Ion species discrimination method by linear energy transfer measurement in Fujifilm BAS-SR imaging plate

The generation of energetic highly charged heavy ions is of great interest for applications in nuclear physics, such as studying heavy nucleus–nucleus collisions. For high energy, high-Z ion accelerators, an ion source that could provide high charge states combined with energies of order 10 MeV/u would be highly attractive. A higher ion energy from the source mitigates the detrimental effect of space charge, which is particularly restrictive for lower energy ions. For example, for a heavy ion synchrotron, increasing the source energy and charge state would allow a reduction in the number of accelerator components in the injector, leading to downsizing of the accelerator and a reduction in the cost.

If high power laser pulses are focused to intensities exceeding 10¹⁸ W/cm²,^1–4 the corresponding electric field at focus can significantly exceed intra-atomic fields, stripping bound electrons and generating plasma. The free electrons are subsequently accelerated to relativistic energies, and the resultant plasma dynamics are governed by collective electromagnetic fields. This relativistic plasma enables the exploration of novel phenomena including laser-driven ion acceleration and related applications such as hadron therapy,⁵ fast ignition of thermonuclear targets with laser-driven ions,^6–8 and nuclear physics.^9,10 Due to recent progress in laser technology, many petawatt (PW) class laser systems are being constructed all over the world for pursuing high intensity physics. The focal intensity of such lasers can reach 10²² W/cm²,¹¹ corresponding to an oscillating electric field at the focus ∼100 TV/m. This is high enough to ionize heavy ions such as Ar up to the K-shell. If the pulses interact with solid density targets, they can generate a quasi-static electric field up to 100 TV/m on the target surface, efficiently ionizing and accelerating ions over a sub-millimeter distance to energies in excess of 10 MeV/u. Therefore, laser-driven ion sources are a promising candidate as an injector in a heavy ion accelerator system. However, before these sources can be seriously considered, it is important to demonstrate that sufficient numbers of highly charged heavy ions can be accelerated in an appropriate energy band for applications.

Laser-driven ion sources typically produce “cocktail beams” comprising ions in multiple charge states and species, all exhibiting broad energy spectra. These unusual beams are typically measured using a Thomson parabola (TP) spectrometer,¹² which separates the ions by the charge-to-mass ratio (Q/M) and disperses them spectrally. However, ions with the same Q/M fall onto the same parabolic curve, and it is therefore impossible to distinguish different ion species with the same Q/M from the position on the detector. A methodology to distinguish ions from different species for the same Q/M is highly desirable.

Although various detectors are commonly used for TPs, Fujifilm imaging plate (IP) is popular due to its extremely high sensitivity and large dynamic range, which allows the identification of individual ions. The energy-dependent IP response to several ion species has been measured using ion beams from both conventional accelerators and laser-produced plasmas. The existing data are limited to protons, deuterium, helium, carbon, and titanium ions,^11–17 and it is highly desirable to extend this to different ion species and large energy ranges. Studies on the IP response have typically either generated empirical formulas^14,15,17 or used simple models^13,14 for the so-called “response curve” relating the Photo-Stimulated Light (PSL) signal recorded on the IPs to the incident energy of the ions. Previous work has implied a dependence on the IP sensitivity due to the stopping power of the incident ions,¹³ resulting in a requirement of independent calibration of the IP response for each species of ions. Recently, a Birks-like¹⁸ behavior was used to describe the non-linear response of the IP as a function of ion LET (Linear Energy Transfer), showing that it was possible to describe the IP response to low-Z ions.¹⁹ However, they were unable to demonstrate the extension of this approach to high-Z ions.

In this manuscript, we investigate the response of Fujifilm BAS-SR IP to different species of heavy ions at different energies between 2.6 MeV/nucleon and 6 MeV/nucleon with dE/dx ≥ 1 MeV/μm range using the Heavy Ion Medical Accelerator in Chiba (HIMAC²¹) located at the National Institute of Radiological Science (NIRS), part of the National Institutes for Quantum and Radiological Science and Technology (QST) in Japan. We compared the response to the previously developed models by Bonnet et al.^13,14 and Lelasseux and Fuchs^19,20 to our dataset in the range of mid-Z carbon (Z = 6) to higher-Z Xe (Z = 54). However, we found that for ions in this range of atomic number, a simple model solely dependent on the energy deposited in the imaging plate fits the data better than the aforementioned models, with no species-dependent sensitivity. We then show that this model allows us to clearly distinguish high intensity laser-driven silver ion tracks from those of mid-Z ions, such as carbon or oxygen, which fall within the neighboring regions on the IP.

