Picosecond pulsed beams of light and heavy keV ions at the Time-of-Flight Medium energy ion scattering system at Uppsala University

A 16 MHz electrostatic beam chopper is implemented at the Uppsala University ToF-MEIS system to complement the 4 MHz sinusoidal scanning, refining the resolution without drift-tube bunching. The system is benchmarked with H 2 + , He + , and Ne + ions in transmission and backscattering geometries. The estimated true pulse duration for 60 keV He + is 34 ps while the direct beam impinging on the detector resulted in measured pulse widths of 295 ps for He ions and 481 ps for Ne ions. In backscattering geometries, ions impinging on the target yield measured pulse durations of 459 ns for H ions, 550 ps for He ions and 810 ps for Ne ions and lead to energy resolution measurements of 2.4 keV (100 keV H), 0.9 keV (60 keV He), and 2.4 keV (160 keV Ne). Discussions cover straggling effects on achievable energy resolution and how to obtain estimates of the true duration of the ion pulse.


Introduction
Energetic ion beams have been used as a probe in materials science for decades, providing accurate composition depth profiles of the nearsurface region in solids [1].In the 1970 s, driven by the ongoing miniaturization in microelectronics, Medium Energy Ion Scattering (MEIS) was developed, originally as an improved version of the already established Rutherford Backscattering Spectrometry (RBS) [2,3,4].MEIS implements ion beams in the keV energy regime, near the stopping power maximum of the projectile to improve depth resolution [5].To improve the energy and consequently the depth resolution, MEIS, employing a toroidal electrostatic analyzer (TEA) has been implemented.The TEA MEIS determines the energy and the scattering angle through the analyzer and achieves a typical energy resolution of δЕ/ E=0.4 % [6].With the term energy resolution, we mean the standard deviation in energy we determine from the spectrum for particles that backscattered in a single collision from the surface of the sample.MEIS studies with a TEA have yielded unprecedented real-space information on the near-surface atomic structure and crystallography through channeling/blocking studies [7,8,9,10], as well as provided elemental depth profiles of thin films for applications in magnetism [11], in nanotechnology [12,13] and electronic applications [14,15].However, the implementation of electrostatic analyzers inhibits the detection of scattered neutrals, the fraction of which increases as the energy of the beam is decreased, affecting elemental quantification.At the same time, the energy range of the analyzer is often limited and hence more grazing geometries may be used to include a wider mass range-leading to reduced information depth [16].In order to tackle these limitations, although with a lower energy resolution, Time-of-Flight (ToF) MEIS setups had been developed since the 1980 s [17,18,19].However, the implementation of ToF-MEIS for characterizing thin films and interfaces close to the surface employing short pulses, which lead to better energy and thus depth resolution, is still largely unexplored.
Employing pulsed beams in a time-of-flight approach in combination with position detectors allows the study of additional fundamental aspects of ion matter interaction, as it enables the detection of other types of secondary particles as well as the time-structure of the response of the solid.Detection of photons [20], electrons [21], secondary ions [22] as well as simultaneous identification of different recoiling target species in a position sensitive ToF detection system has been demonstrated [23].The capability to detect different charge states in combination with the ToF technique enables fundamental investigations of charge exchange processes e.g.Auger processes and their correlation with energy deposition [24,25,26].Most dynamic processes of interest, however, take place on very short time scales of femto-and picoseconds [27,28,29] which has motivated research accessing this regime by shorter ion pulses [30,31,32].
The ToF-MEIS system at the Ångström Laboratory at Uppsala University, employing a combination of an electrostatic chopper and a drift tube buncher with demonstrated beam pulses with a duration as low as 300 ps has been employed for various fundamental studies [33,20,34,35,36] and applied research projects [37,38].Most of the fundamental activities focus on understanding electronic excitations and energy dissipation mechanisms, in the velocity regime around the Bohr velocity, via the implementation of energy-resolved channeling/blocking patterns in crystals, where short pulsing is advantageous.For applications in composition depth-profiling, short pulses yield increased depth resolution.
In this work, we report a recent upgrade of the ToF-MEIS system in Uppsala, featuring an additional, high-frequency electrostatic chopper.Benchmarking experiments testing the performance of the system in terms of achievable time and energy resolution in different experimental geometries along with a detailed discussion regarding the limitations of the present system are included.

