Secondary Logo

Journal Logo


Testing the NASA BioSentinel Pixel Dosimeter Using Gamma-ray and Neutron Sources at the LLNL Calibration Lab

Homan, J.1; Lusby, T. C.1; Ricco, A. J.1; Mintz, J. L.2; Braby, L. A.3; Straume, T.∗,1

Author Information
doi: 10.1097/HP.0000000000001502
  • Open



Measurements in space, using a pixel-based sensor chip identical to the one used in the BioSentinel Pixel Dosimeter (BPD), have indicated an uncertainty in the absorbed dose of 10 to 15% (e.g., Gaza et al. 2017). This relatively large uncertainty may be due to the uncertainty in the actual dose as determined by other instruments in space used for comparison and uncertainties in the modeling of the space radiation environment. To determine the accuracy of the BPD exposed to standard radiation sources (60Co, 137Cs, and 252Cf), a ground-based version of the BPD that will be onboard the NASA BioSentinel mission (Ricco et al. 2020) was exposed at the Lawrence Livermore National Laboratory (LLNL) Radiation Calibration Laboratory (RCL) on 4 December 2019. The findings from those tests are reported here. It is noted that the radiation sources used here do not represent the complex radiation environment in space, which will likely influence the measurement uncertainties and will be described in detail in a separate publication. Only a brief discussion of dosimetry in the space environment is provided here in the “Dosimetry results” section.

The NASA BioSentinel mission, a secondary payload onboard Artemis-1 (NASA 2020), expected to launch in 2022, will evaluate the effects of the deep space environment on a biological test system, yeast (Bhattacharya et al. 2019). The BPD will provide radiation dosimetry to help interpret the biological response. This is a 6 to 9 mo no-return mission that will separate from Artemis-1 on a lunar fly-by trajectory leading to heliocentric orbit several days after launch, and data will be telemetered to Earth. The BPD uses a silicon sensor and provides both dose and LET (linear energy transfer) information for charged particles between 0.2 keV μm−1 and 300 keV μm−1. Charge produced in the silicon is measured by a TimePix chip initially developed at CERN (the European Organization for Nuclear Research) and advanced by NASA for space applications (e.g., Pinsky et al. 2012; Stoffle et al. 2016; Gaza et al. 2017).


The RCL is a low-scatter facility that consists of a 12.2 m × 9.1 m × 6 m concrete-shielded room with suspended Al grid flooring to reduce scatter. The RCL was designed to provide precision irradiations for the calibration of personal and environmental dosimeters. For these tests, three sources were used: a 60Co gamma-ray source (400608), a 137Cs gamma-ray source (400804), and a 252Cf source (SR-3050-OR). The BPD was exposed at selected distances and for selected exposure durations chosen to provide a range of absorbed dose rates relevant to the BioSentinel mission, estimated to be 10 to 20 μGy h−1 from GCR inside a lightly shielded spacecraft such as BioSentinel (Straume et al. 2017) with higher dose rates during significant solar events.

During these test measurements, RCL facility staff operated the sources and provided their facility-based standard dose rate information. The sources were operated remotely via a Hopewell Designs Irradiator model N40 that pneumatically transfers selected sources from shielded locations through an Al tube to the irradiation position. The data were acquired from the BPD via data cable connected to the BioSentinel data acquisition instrument and then to a PC laptop outside the test chamber.

Fig. 1 shows detector positioning during irradiation. The BPD (A) is seen attached to an aluminum frame on the movable table on a rail. The location of the source during irradiation is near the top of the vertical Al tube (B). A linear positioning system provides accurate positioning of the dosimeter relative to the source center, ±1 mm. The BPD was placed at the same height as the source. The source-to-detector distances employed were 100 cm and 200 cm for 60Co, 200 cm for 137Cs, and 200 cm for 252Cf. A data acquisition instrument running the flight software (C) was located on a cart adjacent to the dosimeter stand. A digital camera (D) was mounted above the BPD. Neutron-sensitive and gamma-sensitive pocket dosimeters were used for source monitoring, but not for dosimetry.