Imaging plates are sensitive to electrons, ions, and photons with an excellent spatial resolution of tens of micrometers.^22,23 They also have an extremely large dynamic range, ∼5 orders of magnitude,^22,23 and are therefore suitable for the detection of charged particles in a single particle counting regime. We use the FujiFilm BAS-SR imaging plate, which consists of a protection layer (6-μm-thick 1.27-g/cm³ C₂H₂O), a doped (by Eu) phosphor layer (120-μm-thick 3.1-g/cm³ BaFBr), a support (188-μm-thick 1.27-g/cm³ C₂H₂O), and a magnetic layer (160-μm-thick 3.1-g/cm³ ZnMn₂Fe₅NO₄₀H₁₅C₁₀) (Fig. 1). The principle of the detection of the charged particle on the IP-SRs is as follows: The dopants in the phosphor layer (Eu2+) emit an electron as a charged particle passes nearby, which is then captured in a metastable state in FBr. During the detector readout, the IP is irradiated with ∼2 eV photons, leading to de-excitation and recombination of the electron with an Eu3+ ion, emitting a ∼3 eV photon. The number of photons emitted is directly related to the photo-stimulated luminescence (PSL). In the text, we use two different IP-reader systems, FLA-7000²⁴ and BAS-1800-II,²⁵ for the datasets taken at a conventional heavy ion accelerator, HIMAC,²¹ and an ultra-high intensity short pulse laser system, J-KAREN-P^26,27 laser system, respectively. In these reader systems, the raw signal level is converted to the real signal level that corresponds to the dose of the exposed radiation through the following formulas:

where Gel, Q_L, R, L, and G are the pixel value (raw signal level in the FLA-7000 type), pixel value (raw signal level in the BAS-1800-II type), spatial resolution (R = 50, i.e., 50 × 50 µm² square pixel size), latitude (L = 5), and pixel depth (G = 16),²⁴ respectively. As PSL and Counts are proportional values from the two different scanner systems, we calculated a conversion factor (α) between these two by using the carbon ion events recorded on IP readout by two different systems (Counts = α · PSL, α = 858 ± 84). Hereafter, we use PSL as the unit of signal level for both systems.

The calibration experiments for IP-SR were carried out using a conventional heavy ion accelerator system, HIMAC. The ion beams used were carbon, iron, and xenon, each at three different energies of 2.6 MeV/u ± 0.2%, 4.3 MeV/u ± 0.2%, and 6.0 MeV/u ± 0.1%, which are far higher than the minimum energies detectable by IP-SR of ∼0.375 MeV/u (C), ∼0.161 MeV/u (Fe), ∼0.093 MeV/u (Ag), and ∼0.099 MeV/u (Xe), respectively. For each ion species and energy, one IP-SR sheet was exposed. The flux of charged particles in each irradiation was controlled such that the signal from individual charged particles do not spatially overlap, enabling the measurement of the dose deposited by individual particles as shown in Fig. 2(a). The dependency of the stopping power (dE/dx) for each ion species and energy is shown in Fig. 2(b). The ion beams were provided with an average beam current of 0.02 μeA and then mechanically chopped to a pulse length of 10 µs. We used a single bunch for the irradiation, containing ∼10⁵ ions, with a beam diameter of 5 cm. Tracks with multiple events overlapping were discarded. The different tracks originated from different species showing different brightness. Each track produced a signal that spread over multiple pixels. Therefore, we calculated the PSL value of each track as the summation of the PSL values of 49 pixels (7 × 7) around a pixel that has a local maximum PSL value. We chose the PSL of a nearby pixel that has no signal from any ion tracks as background and subtracted it from the signal pixels before summation.

Unwanted de-excitation occurs between irradiation and readout, causing a fading effect resulting in a time-dependent reduction in PSL. The fading dependence over time has to be quantified in order to know the PSL value at the time of irradiation and is described by a fading function. The fading effect has been examined by using β-particles from a radioactive source.²³ However, the fading function may not be applicable to the case of ion irradiation, especially for heavy ions. The fading function for protons has been found to depend on the incident proton energy.¹³ Therefore, to calibrate the temporal decay of the signal, different IP sheets irradiated under the same condition were read out at several periods after irradiation. Figure 2(c) shows the normalized PSL for C and Xe tracks as a function of time after irradiation. The datasets include different incident energies (shown above, 2.6 MeV/u, 4.3 MeV/u, and 6.0 MeV/u) after appropriate background subtraction. Each data point in Fig. 2(c) is the average value of the PSL of more than 100 tracks. There is no clear energy or Z dependence on the fading function. An empirical fading function is obtained by fitting a double exponential normalized to the earliest scanning time, 15 min, giving the PSL as

where t is the time after irradiation in minutes, as shown in Fig. 2(b). Hereafter, all the PSL values are adjusted to the interpolated value at 15 min after irradiation by Eq. (3).

