1 Introduction

The research on neutrons and neutron-based applications is so extensive that listing them all would be beyond the scope of this paper. In nuclear physics, materials science, biology and medicine, neutrons are utilized primarily for the modification and diagnostics of matter, specifically atoms. Due to the unique properties of neutrons, they enable the investigation of material properties that remain invisible to other measurement techniques.

Early neutron sources were reactor-based, a technology impractical for scalable (civil) research due to reactor safety concerns. The cutting edge of such user research reactors are the ILL (Grenoble, 58 MW) and HFIR (Oak Ridge, 85 MW), alongside numerous smaller reactors worldwide with a thermal power of a few MW. The increase of the neutron flux is made possible by the next generation of sources, including proton accelerator-based spallation neutron sources, such as J-PARC in Japan, SNS at Oak Ridge, or the European Spallation Source (ESS) in Lund. However, the investment (ESS: ~ €2 billion) and operating costs of spallation sources are notably high and not expected to decrease in the future due to the essential component of a high-energy (0.5–2 GeV) proton accelerator.

Low and medium-power cold and thermal neutron sources are now commercially available. The demand for devices, typically at the industrial level (approximately 108–1010 n/sec), is steadily increasing. A survey conducted by the International Atomic Energy Agency (IAEA) indicates [1] growing user demand for neutron sources with significantly higher luminosity and flexibility than those mentioned above. These sources should be available even in smaller laboratories, smaller than facilities such as ESS or ILL. It is important to note that existing (thermal) neutron sources are partly continuous, i.e. their output is constant over time, and partly pulsed. The duration of the neutron pulses is typically a few microseconds, which can be reduced to the ns range by mechanical shutters.

Advances in laser ion acceleration have now enabled PW-class lasers to routinely produce protons with energies of 50 MeV and above. However, this is still considerably lower than the 0.8–2 GeV range utilized in spallation sources. Therefore, neutron generation in laser sources relies not on the fragmentation of large nuclei but rather on nuclear reactions involving small nuclei, such as 2H + 2H, 2H + 3H fusions, or 9Be(p,n)9B, 9Be(d,n)9B reactions [2].

While neutrons can be generated by focusing the laser on a bulk target, we will concentrate on the so-called “pitcher–catcher” arrangement because of its advantages. In this case, the high-intensity laser pulse (> 1018 W/cm2) is first incident on an “ion” target (pitcher) from which an ion beam (typically proton or deuteron) emerges at a well-defined solid angle (Fig. 1). Subsequently, a “neutron” target (catcher) is positioned, which emits neutrons through the desired nuclear reaction involving the proton or deuteron.

Fig. 1
figure 1

Schematic diagram of a catcher–pitcher laser neutron source

Full size image

Unlike spallation, the neutron emission mechanism described above is based on materials containing low-Z elements. Consequently, there are no fission products, and generally, fewer radioactive products are produced. Additionally, the (neutron) target is exposed to much lower thermal power. These factors collectively result in significantly smaller cooling systems and radiation shielding, allowing the sample to be positioned closer to the neutron source. Furthermore, neutrons emitted by laser-induced deuterons are expected to propagate in a quasi-collimated beam with a considerably smaller solid angle than 4π of conventional thermal neutron sources, as demonstrated by experiments conducted so far. Hence, it is expected that the sample to be irradiated may contain an order of magnitude more neutrons in terms of the total neutron count from the neutron source, leading to higher neutron intensity for the same number of neutrons compared to spallation or reactor neutron sources.

The initial pitcher–catcher laser-based neutron irradiation experiment was carried out at the Rutherford Appleton Laboratory in 1998 [3]. Similar experiments conducted then and ever since have been performed almost exclusively with relatively long sub-picosecond laser pulses with energies ranging from several joules to several tens of joules [4,5,6,7,8,9,10,11]. The highest neutron count achieved in a single shot occurred with the PHELIX laser (160 J, 600 fs) at GSI in 2018 [12]. The measured number of neutrons was 1.42 ± 0.25 × 1010/str, corresponding to a dose of 430 ± 50 µSv at 1 m from the source, with a neutron beam aperture angle of about 100°. Figure 2 shows the achieved neutron yield per second as a function of laser intensity. There are only a very few cases, denoted by square marks, where neutron generation was attempted with pulses shorter than 100 fs [11, 13, 14].

Fig. 2
figure 2

Neutron yield per second achievable with laser neutron sources, assuming DD reaction. The USZ–ELI ALPS data from our programme is nearing completion, and the result of the predicted further developments is indicated by the yellow triangle

Full size image

For the sake of completeness, it is worth noting that, apart from the pitcher–catcher scheme, neutron generation can also be achieved by employing a single deuterated target. Here the 2H + 2H reaction takes place between the deuterium present within the target itself and deuterons from the plasma forming on the front surface of the target. This is commonly executed in liquid jets [11], gas jets [15], or thicker deuterated polyethylene [16]. More recently, Knight et al. [17] reintroduced the approach using liquid droplets. However, this process is considerably less efficient and more challenging to optimize compared to the pitcher–catcher scheme.

