We study the competing mechanisms involved in the Coulomb explosion of 2-propanol dication, formed by an ultrafast extreme ultraviolet pulse. Over 20 product channels are identified and characterized using 3D coincidence imaging of the ionic fragments. The momentum correlations in the three-body fragmentation channels provide evidence for a dominant sequential mechanism, starting with the cleavage of a C–C bond, ejecting and CH3CHOH+ cations, followed by a secondary fragmentation of the hydroxyethyl cation that can be delayed for up to a microsecond after ionization. The C–O bond dissociation channels are less frequent, involving proton transfer and double proton transfer, forming H2O+ and H3O+ products, respectively, and exhibiting mixed sequential and concerted character. These results can be explained by the high potential barrier for the C–O bond dissociation seen in our ab initio quantum chemical calculations. We also observe coincident COH+ + C2Hn+ ions, suggesting exotic structural rearrangements, starting from the Frank–Condon geometry of the neutral 2-propanol system. Remarkably, the relative yield of the product is suppressed compared with methanol and alkene dications. Ab initio potentials and ground state molecular dynamics simulations show that a rapid and direct C–C bond cleavage dominates the Coulomb explosion process, leaving no time for H2 roaming, which is a necessary precursor to the formation.
Multiple ionization of isolated organic molecules has attracted significant experimental and theoretical attention.1–10 The ultrafast dynamics that are triggered by the removal of several electrons exhibit charge separation and Coulomb explosion (CE),11,12 as well as structural rearrangement and formation of exotic products such as H3+ ions that were observed in double ionization by fast electron impact13,14 or highly charged ions,15,16 strong-field laser ionization,17–20 and single-photon double ionization (SPDI).21–26
Using 3D coincidence fragment imaging with ultrafast pulses of extreme ultraviolet (EUV) radiation, Luzon et al. have shown that SPDI of isolated methanol molecules can be well described by ab initio nonadiabatic molecular dynamics simulations and that they offer a direct and detailed comparison of time-resolved experiments and theory.24 Molecular dynamics simulations indicate that the C–O bond cannot break on the methanol dication ground state due to a charge-transfer barrier, of a few eVs. At the same time, high-lying excited states exhibit an ultrafast nonadiabatic electron transfer that enables the dominant CH3+ + OH+ Coulomb explosion channel.24 Alternatively, proton transfer from the doubly ionized methyl moiety breaks the C–O bond by forming a minor CH2+ + H2O+ channel. On the low-lying states of the methanol dication (0-2), below the charge-transfer barrier, more complex structural rearrangement dynamics of roaming neutral dominate the CE dynamics H2 and forms the H3+ product.24,27 A study of H3+ yields from strong-field laser ionization by Ekanayake et al. demonstrated that despite the increased number of hydrogens and possible combinations leading to H3+ formation, the yield was reported to decrease as the carbon chain length of isolated organic molecules increases.19 Furthermore, it was suggested that ethanol dimers may suppress the tri-hydrogen formation mechanism.14 Finally, Hoshina et al. showed that H3+ formation by strong-field laser ionization does not require the existence of a methyl moiety in the parent organic species.28
In addition to the C–O bond cleavage and formation of Hn+ products investigated in the case of methanol, the ionization of larger alcohol species can exhibit C–C bond dissociation. Indeed, C–C bond dissociation is a dominant dissociative ionization channel of isolated alcohol systems.29 In particular, for 2-propanol, ejection of a neutral methyl radical and formation of hydroxyethyl cation become the dominant dissociative ionization channels ∼0.3 eV above the photoionization threshold,29 while CH3+ cations are reported to appear only for photon above ∼20 eV photon energies.30 Using strong-field ionization, Mishra et al. initiated complex dynamics on the singly ionized 2-propanol and probed it by a time-resolved CE.8 Double ionization can be expected to trigger cleavage of multiple bonds that can proceed via concerted three-body dissociation mechanisms, e.g., as observed for the case of SPDI of methanol22,24 and of N2O,31 both exhibiting ejection of high speed neutral fragments attributed to concerted three-body breakup dynamics. Furthermore, Hoshina et al. reported a dominant hydronium HnO+ yield from strong-field CE of 2-propanol, indicating nontrivial complex structural rearrangement dynamics beyond H3+ formation.28 In the case of ethanol CE, Wang et al. proposed that H3O+ formation can proceed via an intermediate neutral H2O.32
It is, therefore, appealing to explore the competing C–O and C–C bond cleavage and hydronium formation by SPDI. This work presents the 3D coincidence fragment imaging measurements of 2-propanol following SPDI with ultrafast high-order harmonic generation (HHG) pulses. The double ionization results in a rich variety of dissociation channels, including a dominant ejection of CH3+, and structural rearrangements that form H3+, H2O+, and H3O+. Detailed analysis of the three-body breakup of the 2-propanol dication reveals both sequential and concerted C–C and C–O bond dissociation mechanisms. The measured dynamics are discussed given the presented ab initio calculations of the 2-propanol dication potential landscape and molecular dynamics simulations on its ground state. Furthermore, the sequential three-body breakup of 1-hydroxyethyl ion product is reinforced by the direct experimental observation of its delayed dissociation on the microsecond time scale.
