The generation of laser-driven particle sources is a current hot topic in physics research with implications that go from laser-driven fusion (and in particular the proton-driven fast ignition approach to inertial fusion^[1^]) to the realization of several societal or industrial applications^[2^].

Also, in recent years, high yields of α-particles have been observed from laser-driven hydrogen–boron fusion experiments, opening the possibility to develop a novel approach to high-brightness α-particle sources^[3^–6^]. These experiments are based on the hydrogen–boron fusion reaction^[7^,8^]:

$$\begin{array}{r}p{+}^{11}\mathrm{B}\to 3\left({}^4\mathrm{He}\right)+8.7\mathrm{MeV},\end{array}$$  (1)

In the pitcher–catcher scheme, the laser irradiates a pitcher (usually Al or plastic thin foils), producing a proton beam, which is sent onto the catcher, a secondary boron target where the proton-boron (pB) fusion reactions take place^[13^–16^]. Most experiments within this approach used high-energy, high-power laser beams to produce a bright source of protons through the mechanisms of target normal sheath acceleration (TNSA)^[17–21^].

One critical issue is, however, how to measure the α-yield in a reliable way. The most common diagnostic used in such experiments relies on solid-state nuclear track detectors (CR39)^[22^], for which an α-particle identification might be a problem due to the simultaneous emission of many ion species from the laser-irradiated targets. Many other diagnostics used in these experiments (Thomson parabolas, time-of-flight detectors, …) are also prone to this problem (for a full discussion on the topic see, e.g., Refs. [23–25]).

There is indeed another important issue: CR39 and other diagnostics only measure the α-particles escaping the targets; however, due to their very short propagation range in solid density matter most α-particles are unable to emerge and are indeed trapped inside the boron target.

Therefore, alternative diagnostic approaches are useful to validate experimental results. One of such novel approaches is the detection of radioactive isotopes produced in the targets by secondary nuclear reactions^[23^,26^,27^]. Most of the produced radioisotopes are characterized by γ-ray emission, and therefore the type and number of produced radioisotopes can be characterized by γ-ray spectroscopy, for instance using a calibrated high-purity germanium (HPGe) detector. The number of proton–boron fusion reactions that took place in the target can then be retrieved by the known branching ratio between the pB reaction and the reaction that produced the radioisotopes. Of course, γ-rays can measure the total number of reactions that took place inside the target, so there is no ‘escaping issue’ as for α-particle detection with CR39.

Apart from the diagnostic use, the generation of radioisotopes in laser-driven experiments can be very interesting in itself, in particular for producing radioisotopes for medical applications in therapy or diagnostics, particularly PET (proton emission tomography). Radioisotopes used in medicine are currently produced by neutron irradiation in dedicated research reactors, or by proton irradiation using cyclotrons. In principle, laser-driven sources are able to produce energetic protons and neutrons, and they could be used as a complementary technology to generate radioisotopes for diagnostics and medical treatment^[28^,29^]. In addition, it is also possible to consider radioisotope production using α-particles produced by laser-driven proton–boron fusion. Today, radioisotopes from α-particles sources are little used in the medical domain because, even if often they have very interesting properties, only a few cyclotrons in the world are able to accelerate α-beams with adequate energy and intensity for their production. Usually, the reactions that produce such radioisotopes show maxima in cross-sections for energies higher than 10 MeV^[30^]. Only dedicated cyclotrons such as ARRONAX^[31^] or U-120M^[32^] can produce high-flux of α-particles with energies higher than such energies. The cost and the complexity of such dedicated cyclotrons, as well as the need for extensive radioprotection, strongly limit the spread of such technology and access to related radioisotopes.

This is the case for instance of α-emitters such as ²¹¹At. The current supplies for medically useful α-emitters such as ²¹¹At are limited by naturally isolated by-products from weapons development and the actual level of production is only sufficient for preclinical studies and limited clinical trials. ²¹¹At can also be produced by the irradiation of ²⁰⁹Bi with α-particles,^[33^] which could be realized using laser-driven α-particle sources.