For applications, the number of neutrons produced per unit of time holds greater significance than the number of neutrons produced per shot. Based on the above-mentioned experiments, the neutron yield (neutron/s) achievable with (ultra)short pulse lasers at high repetition rates is comparable to the neutron number produced by high pulse energy lasers (see Fig. 2). It is evident that the yield of neutrons per second correlates with the average power of the laser systems. Currently available high average power laser systems are also the ones with high repetition rates (100 Hz – 1 kHz), still with TW peak powers and relativistic focused intensities.

Hence, our programme is dedicated to the utilization of high-repetition-rate laser systems with few-cycle laser pulses for fast neutron generation. In the initial phase of our programme, we investigated the mechanism of proton acceleration [18] with a few-cycle laser system [19]. Building upon this foundation, we show a study on the choice of neutron generation procedure and neutron converters to achieve the highest possible neutron yield. Subsequently, we describe the versatile experimental setup and neutron diagnostics, report results on the single- and multiple-shot acceleration of deuterons from foils and the subsequent neutron generation. The subsequent sections elaborate on the preparation of a liquid jet target system and provide initial results on proton generation. The paper concludes by describing the preparation of the first user experiment involving in vivo radiobiological research, examining the impact of laser-generated neutrons on living cells.

2 Nuclear reactions for lased-based neutron generation

The first experiments [18] with few-cycle, yet relativistic laser pulses with modest energy revealed that the proton and ion spectra are close to exponential, and the cutoff energy is well below 2 MeV. Taken such data, we need to consider the most important nuclear reactions for neutron production (Table 1).

Table 1 The most important nuclear reactions with light ions (proton, deuteron) of low energy (< 1 MeV) for neutron generation
Full size table

All these reactions are exothermic, which means that they have no threshold energy, and thus all ions are able to produce neutrons. However, the cross-section of these reactions sharply decreases below 100 keV, with ions at higher energies primarily contributing to neutron production. Table 1 illustrates that the most favourable reaction is 2H + 3H, as its efficiency exceeds that of other reactions by orders of magnitude.

The most common targets for neutron production include deuterized polyethylene (dPE), titanium tritide (TiT), titanium deuteride (TiD), heavy water (D2O), and lithium fluoride (LiF). While neutrons can, in principle, also be produced from constituents other than deuterium, tritium, and lithium obtained from these materials, such contributions are generally small.

To compare the expected number of neutrons, as well as their energies and angular distributions for different targets, simulations were performed with the Geant4 software, using the experimentally observed energy and angular distributions of the deuterons. The target thicknesses were adjusted in a way to stop the deuterons completely (Fig. 3). All simulations corresponded to 1015 initial deuterons, and the number of emitted neutrons is indicated after the name of the targets. For the dPE, TiD and D2O targets, the spectra are dominated by the peak originating from the 2H + 2H reaction between 1.8 and 5 MeV. Some contribution from the 2H + 12,13C reaction below and above the main peak is also visible for the dPE target. For the TiT target, the peak corresponding to the 2H + 3H reaction extends from 12 to 18 MeV. For the LiF target, the low-energy part below 5 MeV results from the 2H + 6Li reaction, while the high-energy part above 10 MeV can be attributed to the 2H + 7Li reactions with a small contribution from the 2H + 19F reaction.

Fig. 3
figure 3

Simulation of the neutron spectra of the reactions in Table 1. The legend shows the number of generated neutrons

Full size image

In most applications, neutrons interact with a finite-sized target placed at a distance from the source. Hence, beside the neutron yield, it is also essential to know the spatial distribution of the fusion neutrons. In our simulations, we also investigated the angular distribution of the emitted neutrons (Fig. 4). The results indicate that the 2H + 2H reaction (dPE, TiD, D2O targets) produces a forward–backward peaked distribution, while the distributions for the 2H + 3H and 2H + 6,7Li reactions (TiT, LiF targets) are nearly flat.

Fig. 4
figure 4

Simulation of the neutron angular distribution of the reactions in Table 1

Full size image

As seen from these simulations, TiT emerges as the overall best target. However, due to isotope contamination and safety concerns, only a limited number of laboratories can afford to utilize such targets. The second best candidates overall are those containing deuterium (dPE, TiD, and D2O), yielding neutron numbers comparable to the tritium-containing target. Conversely, the LiF target produces orders of magnitude fewer neutrons.

3 Experimental setup

3.1 Optical path

The SEA Laser of ELI ALPS is able to produce laser pulses with 0.2 to 40 mJ energies, a transform limited pulse duration of 12 fs, and repetition rates from single shot to 10 Hz [19]. Due to the scheme of optical parametric chirped pulse amplification and the modest average power, not only the temporal contrast is excellent, but the pointing of the beam and the focal spot distribution also remain unaffected when the laser energy and repetition rates are varied. These features make the target alignment procedure relatively easy and well reproducible, so the SEA laser is an ideal testbed for exploring laser ion acceleration at various targets, repetition rates, and energies.

