Introduction
When a large cluster ion such as C60 is injected into a solid material, a coincident impingement of multiple atoms composing the cluster induces unique phenomena such as crater formation (Tomaschko et al., Reference Tomaschko, Schurr, Berger, Saemann-Ischenko, Voit, Brunelle, Della-Negra and LeBeyec1995; Yamada et al., Reference Yamada, Matsuo, Toyoda and Kirkpatrick2001), an enhancement of sputtering yield (Hasselkamp and Scharmann, Reference Hasselkamp and Scharmann1983; Brunelle et al., Reference Brunelle, Della-Negra, Depauw, Jacquet, Le Beyec, Pautrat, Baudin and Andersen2001), and an increase of stopping power (Ray et al., Reference Ray, Kirsch, Mikkelsen, Poizat and Remillieux1992; Baudin et al., Reference Baudin, Brunelle, Chabot, Della-Negra, Depauw, Gardès, Håkansson, Beyec, Billebaud, Fallavier, Remillieux, Poizat and Thomas1994; Narumi et al., Reference Narumi, Nakajima, Kimura, Mannami, Saitoh, Yamamoto, Aoki and Naramoto1998). Although a lot of studies on the irradiation effect of high-energy cluster ions have so far been conducted (Dammak et al., Reference Dammak, Dunlop, Lesueur, Brunelle, Della-Negra and Le Beyec1995; Brunelle et al., Reference Brunelle, Della-Negra, Depauw, Jacquet, Le Beyec and Pautrat1999; Tomita et al., Reference Tomita, Murakami, Sakamoto, Ishii, Sasa, Kaneko and Kudo2010), beam energies per nucleon were typically limited to less than 100 keV/u because of the upper limit of the acceleration voltage of electrostatic accelerators. To achieve higher cluster beam energy, Takayama et al. have recently proposed the concept of the induction microtron, which is designed to accelerate large cluster ions from 1 keV/u to more than 1 MeV/u without limitation on mass to charge ratios (Takayama et al., Reference Takayama, Arakida, Iwashita, Shimosaki, Dixit and Torikai2007, Reference Takayama, Adachi, Wake and Okamura2015; Dixit et al., Reference Dixit, Iwashita and Takayama2009). Studies on cluster–material interaction in a more broad energy range are expected to be realized by the induction microtron in the near future.
The induction microtron requires injection of a pulsed cluster ion beam with a pulse duration of typically a few microsecond. Moreover, the input beam current needs to be high enough to obtain a reasonable amount of output cluster ions because charge exchange reactions between cluster ions and residual gas molecules probably cause considerable beam loss in the accelerator. Laser-ablation cluster sources have frequently been used to produce clusters of metals, semiconductors or alloy materials and to investigate physical properties (Brucat et al., Reference Brucat, Zheng, Pettiette, Yang and Smalley1986; Ganteför et al., Reference Ganteför, Gausa, Meiwes-Broer and Lutz1988; Bucher et al., Reference Bucher, Douglass and Bloomfield1991) and chemical reactions (Morse et al., Reference Morse, Geusic, Heath and Smalley1985; Whetten et al., Reference Whetten, Cox, Trevor and Kaldor1985; Alford et al., Reference Alford, Weiss, Laaksonen and Smalley1986) of clusters. A combination of laser ablation and supersonic expansion of high-pressure buffer gas is preferable to produce a short-pulsed intense cluster beam with high directivity. Smalley and coworkers developed a laser-ablation cluster source for the first time to produce an intense beam of small aluminum clusters composed of less than 20 atoms (Dietz et al., Reference Dietz, Duncan, Powers and Smalley1981). Rohlfing et al. introduced a waiting room to their cluster source, which drastically promoted the growth of clusters, and showed that large carbon clusters composed of more than hundreds of atoms were produced (Rohlfing et al., Reference Rohlfing, Cox and Kaldor1984). Milani et al. succeeded in enhancing cluster production by employing a laser-incident channel having a large diameter, which allows them to irradiate the target with a large laser spot and enhance the vapor production (Milani and deHeer, Reference Milani and deHeer1990). Most of these studies have been dedicated to examine the mass spectrum and the yield of generated clusters. On the other hand, the flux waveforms of cluster beams available from a laser-ablation cluster source have not so far been studied sufficiently.