Concerning PET, recently, there has been significant interest in the radionuclides of scandium: ⁴⁴Sc (T_1/2 = 3.87 h) and ⁴³Sc (T_1/2 =3.89 h) as tracers for PET imaging. ⁴³Sc and ⁴⁴Sc can be produced by irradiating natural calcium with α-particles from cyclotrons, as already shown by the IChTJ group^[34^,35^] in the Heavy Ion Laboratory of Warsaw University. Alternatively, ⁴³Sc can be produced by irradiating natural Ca with protons, but in this case ⁴⁴Sc is also produced.

In any case, it is well known that the demand for radioisotopes is rapidly increasing, for both therapy and diagnostics, so there exists a strong societal need for developing new approaches to increasing radioisotope production to meet increasing demand^[36^,37^]. Therefore, it is very important to study the possibility of producing radioisotopes by laser irradiation and, in particular, using laser-driven sources of α-particles based on the proton–boron fusion reaction.

In conclusion, this paper addresses two different but related points: (1) the use of radioisotopes as a diagnostic of α-particle generation in laser-driven proton–boron fusion experiments; and (2) the feasibility of producing medical radioisotopes using laser-generated protons or α-particles.

The experimental investigation was carried out in March 2023 at the Centro de Láseres Pulsados (CLPU) in Salamanca, Spain, using the short-pulse high-intensity laser VEGA-3.

The laser is operating at wavelength λ = 810 nm with the pulse duration τ = 200–250 fs and energy of about 25 J. The laser incidence angle on the pitcher target was Θ = 12^o, providing an on-target focal spot size (full width at half maximum (FWHM)) of 12 μm. The temporal contrast was about 2 $\times$ 10⁻⁵ at 1 ps before the main pulse and below 10⁻⁵ at 5 ps.

Although the laser VEGA-3 can work at repetition rates of up to 1 Hz, we did not use such a high repetition rate in our experiment mainly due to the need for performing an accurate alignment of each pitcher target before the laser shots. We made a shot every 2 min, which compared to many experiments with high-intensity lasers can still be considered as a relatively high repetition frequency.

Notice that we have on purpose opted for a non-optimal temporal compression of the VEGA-3 laser pulse. Indeed, the full compressed duration ( $\sim$ 30 fs) is not optimal for proton acceleration and a longer pulse is more efficient. We experimentally found that the proton energy and proton number were optimized for pulse durations of 200–250 fs^[27^].

The experiment was performed in the pitcher–catcher configuration (see Figure 1). We used a remotely controlled target holder containing tens of pitcher foils producing protons after each shot. Such protons were sent on the same catcher in order to accumulate tens of shots and produce a greater and more easily measurable quantity of radioisotopes. We irradiated several kinds of catcher targets: pure boron (B), boron nitride (BN), ammonia borane (BNH₆) and calcium silicate (Ca₂SiO₄).

Produced radioisotopes were measured using an HPGe γ-ray detector, while other diagnostics were used to characterize the proton and the α-particle generation (including CR39 foil, a Thomson parabola spectrometer, time of flight). The general experimental setup is presented in Figure 1.

The Thomson parabola spectrometer was used to characterize the spectrum of TNSA protons emitted from the pitcher in shots where the catcher was not present. For a detailed experimental setup of the laser and diagnostics see Refs. [27,38], which also describe the results of proton and α-particle generation. The results described in Ref. [38] were acquired during the same experiment. Although the pitcher–catcher scheme was also used, Ref. [38] focuses on the hole-boring scheme (i.e., when the laser directly irradiates the target) and addresses the problem of cross-validation of experimental results and detailed computer simulations.

In this paper, we instead just use the pitcher–catcher configuration, which is more adapted to the production of radioisotopes and in particular we describe the results obtained with ammonia borane and calcium silicate catchers, as an example of the possibilities, and the challenges, offered by the laser-driven approach in the production of medical radioisotopes.

In the experiment, we used an HPGe γ-ray detector equipped with a DSA-1000, 16K channel integrated multichannel analyzer and cooled with liquid nitrogen (LN2) at 77 K. The housing of the detector was made of passive iron shield of 15 cm in all directions, screening it from external radiation sources, and the samples had to be positioned inside the same housing.