The SEA laser is installed in a dedicated laser room with ISO 8 optical cleanliness, while the experimental chamber (LEIA) is in a radiation shielded laboratory (Mid-Shielded Target Area, MTA), about 30 m away. To avoid self-phase modulation of the intense, 35 mm diameter (FWHM) laser beam in the vacuum windows and valves, as well as to relax the damage threshold requirements of the broadband steering mirrors, the laser pulses are only partially compressed (~ 1 ps) when they enter the vacuum beam transport system in the laser room. The other end of the transport system in the experimental hall (MTA) is connected to a compression chamber (CC1, not shown in Fig. 5), where five pairs of chirped mirrors compress the pulses to their transform-limited duration. The next chamber (CC2) immediately preceding the interaction chamber (LEIA), is dedicated to the wavefront correction of the SEA laser beam, with a deformable mirror (DM), two motorized mirrors, and a motorized iris at the output of CC2 (Fig. 5). There is a second motorized iris in the LEIA chamber at the entrance of the beam, ~ 3 m away from the first iris in CC2. In the LEIA chamber, there is a chirped mirror at 7° angle of incidence, providing a small leakage of the incident laser pulse. This beam serves a continuous near-field (NF) and far-field (FF) monitor, as well as wavefront monitoring by the HASO sensor of the DM. One part of the split leaking beam is used for imaging the plane of the input iris of the LEIA chamber onto the NF monitor. The other part of the beam is focused onto the plane of a 0.4 mm diameter pinhole, positioned in the chamber. The plane of the pinhole having the focal point in the middle, is imaged onto the FF monitor. In this way the exact position of the beam entering the LEIA chamber is monitored and maintained, independently of the movement of the chamber itself upon evacuating and airing. The HASO sensor also uses the NF beam, with a flip mirror, so that the DM surface is imaged onto the sensor.

Fig. 5
figure 5

The experimental setup. For the description, please see the main the text

Full size image

To make the preliminary alignments of the optics and targets, a 200 mW diode laser was connected to the CC2. Its diameter was close to the main beam’s diameter and was aligned completely along the main beam path with the use of the NF and the FF diagnostics.

3.2 Off-axis parabolic mirror (OAP) alignment

The precise alignment of an OAP is of crucial importance for maximizing the interacting laser intensity on the target surface. Several, mainly unpublished recipes exist on how to align an OAP to an existing laser beam. To achieve the best possible alignment, we reversed the procedure. First, the OAP was aligned with the help of a HeNe laser beam, then the SEA laser beam was made to co-propagate with the HeNe alignment beam.

First, we aligned the HeNe beam path from the source to the OAP, setting the appropriate interaction angle (Fig. 5). With the Focus imaging objective, the interaction point was designated as the focal point of the HeNe. Therefore, the OAP foci coincided with the focal point of the objective, and hence the HeNe beam leaving the OAP could be collimated. Any deviation from collimation is a sign of the inaccurate alignment of the OAP. To observe collimation, we applied the two-beam interferometry method using two thin plan-parallel plates [20]. In this arrangement, one of the plates served the investigation of horizontal collimation, while the other, the vertical one. As the plates were tilted horizontally and vertically, respectively, the reflected beam from their front and back surfaces produced the well-known interference stripes. In case of perfect collimation, the interference pattern should become a single interference fringe, a homogeneous spot, with an intensity given by the phase shift between the front and back reflections. If the focal point of the OAP is on the alignment beam axis, the stripes are straight and parallel for each plate. If the pattern is curved, then the tip, tilt, rotation, and the focal position of the OAP, mounted onto a five-axis mount, need adjustment. In this way we can align the parabolic mirror’s focal point with an accuracy of ~ 20 µm, while the imaging errors can be reduced as much as possible. Once the HeNe beam is collimated, we steer it back to the entrance of the CC1 chamber so that it overlaps with the SEA laser beam, and the two beams co-propagate.

3.3 Preparation of deuterated thin film targets

Following from the conclusion of Sect. 2 and the experience on proton acceleration with few-cycle laser pulses [18], one can expect efficient deuteron acceleration from deuterated targets with thicknesses around 100 nm. Unfortunately, no such targets are known to be commercially available yet. Hence, first we tried the chemical methods reported so far [21,22,23]. Among these, we found the best results through a modification of the solvent casting method [23].

First, we stirred 3.5 mg of C2D4 granules into 5 ml of xylene in a cup. The mixture was then heated to 160 ℃, close to the boiling point of the xylene, in a double boiler setup. This temperature was maintained for 90 min under precise temperature control and uniform heating, until the C2D4 granules fully dissolved. The clear solution was then immediately and meticulously poured onto microscope slides, resulting in a uniform film upon evaporation. The thin film was removed through a combination of heating/cooling (thermal shock) and water immersion. One cup of solvent was sufficient for eight microscope slides, i.e. to produce eight pieces of deuterated foil with a size of ~ 25 × 50 mm2 each (Fig. 6a).

Fig. 6
figure 6

Our deuterated thin film on the holder frame (a). A thickness measurement relative to the microscope slide with a profilometer (b)

Full size image

Once the films are prepared, the first important question that arises is about their thickness. The rough but simple estimation of the thickness was based on the measurement of the surface and weight of the produced layers, and the density of the material. This calculation provided us with an average value of 230 nm with an error of 18%, without information on the evenness of the sample.

We attempted to measure the thickness directly with a profilometer (Dektak 8 Advanced Profilometer). We left the film to be measured on the microscope slide after the preparation process and left it to dry. We made a clear cut in the middle of the sample vertically (parallel with the shorter side). The sensor needle was driven through the edge of the polymer sheet at ~ 3 mm intervals, resulting in several curves (Fig. 6b). The position of the cut was slightly shifted due to the small deviation from 90 degree of the scan direction relative to the cut. The sheet thickness was defined as the distance between the average height of the sample relative to the cut (level zero). The average film thickness, defined as the average of the several scans, equalled ~ 200 nm, with an error of 12%.