Figure 3(a) shows the examples of the frequency distribution of the PSL values on the IP for 2.6 MeV/u, 4.3 MeV/u, and 6.0 MeV/u Xe ions, with each distribution made from >100 measurements. The distribution was well fit by a log-normal distribution from which the expected value and the standard deviation were calculated. We fit the data of all ion species with an empirical relation of the relative error (R_err) [Fig. 3(b)], given by the ratio of the standard deviation to expected value as a function of incident energy of ions (E_incident), resulting in

We assume this is true for all ions and later use it to estimate the relative error of arbitrary ion species.

Non-linear quenching of the IP at different stopping powers is likely to be important for heavy ion irradiation. Most of the previous studies on IP calibration did not investigate this and instead relied on individual calibration for each different charged particle. Usually, the Birks law is applied to ion stopping in organic scintillators, which are strongly quenched by the large stopping power of high-Z ions. However, the applicability of the model for the case of high-Z ion beam detection by an in-organic scintillator has also been investigated previously, and it was reported that the quenching factor in the Birks law depends on the atomic number of the incident beam.²² Recently, Lelasseux and Fuchs^19,20 applied the Birks law to IP-TR, but they were unable to show scaling up to high-Z ions. In order to make the situation clearer and investigate the elementary process, it is attractive to find the response of the scintillation material for single ions.

We, therefore, checked the validity of using the Birks law to the response function of the IP-SR relating PSL, by assuming quenching effects dominate other possible non-linear responses in the light emitted from the imaging plate, and that the quenching of the trapping of electrons in metastable states in the inorganic phosphor of the IP can be described by the same Birks law used in scintillators,^18,22

where w, dE/dx, S, and kB are the penetration depth of the ion into the phosphor layer, energy deposition to the IP, efficiency of the light emission of the scintillation material, and quenching coefficient, respectively.

We performed Monte Carlo simulation of ions irradiating the IP by using the PHITS code²⁸ including the ATIMA model²⁹ and calculated the stopping power in the phosphor layer of the IP, dE/dx. We carried out the integration of Eq. (5) numerically by using this dE/dx and derived the best-fit curve for the C, Fe, and Xe dataset by assuming that S and kB are independent of the atomic number of the incident ions,³⁰ giving the best-fit parameters of $(10.0±0.791.3)×10−3$ [PSL/MeV] and $(0.0±0.0150.012)$ [μm/MeV], respectively, with a 68% confidence level (reduced-χ² = 1.72). Clearly, the fitting indicates that, as kB ≈ 0, there is no dE/dx dependent quenching in this regime. We, therefore, set kB value to zero all through the analysis below.

Bonnet et al.¹⁴ introduced the exponential factor for explaining the decay of the light within the bulk of the sensitive layer. We also checked the validity of the model including this factor as follows:

In this case, the best-fit parameter is $(12.0±1.01.4)×10−3$ [PSL/MeV] with a 68% confidence level (reduced-χ² = 2.02). The increase in reduced χ² when including the exponential light scattering factor indicates a slightly worse fit to the data.

We, therefore, concluded that we do not need the Birks-like model or an additional depth dependent exponential factor for explaining the response function within our dataset, i.e., the response function is well described by a species-independent model,

Figure 4(a) shows the mean PSL values of the tracks as a function of incident energies. The light blue, orange, and green circles show the mean PSL values and standard deviations accumulated from the tracks on the IP by Xe, Fe, and C beam irradiation with different incident energies. Again, each data point is the average of more than 100 tracks. The overlapped solid lines are the best-fit of and the allowable range of the simplest model shown in the text [Eq. (7)]. The superposed dotted lines for each best-fit curve show the estimated uncertainty of the PSL value given in Eq. (4). For incident energies below ∼2.5 MeV/u, the uncertainty ranges of the PSL of different species begin to overlap with each other showing the limitation of the discrimination ability of this method at low energies. However, at higher energies, it is clear that the PSL value could be used to discern high-Z ions from the low-Z ones.