A fundamental point for the interpretation of the experimental measurement is the characterization and calibration of the spectrometer. We used three different γ-radiation sources: ¹³⁷Cs (emitting a γ line at 661.657 keV), ⁶⁰Co (emitting two lines at 1173.228 and 1332.492 keV) and a mixed source of ¹⁵⁵Eu (86.540 and 105.30 keV) and ²²Na (511.00 and 1274.54 keV).

The first step is the calibration with respect to photon energy, that is, the relation between the analyzer channel (pixel) and the energy of the emitted lines. Figure 2 presents the energy calibration of the HPGe γ-ray detector showing a linear relationship between the photon energy and the channel (pixel).

The second point is establishing a relation between the number of counts and the activity of the sources, that is, determining the detection efficiency, that is, the relation between radioisotope activity and recorded counts. In order to do this, we must take into account the following.

1. (1)The decay of source activity since the sources were acquired:(2) $$\begin{array}{r}N(t)=N_0\exp (-\lambda t)=N_0\exp (-\ln (2)\frac{t}{{\tau }_{1/2}}).\end{array}$$
2. (2)The activity related to each specific γ-ray energy, which is obtained by multiplying the source activity by the γ emission probability, is represented as ‘γ line activity’.

The results of these calculations are shown in Table 1.

We then calculated the peak detection efficiency $D$ as the ratio between the line activity (in kBq) and the recorded number of counts in the γ-ray line, recorded during a 5-min acquisition:

$$\begin{array}{r}D=\frac{\gamma \text{ line activity}}{\text{Number of counts}}.\end{array}$$  (3)

The results of the calculations are shown in Figure 3, where the photon energy is in keV and the line activity is measured in kBq. We see that the peak detection efficiency $D$ depends on γ-ray photon energy and that the relation is practically linear (and passing through the origin, i.e., $D$ is proportional to $h\nu$ ).

We also performed a sensitivity scan by displacing the radioactive source inside the container with respect to the central position (the source placed exactly on the vertical axis of the detector at a fixed distance of 1 cm). Figures 4 and 5, respectively, show the results of displacing the sources in the vertical direction and in the horizontal direction.

The number of recorded counts approximately scales inversely proportionally with respect to the vertical displacement (i.e., $\text{Counts}\approx 1/d$ ) and linearly with respect to the horizontal displacement.

This measurement is important because it allows estimating the error in the number of recorded counts, which corresponds to non-perfect positioning of the source or to different source geometry. For instance, a displacement of 1 cm in the lateral direction implies a reduction of 10% in counts, while a displacement of 1 cm in the vertical direction implies a reduction of 50% in counts.

In order to analyze the γ-ray spectra, we performed an accurate measurement of the background (due to cosmic rays or other sources) with the same detector, and we then subtracted such background from our experimental spectra.

A final question related to the calibration of the spectrometer concerns the impact of Compton scattering on the recorded spectra. Not all the emitted γ-photons at one energy $h\nu$ are found in the corresponding line but many undergo Compton scattering (at an angle $\theta$ ) with the electrons in the detector material, after which they might escape the detector volume and therefore are recorded as photons at  $h{\nu}^{\prime}$ :

$$\begin{array}{r}\frac{1}{h{\nu }^{\prime }}-\frac{1}{h\nu }=\frac{1}{m{c}^2}(1-\cos \theta ).\end{array}$$  (4)

Monte Carlo simulations are needed to give a quantitative evaluation of this effect (depending on the detector size and geometry). However, in our setup, we can get an estimation by using the single-line spectrum emitted by the ¹³⁷Cs source (for multiple-line spectra the situation is more complex due to the superposition of some lines to the Compton shoulder and the superposition of Compton shoulders from different lines).

Figure 6 shows the spectrum of ¹³⁷Cs in logarithmic scale. Although on a linear scale the Compton shoulder seems small, due to its large energy range, it indeed contains more photons than the line peak. The ratio between the total number of counts and the counts in the peak is $\approx$ 3.5.

This factor is anyway naturally included in the calibration, which relates the number of counts recorded in the main γ-ray line to the total activity of the source (i.e., including the decay that will end up in the Compton shoulder).