Here we need to mention that we also tried making thinner films using less C2D4 powder. However, with half of the amount it was challenging to remove them from the microscope slides. On the other hand, doubling or tripling the powder amount allowed for the production of thicker targets.

The detailed procedure guaranteed the reproducibility and reliability of the deuterated foils used in the experiments discussed in the following sections.

3.4 Target positioning

The modest energy of the SEA laser requires tight focusing to surpass the relativistic laser intensity and thereby produce MeV class ion beams. Towards this goal, an F/2 OAP was used to irradiate the target with a spot size of around 3 µm and a Rayleigh range spanning around 35 µm (Fig. 5). Under these tight focusing conditions, to irradiate the target with the highest achievable laser intensity and to maintain shot-to-shot performance stability, each target surface had to be prepositioned at the plane corresponding to the smallest laser focus, with an accuracy of a few micrometres using the same microscopic objective. For targets irradiated under a normal incidence angle, it is most advantageous to use white-light back illumination, as the microscope collects strong signals from back-reflected light [24]. However, for targets shot under oblique incidence, the back-illuminating incoherent white-light source can be replaced with a diode laser or a similar source, as the backscattered signal from the diode laser is much stronger than from a white-light source.

The target surface viewing system comprises a long working distance microscope objective (10 × , Mitutoyo, numerical aperture 0.3) and a 10-bit CMOS camera (Zelux, Thorlab). The broadband white light, a white LED, was incorporated in the target viewing assembly and illuminated the target through a partially reflective mirror (Fig. 7). The coherent source is the HeNe laser itself, used to align the OAP (see above). The optical resolution of this assembly at oblique incidence is around 5 µm, which is sufficient to prealign the ultrathin transparent targets.

Fig. 7
figure 7

Schematic of the target surface positioning arrangement (M, mirror; B.S., beam splitter)

Full size image

A target holder covered with the foils, similar to that in Fig. 6a, was mounted on a rotating wheel. The positions of the first and the last holes in a row were carefully set with the procedure above, while the position of the middle holes were interpolated from them. The target wheel was then moved from one hole to the other, allowing for three shots in a hole.

4 Particle diagnostics

4.1 Thomson ion spectrometers

To characterize the ion beams produced in the experiment, the development of precise diagnostics was essential. A Thomson parabola spectrometer (TPS), equipped with a microchannel plate (MCP) coupled to a phosphor screen, was used to determine the cutoff energy of accelerated particles and, following calibration [25,26,27,28], it also enabled the quantification of the number of accelerated particles.

In our experimental chamber, two identical TPSs were positioned normal to the target surface, in the forward (FWD) and backward (BWD) directions (Fig. 5). Within a TPS setup, a 200 µm pinhole was positioned at the entrance of the TPS chamber, providing a solid angle of 30 × 10−9 sr for ion collection. The magnetic field generated by the permanent magnets was 0.195 ± 0.001 T, while the electric field was set to 3 kV/cm. An MCP coupled to a P43 phosphor screen (Photonis APD75/12/10/8I60:1NR8′′) made the ion traces visible, which were recorded by a 16-bit CCD camera. The distance between the camera’s chip and the MCP surface was 54 cm. The energy resolution of the device for light ions was ΔEi/Ei ∼ 5% above 0.5 MeV, while it approached 1% towards 0.1 MeV.

In our previous study, the Thomson ion spectrometer was calibrated to the number of detected protons using CR-39 plates [29]. Given the importance of determining the count of deuteron ions in neutron generation, a similar procedure was necessary for laser-accelerated deuterons. This calibration procedure enabled a direct “in-situ” comparison of detectors (CR-39 vs. MCP and camera) using the laser-accelerated ion signal.

Here, a slotted CR-39 plate with identical slots spaced at 2 mm was positioned in front of the MCP detector at a distance of 5 mm. A portion of a parabolic trace was intercepted by the CR-39, leading to the formation of ‘shadows’ on the MCP signal corresponding to the CR-39 bars. Simultaneously, the MCP signal was generated through the empty slots. Figure 8a displays the ion traces on the MCP and camera when the deuterated PET foils were irradiated. The deuteron pits on the slots of the CR-39 were calculated with the help of a microscope. By describing the MCP signal and the detected deuteron numbers together (Fig. 8b), a direct comparison could be made between the CCD pixel value and the density of deuteron pits on the CR-39. Their ratio, regarded almost constant (solid line in Fig. 8c), was utilized for evaluating the deuteron counts in Sect. 5.

Fig. 8
figure 8

a Ion traces on the MCP of a TPS detected by a CCD camera. b MCP signal and deuteron pit counts on CR-39 as a function of energy. c The resulting calibration curve of TPS for deuterons

Full size image

4.2 LILITH system

The LILITH neutron spectrometer is designed to detect low-energy neutrons produced by laser-accelerated deuterons, with kinetic energies determined through the time-of-flight (ToF) method. The system comprises four cylindrically shaped EJ-230 fast plastic scintillators (with dimensions ⌀150 mm × 25 mm), coupled to Hamamatsu R2083 photomultipliers. Each detector, positioned on an independent stand, is shielded by a 5 cm lead wall to reduce the impact of gamma rays following each laser shot. The anode signals of the photomultipliers are processed using a CAEN V1751C 10-bit flash ADC waveform digitizer (1 GSample/s), controlled by a self-developed, digiTES-based acquisition system [30] with ToF triggering criterion.

