particle ratio that permits beta radon prediction from alpha radon measurement.

At its simplest, the described methodology is only applied to the uranium series, with the thorium series being deemed negligible because the short half-life (55.6s) of ²²⁰Rn compared to 3.82 days for ²²²Rn means that it is more likely to be trapped in the ground and its daughters never seen by a detector (1,2). However, there is no reason why the previous methodology cannot be applied to the thorium series and even the (very rare) actinium series.

At its very simplest, the beta particle to alpha particle ratio would be calculated externally and loaded into the detector (1,2) as a constant, which when multiplied by the measured number of radon alpha particles predicts the number of radon beta particles to be subtracted from the total number of beta particles detected.

The remainder of beta particles can then be tested for statistical significance to determine if a man-made beta particle release is likely to have occurred (or whether the number is likely to be due to natural variation in the number of background beta particles detected). An alarm is triggered if a man-made beta particle release has been determined.

With reference to Figure 1, in a first embodiment of the detector (1) of the present invention, a temperature sensor (30) is positioned within the funnel (4) close to the detector head (8) where it can measure local temperature and report this value. A microcontroller (31) takes this temperature signal, and uses an internal look-up table to predict detector head (8) noise for that temperature. This number is sent to a DAC (32) and then to an analogue potential divider (33) which in turn presents a suitably scaled direct reference voltage as the LLD threshold value (22) of the LLD (20). The analogue potential divider (33) scales the voltage output of the DAC, which typically swings between 0 and 2.5V, to approximately the range 0 j

- 50mV that is needed for the beta LLD threshold value (22) setting related to electronic poise. This setup allows the optimum LLD threshold value (22) to be used, i.e. setting it just above the system noise.

In this first embodiment of the detector (1) where fitted with a temperature sensor (30), the noise of the detector (1) can be determined as follows:

A representative detector head's (8) noise is measured against temperature, producing a curve that shows noise increasing with temperature. An exponential equation akin tp the Ebers-Moll semiconductor equation is fitted to that curve:

Equation 1 v_n = sqrt(a.e^bT+c²)

Where n_h = RMS noise voltage after shaping and at input to LLD (20) T = absolute temperature (K) e = base of natural logarithms a, b, and c are experimentally determined constants.

The equation allows prediction of detector head (8) noise at any temperature within those measured by the temperature sensor (30).

Ideally, as is the case here, the detector head (8) temperature is measured directly. The temperature sensor (30) should be positioned as close to the detector head (8) as possible. Alternatively the air temperature can be measured through the funnel (4) as an approximation to the detector head (8) temperature.

With reference to Figure 2, in a second embodiment of the detector (2) , noise is measured directly.

Noise would ideally be measured as power, but although this can be done, it may prove cumbersome, so it is more usual to measure and quantify it in terms of voltage, more specifically Root Mean Square (RMS) voltage.

RMS voltage noise measurement requires an additional Analogue to Digital Converter (ADC) (not shown) connected after the shaping amplifier and capable of sampling fast enough and with sufficient amplitude resolution that conversion of its digital record by a perfect Digital to Analogue Converter (DAC) would recreate the original noise waveform leaving the shaping amplifier (12) without significant distortion. Although a separate ADC could be used, it is more convenient to choose a microcontroller (31) incorporating an ADC because this keeps the fast data and calculations within the microcontroller (31), allowing the microcontroller (31) to produce a single slowly-changing parameter that is directly interchangeable with the result of the look-up table used with the first embodiment (1).

It will be appreciated that this additional ADC must sample much faster than the ADC: used for counting radioactive events; a minimum of 500kS/s (500,000 samples per second) or more (still slow by contemporary standards). As a first step in calculating noise, the digital value of each sample leaving the ADC must be squared and a running total kept.

The required time duration of the running total is proportional to shaping time, so ai 10ps shaping time might require a running total duration of 1ms. At the end of the duration, the final running total is divided by the number of samples that contributed to obtain the mean of the squares, and its square root taken to obtain the Root of the Mean of the Squares (RMS). This is the RMS noise voltage of the system at that point, and because it was measured as an RMS voltage, it is one standard deviation of the noise at that point. Because noise has a known distribution, once the standard deviation is known, probabilities of exceeding multiples of that standard deviation can be calculated.