We placed an ammonia borane (BNH₆) pellet on the rear side of the Al pitcher (6 μm in thickness) at a distance of 2 cm. The pellet had a diameter of 12 mm and a thickness of 1.2 mm, and it was produced by Chris Spindloe et al.^[39^] by compression of commercially available BNH₆ powder. A detailed discussion of the targets can be found in Refs. [40,41]. The sample was inclined under 50° to the laser propagation axis. After irradiation the sample was placed in the HPGe detector and acquisitions were done every 5 min (accumulating the signal for 5 min). Figure 7 shows the accumulated γ-ray spectrum recorded after the irradiation over the period of 100 min and measured for 60 min, showing a strong peak at 511 keV.

Since the 511 keV line is due to the annihilation of positrons emitted from radioisotopes with the electron in the material, it is typical of any β⁺ emitter. In order to recognize the origin of such emission we must then analyze the time decay of the line and identify it with the lifetime of a specific radioisotope.

In Figure 8, we recognize two different decay slopes. The first corresponds to a half-life T_1/2 = 20.4 min and the second one to T_1/2 = 109.8 min. This allows to identify the first one as the decay of ¹¹C and the second one as the decay of ¹⁸F. ¹¹C decays as follows:

$$\begin{array}{r}{}^{11}\mathrm{C}\to {}^{11}\mathrm{B}+{\mathrm{e}}^{+}+{\nu }_{\mathrm{e}},\end{array}$$  (5)

and is produced by the following reaction:

$$\begin{array}{r}p{+}^{11}\mathrm{B}{\to }^{11}\mathrm{C}+n-2.765\mathrm{MeV}.\end{array}$$  (6)

¹⁸F decays as follows:

$$\begin{array}{r}{}^{18}\mathrm{F}{\to }^{18}\mathrm{O}+{\mathrm{e}}^{+}+{\nu }_{\mathrm{e}},\end{array}$$  (7)

$$\begin{array}{r}\alpha {+}^{14}\mathrm{N}{\to }^{18}\mathrm{F}+\gamma +4.415\mathrm{MeV}\end{array}$$  (8)

from the α-particles generated by the proton–boron fusion reaction reacting with the nitrogen in BNH₆, or from the following reaction:

$$\begin{array}{r}{}^{18}\mathrm{O}+p{\to }^{18}\mathrm{F}+n-2.44\mathrm{MeV}\end{array}$$  (9)

from the impurities in the sample (i.e., absorbed water). Of course, the quantity of oxygen in our sample is much less than that of nitrogen; however, the flux of protons is much higher than the number of α-particles (usually in this kind of experiment the ratio between α-particles and protons is of the order of 10^{–4 [}12^,15^,16^]). Hence, in our case the most probable origin of ¹⁸F is from oxygen impurities (i.e., from water).

Let us note that in principle we could also observe the 511 keV from the decay of ¹³N, another β⁺ emitter that can either be produced by the reaction α + ¹⁰B ⟶¹³N + n + 1.06 MeV, or by the reaction p + ¹⁶O ⟶¹³N + α – 5.22 MeV. However, due to the very short lifetime of ¹³N (T_1/2 = 9.97 min) we could not see the signature of its decay in our measurements.

Starting from the graph in Figure 8 and the calibration line in Figure 3 we can evaluate the activity and number of produced ¹¹C isotopes. Figure 8 shows that the number of counts recorded during 5 min is N_o $\approx 3.4 \times 10^{4}$ at t = 0. At the photon energy of 511 keV, which is the same as the line from ²²Na, the detection efficiency is as follows:

$$\begin{array}{r}D=\frac{\mathrm{Activity}[\mathrm{kBq}]}{\mathrm{Counts}}\approx 0.0001,\end{array}$$  (10)

from which we recover an activity of 3.4 $\mathrm{kBq}$ obtained with 31 shots on target. The relation among activity, the decay constant $\lambda$ , the lifetime ${T}_{1/2}$ and the number N of radioisotopes is as follows:

$$\begin{array}{r}A=N\lambda ,\lambda =\frac{0.693}{T_{1/2}},\end{array}$$  (11)

which for ¹¹C gives $\lambda =5.66\cdot {10}^{-4}$ s^–1, from which the number of ¹¹C radioisotopes can be estimated as follows:

$$\begin{array}{r}{N}_{31\mathrm{\#}}=6.0\times {10}^6.\end{array}$$  (12)

As specified before, this number corresponds to the beginning of the measurement with the HPGe detector, typically about half an hour after the end of the irradiation (due to the time needed to vent the chamber, extract the sample, etc.). By correcting for the decay rate, we see that the number of counts, which would be recorded just after the end of the irradiation, would increase by a factor $\approx$ 4.34, thereby yielding a count of $\approx 1.48 \times 10^5$ . This corresponds to activity of $\approx$ 15 kBq.