The purpose of this study is to explore the intensity of cluster beams supplied by a laser-ablation cluster source and provide strategies for improving cluster beam flux. The generation and growth of clusters depends on the clustering process, which is closely related to the history of the density and temperature of the laser-ablated vapor plume. The transportation process, in which the ablation vapor containing cluster particles is ejected together with a supersonic helium gas flow through a nozzle, also affects finally available cluster beam fluxes. We investigate the beam fluxes of aluminum atoms and clusters supplied from a laser-ablation cluster source under various operational conditions (backing helium pressure, laser fluence, and nozzle shape), and discuss how these parameters affect the cluster generation and transportation processes.
Experimental setup
Figure 1 shows the geometry of the laser-ablation cluster source developed in this study. A frequency-doubled Nd:YAG laser (λ = 532 nm, 5 ns FWHM, 5 Hz) irradiated an aluminum disk target (ø25 mm) mounted on a motorized rotational stage. The laser spot on the target surface has an elliptical shape (0.1 × 0.05π mm2). The ablation laser fluence (I A) was changed in the range of 5–200 J/cm2 by controlling the Q-switch delay time. The ablation rate of aluminum atoms, which was measured from the mass reduction of the Al target after 105 laser shots, linearly increased with increasing laser fluence as shown in Figure 2. The rotation speed of the target was set to ~20 rpm to avoid the overlap of successive laser shots, leading to good reproducibility of laser ablation. Pulsed helium gas was supplied through a fast opening solenoid valve (open time ~1.5 ms) with a backing pressure (P He) ranging from 0.5 to 3 MPa. To perform laser ablation after helium gas completely filled a waiting room (ø3 mm, L = 15 mm), the time interval between the onset of valve opening and the laser irradiation was fixed to 800 µs. Metal vapor produced by laser irradiation was carried into vacuum through a conical nozzle by a supersonic helium gas flow. A conical nozzle with a throat diameter of 1.5 mm and an open angle of 6° was normally used in this experiment. In addition to this nozzle, a conical nozzle with a different open angle (3°) and a straight nozzle with a throat diameter of 2 mm were also used to examine the nozzle shape effect on the cluster beam flux. All these nozzles have a length of 3.5 cm.
Fig. 1. The cross-sectional view of the laser-ablation-type cluster source developed in this study.
Fig. 2. Dependence of the ablation rate (the number of atoms evaporated per a laser shot) on the laser fluence.
The whole experimental setup is shown in Figure 3. A voltage of 500 V was applied to deflection plates (A) at the outlet of the cluster source to eliminate charged particles and observe only neutral particles downstream. Two skimmers (ø1.5 and ø3) and an aperture (ø3 mm) were coaxially aligned on the beam axis to collimate the cluster beam. A two-stage differential pumping keeps the pressure in the analysis region at ~5 × 10−5 Pa. The distance between the outlet of the cluster source and the skimmer was 7.4 cm. A time of flight mass spectrometer (TOFMS) was located 60 cm downstream from the nozzle exit. An ArF excimer laser (λ = 193 nm, 30 ns FWHM, 5–200 µJ) was used to irradiate a 1 mm × 1 mm area in the acceleration gap of the TOFMS to ionize the clusters. The photon energy of the ArF excimer laser (6.4 eV) is high enough to ionize aluminum atoms and clusters (AlN, N ⩾ 14) by a single photon process (Schriver et al., Reference Schriver, Persson, Honea and Whetten1990). The interval between the ablation laser irradiation and the ionization laser irradiation (τD) was precisely controlled by a pulse delay generator. Ionized clusters were orthogonally accelerated by two-stage acceleration gaps (20 kV in total) and finally focused onto a micro channel plate (MCP) by an einzel lens. To observe relatively small signals of the clusters, a voltage of 500 V was applied to deflection plates (B) behind the acceleration gap to eliminate monoatomic ions and prevent them from disturbing the sensitive measurement of the cluster signals.
