Answer to Question #9411 Submitted to "Ask the Experts"
The following question was answered by an expert in the appropriate field:
The literature recommends that one not use a Geiger-Mueller (GM) detector for a radiation protection survey testing radiation generated by a medical linear accelerator to determine radiation levels in other nearby rooms and on other nearby floors. Rather, it is suggested that an ionization chamber monitor yields more accurate readings.
Assuming that a GM detector was used for such a survey, is there any rule-of-thumb with respect to the degree to which the findings produced may understate the true conditions?
It has been suggested that that the true readings would be no more than 25 percent higher than the readings obtained with the GM detector. Is this a reasonable assumption? Or, might the actual readings be far greater? If the latter is the case, what percentage differences might be likely?
If relevant, the GM readings for areas of concern here were 0.028-0.056 nSv s-1 for certain areas, 0.14 nSv s-1 for other areas, and 0.28 nSv s-1 for yet other areas.
In general, a GM detector is not suitable for measurements from pulsed sources of photon radiation. Please see the answer to question #4511 on the HPS Ask the Experts web site for a discussion of the rationale for this.
Regarding your question as to how accurate the reading from a GM detector might be if the detector were used for a survey outside the LINAC shield, we can look at a simple example of what might be expected. Let us assume for our example that (1) the true dose rate outside the shield was 0.05 mrem h-1 while the LINAC was operating, (2) the machine pulse rate was 100 pps (pulses per second), and (3) the pulse width was 5 microseconds. The GM detector is an event detector that has a typical dead time of 10 s to 100 s of microseconds. With a pulse rate of 100 per second and a pulse duration of 5 x 10-6 seconds, for every second of operating machine time, radiation would be produced for 5 x 10-4 seconds, and no radiation would be produced for 0.9995 seconds of every second—i.e., the pulses are very short duration and there is a relatively long delay of about 0.01 seconds (104 microseconds) between sequential pulses.
This means that when the GM detector views a train of radiation pulses, each pulse being significantly less than the dead time of the instrument, and one pulse separated from the next by a time long compared to the dead time, the detector will record the radiation from each pulse as a single event—i.e., one count—regardless of what the dose per pulse is. Thus, we might expect a count rate as high as 100 counts per second. Commonly used GM detectors often exhibit a response of about 1.8x104 cps per μSv s-1 when calibrated in a constant field of photons. Applying this value, the measured rate could be as high as 5.6 nSv s-1.
The photon detection efficiency of GM detectors is quite low, typically considerably less than 1 percent, and every pulse may not produce a count, depending on how many photons intercept the detector, and the measured rate could be significantly less than this. The point is, however, that there is not a predictable relationship between the actual dose rate and the GM reading when certain pulsed radiation fields are being assessed. Without specific information about what the actual dose rate is, it is not feasible to state that the “measured" dose rate will be greater or less than the actual dose rate by any particular amount.
The above discussion assumes that the GM detector had been calibrated in a constant and continuous photon radiation field. The most common radionuclide source used for calibration is 137Cs (137mBa) , which yields a photon energy of 0.662 MeV. This adds another layer of potential response uncertainty related to photon energy response, independent of the pulsed radiation problems. This has to do with the fact that therapeutic LINACS produce high-energy photons, some in excess of 10 MeV for high-energy machines. The high-energy photons are capable of inducing significant pair production interactions in the GM detector, such events not being possible at energies less than 1.022 MeV. Because the GM detector is typically fabricated of materials with higher atomic numbers than that of soft tissue or air, and because the probability of interaction by the pair production process depends strongly on atomic number of the medium, the extent of such pair production in the detector will likely exceed the extent of production in these lower atomic number media, thus producing falsely high dose rate estimations.
In summary, there are generally too many response inaccuracies and uncertainties associated with the use of a GM detector for measurements around medical linear accelerators. Appropriate ionization chambers, at times integrating chambers that can be exposed for a sufficient time to obtain reliable readings, are desirable and are usually preferable.
I hope this addresses your concerns.
George Chabot, PhD, CHP