ToF is determined by measuring the time between the start signals emitted by a photodiode positioned upstream of the target chamber and the timing of each detector’s signal when the coincidence condition is satisfied. Waveforms, typically ranging from 600 to 1400 ns, are digitized online and analyzed offline to extract ToF data. In Fig. 9, a set of digitized waveforms is presented, and recorded within a 500 ns coincidence time window. The lower signal corresponds to the photodiode (ch0), followed by the responses of four LILITH detectors (ch1-ch4), each exhibiting a gamma peak around 150 ns. Notably, LILITH detectors #1 and #4 detected a second signal from neutrons with distinct time delays.

Fig. 9
figure 9

A typical set of digitized waveforms obtained during the 1 Hz experiment. From top to bottom, the signals for LILITH #1, #2, #3, #4, and the photodiode (time reference) are presented for one laser shot. The baselines are shifted for better visualization

Full size image

In the LILITH system, utilizing the default ToF path of 3 m, the opening angle coverage was determined to be 1.947 ± 0.016 msr. ToF offset calibration was performed by aligning the gamma rays’ peak at 9.9 ns, facilitating the calculation of the ToF path between the centre of the catcher target and the detectors. The detector light outputs and efficiencies were calibrated using 137Cs, 60Co, 22Na, and plutonium–beryllium (PuBe) sources. The efficiencies exhibit variations of less than 5% within the relevant energy range.

5 Proton and deuteron acceleration from ultrathin foils

Once a segment of the target wheel was set according to the description in Sect. 3.4, we started shooting with the SEA laser at 1 Hz repetition rate. The pulse energy on target was 21 mJ, while ~ 15% of the energy was inside the 3.0 × 2.7 µm2 (FWHM) size focal spot. Hence, the interacting intensity is 4 × 1018 W/cm2. Shots were halted after each segment, and each successive segment was pre-aligned. In total, seven segments, hence around 500 shots were made on one wheel, and two to three wheels were completed in a day. The ion spectra were recorded for each shot in both forward and backward directions. In agreement with the observation of proton and carbon ion acceleration from thin foils [18], now the deuteron yield was a few times higher than that of protons, and both the cutoff energy and the ion count were higher in the forward direction than in the backward direction.

In addition to the cutoff energy, the number of charged particles, more precisely, the total energy of the particle bunch serves as a measure of particle acceleration efficiency. The bunch energy was calculated for particles with an energy above 50 keV, taking into account the calibrated spectra, the geometrical parameters of the setup and the divergence of the ion beams. The latter was carefully measured for protons in an earlier experiment [31] and was found to be less than 5°. For the deuteron beam, we assumed the same divergence as the proton beam.

A histogram of accelerated deuterons in the forward direction is displayed for one segment of the rotating wheel. The mean energy and cutoff values are 0.044 mJ and 607 keV, respectively (Fig. 10). The former corresponds to 1.3% conversion efficiency relative to the interacting laser energy. From the analysis of all successful shots during the three days campaign we could reveal that the highest cutoff energy of the accelerated protons and deuterons was around 1.4 MeV and 1.1 MeV, while the conversion efficiencies peaked close to 2.7 and 5.0%, respectively.

Fig. 10
figure 10

Histogram of the cutoff energy (left) and the burst energy (right) of accelerated deuterons from one segment of the rotating wheel target

Full size image

To take advantage of the relatively large number of shots, it is worth looking at the correlations between the cutoff energy and the energy ion bunches (Fig. 11). Interestingly, the best fit functions are steeper in the backward direction than in the forward direction, however the maxima of the values behave in the opposite way. In addition, for both directions, towards higher particle energies (i.e. number of accelerated ions), deuteron dominates, while at low energies protons are prominent. This observation qualitatively agrees with the rough physical picture, according to which the electrons first accelerate the lighter protons, and then the heavier deuterons.

Fig. 11
figure 11

Correlation between the cutoff energy and the total ion bunch energy in FWD a and BWD b directions

Full size image

Since the ion spectra were recorded for each shot also in the backward direction, similarly to [32,33,34], it was worth examining the correlations between the cutoff energies in the FWD and BWD directions (Fig. 12), as the values are easiest to establish from a TPS ion spectrum. The Pearson’s correlation coefficients are 0.31 for protons and 0.47 for deuterons, respectively. The reason of the significantly different correlations is likely due to the different acceleration mechanisms revealed in [18]. Even if the correlation is not superstrong, this may help in estimating the forward accelerated ions, thereby improving the accuracy of target alignment even in cases, when the FWD TPS is obscured by the neutron catcher itself.

Fig. 12
figure 12

Correlation between the FWD and BWD cutoff energies for the shots on the segment shown in Fig. 10

Full size image

6 Neutron generation at 1 Hz repetition rate

In this experiment, fusion neutrons were identified using LILITH detectors positioned at angles of 29 ± 5°, 93 ± 5°, 99 ± 5°, and 153 ± 5° relative to the direction of accelerated deuterons, situated approximately 2 m from the catcher. Then, data captured by the LILITH detector were subjected to an offline waveform analysis.