It will be appreciated that system electronic noise can only be measured accurately in the absence of detected radioactive events, and this is done by temporarily stopping the running total (and its clock) for an appropriate duration whenever a radioactive event is detected by the detector (2). Since typical alpha/beta background is fewer than ten counts per second, a noise running total duration of even 100ms is only likely to interrupted once or twice. Radioactive events are detected by their pulses exceeding a the LLD threshold value (22), and the rise time of the pulse is defined by the shaping amplifier, so there must be some noise samples just prior to any radioactive event detection that are invalid. To prevent these invalid samples being used, a First In First Out (FIFO) buffer memory (not shown) of perhaps ten samples can be placed before the running total calculation that allows invalid samples (and their associated clock time) to be discounted.

Required periodicity of noise measurement is determined by the thermal time constants of the detector head (8) and its coupled masses. Typically, this time constant is of the order of 600s or 10 minutes, and the thermal time constant of the apparatus' (2) environment is likely to be much longer. Thus, measuring noise every 30s would be fast enough to accurately track the changing noise of the detector (8) even in an environment having extreme temperature changes. In a more stable environment, noise measurement could be slower. Realistically, since the noise measurement chain is likely to be dedicated to noise, there is little advantage to running it slower, and it might as well update its noise at the fastest necessary rate of one twentieth of detector (8) thermal time constant.

Referring again to Figure 2, in this second embodiment (2) noise is measured at the signal input (21) of the LLD (20). The signal (40) at the output of the coupling network (13) (which has had any DC offset removed) is fed into the fast ADC within the microprocessor (31), which then calculates the Root Mean Square (RMS) noise voltage (which is a constant voltage related to temperature) from this signal (40). Having determined the RMS voltage noise; each periodic measurement is presented to microcontroller (31) which records it and presents a continuous value to the DAC (32) which via an analogue potential divider (33) sets the threshold value (22) of the LLD (20).

Provided that there is no DC (Direct Current), RMS measurement of noise is directly equivalent to stating that noise's standard deviation (lo). Provided the noise has a Gaussian distribution, lo includes 68% of a two-tailed distribution, whereas 3o includes 99.7%. Or, to put it another way, only 0.3% of events fall outside 3o. The argument is slightly modified for CAMs because noise swings both positive and negative, yet the detector head (8) only produces one polarity of pulses, so the LLD (20) responds only to that polarity, and the statistical argument uses a single-tailed distribution (lo = 84%, 3s = 99.87%). Secondly, system noise must be Gaussian, which is normally ensured by a pulse shaping time constant

<10ps, because the necessary high-pass filter that is part of pulse shaping (i.e. integral within the shaping amplifier (12)) renders system 1/f noise insignificant. Thus, by setting the LLD threshold value (22) in a CAM to a fixed multiple of the system's noise, we have a constant and known probability of false beta particle counting due to noise.

A digital oscilloscope preceded by a bandwidth-defining filter (such as a detector (2)'s shaping amplifier (12)) is capable of measuring electronic noise by invoking RMS (Root Mean Square) measurement and Y amplifier AC (Alternating Current) coupling. Random noise is always measured as RMS. It is necessary that the oscilloscope's sample rate be high enough to avoid aliasing, typically requiring a sample rate more than ten times the shaping amplifier's (12) centre frequency. As an example, 5ps 4^th order shaping requires > 250kS/s. This method can be used to calculate precisely where to set the LLD threshold value (22), rather than through trial and error. Provided that the noise model is valid for the detector head (8) in use, a different detector head could be substituted and the system's false alarm probability would remain unchanged. Mathematically linking predicted RMS noise and LLD threshold value (22) setting to give a fixed and known false alarm probability provides an enhancement over the prior art.

It is assumed that all ambient temperature changes are slow, and that detector head (8) temperature changes slowly.

The detectors of the embodiments (1, 2) of the invention thus provide an improved sensitivity to target particles without increasing false alarms from the detector (1, 2). Modifications and improvements may be made without departing from the scope of the invention.

