Track-Segment Irradiator

The track-segment platform was originally developed in 1976 at Brookhaven National Laboratory (BNL; Upton, NY) (12) and started in operation at the current location at Nevis Laboratories (Columbia University, Irvington, NY) in 1984 (10). It has since been extensively used for broad beam irradiations of thin biological targets (mainly cell monolayers) with ions having a precisely defined linear energy transfer (LET) (13–16). The operation of the platform is described in greater detail elsewhere (12, 17, 18). Briefly, the irradiation protocol consists of rotating a flat irradiation wheel with custom-made sample dishes over the vertical ion beam. The beam is defocused to homogeneously cover the 6 mm × 20 mm vacuum exit aperture. A 2.9-µm-thick havar foil serves as an interface between the vacuum system and the atmosphere. One of the advantages of this platform is a well-designed dosimetry protocol based on three detectors: a proportional counter that measures the average LET of ions in 6 µm of tissue; a surface barrier silicon diode that measures the total beam energy at the target; and an ionization chamber that measures dose to tissue and the dose rate. The proportional counter and the ionization chamber are filled with the methane-based tissue-equivalent gas. All three detectors are mounted on the dosimetry wheel which is exposed to the beam prior to the irradiation of samples. The beam monitor, which is a part of the motor drive system that controls the speed of the wheel rotation, continuously monitors the instantaneous beam current by wiping off a small portion of the beam penumbra. During the dosimetry protocol, the ionization chamber is scanned over the beam exit window instead of the sample dish. Scanning is done with the stepper motor which is controlled by the custom made software. The motor moves the chamber step by step, waiting at each position until the appropriate number of monitor counts is registered. In that way the monitor chamber signal defines the rotation speed of the wheel. Initial number of monitor counts per step can be defined within in the software. After the entire ionization chamber moves over the beam the deposited dose is calculated from the electric charge produced inside the chamber and a correction factor is defined to correct the number of counts per step to obtain the desired dose (input given to the software, usually 2–5 Gy). The chamber is then scanned over the beam with the new number of monitor counts per step and the correction factor is determined again as a ratio between the desired and the measured dose. The procedure is repeated until the correction factor becomes less than 1% different from unity. When this requirement is achieved, the dosimetry protocol is considered to be successfully finished and the rotation speed of the wheel is calibrated in terms of monitor counts/step/Gy. Now the wheel with sample dishes can be positioned on the irradiator and the irradiation protocol started. As an input, the operator provides the doses that should be delivered. The number of monitor counts per step is calculated for each dose according to the result of the dosimetry protocol. Sample dishes are scanned over the beam in the exact same way as the ionization chamber was. Irradiation of each sample dish starts by positioning the dish outside of the exposure field near the edge of the beam exit window. The wheel then rotates step by step, pausing at each position until the appropriate number of monitor counts is registered. The beam current and consequently the dose rate can be easily manipulated by changing the settings of the ion source and/or tuning the defocusing quadrupole magnets which blow up the beam size to homogeneously cover the whole exit window. As the dose rate is increased, it will take less time to reach the same number of monitor counts and the faster rotation speed will be required. The highest dose rate that can be used for irradiations at the track-segment irradiator is actually limited not by the available beam intensity, but by the maximum rotation speed that can be achieved with the current motor drive system. For instance, with the standard beam exit window (6 mm width), the motor system saturates at dose rates of approximately 10 Gy/s. To be able to perform irradiations at even higher dose rates (over 100 Gy/s), we had to narrow the exit aperture down to prevent the saturation of the wheel motor drive system. As a solution, a 100-µm thick nickel foil with a 1-mm-wide line opening was mounted on top of the current exit aperture.

Dosimetry assessment with films. The response of the Gafchromic^™ EBT3 films (Ashland^™ ISP, Wayne, NJ) was previously shown to be dose-rate independent up to the 10⁶ Gy/s range (19), which makes these films suitable for use as absolute dosimeters in FLASH experiments. We compared the actual doses received by the film at various dose rates with the preset dose values defined within the irradiation protocol for the track-segment program to check the dose accuracy of the track-segment irradiator in FLASH mode. Unlaminated EBT3 films, which are one half the thickness of regular EBT3 film and have an exposed active region on one side, were irradiated with the same set of doses (1, 2, 5, 10, 15 and 20 Gy) at instantaneous dose rates of 0.1, 1, 10 and 100 Gy/s. Three pieces of film were irradiated with each dose/dose rate combination. Unlaminated films were used to reduce the energy loss of ions before they reach the active layer since the active layer is directly exposed. The active layer of unlaminated film is also one half the thickness compared to the active layer on the standard EBT3 version. These features of unlaminated film help to keep the LET of the proton beam far from the Bragg peak and nearly constant throughout the active layer. This was important since the signal from Gafchromic films depends on the LET (20–24). Irradiation, handling, scanning and readout of the films were done exactly as described elsewhere by our group (20). The net optical density change, netOD, was used to describe the darkening of the films in the red channel and was calculated as:

where and are the mean pixel values of exposed and unexposed films, respectively, averaged over the selected regions of interest (1 × 1 cm region in the center of the film), and is the background scanner reading. Scanner reproducibility and film uniformity were taken into account by adding 1% error to uncertainty, σ_netOD, that was determined by propagating the errors , and . Final netOD values were determined by averaging over the three readings obtained from films exposed to the same radiation conditions (dose/dose rate). Dose-response data obtained at the lowest dose rate of 0.1 Gy/s were fitted to the expression suggested by the film manufacturer (Fig. 1A):

where netOD represents the change in optical density, D represents the dose and a and b are the calibration coefficients. This calibration was applied to translate the netOD values into absorbed dose for all other dose rates. It should be emphasized that due to the LET dependence, the obtained calibration coefficients are valid only for the LET value in tissue of 9 keV/µm, as measured with the tissue-equivalent proportional counter. The layers between the havar vacuum window and the target plane are similar at both irradiation platforms for FLASH studies, causing the energy of the beam reaching the samples to differ by less than 1%. Therefore, the calibration obtained at the track-segment irradiator with advanced dosimetry protocol can be applied as well to the films irradiated at the FLASH irradiator.

FIG. 1

Panel A: Change in optical density in the red channel of unlaminated EBT3 film as a function of absorbed dose to tissue after 6 Gy/min proton irradiation. The obtained nonlinear fit was used as a general calibration for the film dose response. Panel B: Comparison of measured absorbed doses and preset dose values defined within the track-segment irradiation protocol for various dose rates. Error bars represent the standard deviation in absorbed dose determined with films as described in the section, Dosimetry assessment with films.

Biological application: Clonogenic survival of cancer cells. The first FLASH irradiations of biological samples at the track-segment irradiator were performed with the aim of investigating the role of proton dose rate on the survival of two human and one mouse cancer cell lines. Human lung adenocarcinoma cells A549 (CCL-185; ATCC^®, Gaithersburg, MD) were grown in Dulbecco's Modified Eagle medium (DMEM) supplemented with 12% fetal calf serum (FBS); human myelogenous leukemia cells HAP1 (C859 Horizon) were grown in Iscove's Modified Dulbecco medium (IMDM) supplemented with 10% FBS; mouse mammary adenocarcinoma cells (TSA; kindly donated by Dr. Sandra Demaria from Cornell University, New York, NY) were grown in DMEM supplemented with 10% FBS. For every cell line, the media also contained 100 U/ml penicillin and 100 µg/ml streptomycin. Flasks were stored at 37°C in an incubator filled with humidified air and 5% CO₂. Cell survival was determined using the standard colony formation assay (25). Briefly, cells were trypsinized within 5–10 min of exposure, suspended in growth media, counted, diluted and seeded in 100-mm dishes with numbers resulting in ∼100 clonogenic cells per dish. Three replicates were done for each data point. After 10-day incubation, the cells were rinsed with phosphate buffered saline (PBS), fixed in 95% ethanol and stained with crystal violet. Macroscopic colonies with more than 50 cells were scored as survivors. Measured survival values were corrected for the system plating efficiency which ranged from 60% in the case of the A549 cells to 70% for the TSA cells, as measured with nonirradiated controls.