Fig. 3. A schematic of the experimental setup of the TOFMS and the particle velocity measurement.
Figure 4 shows the survival rates of the cluster ions having initial velocities of 750–2500 m/s in the direction of the cluster beam axis. These survival rates were evaluated by precise particle trajectory calculations in the TOF system. As shown in the figure, heavy cluster ions with relatively high-initial velocities are partly lost during the flight toward the detector because the deflection of these particles is insufficient owing to their large inertia. In contrast, small cluster ions with relatively small initial velocities are also lost because of excessive deflection. One can see that cluster ions with sizes of 50–100 and initial velocities of 1000–2000 m/s can be detected with high efficiency (~100%). Of course, the survival rate varies depending on the deflection voltage, but we fixed it to be 500 V because we focus on the clusters with sizes of ~50–100. The MCP signal was averaged over 64–512 shots to reduce statistical errors and increase signal to noise ratios.
Fig. 4. The survival rate of Al cluster ions with various initial velocities. The voltage applied to the deflection plate (B) is fixed at 500 V.
The velocities of the cluster particles were also measured using an ion detector located on the beam axis 75.5 cm downstream from the cluster source outlet. The detector consisted of ion collection electrodes and a Channeltron. In this measurement, the ArF excimer laser irradiated the center of the acceleration gap of the TOFMS system to ionize neutral particles, but no acceleration voltage was applied to the gap. The ionized particles passed through the gap and reached the Channeltron after being slightly accelerated and focused by the ion collection electrodes and the Channeltron biased to −100 V in total. We evaluated the velocities of the particles from their time of flights (TOFs) from the ionization point to the Channeltron.
Result and discussion
Figure 5 shows a typical mass spectrum of aluminum clusters observed with an ablation laser fluence of 130 J/cm2. The ionization laser fluence was small (0.6 mJ/cm2) enough to prevent the distortion of the mass spectrum due to the photo-dissociation of large clusters. Signals of AlN clusters with sizes up to 200 were observed. The mass spectrum had gentle peaks around N = 40, 55, and 65. Small clusters (N < 15) were not observed because they were largely attenuated by the deflection (see Fig. 4). The inserted figure shows a detailed spectrum of the clusters with sizes of 40–60. Although small peaks of aluminum cluster oxide (AlNO) can be seen between the peaks of AlN clusters, their intensities were much smaller than those of the AlN clusters, which indicates that the impurities in the supplied helium gas were negligibly small and not harmful in the cluster growth process.
Fig. 5. A typical mass spectrum of AlN clusters obtained by TOFMS. A detailed spectrum in a size range from 40 to 60 is also shown in the inserted figure. The ablation laser fluence (I A) was 130 J/cm2 and the ionization laser fluence (I I) was 0.6 mJ/cm2, PHe = 2.0 MPa. τD = 460 µs.
TOF signals observed under various delays of the ionization laser (τD) are compared in Figure 6(a). The shape of the TOF signal was almost independent of τD. The flux waveform of specific size clusters was reconstructed from the TOFMS signals by examining the temporal change of the intensity of the corresponding mass peak. The reconstructed flux waveforms of aluminum atoms and clusters (N = 25, 50, 75, and 100) are shown in Figure 6(b). The yield on the vertical axis was converted from the peak intensity assuming that the MCP gain was 5 × 107 irrespective of the cluster ion mass. Note that the yield of the Al atoms is artificially attenuated by 10−2 in this figure. The arrival time of the Al atoms ranged from 300 µs to 1000 µs and the flux peaked at τD ~500 µs. The fluxes of the AlN clusters also peaked at the same time, but had narrower pulse durations (~100 µs FWHM) than that of the Al atoms (~200 µs FWHM).