The detector of either embodiment (1, 2) described herein may be further enhanced by improving the background beta particle compensation, i.e. improving the method of determining the expected count of beta particles due to naturally occurring radiation. The existing methods of background compensation discussed earlier determine the ratio of (detected) radon alpha particles to (estimated) radon beta particles, and assumed that all emitted beta particles were detected. However, the following losses in a detector (1,2) cause the calculated alpha particle to beta particle ratio to be different to actual alpha particle to beta particle ratio in the (immediate) environment:

1) the beta LLD (20) is not set to OkeV (therefore not all beta particles are counted);

2) self-absorption of alpha particles (therefore not all alpha particles are detected); and

3) alpha particle loss across an air gap (therefore not all alpha particles are detected). The following embodiments improve beta background compensation (removal of background beta particles from the total number of detected beta particles) of the detector (1,2).

This improved method compensates for the variable setting of the LLD threshold value (22) by calculating the detected radon beta particle to detected alpha particle ratio for each LLD threshold value (22), thereby adjusting the predicted beta particle background, and improving accuracy of radon beta particle compensation. Since the daughters of each decay series (uranium, thorium) are known, and their individual beta particle spectra are known, it becomes possible to calculate the proportion of those beta particles that will be detected for each possible setting of LLD threshold value (22), and for each decay series. For a given decay series, the previous process is applied to each beta radon daughter, and the results of all daughters summed for each LLD threshold value (22), making it possible to calculate a detected radon beta particle to radon alpha particle ratio corrected for any setting of the beta LLD threshold value (22) within the chosen decay series. This process is applied individually to the uranium and thorium decay series (and could be applied to the actinium series).

Provided that the ratio of alpha particles attributable to uranium or thorium is known (described below), it becomes possible to split total alpha count into uranium and thorium alpha particles. Once split, it becomes possible to predict uranium beta particles and thorium beta particles, summing the two to predict the total beta particle background. Other possible methods of determining the background beta particle count include pseudo-coincidence whereby the likelihood of a beta particle being present is directly determined by virtue of the presence of an alpha particle.

If the radon (background) alpha particle to radon (background) beta particle ratio has been individually calculated for both the uranium and thorium series, and the actual ratio of alpha particles attributable to thorium to the alpha particles attributable to uranium (henceforth the 'thorium to uranium ratio') is calculated, the number of radon beta particles can be predicted more accurately. Although this technique could be used with a conventional [fixed LLD, its use with a beta LLD (20) that varies dynamically is important if maximum improvement is to be gained. The thorium to uranium ratio may be calculated as follows. The 8.78MeV alpha particle peak is due solely to the thorium series and the 7.69MeV peak is due to the uranium series. The tail of the 8.78MeV peak extends under the 7.69MeV, but once subtracted, the remaining 7.69MeV peak is due to the uranium series. Once the two peak counts are known, it becomes a simple matter to calculate the relative proportions of uranium and thorium, and predict the radon beta particle background. Note that this simplified calculation makes the assumption that secular equilibrium has been achieved - i.e. that the air is old enough that peak amplitudes for each radon daughter have stabilised. It is possible to determine the age of the air using peak counts and the Bateman equations, further increasing prediction accuracy. Nevertheless, provided that the thorium ratio is calculated for each alpha spectrum, its rjesult is good enough.

In practice, building materials containing thorium do not trap all their ²²⁰Rn and solid radon daughters from the thorium series are routinely detected by detectors (1,2). To allow fory this, some existing detectors (such as the Ultra Electronics^® SmartCAM) arbitrarily assume a fixed 10% thorium proportion (relative to uranium) and modify their fixed radon beta particle to radon alpha particle ratio appropriately. Thorium proportion is dependent on the actual composition of building materials, and since transport costs generally mandate that these be locally sourced, thorium proportions vary with location. This means that the fixed proportion method is an approximation and an improved method of background compensation is possible.

Existing methods of beta background compensation as discussed above rely on measuring alpha particles and calculating the expected number of according beta particles. Theoretical physics and Monte Carlo analysis of the efficiency of the detector shows that air gap losses are different for alpha particles and beta particles, which unfortunately means that an alpha to beta particle ratio calculated purely from consideration of half-lives does not match the measured ratio at the filter.