Fig. 1:
Radiation exposure of the BPD at LLNL. The BPD board with its sensor chip and associated electronics (A) is seen suspended from an Al support on the dosimeter stand. During exposure, the radiation source is pneumatically positioned near the top of the Al tube (B). A data acquisition instrument (C) is located on the cart adjacent to the movable stand. A digital camera (D) was mounted above the BPD. Photograph provided by LLNL.

Dosimetry results

Listed in Table 1 are the sources employed, source-to-detector distances, exposure durations, cumulative doses in Si, and dose rates in Si. The background measurement was made in the same position but without the radiation sources. It is observed that background (which would include ambient radiation in the facility and electronic noise) is a small fraction of the dose accumulated during radiation exposure. The dose rates measured in silicon for the sources and distances selected range from 18.1 μGy h−1 to 709 μGy h−1.

Table 1 - Absorbed dose measurements with the BPD (4 December 2019).
Radiation Source ID Distance (cm) Exposure duration (min) Cum dose in Si
Dose rate in Si
(μGy h−1)
Background NA NA 60 0.218 0.218
60Co 400608 100 30 35.6 71.2
60Co 400608 200 30 9.03 18.1
137Cs 400804 200 10 118 709
252Cf SR3050-OR 200 10 6.29 37.7

Listed in Table 2 are comparisons between BPD dose rates and RCL facility standard values for 4 December 2019. For these comparisons, both the BPD values and the facility values were for ICRU (International Commission on Radiation Units and Measurements) soft tissue. The conversion from dose in Si to dose in tissue was calculated from the ratio of the mass energy absorption coefficients obtained from NIST (2004). The gamma-ray energies used for the selection of mass energy absorption coefficients were 1.25 MeV for 60Co gamma rays, 0.66 MeV for 137Cs gamma rays, and nominal 1 MeV for 252Cf gamma rays. For a bare 252Cf source, the spectrum of gamma-ray energies has an approximate mean of 1 MeV (Regnier et al. 2012).

Table 2 - Comparisons with LLNL facility standard values (4 December 2019).
Radiation Source ID Distance (cm) BPD dosesa soft tissue (μGy h−1) Facility dosesb
soft tissue (μGy h−1)
60Co 400608 100 78.6 80 ± 4 −1.8%
60Co 400608 200 19.8 20 ± 1 −1.0%
137Cs 400804 200 782 813 ± 41 −3.8%
252Cf SR3050-OR 200 41.5 125 ± 13 −66.8%
aFollowing background subtraction the dose rate measured in Si was converted to dose rate in soft tissue (ICRU 1989) using ratios of mass energy absorption coefficients.
bBased on NIST traceable calibrations by RCL personnel. Uncertainties were taken as ±5% for the gamma ray sources and ± 10% for 252Cf. Precision from multiple repeat measurements would be better than this but here an attempt was made to consider overall uncertainties. For 252Cf this includes calibration, source strength, contributions from other Cf isotopes, scatter, and effective center of dosimeters. Soft tissue composition based on ICRU 1989.

The facility values for 60Co, 137Cs, and 252Cf in Table 2 are based on benchmark dose measurements for these sources and distances provided by RCL personnel and decayed to 4 December 2019. For 252Cf, separate neutron and gamma-ray contributions to the dose rate were also provided by RCL personnel (Radev 2016) and decayed to 4 December 2019. Comparisons between the BPD results and the facility standard results show good agreement for the gamma-ray sources. However, for 252Cf, the dose rate measured by the BPD differs substantially from the benchmark facility value. The facility dose rate was measured using tissue-equivalent dosimeters (which provide total neutron plus gamma dose), whereas the BPD sensor is silicon, which receives much less dose from neutrons since it does not contain hydrogen. It is observed that the dose to silicon from exposure to the 252Cf source is essentially the dose due to the fission gamma rays only. Based on RCL standard calibrations of the 252Cf source at 200 cm, the total (neutron + gamma) dose rate was 125 ± 13 μGy h−1, the neutron-only dose rate was 81 ± 8 μGy h−1, and the gamma-ray contribution was 44 ± 5 μGy h−1. As seen in Table 2, the BPD measured a total dose rate of 41.5 μGy h−1 at this distance, in agreement with the gamma-ray-only contribution from the 252Cf source.