The experimental setup has been previously described.22 Briefly, 2-propanol vapor sample (J. T. Baker, 99.5% min) is introduced as a skimmed effusive molecular beam to the center of the 3D coincidence spectrometer where it is intersected by an ultrafast EUV pulse. The EUV pulse is produced by HHG in a semi-infinite neon cell with ∼2 mJ, 35 fs laser pulse centered at 800 nm wavelength.33,34 The resulting EUV pulses, with photon energies of upto 100 eV,22 are spatially filtered from the higher divergence near-IR. EUV photons ionize the molecular beam sample and the ions are accelerated up to ∼2 keV by the extraction field in a perpendicular direction to the molecular and laser beams toward a MCP detector equipped with a P46 phosphor anode, through a ∼0.5 m field free flight tube. The 2D positions of the hitting particles are detected using two CCD cameras, which are, respectively, synchronized to odd and even laser pulses at a 1 kHz repetition rate, while the ion times of flight are detected by a fast scope. The time and position data of hitting particles are correlated by the amplitudes on the scope and CCD signals,35,36 allowing one to obtain the ion charge over mass ratios as well as their 3D momenta. If all the fragments are detected, the kinetic energy release (KER) is evaluated in the center of mass frame of the dissociating molecule. In the case of neutral fragments that are not accelerated by the spectrometer fields and cannot be directly detected, the neutral fragment momentum is evaluated by momentum conservation and the measured center of mass recoil in the frame of reference, moving in the average velocity of the molecular beam. Further detail about the three-body breakup analysis method is provided in the supplementary material.
The singlet-state potentials
Of the 2-propanol dication were computed using extended multi-state XMS-CASPT2 (12e, 8o)/aug-cc-pVDZ37,38 with eight-state averaging. We used the same approach for calculating the neutral ground state with two-state averaging and 14 instead of 12 electrons. All calculations used the density-fitting technique within the Brilliantly Advanced General Electronic-structure Library (BAGEL) electronic structure package39 invoking the single-state single reference (SS-SR) contraction scheme40,41 and the vertical shift parameter was set to 0.2Eh.
For the ab initio molecular dynamics (AIMD)
For reasons we describe in the text, we found that AIMD simulations on the singlet and triplet ground state potentials were sufficient for analyzing the mechanisms leading to the Coulomb explosion in the channel. We could not use the above described CASPT2 approach because of computational wall-time constraints and, therefore, resorted to AIMD in the B3LYP42,43/aug-cc-pVDZ level. All calculations used the Q-Chem5.344 software package. The AIMD trajectories used 0.3 fs time steps and were terminated when the inter-fragment velocities reached the asymptotic monotonic behavior (or at a cutoff time of 1.5 ps). At this stage, the effect of the residual long-range Coulomb repulsion on the final velocities is taken into account using the classical equations of motion applied to the center of masses of the resulting cationic fragments.
RESULTS AND DISCUSSION
Single-photon double ionization (SPDI) of (CH3)2CHOH, 2-propanol, results in a dication that dissociates into many different Coulomb explosion channels. Figure 1 shows the measured cation mass coincidences following SPDI with ultrafast EUV pulses, where the logarithmic gray scale indicates the measured yields of correlated ion pairs. The contribution of triply ionized parent cations was neglected because triple cation coincidences are rare, as is consistent with uncorrelated ionization events by the same laser pulse. For clarity, only cations with masses above carbon are presented in Fig. 1, not showing the Hn+ products. The dashed line indicates two-body breakup mass correlations where the sum of the detected cation masses equals 60 amu, the mass of the parent 2-propanol dication. The only significant yield of two-body Coulomb explosion (CE) is of the CH3+ + C2H4OH+, which is observed only in ∼3% of the measured SPDI events. The recoil of the two dissociating fragments toward and away from the 3D coincidence fragment imaging detector results in anti-correlated shifts of the fragment masses. The carefully chosen spectrometer fields allow the unequivocal determination of the detected ions' masses in each dissociation event. Thus, we were able to analyze the three-body fragmentation mechanisms of the dication that produce undetected neutral fragments.