Fig. 6. (a) TOF signals observed with τD = 400–640 µs. I A = 130 J/cm2, I I = 0.6 mJ/cm2, and P He = 2.0 MPa. (b) Reconstructed flux waveforms of Al atoms and clusters (N = 25, 50, 75, and 100).
The horizontal axis in Figure 6(b) represents the elapsed time from the time (τD = 0) at which the target was irradiated with the ablation laser. From the figure, one can see that it takes about 500 µs for the particles constituting the peak of the flux waveform to reach the irradiation point of the ionizing laser (see Fig. 3) after they were generated by laser ablation. However, this result does not necessarily mean that these particles have the same velocity as that obtained by dividing the flight distance (~650 mm) by the arrival time (500 µs). Since the ablation vapor stagnates in the waiting room while mixing with helium gas and then flows through the narrow nozzle throat, the particles expelled from the nozzle at a given moment could generally have different velocities.
To experimentally confirm this, we measured particle velocity using a Channeltron detector located downstream of the TOFMS system. Figure 7 shows the output signal waveforms of the Channeltron obtained while changing the irradiation timing of the ionization laser (τD). The horizontal axis represents the TOF from the ionization location of aluminum atoms and clusters by the ArF excimer laser to the Channeltron (upper axis) and the particle velocity (bottom axis) converted from the TOF by taking into account the acceleration of ions by electric fields induced by the collection electrodes and the Channeltron itself. The diagonal dashed lines in the figure denote the times when the particles were emitted from the nozzle, which were inversely calculated for typical particle velocities. For example, when τD = 400 µs, a peak of the signal appears 40 µs after ionizing laser irradiation, meaning that the average velocity of the particles passing through the ionization point at this moment (τD = 400 µs) was ~2.1 km/s. From the width of the signal waveform and the diagonal dashed line, we found that these particles were emitted from the nozzle of the cluster source between t = 50 µs and 150 µs after the ablation laser irradiation. This result shows that the slower particles (~1.6 km/s) released from the nozzle at earlier time (t ~ 50 µs) and faster particles (~2.5 km/s) released at later time (t ~ 150 µs) reached the ionization point at the same time. Hereafter, we call the time that it takes for the particles such as atoms and clusters to be released from cluster sources “particle release time”. The waveforms for other delay times (τD) have the similar tendencies, but we found that the average velocity of the detected particles gradually decreased with τD. It was also found that the width of the signal, that is the velocity spread of the particles, increased with τD.
Fig. 7. Time-of-flight signals acquired by particle velocity measurement. Signals were averaged over 512 shots. Dashed lines represent the particle release time t at which particles leave the nozzle outlet (t = 0 corresponds to the ablation laser irradiation). I A = 130 J/cm2, I I = 2.6 mJ/cm2, and P He = 2.0 MPa.
Here, again, let us focus on the flux waveforms in Figure 6(b). The average velocity of the particles constituting the peak of the aluminum atomic beam is estimated to be ~1.7 km/s from the waveform of τD = 500 µs in Figure 7, which is slightly higher than the sound velocity of helium (~1.0 km/s at 300 K). This result supports that the aluminum atoms generated by laser ablation were accelerated and transported downstream by the supersonic helium flow formed by the nozzle. On the other hand, the velocity spread of these particles evaluated from the FWHM of the signal waveform in Figure 7 was about ± 0.1 km/s. Since the distance from the nozzle outlet to the ionizing laser irradiation position is 600 mm, the time spread of the flux waveform caused by this velocity spread is estimated to be about 50 µs. As discussed above, considering that it takes about 50–150 µs for the particles to be released from the cluster source, we can roughly explain the observation result that the flux waveform of the aluminum atom had a width of about 200 µs by the velocity spread and the particle release time. Since the particle velocity spread depends mainly on the temperature of the ablated aluminum vapor, it is not easy to reduce it. On the other hand, the particle release time may be significantly shortened if the metal vapor generated by laser ablation can be quickly transported to the nozzle while confining it with high-pressure helium gas and keeping its volume quite small.
