Photoelectron imaging of PtI 2 − and its PtI photodissociation

The photoelectron imaging of PtI 2 − is presented over photon energies ranging from hν = 3.2 to 4.5 eV. The electron affinity of PtI 2 is found to be 3.4±0.1 eV and the photoelectron spectrum contains three distinct peaks corresponding to three low-lying neutral states. Using a simple d-block model, and the measured photoelectron angular distributions, the three states are tentatively assigned. Photodissociation of PtI 2 − is also observed, leading to the formation of I − and of PtI − . The latter allows us to determine the electron affinity of PtI to be 2.35±0.10 eV. The spectrum of PtI − is similarly structured with three peaks which, again, can be tentatively assigned using a similar model that agrees with the photoelectron angular distributions.


Introduction
Platinum halides have complex electronic structures and exhibit unusual bonding, resulting in very high electron affinities (EA) and high formal oxidation states on the platinum core.
These properties have led platinum halides to be investigated as superhalogens and small multiply charged anions. For example, [PtCl4] 2− is the smallest experimentally reported dianion 1 and PtF6 has an EA of ~7 eV and oxidises both O2 and Xe. [2][3][4] To date, the focus has been on platinum fluorides, chlorides andto a lesser extentbromides, with no gas-phase studies investigating the electronic structure of the platinum iodides. [5][6][7] Hence, there are unanswered questions about how the interesting properties that characterise the smaller platinum halides, evolve for larger halogens, especially as changes are observed for more elaborate platinum complexes, such as [PtX3(C2H4)]where X is Cl, Br or I, with larger halide ligands. 8,9 In addition to the increased size of iodine, which may induce steric strain into the molecular framework, iodine has a larger spin-orbit splitting and a higher polarizability, which may change the bonding and electronic structure of the platinum iodides compared to other platinum halides. Whilst building an understanding of the evolution of the electronic structure of the platinum halides is of fundamental interest; platinum halides have also found applications in catalytic processes. For example, perovskites with Pt-I3 active sites are highly efficient photocatalysts for H2 production. 10 Here, we study the electronic structure and photochemistry of the simplest platinum iodides, PtI − and PtI2 − . We employ photoelectron imaging, which probes the electronic structure directly and, through its photoelectron angular distributions, also offers insight into the molecular orbitals involved. Performing photoelectron imaging over a range of photon energies can often provide more insight into the electron loss dynamics, as we and others have shown in several cases, and we use a similar approach here. [11][12][13][14][15][16][17] This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset. PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610 3 The photoelectron angular distributions (PADs) are particularly interesting in the context of metal complexes, where, in principle, the sensitivity to the electronic structure can provide insight into the chemical bonding involved. [18][19][20] Jarrold and coworkers have recorded photoelectron spectra for some transition metal and lanthanide clusters, including NiO − and Gd2O − , at different laser polarisations relative to the detector, in order to gain some information about the PADs and characterise the spectroscopic transitions. 21,22 Here, we record the full PADs, using photoelectron imaging, in order to investigate the symmetry of the molecular orbitals of transition metal complexes. Moreover, for the case of platinum iodides, spin-orbit coupling and relativistic effects are likely to be large and these can have striking influences on the molecular orbitals of such complexes, which again the PADs may be sensitive to. 23,24 Previous studies have considered the electronic structure of platinum fluorides and chlorides, PtFn − and PtCln − (n = 1 -8), using theoretical methods and photoelectron spectroscopy. 5,6 Strong similarities were noted between the platinum fluorides and chlorides.
Both platinum dihalides exhibited linear anion and neutral ground states. The EA for PtF2 was calculated to be between 2.72 -3.13 eV, 6 which is slightly lower than the experimentally measured EA ~ 3.5 eV for PtCl2. 5 The previous photoelectron spectroscopic study of PtCln − (n = 2, 4 and 5) used a single photon energy hν = 4.66 eV; the PtCl2 − spectrum comprised of three distinct bands. 5 Note that the n = 1 diatomic molecule was not observed in this previous work and, to the best of our knowledge, there have been no photoelectron spectra of any platinum halide reported. In general, stable MX molecules are rare, and typically require a d 10 electron configuration and bulky ligands in order to stabilise the cluster. 25 Here, photodissociation of PtI2 − results in PtI − and has allowed us to measure the photoelectron spectrum of a diatomic platinum halide for the first time.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