Figures 13a–c illustrate two-dimensional scatter plots showcasing the relationship between total light output and ToF for gamma rays and neutrons detected by the LILITH detectors at forward (LILITH #1), perpendicular (LILITH #2 and #3), and backward (LILITH #4) orientations. The initial sharp peaks in each case correspond to gamma rays, with ToF offsets calibrated to 6.66 ns. Neutrons were identified beyond 80–120 ns on the right side of the spectra, while events between the gamma peak and the neutron range were attributed to gamma rays produced by scattered neutrons. Directional dependence in neutron ToF distributions was observed for forward, perpendicular, and backward detectors, indicating a shift in the ToF peak towards higher values with increasing laboratory angle, in accordance with kinematic expectations.

Fig. 13
figure 13

Panels a–c display data for forward (LILITH #1), perpendicular (LILITH #2 and #3), and backward (LILITH #4) directions, showcasing directional energy variations. Notably, neutron Time-of-Flight (ToF) distributions exhibit a discernible shift to higher values as the laboratory angles progressively increase

Full size image

Neutron-pile-up events, where two neutrons were detected within the same LILITH detector in less than 10 ns, were observed, accounting for less than 2% of the total detected neutrons across the four LILITH detectors. Out of the 3,128 evaluated shots, a total of 755 neutrons were obtained, with counts of 242 ± 16 (stat) ± 7 (syst), 142 ± 12 (stat) ± 4 (syst), 154 ± 12 (stat) ± 4 (syst), and 217 ± 15 (stat) ± 6 (syst), in LILITH detectors #1, #2, #3, and #4, respectively. Systematic uncertainties primarily arose from challenges in neutron identification via ToF gate and light output cuts. LILITH data indicate an average neutron yield of 1142 ± 59 per laser shot [35], representing the mean yield across the entire 4π solid angle, and accounting for neutron distribution anisotropy.

Determining the kinetic energies of specific deuterons that generated the detected neutrons on an event-by-event basis was not feasible, as the presence of deuterons with varied kinetic energies introduces uncertainty in the ToF measurements of neutrons. This uncertainty arises from the fact that the origin of the gamma peak is the time of interaction between the pitcher and the laser, rather than the interaction between the catcher and the deuteron. To address timing uncertainties, we applied an offline correction factor of 21.9 ns to the measured ToF data. This factor corresponds to the ToF of deuterons with a kinetic energy of 177 keV, serving as the weighted mean value over a path length of approximately 9 cm (catcher–pitcher distance). The neutron kinetic energy spanned a range from 1.5 to 4 MeV.

Figure 14 illustrates the angular distribution of detected neutrons, in accordance with simulation results obtained from the average deuteron spectra of more than 2000 analyzed laser shots. These simulations considered various factors, including kinematic conditions, cross-section, and angular distribution of the 2H–2H reaction, and characteristics of the experimental environment.

Fig. 14
figure 14

Comparison of the detected neutron count at various laboratory angles with the simulated angular distribution

Full size image

The angular distribution is consistent with existing literature on the angular distribution of the 2H(d, n)3He reaction [36, 37], peaking at both forward and backward angles, with a minimum around 90 degrees in the laboratory frame.

7 Outlook to high repetition rates and applications

7.1 Ion acceleration at 10 Hz repetition rate

To increase the repetition rate of the neutron source, a regenerative target system is needed, which can support both laser pulse–matter interaction up to a thousand per second and operation for several hours. A liquid jet is one of the few demonstrated solutions so far [38,39,40], in addition to ultrahigh pressure gas nozzles [41, 42] as well as cryogenic ribbon targets [43], which satisfy this requirement.

Since the thickness of the liquid target needs to be in the range of 100 nm, we launched a project with AMS GmbH for the development of a liquid leaf target to reach the so far unprecedented thickness. The liquid leaf results from the collision of water microjets from two round nozzles. The thickness and the length of the leaf as well as the surface topology depend basically only on the size of the nozzle orifices and the flow rate. The maximum thickness is close to the collision point, while it monotonically decreases towards the other end of the leaf [44].

To measure the thickness of the liquid leaf, also in vacuum, we developed a white-light interferometric arrangement [45]. The almost octave broadband light was steered into the vacuum and sent through the liquid leaf. Spectral interference formed by the beams reflected from the front and back surface of the liquid, and was collected by an objective and sent to a fibre spectrometer. We revealed that a liquid leaf is around 10% thinner in vacuum than in air. The minimum thickness achieved and measured in vacuum was below 200 nm, with an orifice sizes of 11 µm [45].

Using water, we also tested the liquid leaf with the interaction of the SEA laser pulses at 10 Hz repetition rate. The ion spectra were recorded for each shot in a way described in Sect. 5. As seen in Fig. 15, the stability of the cutoff energy of the accelerated protons was 3.5%, while the rms error of the corresponding laser pulse energy was around 0.6%. Such a high stability laser-plasma proton accelerator is clearly unparalleled and has not been demonstrated so far. We believe that with the use of heavy water and a neutron catcher, a high-performance neutron source would be soon demonstrated at a minimum repetition rate of 10 Hz.