In reality such number must also be corrected to take into account that the time needed to accumulate 31 shots was $\approx$ 1 h; therefore, there was a significant decay of ¹¹C during the irradiation itself. The calculation, performed in Appendix A, shows that realizing the 31 shots in a very short time compared to the lifetime (T_1/2 = 20.4 min) would have provided an increase of a factor $\approx 2.42$ , bringing the total production of ¹¹C radioisotopes to $\approx 6.4\times {10}^7$ and the total activity to $\approx$ 36 kBq.

Hence, we estimate that one laser shot thus produces about $2\times {10}^6$ of ¹¹C, or an activity of more than 1 kBq.

Another interesting question is how we compare the number of generated ¹¹C isotopes and α-particles, that is, the number of fusion reactions taking place. This is relevant to establishing the capability of using γ-ray emission from ¹¹C as diagnostics of hydrogen–boron fusion in addition to classical CR39 detectors. The number of created α-particles and ¹¹C isotopes has been evaluated using simple Python software validated against the results of more complex Monte Carlo simulations (as described in Appendix B). The calculation uses the experimentally measured proton spectrum as input (the spectrum is shown in Appendix B).

This shows that in each laser shot $\approx 0.97\times {10}^6$ of ¹¹C particles generated, that is, the experimental result is indeed close to the calculation (within a factor of two).

The experimental evaluation of α-particle yield is more difficult being based on analysis of CR39 foils^[22^,23^,42^], which contains a certain degree of uncertainty and, of course, only measures the α-particles escaping the target. The number of escaping α-particles is evaluated in Appendix B and corresponds to $\approx 1.2\times {10}^6$ particles exiting the target from the front side per laser shot, or about $\approx 0.84\times {10}^6$ α-particles with energy of more than 1.57 MeV.

For comparison, the analysis of CR39 (see Appendix C) provides a number of 5 $\times$ 105 α-particles with energy ≥1.57 MeV per solid angle and per laser shot. Assuming an isotropic generation over the 2π solid angle corresponding to the ‘front side’, we get an estimation of 3 $\times$ 106 α-particles of energy of more than 1.57 MeV. This number corresponds to what is measured when a 5 μm Al foil filter is placed in front of the CR39 detector. This filter transmits α-particles of energy of more than 1.57 MeV but prevents a significant contamination from other laser-accelerated ions.

The theoretical estimation is therefore a factor of 3 below the experimental one (the difference might be largely due to the hypothesis of isotropic generation over a 2π solid angle).

If we look at the ratio of ¹¹C to α-particles (with energy >1.6 MeV), we see the following:

$$\begin{array}{rl}{\left(\frac{\alpha }{{}^{11}\mathrm{C}}\right)}_{\exp }& ={\left(\frac{3\times {10}^6}{2\times {10}^6}\right)}_{\exp }\approx 1.5,\\ {\left(\frac{\alpha }{{}^{11}\mathrm{C}}\right)}_{\mathrm{cal}}& ={\left(\frac{0.84\times {10}^6}{0.97\times {10}^6}\right)}_{\mathrm{cal}}\approx 0.9.\end{array}$$  (13)

This shows that, within the limit of precision of the present experiment, the measured activity of ¹¹C is in fair enough agreement with the measured α-particle yield and can indeed provide a way to estimate the total number of hydrogen–boron fusion reactions. The excess number of α-particles with respect to ¹¹C might be possibly due to the fact that the CR39 measurement is somewhat polluted by the presence of other ions coming from the catcher^[27^]. As mentioned before, the theoretical evaluation from Appendix B predicts $1.2\times {10}^6$ α-particles escaping the catcher from the front side and a total α-particle yield of 1.11 $\times$ 107, that is, in our experimental configuration about 90% of α-particles remain trapped inside the catcher. This is similar to what has already been shown by Scisciò et al.^[27^] in the case of natural B catchers.