Experimental methods
Our work utilises 2D photoelectron imaging and the experimental apparatus has been described in detail elsewhere. 26,27 Briefly the anions were produced via electrospray ionisation of a solution of 2 mMol K2PtI6 in methanol. Anions were desolvated in a capillary, transferred through a series of differentially pumped regions by means of ring-electrode guides, and stored in a ring electrode trap, before being accelerated and mass-selected in a Wiley-Mclaren timeof-flight spectrometer. The ring-electrode guides also serve to perform collision-induced dissociation. 26,27 In the present experiments, PtI6 2− serves as a precursor to form other platinum iodide species, with PtI2 − being one of the most abundant in the mass-spectrum. As described below, PtI − was formed via the photodissociation of PtI2 − . Surprisingly, we did not observe PtI6 2− , suggesting that it may be quite unstable as an isolated dianion.
Photoelectrons were generated through the intersection of the mass-selected anion packet and a nanosecond laser pulse. Pulses of variable photon energies in the visible and UV were produced via a Nd:YAG pumped optical parametric oscillator (OPO). The photoelectrons were imaged on a dual microchannel plate detector in a velocity map imaging configuration, 28 and photoelectron spectra were subsequently obtained from these raw images. Photoelectron imaging also yields the PADs of the emitted electron relative to the polarisation axis, which was parallel to the detector face. Previous studies of platinum halides used a magnetic bottle electron detector, which does not record the PAD and has a very low detection efficiency for photoelectrons with low electron kinetic energy (eKE). 5,21,29 The energy resolution of our photoelectron spectrometer is 5% of the eKE, based on calibration with the photodetachment of iodide.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

Analysis
The raw photoelectron images were deconvoluted with a polar onion-peeling algorithm, 30 which reconstructs the 3D Newton sphere of electrons from the 2D image obtained and consequently recovers both the photoelectron spectra and PADs. The PADs are dictated by the molecular orbital from which the electron is removed in the photodetachment process.

Results
Photoelectron spectra of PtI2 − were recorded with nanosecond (ns) laser pulses at photon energies between hν = 3.2 -4.5 eV. Figure 1 shows two representative photoelectron images of PtI2 − recorded at hν = 3.6 and 4.2 eV, from which subsequent photoelectron spectra have been extracted. All the photoelectron spectra are shown in Figure 2 and reported as a function of electron binding energy (eBE), where eBE = hν -eKE. The spectra are normalised to the most intense feature in each spectrum, averaged using a five-point moving mean and are offset to allow comparison. There are several spectral features that are clearly visible over different ranges of photon energies. These can be broadly classed into three distinct groups: we have colour-coded these in Figure 2 and labelled as A, B, and C, in order of decreasing eBE.
We now consider these in turn.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.  This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
Figure 2: Photoelectron spectra of PtI2 − recorded with hν = 3.2 eV -4.5 eV and presented as a function of electron binding energy (eBE). All spectra are normalised to the largest peak and offset for clarity. Direct detachment channels of PtI2 − are labelled as A and highlighted in green. Photodissociation and subsequent photodetachment of I − is labelled as B and highlighted in blue. Photodissociation and subsequent photodetachment of PtI − is labelled as C and highlighted in red. The photoelectron spectrum of PtI − recorded at hν = 3.5 eV is multiplied by a factor of 10 to accentuate its photoelectron signal and is shown in red.
A, Direct detachment: For hv > 4.1 eV, a cluster of three distinct peaks can be seen with maxima at eBE ~ 3.5, 3.8 and 4.0 eV. The three peaks can be assigned to direct detachment: This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
B, Iodide detachment: For photon energies less than approximately 4.0 eV, a new feature emerges peaking at eBE = 3.06 eV. This feature grows in relative intensity as the photon energy decreases, but this is likely just a reflection of the decreasing cross section of the direct features (A) discussed above. The feature becomes much less prominent for hv > 3.9 eV and remains visible down to 3.2 eV. The spectral shape, binding energy and β2 parameters of peak B are consistent with the well-known photoelectron spectrum of I -. 33 This detachment peak is presumably formed via a multiple-photon process involving the two steps of photodissociation and subsequent photodetachment (here, we will use the term two-photon to describe such a process), The presence of two-photon photodissociation and photodetachment of the resulting I − , below the ADE of PtI2 − , requires the presence of at least one bound electronic state of the anion, which is excited by the first photon. Dissociation may occur on the excited state potential energy surface or on the ground state surface following internal conversion; we cannot obtain information about the mechanism from our current experiments. As hν increases, direct detachment also becomes possible so that the anion excited states are now in the detachment continuum (i.e. resonances) and dissociation will compete with autodetachment. 17,[34][35][36][37] The relative decrease of the intensity of the I − feature compared to the direct detachment channels of PtI2 − can arise from a number of reasons in addition to the one noted above: the favourability of a one-photon process over two-photon process, the large apparent photodetachment crosssection of PtI2 − , and the absorption profile to the excited states of PtI2 − . At the low-energy spectral end, we did not succeed in acquiring a spectrum at hv = 3.1 eV, in part because our OPO has very weak output here and the cross section for detachment from iodide is low (near-This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset. extracted from the hv = 3.5 eV spectrum. However, we do note that the signal level for these features is very low, such that the spectral structure and the PADs, should be taken as qualitative rather than quantitative measurements.