Fig. 15
figure 15

Cutoff energy (blue) of laser accelerated protons from a water sheet target and the corresponding laser pulse energy (yellow) during a 15 min operation at 10 Hz repetition rate

Full size image

7.2 Neutron detection with PSD

There can be experimental arrangements where the feasibility of Time-of-flight (ToF) spectrometry is limited. In such cases, a well-characterized neutron sensitive scintillation detector, the signals of which are evaluated with the Pulse Shape Discrimination (PSD) and the Pulse Height Response Spectrometry (PHRS) techniques in combination with a derivative unfolding method can be used to obtain information on the neutron spectrum [46, 47].

The principle of the method is shown in Fig. 16. The X-photons and neutrons emitted by the laser-based neutron source were detected by a 2″ × 2″ cylindrical EJ-301 organic liquid scintillator. The X-photons and the neutron induced recoils (protons, carbon and oxygen nuclei) generated light pulses in the scintillator. The light pulses were converted to electronic pulses by a HAMAMATSU R7724 PhotoMultiplier Tube (PMT). The output signals of the PMT were processed by a CAEN DT-5751 desktop digitizer via Digital Pulse Processing (DPP). The decay time of the electronic pulses generated by the X-photons and the recoils was different. The charge integration method was used for PSD and identification of the X-photon and neutron induced detection events.

Fig. 16
figure 16

The sketch of the PHRS system tested at 1 Hz repetition rate on the laser-based neutron source

Full size image

A lead shielding with a 6 cm thick cylindrical wall and a 6 cm thick front slab was available to attenuate the intensity of the X-photons that reached the scintillator. During the test at a repetition rate of 1 Hz, the scintillator was behind the lead shielding. It was found that even the 6 cm thick lead (Pb) shielding was not sufficient to fully eliminate all the pulses induced by energetic X- or gamma photons. Thus, the applicability of the self-triggering mode of the DPP-PSD firmware of the digitizer was limited. Therefore, we externally triggered the digitizer with laser pulses and recorded the waveforms. Figure 17 shows typical waveforms with pulses induced by X- and gamma photons and neutron recoiled nuclei. The waveforms were analyzed offline after the experiments. The X-photon and neutron induced detection events could be distinguished with the charge integration method (Fig. 18). The statistics of the low numbers of the observed detection events did not enable neutron spectrum unfolding. At the same time, it can be concluded that the obtained results are valuable for the further development of the method for applications at laser-based neutron sources that operate in higher frequency laser pulse repetition modes and produce neutrons with higher intensities.

Fig. 17
figure 17

Typical waveforms with electronic signal pulses induced by the primary X-photons and neutron recoiled ions (protons, C and O nuclei)

Full size image
Fig. 18
figure 18

The PSD value—signal amplitude distribution of the neutron detection events observed during the 1 Hz mode of operation. A 6 cm thick lead shielding was used around the scintillator of the PHRS detector. Waveforms were recorded and processed offline after the experiment

Full size image

7.3 Calibration of a neutron bubble detector spectrometer

The neutron detection systems discussed so far can measure the neutron spectra continuously, even in the case of a low neutron count. Many applications, for example radiobiological and medical ones, require that the neutron dose should fall into a certain energy range. Although the dose can be calculated from the ToF or the PSD (individual) events over a certain time, it is worth applying a device which is robust, and insensitive to electromagnetic pulses (EMP) and gamma radiation, both of which occur upon the acceleration event. A bubble detector spectrometer (BDS) is one of the best candidates for this task.

A BDS made by Bubble Technology Industries Ltd. consists of six bubble detectors (BD) with different neutron energy thresholds (0.01, 0.10, 0.6, 1.0, 2.5 and 10 MeV). A BD utilizes a gel with superheated droplets. The interaction of a neutron with the gel causes vaporization, i.e. the formation of microbubbles. If there are sufficient neutrons within a volume, the microbubbles form a macroscopic bubble. These bubbles, stable within the gel, are visible. If their number is above typically 100 per BD, then they are linearly proportional to neutron dose equivalents [48,49,50,51,52]. However, if the neutron flux is relatively low, the formation of the macrobubbles become uncertain, and linearity is impaired.

The sensitivities of the BDs (bubbles per ambient dose equivalent) were studied at the (5.3*106 ± 10%) n/s activity Pu-Be neutron source of ATOMKI. The broad energy spectrum of the neutrons emitted by the source covers the En = 0–13 MeV region [53]. The BD, placed into vacuum-tight aluminium containers, were irradiated with the Pu–Be source, the front surface of which was approximately 70 cm far from them. The distance between each BD pair was 5 cm. The irradiations lasted for 15 min, 30 min, 1 h, 2 h, 4 h and 8 h. After each irradiation, the bubbles that formed in the BDs were counted with the BTI-III reader of the BDS. These values are referred to as counted bubbles.

Due to the different angles between the Pu–Be source and the detectors, the neutron dose reaching a single BD was calculated from the known angular and spectral characteristics of the PuBe source. The ambient neutron dose equivalent rate at 70 cm was calculated using the neutron spectrum and the energy dependent fluence rate to ambient dose equivalent rate conversion factors recommended by the United States Nuclear Regulatory Commission [54]. For each BD, the so-called expected number of bubbles was estimated based on their sensitivities provided by the supplier. The results shown in Figs. 19 and 20 indicate the increase of the BDs’ sensitivity when bubble formation is low. Therefore, corrections are always necessary when the bubble count is below 100.