FLASH Irradiator

The FLASH irradiator was assembled as an extension to the horizontal part of the track-segment beamline. Approximately 1 m after the switching magnet that directs the ions towards the appropriate beamline, monoenergetic protons encounter a beam deflector comprised of two parallel 0.6-m-long metal deflector plates mounted 3 cm apart. Deflector plates are directly coupled to the output of the TREK^® voltage amplifier unit (Lockport, NY) with an amplification factor of 100. The amplifier is run with an offset voltage of 1,000 V applied between the plates. The transverse momentum introduced by this electric field bends the trajectory of the beam enough to make protons hit the slits closed down to 3 mm around the middle of the beam optical axis and positioned approximately 4 m behind the deflector. To expose the samples to intense proton beam pulses, the deflector gets discharged by sending a square 10-V pulse generated by a waveform generator (B&K Precision Corp., Yorba Linda, CA) to the TREK unit. The duration of the discharging pulse is equal to the beam-ON time and, in combination with dose rate, determines the total dose that will be delivered. When the deflector is discharged, the beam can reach the end of the beamline and exit into the atmosphere through a 2.9-µm-thick havar foil. The foil is suspended over the 10-mm-wide exit aperture and is strong enough to withstand a one-bar pressure difference applied over that area. An ionization chamber, placed between the havar foil and the target, is used as an online monitor of proton current. Samples are mounted on a motorized stage that moves in a 2D plane perpendicular to the beamline axis. The motion of the stage is controlled by the computer software (AMICron version 5.81) paired with a motor controller (Semprex^® Corp., Campbell, CA). A standard webcam (Logitech, Lausanne, Switzerland) is used for positioning of the samples in front of the beam. Sample holders are custom designed and manufactured according to the needs of each experiment, which makes the irradiation platform easily adaptable to samples of various shapes and biological complexity. The end-station is described further in Results and Fig. 3.

FIG. 3

Panel A: Schematic image of the FLASH irradiator end-station. Panel B: The shape of the exposure area of the FLASH irradiator as recorded by irradiating an EBT3 film with a dose of 10 Gy. Panel C: Horizontal dose profile along the white line shown in panel A with ±3% dose limits shown by intermittent lines. Panel D: The depth-dose profile of 5.03-MeV protons in water simulated by TRIM Monte Carlo software. Percentage depth dose (PDD) increases by 30% at a depth of 160 µm.

Dosimetry protocol. The dosimetry protocol at the FLASH irradiator platform is based on two devices: a thin parallel plate ionization chamber and unlaminated Gafchromic EBT3 films. The ionization chamber is used to set the dose rate and to monitor the beam current during the experiment while films are used to confirm that the appropriate dose was delivered to the target. Before the actual irradiation, the sample is moved away from the irradiation field and the beam is allowed to propagate unobstructed and reach the ionization chamber mounted in air, 2 mm apart from the havar foil. With continuous beam on, a current measured by the ionization chamber is directly proportional to the instantaneous proton dose rate. The beam intensity is then tuned until the desired reading from the ionization chamber is achieved. Characteristics and the calibration of the ionization chamber are described in the section on dose-rate measurements, below. Once the dose rate is set, the beam is deflected by the offset voltage on the TREK unit. A sample can be positioned in front of the exit aperture and irradiated by pressing the trigger button on the waveform generator. The duration of the voltage pulse used to discharge the deflector plates depends on the required dose and can be easily controlled in the settings panel of the waveform generator.

1. Absolute dosimetry with films. Doses that are delivered by high-intensity proton pulses during an experiment are verified by unlaminated Gafchromic EBT3 films to ensure that the LET in the active layer of the film remains nearly constant. A piece of film is mounted on the stage together with samples and positioned so that the active layer and the surface of the sample are at the same distance from the vacuum window. Each film is irradiated three times for each dose rate/pulse duration combination that is used within the experiment. Irradiated films are handled and processed as described above, in the section on track-segment film dosimetry. A dose-response calibration curve obtained with the track-segment facility at a dose rate of 0.1 Gy/ s was used to read the dose from the film signal. An LET correction of the form,

with being a dose-averaged LET in the sample, is required for thicker samples to account for an increase in LET throughout the sample. The value of 9 keV/µm corresponds to the LET in tissue that was used to obtain the calibration curve at the track segment. The final dose is determined as the mean of three film readings obtained with the same pulse conditions while dose uncertainty is obtained by propagation of the corresponding errors.

2. Dose-rate measurements with ionization chamber. Dose rate of the proton beam is monitored online by the custom-made ionization chamber, which consists of two parallel circular electrodes made from 0.1-µm-thick titanium microfoil (Goodfellow Corp., Coraopolis, PA) supported by 3.5-µm-thick mylar. Electrodes are fixed 3 mm apart and the interior of the chamber is filled with methane-based tissue-equivalent gas at ambient pressure. The ionization current generated by traversing protons is read with the electrometer and normalized to standard pressure and temperature by applying the correction factor,

where T is the temperature in degrees Celsius, and P pressure in mmHg at the time of measurement. The corrected current reading is independent from the ambient conditions and was calibrated in terms of the dose rate with the help of the unlaminated EBT3 film. This was done by exposing films to proton pulses of various duration while keeping the current through the ionization chamber fixed. Analysis of the irradiated films provided the relationship between the pulse length and the absorbed dose with dose rate as a factor of proportionality. The procedure was repeated for several dose rates spanning a range of five orders of magnitude (from 0.02 Gy/s to 1,000 Gy/s) while the lengths of the irradiation pulses were scaled to deliver approximately the same doses with different beam intensities. The obtained dose-rate calibration is again valid only for LET of 9 keV/µm. Experiments with thicker samples require LET correction, defined by Eq. (1).