This demonstrates that the BPD performs well as a dosimeter for gamma rays, i.e., primarily secondary electrons from gamma-ray interactions. It is also expected to perform well for other charged particles (for example, cosmic ray protons and solar protons) and for heavier charged particles with stopping power less than that which would saturate the individual pixels of the detector. This has been confirmed in space using an identical pixel-based sensor chip (the Battery-operated Independent Radiation Detector, BIRD), which was onboard the first unmanned NASA exploration flight test (EFT−1) of the Orion Multipurpose Crew Vehicle (Bahadori et al. 2015; Gaza et al. 2017). They compared mission-absorbed dose detected by BIRD with those from several TL, OSL, and CR-39 detectors and demonstrated agreement within about 12% (Gaza et al. 2017). During the 4.5 h flight test, EFT-1 orbited Earth and made passes through the Van Allen belts. In contrast, for BioSentinel, the BPD will measure the deep-space environment beyond the Van Allen belts where water-based fluid will be added to activate desiccated yeast (Bhattacharya et al. 2019). The conversion of absorbed dose measured in silicon to absorbed dose in the yeast cells will assume that the average dose to the yeast is the same as the average dose to the water-based solution containing the yeast, Dt,yeast = Dt,H2O = Dt,Si × C (dE/dxt), where Dt,yeast is the absorbed dose to the yeast cells during time interval t of the mission, Dt,Si is the absorbed dose measured in Si during time interval t, and C is an energy dependent factor converting dose in Si to dose in water for space radiation (Bahadori et al. 2015; Gaza et al. 2017). For EFT-1, C was calculated using an asymptotic limit of 1.23 from the ratios of stopping powers for water and silicon based on dE/dx for individual tracks obtained during the mission (Bahadori et al. 2015). For BioSentinel, C will be calculated based on the relevant space radiation environment, i.e., deep space beyond the Van Allen belts, and considering the positioning of the Si sensor within the BioSentinel (the BPD is placed directly on the box containing the yeast). This will be described in a follow-on publication. From available space-related tests and evaluations of the pixel-based sensors, the overall uncertainty for absorbed dose to the yeast cells is likely to be within 15%.

The insensitivity of the BPD to neutrons is consistent with observations from other silicon based detectors exposed to 252Cf radiation (Straume et al. 2016). It is noted that for the BioSentinel mission, secondary neutron fluence is expected to be low due to the small mass of BioSentinel (~14 kg) and therefore neutron detection may not be an important consideration for this mission.

The directional dependence of this device was not evaluated here. Radiation exposures were directed orthogonally to the pixel chip surface. It is expected that this detector may have some directional dependence. However, it is noted that GCR in deep space is essentially omnidirectional, minimizing the significance of directional response.

Spectrometry results

The algorithm installed in the BPD determines the LET of each event by analyzing the cluster of pixels assumed to be produced by a single charged particle depositing energy in the silicon sensor. The dimensions of the cluster of activated pixels are analyzed to determine the length of the particle track in the sensor assuming that the particle had sufficient range to exit the sensor; the charge recorded by each of the pixels in the cluster is summed to determine the energy imparted. The LET of the particle, averaged over its path length in the sensor, is the total energy imparted divided by the path length. The number of events at each value of LET is summed and stored for telemetry. Fig. 2 shows the results from our ground-based test measurements at LLNL. A comparison of the LET spectra shows that 60Co gamma rays at 100 cm and 200 cm and 252Cf gamma rays at 200 cm are essentially parallel on the plot, indicating similar LET distributions. This is consistent with their similar gamma ray energies (60Co 1.25 MeV mean, 252Cf fission gamma about 1 MeV mean). Also, 60Co at 100 cm results in about 4 times the counts observed for 60Co at 200 cm, as expected from the “inverse square law.” The complete LET spectrum (not seen in Fig. 2) also confirms that the BPD detects only the gamma-ray component of the radiation from the 252Cf source. Counts from neutron interactions would have been clearly observable at higher LETs if present.