The most frequent product channels are C2H3O+ + CH3+ and C2H3+ + CH3+, observed in ∼12% and ∼10% of all the measured CE events, respectively, where the associated neutral H2 and H2O fragments cannot be directly detected in the mass spectrum and can, therefore, be either intact molecules or two neutral fragments. Nevertheless, the coincident masses still tend to result in anti-correlated mass shifts due to the strong recoil of the coincident ions toward and away from the detector. The cleavage of two C–C bonds can produce CH3+ + COH+, observed in ∼8% of CE events, which can be similarly assigned to four-body fragmentation or to a structural rearrangement by H migration to form methane. Similar to the dominant CH3+ + X+ products of C–C bond cleavage, we observe significantly weaker CH2+ + X+ products that are tentatively attributed to dissociation of the methyl cation. Cleavage of the C–O bond is less abundant and is dominated by proton transfer, forming H2O+ and double proton transfer forming H3O+ ions. Another intriguing example of nontrivial structural rearrangement is the formation of mass 29 amu, COH+, in coincidence with mass 27 or 26 amu, corresponding to C2H3+ or to C2H2+. If this process initiates from the Frank–Condon (FC) geometry of the 2-propanol, it requires the ejection of COH+ after forming a new bond and closing the transient three-carbon ring.
Figure 2 summarizes the relative yields of C–C bond breaking CHn+ + X+ coincidences, C–O bond breaking coincidences that result in HnO+ + X+ products, and COH+ + C2Hn+ coincidences and the minor yields of Hn+ + X+ from the Coulomb explosion of doubly ionized 2-propanol. In agreement with Hoshina,28 the branching ratio toward H3+ formation from 2-propanol is significantly suppressed compared with methanol, which is formed via a roaming H2 mechanism.19,23 Furthermore, in the case of methanol dication, direct dissociation of the C–O bond to form OH+ is an order of magnitude more efficient than the proton transfer assisted dissociation that generates H2O+.22,24 In contrast, the relative yields in Fig. 2 show that dissociation of the C–O bond of the 2-propanol dication is dominated by proton transfer and significant double proton transfer that form water and hydronium ions, respectively. The measured branching ratios emphasize that the Coulomb explosion of the isopropanol dication is dominated by cleavage of the C–C bond and ejection of a CH3+ cation, typically accompanied by additional fragmentation and formation of neutral products.
The momentum correlations of a three-body breakup can provide valuable insight into the mechanism of breaking multiple bonds.31,45 Although our 3D fragment imaging spectrometer cannot detect neutral SPDI products, we can still infer their mass by completing the total detected ion mass to that of the parent dication (60 amu). Furthermore, we can estimate the kinetic energy carried away by the neutral fragment (or fragments) from momentum conservation (neglecting the velocity spread in the parent molecular beam).31,45 Thus, it is possible to determine the three-body momentum correlations between the two detected cationic fragments and the missing neutral mass. Due to velocity map imaging, the center of mass recoil resolution in the 2D detector plane is superior to the recoil along the time-of-flight axis. Therefore, three-body momentum correlations are obtained from the projected momenta in the detector plane using the weighted-Dalitz plot method.46,47 The measured momentum correlations of two representative C–C breaking channels are presented in Fig. 3. Extending the original concepts, considering unimolecular dissociation into three equal masses,48 we developed the generalized mass-scaled Dalitz plot,24 where each point within the unit circle corresponds to a particular energy partitioning between the three fragments with a specific geometry of dissociation. The plot's axes correspond to the kinetic energy fraction of each product (e.g., εCH3+, εH3O+, εC2H2), scaled by the maximal fraction allowed by momentum conservation. The points corresponding to uncorrelated three-body dissociation observing total energy and momentum conservation appear as a uniform distribution within the unit circle. We provide a detailed description of different three-body breakup geometries mapping onto the Dalitz plot in the supplementary material.