Biochemical Application

1. The microfluidic flow-through system. An additional feature available at the FLASH irradiator is a microfluidic flow-through system that was manufactured to allow for larger volumes (up to 1 ml) of liquid samples to be irradiated with short-range protons. The system is convenient for experiments with blood samples, suspended cells or chemical solutions. It is based on a microfluidic device containing a 100-µm-deep and 1-mm-wide channel through which the sample is forced to flow by a standard syringe pump (New Era Pump Systems Inc., Farmingdale, NY). During the flow, media is exposed to a continuous beam of protons at a desired dose rate. The microfluidic device was fabricated using traditional Polydimethylsiloxane (PDMS) soft microfluidic fabrication methods (26), with the exception of our mold fabrication which was made using a craft cutter similar to the work by Taylor et al. (27). Briefly, we defined a pattern using AutoCAD^® (AutoDesk^® Inc., San Rafael, CA) which was then exported to a Silhouette Cameo^® craft cutter (Silhouette America, Lindon, UT). The craft cutter was used to cut a piece of clear polyester double-sided tape (Adhesives Research^® Inc., Glen Rock, PA). After cutting the desired pattern, the liner was removed from one side, leaving a total thickness of approximately 100 µm, and the pattern was adhered to a glass slide to act as a negative mold. The mold was covered with uncured PDMS and allowed to cure overnight at room temperature. The cured PDMS was released from the mold, inlet and outlet holes were punched using a 1.25-mm biopsy punch, and the PDMS structure was sealed to a 30-µm-thick piece of glass (Schott, Elmsford, NY) using oxygen plasma bonding. The ring structure used for this work had an outer diameter of 4.5 mm and an inner diameter of 3.5 mm, thus with a 100-µm height of the channel the total volume in the target area was approximately 2.5 µl.

2. Irradiations of Fricke solution at different oxygen concentrations. The microfluidic flow-through system described above is convenient for investigating proton beam-induced oxygen depletion in water-based solutions at FLASH dose rates. To demonstrate this, we exploited the dependence of the dose response of the Fricke chemical dosimeter (28–30) on the presence of oxygen in the solution. One liter of Fricke solution was prepared by mixing 0.392 g of ammonium ferrous sulfate hexahydrate [(NH₄)₂Fe(SO₄)₂], 0.058 g of sodium chloride [NaCl] and 0.4 M of sulfuric acid with ultra-pure distilled water (Invitrogen^™, Carlsbad, CA). Dosimetric properties of the ferrous sulfate dosimeter are based on the radiolytic oxidation of ferrous (Fe²⁺) to ferric (Fe³⁺) ions and subsequent change in optical density (OD) at 303 nm (28–33). However, the yield of Fe³⁺ ions postirradiation is reduced if all of the molecular oxygen is consumed from the solution by the products of water radiolysis (32, 34, 35). Oxygen depletion during proton FLASH irradiation was studied with Fricke solutions containing 21% O₂ and 4% O₂. The higher oxygen content corresponds to atmospheric conditions, while the lower value was chosen as a representative of tissue conditions, although the partial pressure of O₂ varies significantly from organ to organ (36). Solutions with different O₂ concentrations were prepared from the same stock by equilibrating the samples at either room air (for 21% O₂ solution) or using a gas mixture containing 4% O₂ and 96% N₂. Before every exposure, the microfluidic channel was washed twice with fresh Fricke solution and dried by flushing the appropriate gas (air or 4% O₂ mixture) through the channel. A syringe pump was used to propel the Fricke solution in the microchannel, which was exposed to a continuous proton beam at 1,000 Gy/s. It should be noted here that due to the parabolic velocity profile across the channel width (1 mm), the maximum medium velocity in the middle of the channel is approximately 50% higher than the velocity near the channel edges (37). As a consequence, each irradiated sample was exposed to a distribution of doses. The final dose assigned to the sample was actually the average dose that corresponds to the product of the average flow rate (set by the syringe pump) and the LET-corrected dose rate. Average LET in the media required for the correction was calculated by Monte Carlo simulations described below. Doses used in the experiment ranged from 0 to more than 1,000 Gy. At every flow rate, three irradiated samples of 1-ml volume each were collected in vials and sealed. Samples were transferred into a quartz cuvette and read 24 h after irradiation using the Spectronic GENESYS 7 spectrophotometer (Thermo Electron Scientific Instruments LLC, Madison, WI) set to measure the absorbance at 303 nm. The same cuvette was used for all samples. During the sample exchange, the cuvette was washed twice with ultra-pure water.