Experimental setup and method
The ToF-MEIS setup at Uppsala University is schematically illustrated in Fig. 1.The beamline is connected to a 350 keV air-insulated accelerator from Danfysik and it features two chopper systems, a drifttube buncher and an ultra-high vacuum experimental chamber.Details regarding the beamline electronics, the drift-tube buncher, the charge deflection unit mounted inside the experimental chamber and the data collection can be found elsewhere [39,40].Ion beams for a plethora of chemical elements with masses between 1 and 51 atomic mass units and energies between 5 to 330 keV for singly charged ions have already been produced by the ion source and have already been delivered to the MEIS chamber, located at a distance of eight meter from the sample [39].The produced ion beams are pulsed by a combination of an electrostatic chopper with a horizontal 4 MHz sinusoidal scanning with a vertical 1/n * 4 MHz electrostatic gating (n = 4,8,16,32), resulting in typical pulse lengths of 1-2 ns [39,40].By employing the sets of slits before and after the choppers, and considering the high intensity of the ion beams originating from the ion source, only a narrow parallel fraction of the beam can be selected.A further compression of these pulses in time, on the expense of precision in energy, can be achieved by using the drift tube buncher [39].The slits located after the two choppers are used as apertures on which the beam is scanned while the slits located at the entrance of the low-frequency chopper and the aperture located at the entrance of the high-frequency chopper are used to define the shape and the position of the beam.The buncher is optimized for 100 keV H with the best achieved time resolution of 300 ps [38].Nevertheless, the successful implementation of the buncher is confined to a very limited number of ion-energy combinations due to the necessity of satisfying the criterion T(n + 1/2), where T represents the period of the buncher voltage and n the 0 or a positive integer [39].The most recent upgrade of the system and another way to further compress the produced pulses, for all ion-energy combinations is by employing an additional, high-frequency electrostatic chopper operating at the frequency of 16 MHz (shown in Fig. 1) located ~2.74 m from the sample.The chopper features two horizontal plates with a length of 82 mm mounted at a distance of 13 mm.Before and after the plates, two circular apertures are placed with diameters of 10 mm and 20 mm, respectively, at a distance of 113 mm.The combination of the two choppers offers a time window sampling rate from ~30 kHz to MHz.Before the highfrequency chopper, a quadrupole triplet magnet is employed for beam focusing.After the high-frequency chopper, a 7 • bend of the beamline facilitates the suppression of neutral particles.A second electrostatic steering unit directs the ions in the bent beamline and through the last set of independent slits located at the entrance of the chamber.The final beam impinging on the sample as delivered by the relevant ion optics of the beamline has an angular divergence of < 0.056 • .The beam spot smaller than 1 mm 2 and the primary currents can be reduced to fA on the pulsed mode enabling, on average, single ion impact per pulse.The start signal of the total time of flight signal of the detected particles is measured from a delta pulse generated at a function generator and synchronized with the gating signal on the lower frequency electrostatic chopper.The pressure of the beamline is kept at ~2 * 10 -6 mbar.
The ultra-high vacuum target chamber features a precise 6-axis goniometer equipped with a tungsten filament for the sample annealing and two position-sensitive microchannel plate detectors (MCP) coupled with delay lines [41,42].The delay line detector with a 120 mm diameter (DLD120 from Roentdek with a temporal resolution of < 0.2 ns [42]) employed in the present study (marked in Fig. 1 as Detector A) can be rotated around the sample and measure at scattering angles between − 30 • and 160 • relative to the direction of the beam at a distance of 0.290 m (solid angle of 130 msr).The scattered or transmitted ions are detected along with the emitted photons, recoils and desorbed species.Electrons [21] and secondary negative ions can also be detected by changing the polarity on the detector A. A second delay line detector Fig. 1.Schematic illustration of the TOF-MEIS experimental setup at Uppsala University modified from [40], featuring an ~8 m long beamline with two electrostatic choppers and a resonant drift tube buncher along with the relevant ion beam optics.The experimental chamber at the end of the beamline features a 6-axis goniometer and two Delay Line Detectors.
(marked in Fig. 1 as Detector E) is mounted at a backscattering angle of 135 • , at a distance of 1050 mm.The increased flight distance from sample to detector E compared with detector A results in enhanced energy resolution, at however reduced count rate [40].
For the benchmarking of the setup, experiments with H 2 + , He + and Ne + ions were employed in both transmission and backscattering geometries with the implementation of two samples: a thin Au layer sputtered on top of a Si (1 0 0) bulk crystal and a 50 nm Si (1 0 0) singlecrystalline, self-supporting membrane (thickness measured in [43]).The 50 nm Si (1 0 0) crystal is implemented to estimate the width of the ion pulse in time.For the backscattering measurements, Au was selected for a thin layer since it is monoisotopic, chemically inert and has low surface energy.The thin Au layer on top of the Si crystal was characterized by RBS to be 23 nm thick.The thickness measurement was performed at the 5 MV 15SDH-2 Pelletron Tandem accelerator at Uppsala University [44] employing a 2 MeV 4 He + beam and measuring the backscattered ions at 170 • (relative to the beam direction) with a passivated implanted planar silicon (PIPS) detector.The sample was transferred into UHV immediately after Au was sputtered on the Si crystal to introduce minimal surface contamination from hydrocarbons.
