Gamma Radiation-Induced Oxidation, Doping, and Etching of Two-Dimensional MoS2 Crystals

Two-dimensional (2D) MoS2 is a promising material for future electronic and optoelectronic applications. 2D MoS2 devices have been shown to perform reliably under irradiation conditions relevant for a low Earth orbit. However, a systematic investigation of the stability of 2D MoS2 crystals under high-dose gamma irradiation is still missing. In this work, absorbed doses of up to 1000 kGy are administered to 2D MoS2. Radiation damage is monitored via optical microscopy and Raman, photoluminescence, and X-ray photoelectron spectroscopy techniques. After irradiation with 500 kGy dose, p-doping of the monolayer MoS2 is observed and attributed to the adsorption of O2 onto created vacancies. Extensive oxidation of the MoS2 crystal is attributed to reactions involving the products of adsorbate radiolysis. Edge-selective radiolytic etching of the uppermost layer in 2D MoS2 is attributed to the high reactivity of active edge sites. After irradiation with 1000 kGy, the monolayer MoS2 crystals appear to be completely etched. This holistic study reveals the previously unreported effects of high-dose gamma irradiation on the physical and chemical properties of 2D MoS2. Consequently, it demonstrates that radiation shielding, adsorbate concentrations, and required device lifetimes must be carefully considered, if devices incorporating 2D MoS2 are intended for use in high-dose radiation environments.


Compton Scattering Calculations
Where and represent the wavelength of the scattered and incident photon, respectively, is the Planck ′ ℎ constant, is the electron rest mass, is the speed of light and is the scattering angle. As valence electrons are weakly bound and the energy of the incident photons produced by 60 Co decay is known, the energy of the scattered photon and kinetic energy of the recoil electron can be calculated as a function of the scattering angle ( Figure S1(a)). The inverse relationship between the energies of scattered photon and recoil electron is easily conceptualised as maximum energy transfer will occur during a head-on collision i.e. when . = 180°U nlike photons, as the energetic recoil electrons have a finite mass, they can engage in elastic collisions with the Mo and S atoms in the lattice. Hence, vacancies can be produced in MoS 2 crystals under gamma irradiation, providing that the kinetic energy transferred to the Mo or S recoil atom is greater than its threshold displacement energy. The threshold displacement energies of Mo and S atoms in MoS 2 are calculated to be 20 eV and 6.9 eV, respectively, corresponding to required electron kinetic energies of 560 keV ( and 90 keV ( ). 1 Hence, Figure S1(a) shows that via Compton scattering, 60 Co decay ) produces recoil electrons with sufficient kinetic energy to create Mo and S vacancies in MoS 2 crystals. However, the cross section for Compton scattering depends strongly on the scattering angle i.e. the probability of producing a recoil electron with a given kinetic energy is characterised by a distribution.
Therefore, the normalised cross section for Compton scattering, as a function of , provides insights into S3 the relative yields of Mo/S vacancies produced in MoS 2 crystals under 60 Co gamma irradiation and can be calculated using the Klein-Nishina formula: Where represents the differential scattering cross section per unit of solid angle, is the classical Ω electron radius, and represent the frequencies of the scattered and initial photons, respecetively, is ′ ℎ the Planck constant and is the scattering angle ( Figure S1(b)). For 1.17 MeV photons, scattering angles of 53 o and 16 o produce recoil electrons with energies corresponding to and , respectively. Therefore, Figure S1(b) shows that S vacancy formation is significantly more likely compared with Mo  Hence, Figure S2(b) definitively proves that MoS 2 oxidation is driven to a large extent by the products of water radiolysis. Moreover, as oxidation of MoS 2 via heavy ion irradiation and O 2 adsorption does not seem to yield Mo V S y O x , 3,4 Figure S2(b) suggests that Mo V S y O x intermediates are the sole product of reactions between MoS 2 surface atoms and the products of water radiolysis.