For comparison, 1 ml of Fricke solution samples were X-ray irradiated [160 kVp, 1-mm-copper half-value layer (HVL), 0.3-mm copper filtering] at a dose rate of 8 Gy/min. Dosimetry was performed immediately prior to irradiation, using a NIST-traceable ionization chamber (Radcal^®, Monrovia, CA) calibrated to air kerma. Readout of the irradiated samples was performed 24 h postirradiation, as described earlier.

3. Monte Carlo simulations. The LET of protons inside the microfluidic channel during the experiment with the Fricke dosimeter was determined from simulations done using the TRIM module of the Monte Carlo software SRIM (38). Protons with nominal energy of 5.4 MeV pass through a total of 2.9 µm of havar foil, 4 mm of air, 3 mm of methane-based tissue-equivalent gas, 30 µm of glass and 100 µm of Fricke solution which was represented with water. TRIM software was used to simulate the energy of the beam right before and after traversing the Fricke solution. The difference between those energies gives the energy deposited in the solution. The average LET was defined then as a ratio of the deposited energy and the thickness of the media (100 µm).

Performance of the Track-Segment Irradiator

The track-segment irradiator is optimized for irradiations of cell monolayers. Its dosimetry protocol measures dose, dose rate and LET of particles in the first several micrometers of tissue. With the exit aperture narrowed down to 1 mm, the track-segment irradiator is capable of delivering doses as small as 2 Gy at a dose rate of 100 Gy/s, which falls under the range for which the FLASH effect was observed with electron beams. Exposures at higher dose rates are also possible, but require a proportional increase of the lowest deliverable dose, which is limited by the maximum rotation speed of the wheel.

The performance of the irradiator at high dose rates was tested with unlaminated EBT3 films. The irradiation protocol was tuned to deliver the same set of doses at dose rates of 0.1, 1, 10 and 100 Gy/s. Figure 1A shows the comparison between the absorbed doses in tissue measured with films and the preset dose values defined in the control software. For all dose/dose-rate combinations the difference between delivered and preset doses is less than 3%, confirming that the irradiator, with its standard dosimetry and irradiation protocols, can be reliably used for irradiations at high dose rates.

Dose-response data obtained for the lowest dose rate of 0.1 Gy/s were used to calibrate the film (Fig. 1B). Data were fitted to Eq. (2) and the resulting fitting parameters (a = 0.7, b = 18.5) were used for transferring the netOD values into dose for all films irradiated at track segment and the FLASH irradiator.

Survival of cancer cells. In our previously published work on the biological effects of proton FLASH radiation in normal cells, we investigated the role of proton dose rate on the survival of normal lung fibroblasts, IMR90 (39). Here, we focused on a similar experiment involving three cancer cell lines. Figure 2 shows the relationship between the survival fraction and the absorbed dose for human A549 and HAP1 and mouse TSA cells that were proton irradiated at dose rates of 0.1, 10 and 100 Gy/s. To analyze the data, survival fraction (SF) values were natural log (ln) transformed. The nonlinear regression was applied to fit the ln(SF) versus dose relationship to the second-order equation:

where α and β are the standard coefficients of the linear-quadratic model. Calculated α and β values are shown in Table 1. As shown in Fig. 2, for each cell line there is no difference in the survival trends regardless of the dose rate. Consequently, the fitted α and β values show no significant variation among different dose rates, as tested using the Student's t test.

FIG. 2

Survival curves for A549, HAP1 and TSA cancer cell lines irradiated at the track-segment irradiator with dose rates of 0.1, 10 and 100 Gy/s.