Each time the detector (1,2) assembles a new alpha particle spectrum, the thorium to uranium ratio is re-calculated, and detector noise measured or predicted to determine appropriate beta LLD threshold value (22) setting, enabling the current beta particle to alpha particle ratio to be calculated, finally enabling a more accurate radon beta subtraction.

The detector (l,2)can be further enhanced by compensating for the self-absorption of radon alpha particles. As dust (7) builds up on the filter (3), self-absorption within the dust (7) reduces the apparent alpha particle emission of radon daughters, but because beta particles interact less with matter, detected beta particle emission is hardly affected. If the degree of dust (7) build-up on the filter (3) is known, this loss can be estimated and compensated.

It is usual for a detector (1,2) to have air flow and pressure measurement as a means of detecting filter (3) problems (tears, clogging), but by combining the measurements, filter (3) air resistance may be calculated from the air flow through it and the pressure drop across it. Resistance is calculated by:

Equation 2

pressuredrop across filter

resistance = - - ^£ - flow through filter

New filters can only be inserted with the gate (not shown) open and air flow turned off, so a detector (1,2) may reset its systems upon gate closure. Thus, a reference measurement of air resistance can be taken automatically shortly after insertion of a new filter (3), and any subsequent increase may be attributed to dust (7) and is directly proportional to the loss of detected alpha particles. Thus, the increase in air resistance can be used to correct the prediction of detected radon beta particles thereby further enhancing the accuracy of the radon beta particle compensation over time.

The detector (1,2) can be further enhanced by compensating for air gap losses. The dust- collecting filter (3) is a planar source and is separated from the planar detector by an air gap. Source/detector geometry places a geometric limit on the proportion of radioactive particles emitted by the source that can directly reach the detector head (8), and therefore there is a loss in the air gap (i.e. difference between actual and detected particles). If 50% of beta particles known to be emitted are detected, we would state a beta efficiency of 50%; similarly a figure of alpha efficiency could be calculated, both purely due to geometry.

In practice, alpha particles interact more with matter than beta particles, and radon alpha particles typically lose IMeV taking the shortest path across the air gap, but oblique paths expend even greater energy, perhaps stopping an alpha particle from completing a geometrically viable path and reaching the detector, reducing alpha efficiency below its geometric value. Conversely, beta particles are rarely stopped by the air gap, and approach the geometric limit of efficiency. Thus, the air gap significantly reduces alpha efficiency, but leaves beta efficiency substantially unchanged.

As an example, a 25mm filter facing a 25mm detector head (8) across a 5mm air gap had a measured alpha efficiency (²³⁹Pu) of 23.1%, but the pure geometric efficiency predicted by Monte Carlo analysis for this geometry was 30.5%, so the measured loss from geometric efficiency is caused by the intervening air, and this effect is well known. The measured beta particle efficiency of this configuration to ³⁶CI was 27%, but the LLD threshold value (22) was set to 60keV, resulting in only 88% of beta particles reaching the detector head (8) being counted. 30.5% x 88% = 27%, showing that the intervening air has negligible effect on beta particles. Far more significantly, it is common for a detector to quote similar alpha and beta efficiencies, which seems logical to the user, but this is really a coincidence caused by the typical lOOkeV choice of beta LLD threshold value (22).

This new technique can be used to incorporate the ratio of geometric efficiency (assumed to be beta particle efficiency) to measured alpha efficiency as a modifier to more accurately predict the number of detected radon beta particles. Whereas existing air gap compensation techniques result in a difference of nearly 40% between measured and actual radiation, on test this new technique reduced the error to 8%, i.e. a by factor of five. The detector (1,2) can be further enhanced by compensating for age of air. Accurate! beta particle detection splits in to two main problems, namely the deficiencies with the measurement / detection system and the age of the air. The first problem has been addressed but there are existing known techniques for compensating for the age of the air which can be used in addition to this invention to further increase the accuracy of detection.