All the ToF experimental spectra presented in this work feature a photon peak.The photons are created upon ion impact and during the whole trajectory in the sample, due to transitions within the valence and the conduction bands ("interband transitions") of the irradiated sample and due to electronic de-excitations to deeper levels [20].Photons can be emitted on different timescales depending on the de-excitation mechanism.The yield of the photons depends on the type and the energy of the incident ion, the detection angle as well as the atomic number, the band structure and the thickness of the sample [20].Besides the above-mentioned physical processes, the time distribution of the recorded photon signals includes contributions from the true pulse duration, the time resolution including time-jitter in chopper electronics and MCP-timing and the energy spread of the beam.Thus, the full width at half maximum (FWHM) of the photon peak yields an upper limit of the time structure of the ion pulse on the sample, as well as the intrinsic time structure of the signal processing and detection electronics, which is broader than the pulse width itself.

H 2
+ ions are employed in the benchmarking process to compare the time and energy spread of the resulting ion pulses by the implementation of the high-frequency chopper with the time resolution achievable by the drift tube buncher [38].In Fig. 2 (a), the ToF spectrum of detected backscattered protons from a 200 keV H 2 + primary beam impinging on the gold thin film is shown.The detector was positioned at a backscattering angle of 160 • .The spectrum includes the photon peak and the backscattered protons from Au and from Si.The photon peak can be used to derive the time structure of the ion pulse on the sample by fitting a Gaussian function to the peak and determining the FWHM value.In this particular experiment the pulse width, as derived from the photon peak, was 459 ps, which is exceeding the 300 ps, achieved in the past in this setup for the same projectile by implementing the drift-tube buncher [38].For a full comparison, it is important to consider that the buncher is working only for specific projectile velocities and furthermore compresses the pulses in time.Thus, bunching introduces an energy spread of the beam setting a limit for the energy resolution even for the shortest possible pulses [39].The chopper, on the other hand, minimizes the width of the ion pulse in time at a distance of ~3 m from the sample and there is no physical limitation for achieving shorter pulses besides the physical properties of the chopper.In Fig. 2 (b), the energy spectrum, as converted from the experimental ToF spectrum, is presented along with a Gaussian error function fitted to the high-energy edge.The energy resolution, derived by fitting an errror function to the measured data is 2.4 keV.This indicates a relative energy resolution of δE/E=0.024.
The evaluation of the time and energy resolutions for He + and Ne + ions included three measurements: In the first one, the beam directly hits the detector and we can evaluate its temporal dispersion.In the second one, the ions impinge on the single crystalline Si membrane in a transmission geometry and in the third one the ions impinge on the Au layer on Si in a backscattering geometry with the detector set at 160 • .In backscattering geometry, the chamber slits are more opened than in transmission to allow higher beam current.
In Fig. 3 (a) we show that the temporal upper limit of the pulse width for the 60 keV He beam impinging directly on the detector is 295 ps.For the ToF measurements of the 60 keV He beam transmitted through the 50 nm Si membrane, the ion pulse width becomes wider since we open the entrance slits to allow more beam to reach the sample as compared with the direct beam measurements and the resulting spectrum is shown in Fig. 3 (b).The pulse width derived from the photon peak is determined to be 440 ps.In Fig. 3 (c) and (d), the ToF and the energy spectrum are shown, respectively, for a 60 keV He beam impinging on the 23 nm Au layer on the top of Si.The pulse width derived from the photon peak is again changed to 550 ps due to the further opening of the slits.The energy resolution deduced from the energy spectrum is 0.9 keV which yields a relative energy resolution of δE/E=0.015.