Considering that a Birks-like relation can explain well the IP response for low-Z ions,¹⁹ our observation of no species-dependent response over such a large range of atomic numbers at high-Z (for which the peak dE/dx in the IP varies by over an order of magnitude) seems initially surprising. However, the sensitivity S is significantly lower than the species-dependent sensitivity for low-Z ions measured previously, as shown in Fig. 4(b).^14,31 Here, the data shown in red open triangles are obtained by using the model used in Ref. 14, and the data in black open triangles are obtained by using our simplest model [Eq. (7)]. The sensitivity shows a sudden decrease starting from the nominal quenching-free sensitivity given by electron irradiation as Z increases from 1 to 6, after which it stays almost constant (or just slightly decreases) for Z ≥ 6. The data from Strehlow et al.¹⁷ [red filled circle in Fig. 4(b)], which were unable to be fit by the Birks-like model of Lelasseux and Fuchs,^19,20 lie in the high-Z region, consistent with our observation. It therefore seems likely that the Birks law ceases to be valid in the range of extremely high stopping powers. This has been demonstrated before for organic scintillators. For example, Ogawa et al.³² showed that the Birks law is not valid for a plastic scintillator when irradiated by ions with a high dE/dx (>0.1 MeV/μm). Although each scintillator will respond differently, we note that all of the data used in this study fall in the range of dE/dx > 0.1 MeV/μm. It has been proposed, for organic scintillators, that this can be explained by separation of the scintillation into two components.^33,34 The principal component is in the densely ionized region around the individual particle track, which becomes heavily quenched and eventually saturates. A secondary component is due to the energetic secondary electrons (delta rays) generated by the stopping particle, which can travel outside the densely ionized region and cause scintillation in a larger volume, resulting in scintillation that is linear with energy deposition even at high stopping powers.

Therefore, it may be the case that while a Birks-type relationship can fit the IP response for low-Z ions,¹⁹ at high-Z, the response is estimated using a simple single-parameter model [Eq. (7)]. Although the range of atomic numbers used in this study prohibits an independent fit over all atomic numbers, it is likely a global model will have the form

The first term dominates for low stopping powers, while the second term dominates for extremely high stopping powers where (S₁/kB · dE/dx) < S₂. In this case, S₂ can be given by our data, $(10.0±0.791.3)×10−3$ PSL/MeV, as calculated above. We cannot calculate S₁, L, and kB from our data, but S₁ and L should be equal to the sensitivity for electrons/photons and IP absorption length, given by Bonnet et al.¹³ to be (0.33 ± 0.08) PSL/MeV and (118 ± 16) µm. Lelasseaux and Fuchs^19,20 estimated for IP-TP a quenching factor kB = 15 µm/MeV.²⁰ As the phosphor compounds between IP-TR and IP-SR are slightly different, the kB value for IP-SR will vary somewhat, but using the above values suggests that the first and second terms of Eq. (8) may become of comparable order when dE/dx ∼ 1 MeV/μm, corresponding approximately to the stopping power in the Bragg peak of carbon stopping in IP-SR. However, it should be noted that combining data from different studies may be affected by the difference in the methodology or the IP scanner behavior, and so the proposed model and relevant parameters need to be verified in future studies, encompassing a larger range of atomic numbers.

We now demonstrate the effectiveness of our model to enable discrimination between different ion species in the multispecies cocktail beam generated in a laser-driven ion accelerator. The experiment was carried out at the Kansai Photon Science Institute (KPSI), National Institutes for Quantum and Radiological Science and Technology (QST), Japan, by employing the J-KAREN-P laser system,^26,27 capable of providing PW-class peak power at 0.1 Hz on target. In this experiment, laser pulses with a central wavelength of 800 nm, a pulse duration of 35 fs (FWHM), and an energy of 10 J were focused by an off-axis parabolic mirror (F/1.4) to a maximum peak intensity of ∼5 × 10²¹ W cm⁻² onto thin silver foils of varying thickness at an incidence angle of 45° with p-polarization.

The energetic ion beam produced by the interaction between the laser pulses and thin-solid targets contained different ion species and charge states. In addition to the silver ions from the bulk material, proton, carbon, and oxygen are also accelerated from a contaminant layer on the target surface. In order to separate the ions by charge-to-mass ratio, a TP was used in which co-linear electric and magnetic fields disperse the ions according to their kinetic energy and different charge-to-mass ratios. Figure 5(a) shows the experimental setup. The TP is placed in the target normal direction with a collimator of diameter 250 µm placed 340 mm from the target. The size of the collimator was chosen such that the number of ions was small enough to be measured in a single event mode in which the PSL distributions corresponding to each event are separated in space with no overlap. A 0.77 T magnetic field was placed 50 mm after the collimator and applied over 50 mm. After 120 mm drift space from the end of the magnet, an electric field of ∼20 kV/1.5 cm was applied over 300 mm. The IP-SR was placed after 940 mm to detect the ions.