S5
Relative to the 100 kGy sample ( Figure S2(a)), the structure of the S 2s signal in the 100 kGy (H 2 O) sample exhibits notable asymmetry at higher binding energy ( Figure S2(b)). Deconvolution of the signal suggests that a higher concentration of organosulfur species are produced during gamma irradiation of MoS 2 crystals in the presence of elevated levels of adsorbed water. The deconvoluted S 2p x-ray photoelectron spectrum of the 100 kGy sample (Figure S3(a)) exhibits two doublets associated to S 2and organosulfur environments, as described in the main text. However, in the 100 kGy (H 2 O) sample, the binding energy of the organosulfur signal is ~ 1.2 eV higher such that the S 2p 3/2 and S 2p 1/2 photoelectron lines exhibit values of 163.8 eV and 165.0 eV, respectively ( Figure S3(b)).
This observation is in agreement with the more pronounced asymmetry of the S 2s signal in the Mo 3d spectrum of the 100 kGy (H 2 O) sample ( Figure S2(b)).
Moreover, similarly to the presence of Mo V S y O x species in the Mo 3d spectrum, the binding energies of the organosulfur species in the S 2p spectrum of the 100 kGy (H 2 O) sample agree well with those exhibited after 500 kGy irradiation of MoS 2 crystals stored under ambient conditions (Figure 1(c), main text).
Hence, the Mo 3d ( Figure S2) and S 2p ( Figure S3) x-ray photoelectron spectra provide conclusive S6 evidence that the products of water radiolysis directly influence and accelerate the oxidation of MoS 2 crystals. This is further evidenced by the relative concentration of Mo (Table 1) and S (  Table 2. Percentage of S atoms in a given chemical state, relative to the total integrated intensity of the S 2p photoelectron signal, for both 100 kGy and 100 kGy (H 2 O) samples.

X-ray Photoelectron Spectroscopy of Polycrystalline Monolayer MoS 2 Films
The S 2p x-ray photoelectron spectrum of the pristine polycrystalline monolayer (1L) MoS 2 film, produced by chemical vapour deposition (CVD), exhibits a doublet corresponding to the S 2environment in which the S 2p 3/2 and S 2p 1/2 photoelectron lines possess binding energies of 162.4 eV and 163.5 eV, respectively ( Figure S4(a)). These values are ~ 0.6 eV higher than those observed for MoS 2 single crystals deposited via MME (Figure 1(a), main text). This is in agreement with the common observation of ndoping in MoS 2 crystals produced by CVD. 5 Moreover, a second doublet is observed in the S 2p spectrum of pristine 1L MoS 2 , produced by CVD¸ in which the S 2p 3/2 and S 2p 1/2 photoelectron lines exhibit binding energies of 164.3 eV and 165.5 eV, respectively. These values do not agree with those reported for sulfoxide compounds, 6,7 aliphatic sulfides 7 or thiols. 8 Moreover, formation of these species is highly unlikely as CVD process is carried out under inert atmosphere and, obviously, the pristine sample has not S7 yet been exposed to gamma radiation. However, the binding energy of the second doublet agrees well with the values reported for the S 0 environment of elemental sulfur 9 which could be deposited during sulfidation step of CVD. However, as the sample was purchased from a commercial retailer, the exact deposition protocol is unknown. Upon irradiation to an absorbed dose of 40 kGy the signal attributed to the S 0 environment of elemental sulfur is no longer observed (Figure S4(b)). It is likely that the elemental sulfur is oxidised to yield sulfate species via reactions involving the products of water radiolysis. 10 Indeed, a new doublet is observed in which the S 2p 3/2 and S 2p 1/2 photoelectron lines exhibit binding energies of 168.8 eV and 169.9 eV, respectively, which agree well with those reported for sulfate compounds. 11