Fig. 2:
LET spectra obtained by the BPD exposed to radiation sources at LLNL. The LETs measured are from the secondary charged particles produced by the interactions with the gamma rays in Si.

The 137Cs gamma ray response is notably different than the responses to 60Co and 252Cf gamma rays (Fig. 2). The spectrum for 137Cs is higher at low LET, consistent with the higher accumulated dose in the Si sensor (about a factor of 3 higher than that for 60Co at 100 cm, see Table 1) but decreases more rapidly with increasing LET, falling below the 60Co data for LET values above about 1.5 keV μm−1. Two potential causes for this difference are apparent: the high dose rate from the 137Cs source and the lower photon energy of 137Cs gamma rays.

The BPD operating system uses an adjustable time frame, or integration time, for acquiring data. Charge, produced by the interaction of radiation with the silicon detector, is collected on pixels until the end of the current time frame. The charge on each pixel is then read out and recorded before the next frame is initiated. If a large number of charged particle events occur in the silicon sensor during one frame, there is a chance that clusters of activated pixels produced by two independent charged particle tracks will overlap. Overlapping pixel clusters would be interpreted as a single track with longer path length in the detector and higher energy deposited than was produced by either of the interacting particles, thus resulting in an erroneous estimate of LET for that event. To avoid this type of error, the operating system monitors the number of pixel clusters in each time frame, and if a specified number is exceeded—roughly 3% of the total number of pixels—indicating a high dose rate, it decreases the duration of subsequent time frames. In order to avoid unnecessary dead time, if the number of activated pixel clusters falls below a specified value, the duration of the next frame is increased. The operating system corrects for dead time when calculating the absorbed dose. A few overlapping clusters may occur before the frame length is changed. Typically, when cluster overlap occurs, the energy imparted is recorded as the sum of the energies of the two particles, but the path length deduced from the dimensions of the cluster is less than the sum of the path lengths of the two particles, resulting in overestimating the average LET. This would be the opposite of what is observed, so it appears that the adjustment of the time frame for data acquisition has successfully prevented errors due to the high dose rate provided by the 137Cs source.

The lower photon energy of 137Cs results in a lower average secondary electron energy, which results in higher average LET and shorter average track length. Because of Compton scattering and the photoelectric effect, the electron energy spectrum is quite broad, including many low-energy electrons. These low-energy electrons have relatively large stopping powers and short paths. For example, a 40 keV electron has a track length of 38 μm in silicon. These short tracks stop within the detector and produce small clusters of activated pixels. Since the algorithm is based on the assumption that the track exits the detector, the path length calculated exceeds the actual path length, and the calculated LET is therefore less than the stopping power averaged over the actual track length. Consequently, the measured values for low-energy electrons, which are expected to have the higher LET values, are shifted to the left, resulting in the rapid decrease with increasing LET for the 137Cs data. There is probably a shift in the high LET values for the 60Co and 252Cf gammas as well, but it is less noticeable due to the higher mean energy of the electron spectra.


Based on comparison with NIST-traceable standards, it is evident that the BPD measures absorbed dose in silicon due to low-LET charged particles with an accuracy of better than 5%. This is well within the uncertainties expected for measurement of absorbed dose in space during the BioSentinel mission, which are likely to be in the 10% to 15% range (Gaza et al. 2017).

The BPD is insensitive to neutrons because of low H in Si and the energy of Si recoil ions is too low to be detected by this device. Evaluation of charged particle average LET is dependent on the range of the particles, with the LET of short-range particles being undervalued. Errors in evaluating average LET result from errors in the charged particle path length, not from errors in evaluating energy imparted, and therefore do not result in errors in absorbed dose.