Taking into account that one fusion reaction releases three α-particles, we estimate that in this configuration we have produced $\approx 4\times {10}^6$ hydrogen–boron fusion reactions per laser shot.

The irradiation of calcium can result in the production of scandium radioisotopes, which are very interesting for present, and even more for future, medical applications. In particular, ⁴⁴Sc and ⁴³Sc are β⁻ emitters with a short lifetime ( $<4$ h) and with simultaneous emission of γ-rays only at relatively low energy. For these characteristics, they will produce little collateral damage to healthy cells and are therefore considered as optimal radiation sources for PET imaging.

In the experiment we irradiated calcium silicate samples (Ca₂SiO₄, sample size: 5 cm $\times$ 5 cm $\times 0.$ 5 cm). A thin layer of ¹¹B was deposited on the samples using a PLD technique at Politecnico di Milano. The layer thickness was ≈3 μm and its atomic composition was approximately equal to 95%–96% ¹¹B and 4%–5% O. More details about the PLD system and the characteristics of the boron films can be found in Refs. [43,44]. The goal of this layer was of course to absorb part of the incident proton flux and produce α-particles by the proton–boron fusion reactions. Such α-particles could then induce the formation of radioisotopes in calcium silicate.

The samples were placed behind the Al pitcher (6 μm) inclined by 12° with respect to the laser propagation axis, at a distance of 2 cm, and irradiated with 31 laser shots. After irradiation the sample was placed in the HPGe detector and acquisitions were done every 5 min (accumulating signal for 5 min). The accumulated spectrum is shown in Figure 9 in the range $950-1700\;\mathrm{eV}.$ It shows the presence of several γ-ray lines from ⁴⁴Sc and ⁴⁸Sc. At lower energy we find the line at 511 keV, due to positron annihilation.

Similar spectra have already been identified in the literature^[45^–47^] (but not from laser-generated radioisotopes). As an additional proof, we measured the decay time of the line at 1157 keV, as shown in Figure 9 (right). The measured lifetime agrees well, within error bars, with the lifetime of ⁴⁴Sc.

Scandium radioisotopes can be produced by irradiation of natural calcium with protons or α-particles, as shown in Table 2^[48^].

⁴⁴Sc is produced by the reaction of protons with ⁴⁴Ca. In our case, the isotope ⁴⁸Sc is likely produced also by protons rather than α-particles, first of all because, as written before, in laser-driven proton–boron fusion experiments the ratio of produced α-particles to protons is $\approx$ 10^–4, and secondly because ⁴⁸Ca is much more abundant than ⁴⁶Ca, representing only 0.004% of natural calcium (see Table 3).

As done for ¹¹C we can evaluate the number of produced radioisotopes starting from the recorded γ-ray spectra. The number of counts recorded at the initial time (see Figure 9) is $\approx 200.$ In this case since the lifetime of ⁴⁴Sc is not as short as that of ¹¹C, the corrections taking into account the decay time are not so important. The 200 counts obtained at the beginning of the measurement with the HPGe correspond to $\approx$ 226 counts half an hour before (i.e., at the end of the irradiation) and the correction taking into account the irradiation time implies an additional factor of $\approx 1.05$ (see Appendix A). This would bring the total number of counts to $\approx$ 240.

The detection efficiency approximately corresponds to that obtained from ²²Na at energy of 1274.5 eV (which is close to the 1157 keV line of ⁴⁴Sc), that is, $D\approx$ 0.00027 (for 5 min accumulation), which corresponds to an activity $A$ as follows:

$$\begin{array}{r}A\left[\mathrm{kBq}\right]=D\times \mathrm{Counts}\approx 0.065\mathrm{kBq},\end{array}$$  (14)

with 31 shots on target. The decay constant for ⁴⁴Sc is $\lambda =4.76\times {10}^{-5}$ s^–1, from which we get the following:

$$\begin{array}{r}{N}_{31\mathrm{\#}}=\frac{65}{4.76\times {10}^{-5}}\approx 1.4\times {10}^6.\end{array}$$  (15)

Hence, we can estimate a production of $\approx 4.5\times {10}^4$ radioisotopes per shot.