Figures 5(b) and 5(c) show an example of the Thomson parabola traces recorded on the IP-SR from a single laser shot onto a sub-micrometer thick silver target. The TP curves resulting from ^AAg^Q+ (A = 107 or 109, Q = 36–45) are selected for model validation because they are well separated from them and other nearby traces and have a sufficiently low beam flux that single events are visible on the IP. These single events are identified along the curve given by the relative charge-to-mass ratio [Fig. 5(b)]. Tracks that are overlapping with the traces of silver are far brighter than the tracks that are overlapping with the nearby C⁵⁺ parabola trace. The PSL values for each silver track are calculated, followed by the appropriate background subtraction mentioned above.

Figure 6(a) shows the measured PSL values for “Ag” ions with red triangles; here, each point corresponds to the PSL value from one track. The data (red triangles) in this figure are plotted by assuming that the tracks are generated by silver ions, which will later be demonstrated. For silver target shots, ion tracks were seen along the trace corresponding to charge states +38 to +45 [Fig. 5(c)]. Again, the traces of these charge states (Q = 38–45) partially overlap due to the finite size of the collimator. The typical error in the charge state (Q) and the energy of the silver ions caused by the finite collimator size were ±2 MeV/u and ±1 MeV/u (39 < Q < 45), and 1 MeV/u and ±0.5 MeV/u (34 < Q < 40), respectively. These traces could also have been generated by other partially charged ions with similar Q/M. The superimposed curve in Fig. 6(a) is the calculated response function for silver (Q = 40), with the model described above. Here again, the uncertainty range is shown by the dotted line. All the “Ag” datasets fall within the range of the calculated response curve within the uncertainty range for silver, despite there being no new free fitting parameters. As well as demonstrating the universality of our model to high-Z ions, this supports the assertion that the accelerated ions are indeed silver ions from the target bulk.

Additional evidence of this is given by the impurity measurements of the target foil. Figure 6(b) shows the typical atomic composition of the Ag target bulk measured by a scanning electron microscope (SEM: Zeiss Sigma) with an energy-dispersive x-ray spectrometer (EDX: Oxford Instruments X-max 80 mm²). EDX spectra were accumulated from ten different randomly chosen positions on the target. The EDX analysis shows the peaks corresponding to Ag, Si, and Al atoms. The chemical composition of the Ag target bulk was estimated to have a metallic purity 99.84 ± 0.10 wt. %, where the error shows the standard deviation of the results from ten measurement points. The impurity from other species, especially those having a higher atomic number than silver, is negligibly small. Therefore, the possible ion species other than Ag in the TP traces are Al and Si from the target foil and H, C, and O from the contaminant layer on the target surface. The blue lines in Fig. 6(a) show the predicted range for Si, which has the largest atomic number of the measured impurities. The “Ag” ion dataset does not overlap with the region of Si in the energy range > 2 MeV/u. Therefore, the assumed Ag tracks having incident energies higher than 2 MeV/u could only have been generated by silver ions.

We used a novel technique of single ion counting to describe the response of the imaging plate (IP-SR) to ion species over a large range of atomic numbers, from carbon (Z = 6) to xenon (Z = 54). We showed that for heavy ions, Z ≥ 6, the IP response was linearly dependent on the energy deposited in the IP sensitive, indicating that the dedicated calibration experiments for different heavy ion species and incident energies are unnecessary in this regime. We then applied this model to unambiguously identify tracks caused by silver ions in a laser plasma acceleration experiment, which could otherwise be due to other lighter species with similar charge-to-mass ratios. This identification method, therefore, allowed us to conclusively demonstrate the acceleration of a beam of silver ions generated by a high intensity short pulse high contrast laser interacting with a sub-micrometer thick silver target.³⁵ Our generation and technique for accurate diagnosis of such beams will be of great interest for both laser ion sources and future high-Z accelerator design.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

This work was supported by the JST-Mirai-Program, Japan (Grant No. JPMJMI17A1). M.N. was supported by the JST PRESTO (Grant Nos. JPMJPR16P9 and Kakenhi 10K05506). M.N. and N.P.D. were supported by the JSPS Post-doctoral Fellowship and Kakenhi Project No. 15F15772. M.N. and H.S. were supported by the Mitsubishi Foundation (ID28131). This research was partially supported by QST President’s Strategic Grant [QST International Research Initiative (AAA98) and Creative Research (ABACS)].