The work was performed under the auspices of the NASA Ames Research Center in support of the BioSentinel mission. We thank Lillian Gavalas and the NASA JSC Human Health and Performance Directorate for providing the BPD to NASA Ames (also known as the LET Spectrometer). We especially thank the Lawrence Livermore National Laboratory for allowing us to expose our dosimeter to their calibrated radiation sources in the LLNL Radiation Calibration Laboratory and their review (LLNL-JRNL-827053). We thank Sharmila Bhattacharya and Sergio Santa Maria for their comments and edits on the manuscript, and also thank Daniel Stone of LLNL for discussions regarding RCL dosimetry.


Bhattacharya S, Ricco AJ, Hanel RP, Crusan J. BioSentinel [online]. 2019. Available at Accessed 11 June 2021.
Bahadori AA, Semones EJ, Gaza R, Kroupa M, Rios RR, Stoffle NN, Campbell-Ricketts T, Pinsky LS, Turecek D. Battery-operated independent radiation detector data report from Exploration Flight Test 1. Houston, TX: NASA; NASA Report: NASA/TP-2015-218575; 2015.
Gaza R, Kroupa M, Rios R, Stoffle N, Benton ER, Semones EJ. Comparison of novel active semiconductor pixel detector with passive radiation detectors during the NASA Orion Exploration Flight Test 1 (EFT-1). Radiat Meas 106:290–297; 2017.
International Commission on Radiation Units and Measurements. Tissue substitutes in radiation dosimetry and measurement. Bethesda, MD: ICRU; ICRU Report 44; 1989.
    NASA. NASA’s Lunar Exploration Program overview [online]. 2020. Available at Accessed 12 January 2021.
    National Institute of Standards and Technology. NIST Standard Reference Database 126. Tables 3 and 4 [online]. 2004. Available at Accessed 10 January 2021.
    Pinsky L, Empl A, Hoang S, Stoffle N, Jakubek J, Vykydal Z, Turecek D, Pospisil S, Kitamura H, Ploc O, Uchihori Y, Yasuda N, Amberboy C, Hauss J, Lee K, Semones E, Zapp N, Parker R, Cooke D. Preparing for the first Medipix detectors in space. In: Proceedings of the IEEE Aerospace Conference Big Sky, Montana; New York:IEEE Publishers; pp. 1–6; 2012.
    Radev R. Neutron spectra, fluence and dose rates from bare and moderated 252Cf sources. Livermore, CA: Lawrence Livermore National Laboratory; Report, LLNL-TR-688699; 2016.
    Regnier D, Litaize O, Serot O. Monte Carlo simulation of prompt fission gamma emission. Physics Procedia 31:59–65; 2012.
    Ricco AJ, Santa Maria SR, Hanel RP, Bhattacharya S. BioSentinel: a 6U nanosatellite for deep-space biological science. IEEE Aerospace Electronic Systems Mag 35:6–18; 2020.
    Stoffle N, Pinsky L, Kroupa M, Hoang S, Idarraga J, Amberboy C, Rios R, Hauss J, Keller J, Bahadori A, Semones E, Turecek D, Jakubek J, Vykydal Z, Pospisil S. Timepix-based radiation environment monitor measurements abord the International Space Station. Nucl Inst Meth Phys Res A 782:143–148; 2016.
    Straume T, Mertens CJ, Lusby TC, Gersey B, Tobiska WK, Norman RB, Gronoff GP, Hands A. Ground-based evaluation of dosimeters for NASA high-altitude balloon flight. Space Weather 14:1017–1031; 2016.
    Straume T, Slaba T, Bhattacharya S, Braby LA. Cosmic-ray interaction data for designing biological experiments in space. Life Sci Space Res 13:51–59; 2017.

    TimePix; detector, silicon; dosimetry; space radiation

    Written work prepared by employees of the Federal Government as part of their official duties is, under the U.S. Copyright Act, a "work of the United States Government" for which copyright protection under Title 17 of the United States Code is not available. As such, copyright does not extend to the contributions of employees of the Federal Government.