We also looked for the signature of ⁴³Sc, which emits a γ-ray line at 373 keV. The line was indeed present but it was very weak and superimposed on the Compton shoulder. In order to get an estimation, we had to remove the contribution of the Compton shoulder and smooth the data to remove noise. The original spectrum and the treated one are shown in Figure 10.

The low production of ⁴³Sc in comparison to ⁴⁴Sc is explained by the lower abundance of ⁴³Ca in comparison to ⁴⁴Ca within the natural calcium material used in the experiment. In addition, the production of ⁴³Sc from α-particles is negligeable, again because of the much smaller number of α-particles with respect to protons, which balances the fact that ⁴³Ca is only 0.135% of natural calcium (see Table 3).

We also observed a weak emission of γ-ray lines at 477 keV from the isotope ⁷Be, which is produced by the reaction p + ¹⁰B $\to$ ⁷Be + α + 1.15 MeV (T_1/2 = 53.22 days).

We have successfully shown the production of radioisotopes in laser-driven experiments. This can be useful potentially for the possibility of producing radioisotopes of interest for medical applications and also for diagnostics purposes (i.e., to infer the total number of nuclear reactions taking place in the target). In this case, of course, we need to greatly improve the yields from laser experiments in order to become competitive with existing tools for radioisotope production.

Currently radioisotopes are produced either in nuclear reactors (by neutrons) or in dedicated cyclotron systems (by protons). A few specific radioisotopes are also produced using large heavy ion cyclotron systems (such as the cited ARRONAX or U-120M), which accelerate He nuclei, that is, α-particles. For instance, ARRONAX can produce a current of 70 μA of α-particles. A current of 10 μA corresponds to a flux of α-particles:

$$\begin{array}{r}10\mu \mathrm{A}\to N_{\alpha }[{\mathrm{s}}^{-1}]=\frac{{10}^{-5}\mathrm{C}/\mathrm{s}}{2\times 1.6\times {10}^{-19}\mathrm{C}}=3\times {10}^{13}{\mathrm{s}}^{-1}.\end{array}$$  (16)

Reaching performances of the order of a few 10 μA of α-particles with lasers is extremely challenging. Today, laser experiments show a maximum of 1011 α-articles per sr per shot or about a maximum of ${10}^{12}$ α-particles per shot^[11^,12^]. In order to be comparable, such a laser-driven source must then work at a repetition rate as follows:

$$\begin{array}{r}f=\frac{3\times {10}^{13}\mathrm{per}\mathrm{second}}{{10}^{12}\mathrm{per}\mathrm{shot}}=30\mathrm{Hz}.\end{array}$$  (17)

This requires developing a new generation of $\approx$ 100 Hz petawatt laser systems. Although this is a challenging goal, laser technology is indeed already moving in this direction^[49^]. Also notice that in our experiment, the α-particle yield was much lower than ‘record’ yields.

In reality, even operating lower performing laser-driven α-particle sources could still be interesting if laser-driven sources are cheaper and more compact so as to be installed in more medical centers and be more diffused in the territory (this is particularly important for short-life radioisotopes, which are the most interesting for medical applications), serving as ‘in-hospital’ isotope manufacturing for fast administration of short-lived isotopes.

At the same time, in order to increase the number of produced radioisotopes for laser shots, we need to carefully choose target materials. For instance, while in our experiment we produce $4.5 \times 10^{4}$ of ⁴⁴Sc radioisotopes per shot with natural Ca (a mixture of ⁴⁴Ca and other Ca isotopes), we could increase the production by using a target containing only ⁴⁴Ca. This is likely to increase radioisotope production by a factor 1/2.09% $\approx$ 50. An even larger increase would be obtained by using a pure ⁴⁴Ca target instead of a calcium silicate.

It is also clear that for radioisotopes produced by α-particles, a three-step process (pitcher $\to$ catcher $\to$ target material) is very ineffective. Many particles are lost at each step, and, above all, most α-particles are confined in the catcher due to the very short propagation range in solid density matter ( $\approx$ micrometers). We therefore need to use mixed targets, where boron (to produce α-particles from protons) and the target material (generating the radioisotopes) are in direct contact. Concerning Sc radioisotopes, interesting materials are indeed calcium silicate or even better calcium hexaboride (B₆Ca).