The Dalitz plot can provide insight into the three-body breakup dynamics. For example, sequential bond dissociation mechanisms can be distinguished by characteristic momentum correlations. In particular, the signature of a high KER Coulomb explosion that is followed by a secondary low KER dissociation of one of the molecular ions can be revealed by the intact CE product carrying most of its allowed kinetic energy fraction. If the secondary dissociation occurs on a time scale comparable to or larger than the rotations of the dissociating subsystem, the energy partitioning between the secondary products will be uncorrelated. Such a signature, corresponding to Coulomb explosion of a C–C bond that releases a high kinetic energy CH3+ followed by dissociation of the C2H4OH+ ion, clearly dominates the Dalitz plots of most CH3+ + A+ + B channels. Figures 3(a) and 3(b) show two such channels that exhibit a clear signature of a sequential dissociation mechanism. Here, in both figures, the vertical coordinate, defined as η2 = 2ɛCH3+ − 1, represents the kinetic energy of the CH3+product.
By comparing the yields of two-body and three-body breakup events with εCH3+ > 70% (equivalent to , assigned to sequential breakup, we can conclude that over ∼90% of the nascent two-body CH3+ + C2H4OH+ Coulomb explosion is followed by a secondary dissociation of the hydroxyethyl cation. For these events, the KER in the initial CE step is evaluated based on the measured 3D ion recoil velocities, where the mass of the undetected neutral fragment is added to the mass of the secondary ion, neglecting the low recoil in the second dissociation step. The bottom curve in Fig. 4 shows the KER distribution measured for two-body CH3+ + C2H4OH+ CE events, peaking at ∼4.8 eV. The two-body KER distribution is directly compared with the KER in the initial CH3+ + C2H4OH+ breakup for different C2H4OH+ dissociation channels. For most channels, the KER is found to be similar to the distribution recorded for the two-body breakup channel, thus supporting a purely sequential mechanism in which the fate of C2H4OH+ is not correlated with the initial Coulomb explosion step. In contrast, C2H4OH+ breakup into HCOH+ + CH3 is associated with significantly lower KER, peaking at ∼3.8 eV. The lower KER may indicate that the formation of the hydroxymethylene radical cation occurs on higher-lying states of the C2H4OH+ intermediate whose formation may be suppressed for events with a high KER in the initial dissociation step.
In addition to the prominent high εCH3 feature, attributed to the sequential CE of a C–C bond that is followed by dissociation of the C2H4OH+product, the Dalitz plot in Fig. 3(a) shows a minor feature of CH3+ + H3O+ + C2H2 breakup, characterized by a nearly maximal εH3O KER fraction. These events can be attributed to an initial CE of the C–O bond, followed by a secondary ejection of the methyl cation. However, the momentum correlations of the main C–O bond breaking channels do not exhibit distinct sequential signatures. Within the statistical error associated with the lower abundance channels, we identified two additional channels that can be attributed to a sequential mechanism initiated by dissociation of the C–O bond. Figure 5 compares the KER of the initial Coulomb explosion of the three sequential C–O bond breaking channels: direct dissociation of OH+, proton transfer to form a water cation, and with double proton transfer yielding the H3O+ hydronium ion. Interestingly, the three processes exhibit a similar KER in the initial cleavage of the C–O bond that peaks at ∼4.1 eV.
Figures 6(a) and 6(b) compare the Dalitz plots of the minor H2O+ + C2H3+ + CH3 channel, attributed to a sequential mechanism, with the momentum correlations measured for OH+ + C3H3+ + 4H, the predominant C–O breaking channel. The Dalitz plot of the latter, more significant, channel exhibits nontrivial energy partitioning between the ionic and neutral fragments, indicating a concerted breakup mechanism. Furthermore, Fig. 6(c) shows the Dalitz plot measured for the surprising COH+ + C2H3+ + 4H channel. As mentioned above, considering the FC geometry of 2-propanol (shown in the inset of Fig. 7), the ejection of either COH+ or C2H3+ from the parent dication would remove the central carbon atom and disintegrate the remaining C–C and C–O bonds, thus preventing the formation of the second ion product. We, therefore, propose that the mechanism should involve closing of a three membered carbon ring before ejection of COH+. Indeed, the associated Dalitz plot shown in Fig. 6(c) exhibits nontrivial momentum correlations that cannot be attributed to a dominant sequential breakup. It is interesting to note also the feature corresponding to high KER fraction carried away by sum of the four H atoms. As it is unlikely that four neutral hydrogen atoms are ejected in the same direction to cause such significant recoil of the cation center of mass, this statistically significant feature is assigned to a minor contribution from traces of isopropanol dimers that can form in the effusive beam target. The measured recoil momentum of these events can, therefore, be associated with the combined mass of the 4H with another isopropanol molecule and correspond to a 16 times lower kinetic energy of the neutral fragments. In this case, the correlated formation of COH+ and C2H3+ can proceed not by ring-closing but by a reaction between the two isopropanol molecules within the dimer.