In Fig. 4 (a) a 160 keV Ne beam directly impinges on the detector and yields an upper limit of the pulse width of 481 ps.In Fig. 4 (b) the ToF spectrum of a 160 keV Ne beam transmitted through a 50 nm single crystalline Si membrane is presented along with the ion pulse as extracted from the photon peak yielding 680 ps.The Ne ions

Discussion
The above-mentioned values of the energy resolution for the three different ions are derived from the energy converted ToF-spectra.The contribution of multiple scattering smoothens the high energy edge for heavier projectiles, as shown in Fig. 4 (d).The high-energy edge of the spectrum includes contributions from the intrinsic energy distribution of the beam particles, the energy distribution acquired in the large-angle collision at the backscattering depth as well as a contribution to the perceived energy uncertainty stemming from the duration of the beam pulse impinging on the sample.The energy distribution from the large-angle collision includes the geometrical straggling (kinematic broadening) and the electronic energy loss straggling from the close collision.Moreover, for the present measurements it will include the contribution of the energy loss straggling in surface adsorbents since no in-situ cleaning of the sample surface could be conducted.This contribution is expected to be most significant assuming as contaminants on a level of 5-10 Å will introduce an electronic energy loss straggling of a few hundred eV, with larger contributions for heavier ions [46].A lower limit for the straggling value for 160 keV Ne + ion impinging on clean Au can be estimated from the CasP program [47,48] as 217 eV.
For H ions, we can compare the energy resolution as extracted from the backscattering spectrum (2.4 keV) with the expected energy resolution which we can analytically calculate based on the photon peak in the ToF backscattering spectrum, which is 1.4 keV.Besides the surface contamination layer, which is the most significant contribution to the broadening of the measured energy resolution, the energy-loss straggling commonly referred to as Coulomb broadening σ C [49] for H 2 ions affects the energy resolution, even though at 100 keV/nucleon the Coulomb broadening is weaker than for higher energies e.g. 150 keV/ nucleon [50].Contributions from geometrical straggling were calculated to be of minor relevance.
For He + ions, the calculated energy resolution based on the pulse width extracted by the photon peak is 0.4 keV, which yields a relative resolution of δE/E=0.0067.Again, the difference between the analytically calculated value and the experimental one (0.9 keV) can be attributed mainly to the contribution of the surface contamination layer to the electronic energy loss straggling.
For Ne + ions, as a more heavy, complex projectile, the photon peak features a detectable tail towards higher time values, expected to be Fig. 3. ToF-spectra of (a) a direct He 60 keV beam impinging on the detector.The Gaussian fit shows the upper limit of the pulse width of the direct beam.(b) a 60 keV He beam impinging on a single crystalline Si membrane in a transmission geometry.The Gaussian fit on the photon peak, magnified in the inset, displays the pulse width deduced from the photon peak of the system.(c) of a 60 keV He beam impinging on a 23 nm Au layer sputtered on [0 0 1] Si crystal (d) energy spectrum corresponding to (c).The pulse width is extracted by a Gaussian fit to the photon peak of the ToF spectrum while the energy resolution is extracted by fitting a Gaussian error function to the high energy edge of the energy spectrum.caused by complex excitation processes which can produce long lived states e.g.defect-induced photoluminescence from Si [51] or projectile excitation.The calculated energy resolution based on the pulse width extracted by the photon peak is 1.1 keV, which yields a relative resolution of δE/E=0.0069.The comparison of achievable resolution for He and Ne suggests that, with the present ToF-MEIS system, implementing a Ne beam proves being more beneficial for near-surface applications, since it facilitates superior mass and depth resolution compared to using a He beam.However, the effect of multiple scattering is affecting mainly the spectra recorded for the Ne beam since the scattering cross section is much larger at approximately similar energies but larger Z.
Comparing the deduced energy resolution for both He and Ne ions results in a relative resolution δE/E of 0.007 which is, despite the upgrades, not competitive with the best relative energy resolution obtained with the MEIS setup with TEA reported as low as 0.001 [52].However, a large number of MEIS setups operating with a TEA implement a grazing incidence geometry to include a wide mass range in the limited energy window of the TEA.The grazing incidence geometry however results in an increased ion path length during the incident and the exit path, thereby increasing the geometrical and the electronic energy loss straggling resulting to a limited depth resolution and accessible depth [16].Furthermore, for the same arguments of surface adsorbents deteriorating the perceived energy resolution, the sub-surface resolution is quickly governed by energy loss straggling.In such a scenario, i.e. an interest of subsurface profiling, the present approach will present advantages due to charge-integrated detection and thus large flexibility in probing beam.