The other important question concerns the purity of the produced radioisotopes. For instance, in the case of Sc, it would be preferable to produce a single radioisotope rather than a mixture of ⁴⁴Sc, ⁴⁸Sc and ⁴³Sc. ⁴⁴Sc and ⁴³Sc are both used in medicine as a source for PET imaging. However, ⁴³Sc is often considered as the ‘radioisotope of the future’ in PET because of its short lifetime and the absence of simultaneous γ-emission at energies high enough to possibly cause radiation damage to human cells (⁴³Sc emits γ-rays at 373 keV, which is even lower than γ-rays emitted by ⁴⁴Sc at 1157 keV). Pure ⁴³Sc could be obtained by irradiating natural calcium with α-particles. Unfortunately, in experiments with laser-driven α-particle sources, the α-particles are always accompanied by a much larger flux of protons that, as in the present experiment, can produce ⁴⁴Sc by reaction with the isotope ⁴³Ca. Indeed, although ⁴³Ca is only 2.09% of natural calcium, there are far fewer α-particles than protons, so this reaction channel is dominant. Using isotopically pure ⁴⁰Ca would prevent this; however, ⁴⁰Ca is present at 96.9% in natural Ca and it is very difficult to get isotopic concentration above 99.5%^[50^].

Instead, pure ⁴⁴Sc could be obtained using protons on isotopically pure ⁴⁴Ca (the production of ⁴³Sc by the 2n reaction being largely minor). In this case, one should consider that the present compact cyclotrons used for production of medical radioisotopes have currents of the order of 100–150 μA, which is a factor of $\approx$ 2 above the α-particle current produced by ARRONAX. However, as we said before, the ratio of α-particles to protons in laser-driven experiments for the production of particle sources is typically a factor $\approx$ 104 in favor of protons. To produce 100 μA, a laser that works at 100 Hz repetition frequency should accelerate $\approx 6 \times 10$ 12 protons per laser shot. If the protons have an average energy of 10 MeV, the total energy in the proton beam is about 1 J, which could imply a laser energy per shot of 10 J assuming a laser-to-proton conversion efficiency of 10%. Indeed, these numbers represent a reasonable (although optimistic) extrapolation of current performances, which shows how laser-driven proton sources could also be interesting for the production of medical radioisotopes.

In our experiment, we have successfully identified γ-ray lines from multiple radioisotopes created by irradiation using laser-generated α-particles or protons. These includes ⁴³Sc, ⁴⁴Sc, ⁴⁸Sc, ⁷Be, ¹¹C and ¹⁸F. With respect to our previous work^[28^], here we focused on radioisotopes of medical interest (⁴³Sc, ⁴⁴Sc) and we provided an evaluation on how laser-driven sources could become competitive for radioisotope production. We also described the HPGe detector and the calibration procedure in detail, and we have also shown a fairly good agreement between data obtained from γ-ray spectroscopy and other diagnostics (CR39). This shows the possibility of using γ-ray spectroscopy as a reliable diagnostic tool to measure the activation of target materials and reaction rates. This can be important, for instance for proton–boron fusion experiments where measuring α-particle yields with CR39 has always been an issue.

We have also shown the production of $\approx$ 6 × 106 of ¹¹C radioisotopes and $\approx$ 5 × 104 of ⁴⁴Sc radioisotopes per laser shot. Again, a proper choice of target material can considerably increase the production rate: this could be pure ¹¹B for ¹¹C production or pure ⁴⁴Ca for ⁴⁴Sc production. This result can open the way to develop laser-driven radiation sources of radioisotopes for medical applications.

In this context in the future, we will need to perform experiments that produce a separable quantity of radioisotopes (first step), and that match the doses (MBq) used in medicine (second step). These objectives can be achieved by accumulating a much larger number of shots (i.e., working at a high repetition rate), by optimizing laser parameters and by developing new target materials (e.g., for radioisotopes produced by α-particle targets mixing boron and the precursor material of the radioisotope to be produced).