Figures 7(a) and 7(b) compare the ab initio potential energy curves of the low-lying dication states, calculated as a function of the C–O or C–C bond extension from the initial FC geometry of neutral 2-propanol. Similar to the C–O bond extension in the methanol dication,22,24 the potential curves of the ground and excited states shown in Fig. 7(a) rise with the extension of the bond distance and prevent direct dissociation of the C–O bond. In contrast, extension of the C–C bond shown in Fig. 7(b) exhibits a barrierless descent on the singlet ground state toward separation into the dominant CH3+ + CH3CHOH+ breakup in agreement with the experimental observation that the predominant initial step of the CE in 2-propanol is the C–C bond cleavage. On the other hand, electronically excited states exhibit barriers also toward dissociation of the C–C bond. Single-photon double ionization using photons of up to ∼100 eV in the HHG pulse is expected to populate all the low-lying states of the dication.9 It is, therefore, proposed that the observed predominance of C–C bond dissociation occurs by efficient nonadiabatic transition to the ground state of initially electronically excited dications. For example, Fig. 7(b) shows evidence for avoided crossing between the ground and first excited state near the FC geometry that may facilitate the proposed nonadiabatic transition.
The proposed mechanism was further tested by simulating 100 AIMD trajectories on the dication ground state, initiated by sampling the neutral ground state geometry distribution at 300 K. Figure 8(a) shows the simulated KER spectrum of the direct C–C bond dissociation on the singlet ground state that was observed in over 90% of trajectories, in rough agreement with the most probable experimental KER indicated by the dashed vertical line. Figure 8(b) shows the simulated KER distribution obtained by initiating the AIMD trajectories on the triplet ground state that exhibits C–C bond dissociation on the 600 fs time scale and significantly lower KER compared with the singlet-state dynamics. Within the simulated time scale, three-body breakup was observed only in two triplet trajectories that exhibit a secondary C–C bond breakup in the hydroxyethyl cation product. Secondary dissociation into HCOH+ + CH3 on the triplet ground state, corresponding to lower KER in the initial CE, is in agreement with the lower initial KER that is experimentally measured for this secondary breakup channel, as shown in Fig. 4. Nevertheless, as previous measurements indicated a significant nontrivial preference of singlet over triplet formation by SPDI of methanol,25 we cannot dismiss the possibility of its formation on excited singlet states. Simulation of the excited states dynamics, which are also attributed to the relatively rare events with initial cleavage of the C–O bond that cannot proceed directly and involve significant structural rearrangement, requires higher level nonadiabatic AIMD calculations. Representative AIMD trajectory simulations can be viewed in the supplementary material movie file.
Both the sequential and the concerted mechanisms described in the above for the cleavage of C–C and C–O bonds occur on ultrafast sub-nanosecond time scales, such that the measured times and positions of particles impinging on the detector can be directly associated with their masses and recoil from the center of mass of the parent molecule. In contrast, longer delayed, ultraslow fragmentation on the μs time scale during the acceleration of ionic products in the photofragment spectrometer can result in directly measurable time shifts that depend on the dissociation time-delay as well as on the fragmentation products. Figure 9 shows a section of the measured cation mass coincidence plot that exhibits the clear signature of such delayed dissociation events, which appear as streaking lines across the mass coincidence plane. The lower part of the symmetrized coincidence map is overlayed with the dashed red curve, generated by simulated delayed dissociation of an instantaneously produced mass 45, CH3CHOH+ photoionization product, into H3O+ + C2H2 (see the supplementary material for representative SIMION simulations). Here, different dissociation times are mapped to different mass correlations. Several times are specifically indicated by the red circles within the observed range corresponding to dissociation time-delays starting from ∼1 µs, before which the neutral fragment arrives with little velocity and is not detected, and up to ∼1.4 µs, after which the parent ion leaves the acceleration region and both fragments travel to the detector with the speed associated with the parent 45 amu cation mass. The atomic mass resolution of the secondary product assignment is demonstrated by the clearly distinguished dashed blue curve, simulated for the possible CH3CHOH+ → H2O+ + C2H3 channel that requires less structural rearrangement. An additional though significantly weaker delayed dissociation feature can be attributed to delayed CH3CHOH+ → COH+ + CH4, whose simulation is shown by the dashed green curve.