Fig. 19
figure 19

The results obtained at the Pu-Be source for the BDs with different En;th neutron energy thresholds (En;th = 0.1 MeV, 0.6 MeV, 1 MeV, 2.5 MeV, 10 MeV). The counts were normalized to SSD = 70 cm. The bubble count in the BDs after the irradiations (a). The expected bubble count after the irradiations, the error for each point is one bubble (b)

Full size image
Fig. 20
figure 20

The ratio of the counted and expected bubble numbers as a function of the number of the bubbles detected. The Enth detection thresholds of the BDs are a 0.6 MeV, b 1 MeV, c 2.5 MeV, and d 10 MeV

Full size image

7.4 Innovative solutions in laser-driven neutron generation for radiobiological investigations

High-repetition-rate laser driven neutrons provide opportunities to conduct radiobiological investigations using a neutron beam with unique characteristics such as short bunch duration and high dose rate. Among the few possibilities [55], we chose two biological systems as the subject of our investigations, i.e. zebrafish embryos (ZFEs) aged 6 and 24 h post-fertilization (hpf) [56, 57] and the U251 glioblastoma-derived human cell line. Due to the relatively low number of neutrons per shot and their angular distribution discussed in Sect. 6, these biological samples can be irradiated with a sufficiently high dose if they are placed in the immediate vicinity of the neutron converter in vacuum. This requirement poses technical and dosimetry challenges which we discuss here together with the preliminary solutions.

Our biological samples require the provision of standard environmental conditions, like atmospheric pressure, sufficient oxygen, humidity, and temperature. These conditions need to be maintained for at least four hours, as we need that much time to open and close the experimental chamber (Sect. 3), and to fire enough laser shots to achieve a sufficiently high neutron dose. A typical set of ZFEs consists of about 400 individuals, so an aluminium tube with an inner diameter of 35 mm and a height of 15 cm would provide them with enough oxygen. The tube is sealed with a vacuum-tight screw cap (Fig. 21a).

Fig. 21
figure 21

A vacuum-tight aluminium sample holder (a). The insert for cell culture consists of two polystyrene adhesion slides inside a 50 ml sterile tube, where the cross marks the position of the colony (b). The insert for zebrafish embryos is a 3D-printed tube rack, accepting 4 × 1.5 ml Eppendorf tubes, each containing 100 embryos (c)

Full size image

For human cell cultures, we selected a sterile 50 mL polypropylene tube, which vertically accepts two polystyrene adhesion slides joined back to back. We seeded 2 × 104 cells on the slides in a patch of 1 cm in diameter and facing the direction of the beam. The tubes were filled with 25 ml CO2 independent culture media to cover the cells (Fig. 21b).

For ZFEs, a rack accepting 4 × 1.5 ml Eppendorf tubes was 3D-printed for precise sample positioning. Assuming the aluminium tube was placed at a distance of merely 1.2 mm from the neutron converter (catcher), the Monte-Carlo simulations proved that in this arrangement the ratio of the highest and lowest doses is close to 1:7. Hence, with one irradiation campaign four different doses can reach the zebrafish embryos. Each 1.5 mL Eppendorf tube can optimally store 100 ZFEs covered with embryo medium (Fig. 21c). To maintain the temperature of the samples inside the aluminium holder, cylindrical copper blocks were added.

To test the viability of the above-described setups, survival assay was made with 4 × 100 ZFEs (aged 6 or 24 hpf) sealed inside the aluminium holder along with a preheated copper insert (27 ℃). It was then placed in the vacuum chamber for 4 and 24 h. Younger embryos were more susceptible after 24 h of incubation than the older ones, but mortality was still kept under 20% (Fig. 22a. Furthermore, we tested the viability of cells inside the sample holder, placed in vacuum for 1 h (Fig. 22b. The viability of the cells showed no significant difference after 1 h incubation at room temperature (24.5 ℃) versus 37 ℃. Therefore, the temperature (< 30 ℃) maintained by the copper insert during an experiment should be sufficient.

Fig. 22
figure 22

The survival rates of ZFEs aged 6 and 24 hpf inside the aluminium holder remained above 80%. The plot shows the mean values and standard error of three independent experiments (a). There was no statistical difference in the viability of human cells in CO2-free medium in a 37 ℃ incubator and inside the aluminium sample holder at room temperature (R.T.). Values are presented as means and standard deviation. Unpaired t-test was used with Welch’s correction (p = 0.0535) (b)

Full size image

We can hence conclude that ZFEs aged 24 hpf are the best candidates for use in the neutron irradiation experiments. Reliable biological endpoints to be investigated will include survival assays, apoptosis assay, labelling of γH2AX foci, histological and gene expression analysis [58,59,60].

8 Conclusion

We have presented the first neutron generation results of the National Laser-Initiated Transmutation Laboratory of the University of Szeged. While 1 Hz repetition rate operation was demonstrated, preparation for 10 Hz repetition rate have been discussed. Unlike most of the previous experiments worldwide, here we have used and will be using few-cycle laser pulses, which allow for the acceleration of deuterons to a few MeV range only. However, the use of already existing lasers at 1 kHz repetition rate and an average power of 100 W would offer a unique pulsed fast neutron source of 108 neutrons/second, which is attractive for many applications in various fields. Once the average power of the lasers is further pushed to the tens of kW range, we will be able to reliably generate enough neutrons to drive a subcritical nuclear reactor for the transmutation of minor actinides [61] and even for energy harvest [62].