From the flux waveforms of the cluster beams in Figure 6(b), it can be seen that their pulse widths (FWHM) were much smaller than that of aluminum atoms. Assuming that cluster particles of different sizes are in equilibrium with each other and have the same temperature, the velocity spread of the cluster particles should decrease with its size (mass). The experimental result that the pulse width of the cluster beam was independent of the size indicates that what mainly determines the pulse width of the cluster beam flux is not the velocity spread of cluster particles, but the particle release time. Cluster formation does not necessarily occur in the entire volume of the aluminum vapor plume, but is generally limited to a region where thermodynamical conditions suitable for atomic aggregation hold. Therefore, it is reasonable that the particle release time of clusters was much smaller than that of aluminum atoms, resulting in the shorter pulse length of the cluster beams.
Backing pressure dependences of the beam fluxes of Al and Al50 are presented in Figure 8(a) and (b), respectively. The largest fluxes were obtained with a backing pressure of 2 MPa for both Al and Al50. This result indicates that when P He ⩽ 2 MPa, the increase in the backing pressure improved the transport efficiency of the aluminum vapor containing clusters. On the other hand, when P He > 2 MPa, the increase in the residual gas pressure in the chamber probably disturbs and decelerates the supersonic helium gas flow and largely reduces the transport efficiency of the particles. The deceleration of the supersonic jet can be minimized by shortening the distance between the nozzle outlet and the skimmer or intensifying the evacuation speed. As shown in Figure 8(b), the tail of the Al50 flux waveform observed with a backing pressure of 3 MPa is much longer than those with lower pressures. This result implies that the cluster condensation efficiency was enhanced by the strong confinement of the metal vapor in high-pressure helium gas, leading to the increase in the total amount of clusters.
Fig. 8. Backing pressure dependences of flux waveforms of (a) Al and (b) Al50. I A = 130 J/cm2, I I = 0.6 mJ/cm2.
Figure 9 compares the flux waveforms of Al and Al50 observed with two conical nozzles and a straight nozzle. In both cases of Al and Al50, the particle fluxes were largely enhanced by the conical nozzles. As for the conical nozzles, more intense beam fluxes of Al and Al50 were obtained with a larger open angle (6°). This is probably because the Mach number of the jet (the directivity of particles) increased as the ratio of the cross-sectional area of the nozzle outlet (A) to that of the nozzle throat (A*) increased. Assuming that the temperature of the helium gas in the waiting room is 300 K, the flow velocities at the nozzle outlet are calculated to be 1680 m/s and 1725 m/s, respectively for the 3° conical nozzle (A/A* ~ 12) and the 6° conical nozzle (A/A* ~ 35). For the conical nozzles, the flight time from the nozzle outlet to the TOFMS is ~350 µs, and considering that the particle release time was 50–150 µs, the arrival time of Al50 is 400 to 500 µs, which is in good agreement with the experimental result.
Fig. 9. Dependence of Al and Al50 beam flux waveforms on the nozzle shape. Two conical nozzles with different open angles (6° and 3°) and a straight nozzle were used. I A = 130 mJ/cm2, I I = 0.6 mJ/cm2, and P He = 2 MPa.
TOF signals and flux waveforms of Al and Al50 with different ablation laser fluences are compared in Figure 10. The aluminum atom flux increased with the ablation laser fluence as shown in Figure 10(b). This is clearly due to the enhancement of the amount of ablation vapor as shown in Figure 2. The Al50 flux was drastically enhanced when the ablation laser fluence increases to 40 J/cm2. However, when I A ⩾ 40 J/cm2, the amount and the size distribution of clusters hardly changed [Fig. 10(a)] and the peak values of the Al50 cluster flux were almost the same as shown in Figure 10(c). This result suggests that the cluster formation and growth was suppressed by the temperature increase of the ablated vapor due to increased laser fluence.