If the air in a room (or any space from which it is being collected to be analysed) is being interchanged at a given rate, it is possible the air has been interchanged before the two rjadon decay series to have reached stability (secular equilibrium), meaning that some decay daughters (and their associated alpha particles and beta particles) may not be seen before being extracted from the room or space. This loss is predictable using the Bateman equations from measurements on radon alpha particles. The Bateman equations are described in more detail at the following reference: Bateman H 1910 - solutions of a system of differential

I

equations occurring in the theory of radioactive transformations. Proceedings of Cambridge

Philosophical Society 15 pp423 -427.

Claims

Claims

1. A detector for detecting airborne beta particles which generates an electronic noise in use, said detector comprising a detector head for measuring energy, means of determining an electronic noise value of the detector, a low level discriminator threshold, and a means of adjusting the low level discriminator threshold, wherein the low level discriminator threshold is configured to an initial value, the means of determining an electronic noise value of the detector detects the electronic noise value of the detector, the electronic noise value is compared against the low level discriminator threshold value, and if the electronic noise value differs from t ie low level discriminator threshold value, the low level discriminator threshold value is varied, and energy measured by the detector i head below the low level discriminator threshold value is disregarded.

2. A detector according to Claim 1 further comprising a temperature sensor for measuring at least one temperature of the detector, and a means for deriving the electronic noise value from the at least one temperature measured by the temperature sensor.

3. A detector according to Claim 1 or Claim 2 wherein the low level discriminator threshold value is configured to be variable according to a multiple of the standard deviation of the electronic noise value.

4. A detector according to any preceding claim configured to determine the electronic noise value continually thereby permitting the low level discriminator threshold value to be varied continually over time.

5. A detector according to Claim 1 wherein the detector comprises! a means for directly measuring the electronic noise value of tlie detector, and the low level discriminator threshold value is variable according to the electronic noise value.

6. A detector according to Claim 5 wherein the low level discriminator threshold value is configured to be variable according to a multiple of the standard deviation of the electronic noise value.

7. A detector according to Claim 5 or Claim 6 configured to determine the electronic noise value continually thereby permitting the low level discriminator threshold value to be varied continuously over time.

8. A detector according to any previous Claim which further comprises a means of compensating for an expected count of beta particle's due to naturally occurring radiation.

9. A detector according to Claim 8 further comprising a means of modifying the expected count of beta particles due to naturally occurring radiation according to the low level discriminator threshold value.

10. A detector according to any preceding Claim further comprising a filter for collecting dust.

11. A detector according to Claim 10 further comprising a vacuum pump for forcing air to flow through the filter.

12. A detector according to Claim 10 or Claim 11 further comprising a means of measuring an air resistance across the filter and a means of adjusting the expected count of beta particles due to naturally occurring radiation according to the measured air resistance across the filter.

13. A detector according to any of Claims 10 to 12 further comprising an air gap between the filter and the detector head, and a means of modifying the expected count of beta particles due to naturally occurring radiation to compensate for losses due to the air gap.

14. A detector according to any preceding Claim further comprising a means of compensating for age of air.

15. A method of detecting airborne beta particles comprising the steps of:

- taking a detector according to any of Claims 1 to 14 ;

- determining an electronic noise value of the detector;

- setting the low level discriminator threshold to an initial value;

- determining the electronic noise value of the detector;

- comparing the low level discriminator threshold value to the determined electronic noise value;

- setting or varying the low level discriminator threshold value according to the determined electronic noise value.

16. A method according to Claim 15 wherein the step of determining the electronic noise value of the detector comprises:

- determining the electronic noise value of the detector by measuring the electronic noise directly.

17. A method according to Claim 16 wherein the steps of:

- determining the electronic noise value of the detector by measuring the electronic noise directly;

- setting the low level discriminator threshold value according to the electronic noise value;

are repeated two or more times.

18. A method of detecting airborne beta particles comprising the steps of: - taking a detector according to any of Claim 2, 3 or 4;

wherein the step of determining the electronic noise value of the detector comprises:

- measuring at least one measured temperature of the detector;

- determining the electronic noise value of the detector by estimating the noise from the measured temperature of the detector.

19. A method according to Claim 18 wherein the steps of:

- measuring a measured temperature of the detector;

- determining the electronic noise value of the detector by estimating the noise from the measured temperature of the detector;

- setting the low level discriminator threshold value according to the electronic noise value;

are repeated two or more times.