SPDI of isolated 2-propanol molecules results primarily in Coulomb explosion of one of the C–C bonds and formation of CH3+ + CH3HCOH+. This is in accord with the calculated ab initio potential landscape, showing barrierless dissociation of the C–C bond on the dication ground state. Excited states of the dication that do exhibit a barrier toward dissociation of the C–C bond are expected to dissociate via nonadiabatic coupling to the ground state. The larger, hydroxyethyl cation is mostly unstable and undergoes a secondary fragmentation resulting in the observed three-body momentum correlations that are characteristic of sequential breakup mechanisms, where except for HCOH+ + CH3+ + CH3 breakup, the KER in the initial step of CH3+ ejection is found to be similar to the KER of the two-body CH3+ + C2H4OH+ channel. The measured KER in the initial Coulomb explosion of the C–C bond is in agreement with the simulated dynamics on the singlet ground state of the dication, while the lower KER of events followed by the secondary formation of HCOH+ is characteristic of simulated C–C dissociation on the triplet ground state. The sequential mechanism is further reinforced by the observation of two delayed dissociation pathways of metastable CH3HCOH+ products to H2O+ + C2H3 and COH+ + CH4, occurring on the microsecond time scale.
In contrast to dissociation of the C–C bond, ab initio potentials and ground state AIMD simulations indicate no direct cleavage of the C–O bond, in agreement with the observed preference to proton transfer and double proton transfer yield of the observed water and hydronium ions as opposed to OH+ products. C–O bond dissociation results in three-body breakup, and three of the smaller channels were found to exhibit clear indication of sequential breakup, while the major HnO+ forming channels exhibit concerted nature. Other channels, including the C2H3+ + COH+ + 4H breakup that would require transient formation of a three-carbon ring on an isolated molecule, also exhibit nontrivial correlations of the two cations and neutral center of mass indicating complex concerted breakup dynamics, where a minor contribution to this surprising yield of correlated C2H3+ + COH+ ions can be attributed to intermolecular reaction within a 2-propanol dimer, based on the particularly high recoil of the ion center of mass.
In agreement with previous studies, we find that although 2-propanol has a higher number of H atoms, it exhibits relatively low ∼0.1% yields of H3+, for example, in comparison to ∼10% H3+ yield from double ionization of methanol.24 We propose that efficient C–C breaking on the 2-propanol ground state can quench tri-hydrogen production via roaming H2 dynamics. This can help explain the previously reported trends of decreasing H3+ production with increasing carbon chain length.19 Further theoretical and experimental studies are required to better understand the effect of C–C bond dissociation on H3+ production as well as the C–O bond dissociation accompanied by both proton transfer and double proton transfer.
See the supplementary material for detailed explanation of the mass-scaled Dalitz plot representation and the spectrometer potentials. The supplementary material video file shows typical simulations of C–C breaking dynamics.
We acknowledge the support provided under ISF Grant Nos. 674/21 and 800/19 and the use of equipment provided by the Wolfson foundation. D.M.B. acknowledges support provided by the Lady Davis Fellowship.
Conflict of Interest
The authors have no conflicts to disclose.
Dror M. Bittner: Data curation (equal); Investigation (equal); Writing – original draft (lead). Krishnendu Gope: Data curation (equal); Investigation (equal); Writing – review & editing (equal). Ester Livshits: Data curation (equal); Investigation (equal); Writing – review & editing (equal). Roi Baer: Funding acquisition (equal); Investigation (equal); Supervision (equal); Writing – review & editing (equal). Daniel Strasser: Data curation (equal); Funding acquisition (equal); Investigation (equal); Supervision (equal); Writing – original draft (equal).
The data that support the findings of this study are available from the corresponding authors upon reasonable request.