PLEASE CITE THIS ARTICLE AS
As only Pt and I is present in PtI2 − , we first considered the possibility that the photodissociation product corresponds to Pt − , which is subsequently detached. However, the electron affinity of Pt has been measured as 2.12 eV, and the previously reported photoelectron spectrum shows different structure and relative intensity, 38, 39 so the spectrum is not consistent with this explanation. The photoelectron spectrum measured is also not consistent with that of I2 − , which has been extensively studied. 40 This leaves us with the only other alternative, PtI − .
To the best of our knowledge, this diatomic has not been characterised, either as an anion or This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset. Analogous to feature B, the most likely explanation for this feature is the two-photon process:

PLEASE CITE THIS ARTICLE AS
PtI2 − + hν → PtI − + I , Hence, the excited anionic states accessed in PtI2 − appear to lead to a competition between dissociation leaving the negative charge on the PtI or I, with the latter apparently dominating (although this is difficult to verify without knowledge of relative photodetachment cross sections of the anions). It may also be possible that PtI − undergoes dissociation (either spontaneous or by absorption of a further photon) to form I − , and this would be indistinguishable spectroscopically from the two-photon process of photodissociation of PtI2 − to form I − , although a photodissociation would be unlikely by the small probability of a threephoton process required.
The three peaks associated with the photodetachment from PtI − with differing β2 values are similar to the photodetachment from PtI2 − (feature A). The spectral structure of the direct detachment channels for PtI − and PtI2 − both exhibit three peaks, with a β2 < 0, > 0 and > 0, and a total width of ~1 eV. This is perhaps unexpected as the electron configurations of the two molecules are different: PtI − is an even-electron species whereas PtI2 − is an odd-electron species.