TABLE 1

Alpha and Beta Coefficients Obtained by Nonlinear Fit of Survival Data to Standard Linear-Quadratic Model

The FLASH Irradiator

Contrary to the track-segment platform, where the samples are irradiated by scanning the sample dish over the beam, the FLASH irradiator delivers sub-second proton pulses to the whole target area at once. The diameter of the exposure field is limited by the exit aperture and measures approximately 12 mm, as can be seen by examining the piece of unlaminated EBT3 film exposed to one FLASH pulse delivering 10 Gy (Fig. 3A). Dose is delivered homogeneously over this whole area (Fig. 3B). The energy of the beam at the target plane is equal to 5.03 MeV as simulated by SRIM. Protons with this energy stop in 360 µm of tissue (Fig. 3C). However, due to the increase in stopping power towards the end of the ion range, absorbed dose in tissue at a depth of 160 µm is already 30% higher than the entrance dose. Thus, the variation in dose with depth should be taken into account when planning for the experiment.

The custom-made ionization chamber used for monitoring the beam current during the experiment was tested for possible ion recombination at high dose rates. This was done by measuring the ionization current in the chamber at different operating voltages during exposure to beam at dose rates of 100 Gy/s and 1,000 Gy/s. The current-voltage (I-V) characteristics showed a plateau region in both cases, starting at approximately 300 V for 100 Gy/s and 800 V for 1,000 Gy/s (Fig. 4A). Based on these results, 1,000 V was used as a standard operating voltage for the ionization chamber. At this voltage, the recombination of ions in the ionization chamber was estimated to be less than 1% at 1,000 Gy/s using the 2-voltage method (40).

FIG. 4

Panel A: I-V characteristic of the custom-made ionization chamber at dose rates of 100 Gy/s and 1,000 Gy/s. Panel B: Dependence of the absorbed dose on the time duration of radiation pulses at the FLASH irradiator for corrected ionization chamber current readings of 1 and 10 µA. Panel C: Dose-rate values that correspond to several corrected ionization chamber current readings. All lines represent linear fits to data. Error bars represent standard deviation of the absorbed dose values determined using films as described in the section on dosimetry assessment with films.

Dose rate measurements. During an experiment at the FLASH irradiator, the instantaneous dose rate is measured by the custom-made ionization chamber. To calibrate the chamber, we measured dose as a function of a pulse duration at different electrometer readings. Figure 4B shows an example of such relationships for electrometer readings of 1 µA and 10 µA (dose rates of approximately 94.3 Gy/s and 928 Gy/s, respectively). Corresponding dose rates are equal to the slopes of the plotted relationships. Figure 4C shows calculated dose-rate values for ionization currents spanning six orders of magnitude. Linear fit of data points in Fig. 4C yielded a final dose rate calibration of (92.9 ± 0.2) Gy/s/µA.

Measurements of oxygen depletion by water radiolysis. The Fricke ferrous dosimeter was used to investigate depletion of dissolved oxygen by short intense pulses of proton radiation. Figure 5A and B shows the change in the absorbance of the Fricke solution equilibrated either at room air or with the gas mixture consisting of 4% O₂ and 96% N₂, respectively as a function of the average dose delivered at 1,000 Gy/s. It is evident that the initial rate of increase of absorbance becomes less steep after a certain dose. The rate of the increase of absorbance at low doses (<80 Gy for 21% O₂ and <40 Gy for 4% O₂) and high doses (>120 Gy for 21% O₂ and >70 Gy for 4% O₂) was determined by linear regression. Fitting results are shown in Fig. 5 as well. It is likely that the intersection of linear fits at low and high doses gives a dose value after which all oxygen gets depleted and consequently, the trend of change in absorbance breaks towards a lower rate. For Fricke solution equilibrated with 21% O₂ breaking occurs at a dose of 107 Gy, while for the solution equilibrated with 4% O₂ this effect occurs at a dose of 56 Gy. For comparison, samples taken from the same stock solution that received X-ray irradiation at low dose rates of 8 Gy/min show no break in response up to 600 Gy (Fig. 5C). After 800 Gy the response of the Fricke dosimeter saturates.

FIG. 5

Measured absorbance of the Fricke solution as a function of dose for samples equilibrated with 21% O₂ (panel A) or 4% O₂ (panel B) and proton-irradiated at a dose rate of 1,000 Gy/s. Panel C: Samples equilibrated with 21% O₂ X-ray irradiated at a dose rate of 8 Gy/min. Dashed lines represent linear fits to data at low (red) and high (blue) doses. The corresponding fitting results are shown inside the graphs. Black arrows indicate doses at which the breaking from the linear trend in the response occurs. Error bars represent standard errors of the mean.