Fig. 10. Dependence of Al cluster generation on the ablation laser fluence I A. I I = 0.6 mJ/cm2. P He = 2 MPa. (a) TOF signals. τD = 500 µs. (b) Flux waveforms of aluminum atom. (c) Flux waveforms of Al50.
The influence of the ionization laser fluence (I I) on the cluster ion production is shown in Figure 11. The TOF signals of Al cluster ions shown in Figure 11(a) were slightly shifted to a lower side and the gentle peaks on the signals gradually disappears as the ionization laser fluence increases. This result shows that the photo-dissociation of clusters became remarkable when the ionization laser fluence was increased. Small cluster ions (N ⩽ 14) observed under high-ionization laser fluences are probably the product of the photo-dissociation of large-cluster ions. Although the cluster beam flux increased with increasing the ionization laser fluence, the shape of the flux waveforms of Al50 clusters was less sensitive to the fluence as shown in Figure 11(b). To compare the cluster ion production efficiency, total yields of clusters were evaluated by the time integration of the flux waveforms [Fig. 11(c)]. The total yields of aluminum atoms and clusters (N = 50 and 100) almost linearly increased with increasing laser fluence when I I ⩽ 10 mJ/cm2. When the ionization laser fluence exceeds 10 mJ/cm2, the total yields of the aluminum atoms and relatively small clusters (N = 14) were largely enhanced probably due to the increase in the amount of photo-fragments. On the other hand, the total yield of large clusters (N = 50 and 100) saturated with laser fluences over 10 mJ/cm2. The photoionization cross-section of aluminum atoms is ~50 Mb (0.5 Å2) at a wavelength of ~193 nm (Kohl and Parkinson, Reference Kohl and Parkinson1973) and large aluminum clusters have larger photoionization cross-sections than this value. Thus, the observed saturation in the yield of the Al clusters may be due to that almost all the aluminum clusters were ionized under laser fluences over 10 mJ/cm2 (~1 photon/Å2).
Fig. 11. Dependence of cluster ion production on the ionization laser fluence (I I). I A = 130 J/cm2. P He = 2.0 MPa. (a) TOF signals. These signals are normalized by the signal intensities. τD = 480 µs. (b) Flux waveforms of Al50. (c) Total yields of each particle (aluminum atom and Al14,50,100) obtained by the time integration of the flux waveform.
In our cluster source the laser spot size was limited by the narrow-laser incident channel. Thus, increasing the ablation laser energy resulted in excess vapor heating. For increasing the cluster flux, it is probably effective to increase the spot size while maintaining the fluence. The pulse width of the cluster beam could be shortened by improving the cluster source structure so that the generated vapor can be transported to the nozzle in the waiting room while keeping the vapor volume small.
Conclusion
To design high-flux cluster beam sources required for high-energy circular cluster accelerators, we conducted an experimental survey on how the operational parameters of the laser-ablation cluster source affect the cluster size distribution and flux waveforms. We observed aluminum clusters with sizes up to 200 and found that the pulse width of the cluster beam was typically about 100 µs. Increasing the backing pressure enhanced the cluster generation and improved the transportation efficiency of the clusters. The nozzle shape also affected the cluster flux because the directivity of the cluster beam was determined by the acceleration in the nozzle. The largest cluster flux was obtained by the conical nozzle with an open angle of 6°. When the ablation laser intensity increased beyond 40 J/cm2, the cluster flux saturated, which may be caused by an increase in the vapor temperature. Increasing the laser spot size while keeping the fluence is probably effective to enhance cluster production. The pulse width of the cluster beam was found to be about 100 µs, which is mainly determined by the particle release time from the source. To enhance the cluster beam flux, the correlation between the cluster source structure and the pulse width of the cluster beam should be investigated in more detail.
Acknowledgement
This work was partly supported by JSPS KAKENHI Grant Number JP16H03904.