Discussion
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

11
The direct photoelectron spectrum of PtI − shares many characteristics of that seen for PtI2 − . Our measured ADE (and VDE) are also similar to those calculated values for PtCl − (VDE = 2.21 to 2.70 eV, depending on level of theory), 5 the only other diatomic platinum halide studied to date. As we were not able to form an ion beam of PtI − directly via electrospray ionization, it is difficult to determine whether it undergoes photodissociation. However, this may be unlikely as ionic photodissociation could result in a neutral Pt atom, which is not a favoured oxidation state of platinum.
Previous photoelectron spectroscopy of PtCl2 − determined its ADE = 3.5 eV and its VDE = 3.83 eV, 5 while computational work predicted the electron affinity of PtF2 to be between 2.72 -3.13 eV depending on the level of theory used. 6 Our measured VDE = 3.5 eV for PtI2 − , is therefore of similar magnitude to these lighter halides. Clear similarities are also observed between the photoelectron spectra of PtI2 − and PtCl2 − . 5 The hν = 4.66 eV photoelectron spectrum of PtCl2 − exhibits three distinct peaks due to direct detachment. The highest eBE peak was near threshold and its low relative intensity may be skewed by threshold effects, as we noted in the hv = 4.1 eV spectrum in Figure 2 for example. The two other peaks at lower eBE are well-resolved and the overall width of the observed photoelectron signal spanning the three peaks is approximately 1 eV. This overall appearance is similar to the photoelectron spectra observed for PtI2 − (e.g. see Figure 2 spectrum at hν > 4.1 eV, which has three sharp features with a total spectral width of ~1 eV). The ~1 eV total spectral width of the direct detachment bands for the dichloride and diiodide platinum complexes indicates that the electronic states in the neutral are not just spin-orbit split states arising from the spin-orbit coupling involving the halide.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset. This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
13 by overlap with the ligand field. With reference to Figure 3(a), the dx2-y2 and dxy on Pt cannot interact with the atomic orbitals on I and forms a pair of degenerate non-bonding (n) MOs. The overlap between the dxz and px as well as the dyz and py or it l will form ou ly ege er te π π MO . In the absence of orbital mixing, the overlap of the dz2 and ligand localised pz orbitals may be expected to le to σ σ MO . However, the 6s orbital of Pt is close in energy to the 5d orbitals, as evidenced by the 5d 9 6s 1 ground state electron configuration of Pt, and of the correct symmetry to mix with the 5dz2 orbital, leading to a hybrid s/d orbital. This hybrid orbital in turn interacts with the pz orbital on I to lead to three MOs: one bonding, one antibonding and one non-bonding, as shown in Figure 3(a) in red. These MOs, while being hy ri , effectively e r σ-type MOs.
Taking the above arguments and applying it to the case of a linear ML2 molecule with to be an oversimplified picture of the MOs, meaning the energy difference between the n orbitals, one of which is a hybridised MO, is likely to be larger than implied in the simple This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
model. Alternatively, the n orbitals are near-degenerate and both contribute to the highest eBE peak.
We additionally have measured the PADs for the three individual direct detachment peaks. Detailed qualitative and semi-quantitative models have been developed for the prediction of the β2 parameter. [41][42][43] Given the qualitative picture presented above, we continue along such lines and consider the qualitative PADs that might be expected for the three different detachment channels. Qualitatively, β2 parameters are expected to be positive or negative e e i g o the ture of the or it l from which the electro i et che . Ge er lly, for σ orbital, the outgoing wave can be approximated as a p-wave resulting in β2 > 0; in contrast, for π or it l, the outgoing wave will have a mixture of s-and d-waves which results in β2 < 0.
For the three channels observed in the experiment, we observe β2 = −0.9 for removal from the π MO; β2 = +0.3 eV for the remov l from the σ hy ri MO; β2 = +0.4 eV for the removal from the n MO. These observations are in qualitative agreement with the expectation assuming that the highest eBE peak arises from the n-hybrid MO. The PADs therefore offer additional support that the proposed assignment and simple d-block picture is representative of the electronic structure of PtI2 − . It should be noted that the origin of the structure of the PtCl2 − photoelectron spectrum was not discussed by Joseph et al.. 5 In this simple d-block model we have not accounted for the role of spin multiplicities, which would allow the observed spectral structure and PADs for PtI2to be explained via photodetachment from just two MOs, instead of the three suggested by the above model.
Specifically, the three bands could arise due to remov l of electro from the π , which would produce a 1 Σ eutr l t te, or from the σ , which would produce 1 Π 3 Π neutral states. The observed PADs would also be consistent with this model, as the two bands originating from electro remov l from the me σ woul oth h ve β2 > 0.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
However, the d-block model shown in figure 3 45 , it may be expected that the different spin-or it t te of Π symmetry will have significantly different energies but, as the Π states originate from electron ejection from the same σ orbital, the PADs may be expected to be similar (as in the spin multiplicities model described above). Therefore, this combined d-block model with spin-orbit coupling could also explain the origin of the three observed bands and the measured PADs for PtI2 -.
However, from this picture we may expect more bands in the photoelectron spectra of PtI2than are observed (overlapping bands may not have been resolved). In addition, it is more difficult to explain the strong similarities between the photoelectron spectra of PtCl2and PtI2 -, using this picture, as the Cl and I have significantly different spin-orbit splittings.
On a more fundamental note, it is questionable how valid the use of LS coupling is for PtI2 -, when heavy atoms such as Pt are best described by J-J coupling and the d-block is localised on the metal core. It should be noted that J-J coupling would also result in four spinorbit split neutral states following electron removal from the σ . One further consideration is the effect of spin-orbit coupling on the PAD, particularly in the limit of J-J coupling, where the orbital angular momentum quantum number (L) is no longer a good quantum number. PADs are often qualitatively interpreted in terms of the L of the orbital from which the electron is lost, and therefore if J rather than L is well defined, it may be challenging to predict the PADs associated with specific photodetachment channels.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
A very similar three-state d-block picture can be constructed for PtI − (Figure 3(b)). The highe t MO ei g π , followe y σ hy ri MO the hy ri MO. Formally, the dblock would also contain a lower energy n MO, but this is not observed here, either because it is too low in energy or because the peak in the photoelectron spectra is obscured by the much higher intensity I − feature (B in figure 2). The hybridisation in PtI − is slightly more complicated because there is likely to also be mixing with the 6p orbitals of the Pt atom. But overall, given the similarity between the photoelectron spectroscopy of PtI2 − and PtI − , our proposed picture appears appropriate. In Figure 3(b), the MO occupancy is also shown, along with the likely photodetachment channels that contribute to the photoelectron spectrum. Analogous to the PtI2 − detachment, the three peaks come about from the et chme t of the π , σ hy ri , hybrid MOs. The PADs are expected to be similar again: β2 < 0, β2 > 0 and β2 > 0, respectively. This is again in excellent qualitative agreement with experiment. It should be noted that addition of spin multiplicities and spin-orbit coupling to this three-state d-block model does not readily describe the observed spectral structure of PtI -, as LS coupling would predict two bands at the highest eBE with β2 < 0.
Note that we specifically refrained from performing electronic structure calculations.
This was done because of the lack of confidence we have in these, particularly for predicting the MOs that may be contributing to the detachment. Specifically, spin-orbit coupling, which in reality is likely to be intermediate in character between the limits of LS and J-J coupling, can lead to strong mixing of angular momenta. However, despite the likely large role spin-orbit effects have on the electronic structure of PtI2and PtI -, it is very challenging to accurately account for these effects in electronic structure calculations. Therefore, instead of attempting to provide interpretations based on electronic structure calculations in which we have little confidence, we prefer the simple d-block picture which we feel offers much more chemical insight too.
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
In addition to the direct photoelectron spectra of PtI2 − , we also observe secondary (twophoton) detachment from I − or PtI − , indicating that bound (with respect to electron loss) electronically excited states for PtI2 − exist (see Figure 2, feature B and C) and lead to two possible dissociation channels: I + PtI − or I − + PtI. From our experiments, we cannot gain any insight into the dynamics that leads to dissociation, and we cannot determine whether the dissociation occurs on the ground or the excited state. However, assuming that photodetachment from I − and PtI − have similar overall cross-sections, the photoelectron spectra in Figure 2 suggest that the I − + PtI channel dominates. This observation is consistent with the fact that the electron affinity of iodine (3.059 eV) is larger than that of platinum iodide (2.35 eV).