In the following, we attempt to estimate the true duration of the ion pulse in the time domain for the 60 keV He beam in our system.Such approximation requires deconvoluting the direct beam signal and the contributions from the time-jitter in chopper electronics and the MCP timing and flight time spread due to the intrinsic energy spread of the beam particles.Deconvolution of the time difference among particles landing on opposite edges of the channel of the MCP plate is required, and for the case of 60 keV He, this term is ~104 ps, estimated from the pore size diameter (25 μm) and angle of the MCP (8 • ).This time difference represents the minimum possible detection jitter, which is only reached if all ions land in the same MCP channel.The time-jitter in chopper electronics is estimated ~210 ps, by measuring the individual and independent jitter from the start signal and the detector signal (MCP+preamplifier) together with the corresponding constant fraction discriminators to be each around 150 ps.This value (~210 ps) is in agreement with the value given in the manual of the Roentdek for the temporal resolution of the MCP and delay lines (<0.2 ns).To estimate the energy spread of the beam we examine the limitations originating from the ion source itself, since the analyzing magnet, in the absence of narrow entrance and exit slits, does not present any limitations.The intrinsic energy spread from the source and the energy spread introduced by the ripple of the extraction voltage at the platform are estimated to be on the order of 6 eV (corresponding to an energy stability of 1⋅10 -4 ) for a 60 keV He beam.The timing spread resulting from these values and the distance between chopper and target results in 176 ps, assuming instantaneous chopping.Although several of the aforementioned contributions may not follow a Gaussian distribution, the deconvolution process was simplified by assuming all contributions to be Gaussian.By deconvoluting these three contributions, namely the 104 ps from the difference in the MCP position, the ~210 ps from the time-jitter and the 176 ps from the energy spread, from the time resolution of the 60 keV direct beam (295 ps from Fig. 3 (a)) we estimate the maximum true pulse duration for our system to be ~34 ps, i.e. assuming a Gaussian time profile σ = 17 ps.Note, that 60 keV He ions travel only a distance of 58 μm during this time period, which indicates that even subtle flight path differences between ions created in an even shorter pulse can sum up to the observed width.The estimated value based on the deconvolution of 34 ps is very well aligned with the SIMION [53] predictions which pointed to a minimum pulse duration as low as 30 ps.True pulse durations for other ions in the present study are expected to be ~200 ps, resulting in σ ~100 ps.

Summary
We described and benchmarked the high frequency chopper that was added to the ToF-MEIS beamline at Uppsala University.The benchmarking of the system in terms of achievable time and energy resolution, was performed with H 2 + , He + and Ne + ions impinging directly on the detector, on a self-supporting single crystalline Si membrane in transmission geometry and on a Au layer sputtered on top of a bulk Si crystal in a backscattering geometry.The direct beams yielded measured pulse durations (FWHM) of 295 and 481 ps for He + and Ne + ions, respectively.These results confirm the capability of the new system to provide comparable pulse length for a wide range of ion species and energies, overcoming earlier limitations of the drift tube buncher.The true duration (FWHM) of the ion pulse at the sample position, for 60 keV He ion, was estimated to be less than 34 ps, after deconvoluting the contributions from the electronics and the energy spread introduced by the ion source and platform.We also briefly discussed the contribution of straggling to the broadening of the high energy edge in the energy spectra and concluded that electronic energy loss straggling is the most significant contribution to the observed broadening of the energy edge.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fitting a
Gaussian distribution to the photon peak provides information on the pulse width while the high energy edge of the energy spectrum is fitted with a Gaussian error function to extract the energy resolution.backscattered from Si appear in a double peak distribution because of different trajectories inside the Si crystal; the faster peak corresponds to channeled trajectories while the slowest to random trajectories.In Fig.4(c) and (d), the ToF and the energy spectrum are shown, respectively, for a 160 keV Ne beam impinging on the 23 nm Au layer on top of Si.The width of the ion pulse calculated from the photon peak is further changed to 810 ps due to the further opening of the slits.The Gaussian fitting to the photon peak is complicated by to the tail on the right side of the peak, the origin of which is discussed at a later instance.In Fig.4 (d)the contribution of the multiple scattering at the high energy edge of the energy spectrum is prominent.Since we are interested in the slope of the obtained high energy edge due to single scattering only, we perform a linear fit to the plural scattering contribution[45] (pink dotted line at Fig.4 (d)) which we subsequently subtract.The resulting high energy edge is plotted at Fig.4 (d) with a green dashed line and is fitted with a Gaussian error function.This process results in an energy resolution of 2.4 keV corresponding to δЕ/Е = 0.015.

Fig. 4 .
Fig. 4. ToF-spectra of (a) direct Ne 160 keV beam impinging on the detector.The Gaussian fit shows the upper limit of the pulse width of the direct beam.(b) a 160 keV Ne beam impinging on a single crystalline Si membrane in a transmission geometry.The Gaussian fit on the photon peak, magnified in the inset, displays the pulse width deduced from the photon peak of the system.(c) of a 160 keV Ne beam impinging on a 23 nm Au layer sputtered on [0 0 1] Si crystal and (d) energy spectrum corresponding to (c).The pulse width is extracted by a Gaussian fit to the photon peak of the ToF spectrum while the energy resolution is extracted by fitting a Gaussian error function fit to the high energy edge of the energy spectrum.