Conclusion
A photoelectron imaging study of PtI2 − i re e te i the hoto e ergy r ge 3.2 ≤ hv ≤ 4. eV. The V E of PtI2 − is measured to be 3.5±0.1 eV and the electron affinity of PtI2 is 3.4±0.1 eV. Three peaks contribute to the direct photoelectron spectrum, each with a distinct photoelectron angular distribution (PAD), which we can assign to the direct detachment from the anion to the three lowest lying electronic states of the neutral. Using a d-block molecular orbital model in which the 5d and the 6s orbitals on Pt interact with the 6p orbitals on I, a molecular orbital picture is constructed that is consistent with the observed spectrum including the PADs. The use of PADs to assign and understand the electronic structure of transition metal complexes holds significant potential in building up comprehensive pictures of such complexes.
PtI2 − is also observed to undergo photodissociation to produce I − predominantly, as evidenced by the photodetachment from iodide. This feature is visible over a large range of photon energies, including below the detachment threshold, indicating that PtI2 − has at least This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0085610
one bound excited state with respect to electron loss. In addition to the loss of I − , additional features are seen over a spectral range near hv ~ 3.5 eV. These have been assigned to PtI − and a d-block model analogous to that for PtI2allows us to assign the features and their PADs. The electron affinity of PtI is found to be 2.35±0.10 eV.