S8
At higher absorbed doses of 60 kGy (Figure S4(c)) and 150 kGy (Figure S4(d)), the integrated intensity of the sulfate photoelectron signal increases significantly relative to the S 2environment. As there is no elemental sulfur present after irradiation to an absorbed dose of 40 kGy ( Figure S4(b)), this suggests that the increase in the concentration of sulfate species at higher adsorbed doses is due to radiolytic oxidation of the 1L MoS 2 film via reactions involving the products of water radiolysis.
As detailed in the main text, the radiolytic oxidation of S atoms in MoS 2 crystals deposited by MME progresses via the formation of organosulfur intermediates to eventually yield sulfate species at absorbed doses of 500 kGy (Figure 1, main text). Therefore, upon irradiation of 1L MoS 2 produced by CVD to ≥ absorbed doses between 40 kGy and 150 kGy, the absence of organosulfur intermediates and high concentrations of sulfate species suggests that the rate of radiolytic oxidation is considerably higher for samples produced by CVD with respect to MME. This may be rationalised by the morphology of the material produced by CVD. As the 1L film is polycrystalline, the fraction of atoms at grain boundaries and crystal edges verses those within the basal plane is likely to be significantly higher with respect to MoS 2 single crystals deposited by MME.
Therefore, similarly to the mechanism of edge-selective radiolytic etching, discussed in the main text, the radiolytic oxidation of polycrystalline 1L MoS 2 produced by CVD is likely proceed at lower absorbed doses due to the high reactivity of grain boundaries and active edge sites. [12][13][14] Prior to irradiation, the Mo 3d x-ray photoelectron spectrum of the polycrystalline 1L MoS 2 films produced by CVD contains three doublets and two singlets (Figure S5(a)). The S 2s singlet exhibiting a binding energy of 228.8 eV corresponds to the S 0 environment of elemental sulfur as rationalised earlier.
The photoelectron lines corresponding to the Mo IV and S 2environments of MoS 2 are ~ 0.6 eV higher than those observed for MoS 2 single crystals deposited via MME (Figure 2 Upon irradiation to absorbed doses of 40 kGy (Figure S5(b)), 60 kGy (Figure S5(c)) and 150 kGy ( Figure S5(d)), the integrated intensities of the Mo VI S y O x and Mo V S y O x signals increase whilst the MoS 2 (Mo 4+ and S 2-) signals decrease as a function of radiation exposure. This agrees with the dependencies of the SO 4 2and S 2signals in the S 2p spectra ( Figure S4). Also akin to the S 2p spectra, the S 0 environment S10 of elemental sulfur is not observed in the Mo 3d spectrum of the 1L film after irradiation to an absorbed dose of 40 kGy ( Figure S5(b)). Accordingly, a considerable increase in the S 2s singlet at 233.3 eV, corresponding to the sulfate environment, is observed.
These findings show that the rate at which the 1L film becomes oxidised, to produce Mo VI S y O x and sulfate species, is the higher for CVD samples relative to MoS 2 single crystals deposited by MME. As outlined earlier, this could be attributed to the polycrystalline nature of 1L MoS 2 films produced by CVD i.e. more atoms are located at grain boundaries and crystal edges. Moreover, the considerably higher concentration of Mo VI S y O x and sulfate species in polycrystalline 1L films could be attributed to its large surface area to volume ratio. For instance, in few-layer crystals, such as those deposited by MME, the rate of reactions between the products of water radiolysis/oxygen and buried MoS 2 layers far from the crystal-atmosphere interface are expected to be lower leading to slower oxidation of the MoS 2 crystal. Conversely, in 1L MoS 2 films produced by CVD, a higher fraction of atoms will reside at the surface of the crystal, leading to faster oxidation of the nanostructure.

Optical Microscopy of SiO 2 /Si Substrates
In order to assess whether the circular features formed during gamma irradiation of MoS 2 /SiO 2 /Si samples are a product of MoS 2 radiolysis, a freshly cleaned SiO 2 /Si wafer was irradiated to an absorbed dose of 391 kGy in the absence of MoS 2 crystals. Prior to irradiation, the SiO 2 /Si wafer was sonicated for 20 minutes in a solution of propan-2-ol (99.9%, Sigma Aldrich). The wafer was sonicated for a further 10 minutes in acetone (99.9 %, Sigma Aldrich), following which, residual solvent was removed using N 2 airflow. Upon drying, the wafer was crimp-sealed inside a borosilicate vial and exposed to gamma radiation.
"Circular features", akin to those reported by Ozden et al on the surface of gamma irradiated 2D MoS 2 films produced by CVD, 28 were observed on the SiO 2 surface of the wafer after irradiation to an absorbed dose of 391 kGy ( Figure S6). Therefore, the circular features are not a product of MoS 2 radiolysis, rather, S11 they are likely to be carbonaceous aggregates formed via radiolytic reactions involving adventitious carbon which is ubiquitous on all surfaces exposed to ambient conditions. 29 Figure S6: Optical micrographs of a SiO 2 /Si wafer: (a) prior to irradiation and (b) after irradiation to an absorbed dose of 391 kGy. All scale bars correspond to a length of 50 μm. Optical micrographs of a two-dimensional (2D) MoS 2 crystal exhibiting 1L, 2L, tri-(3L) and quadri-layer (4L) domains were acquired prior to irradiation (Figure S7(a)) and after irradiation to absorbed doses of 500 kGy (Figure S7(b)) and 1 MGy (Figure S7(c)). Crystal domains of different thicknesses are S13 identifiable on account of their distinct optical contrast. However, the optical micrograph of the pristine 2D crystal has been annotated to emphasise the contrast of the 1L, 2L, 3L and 4L domains (Figure S7(d)).

Correlative Optical Microscopy and Raman/PL Mapping of Few-Layer MoS 2 Crystals
Due to the symmetry of the out-of-plane mode, S atoms in adjacent MoS 2 layers exert an effective 1 restoring force on one another, leading to a redshift of the mode in thinner MoS 2 crystals on account 1 of the lower force constant of the vibration. 18 Unlike the mode, the in-plane mode does not 1 1 2 redshift in thinner pristine crystals due to reduced effective restoring forces. Instead, a blueshift is observed and attributed to an increase in the surface force constant. This is related to a slight charge redistribution caused by the absence of neighbouring MoS 2 layers. 19,20 Hence, the -frequency difference is a reliable metric commonly used to determine the thickness of 2D MoS 2 crystals that has been rigorously verified both experimentally and theoretically. 18,19 Therefore, in order to assess whether edge-selective radiolytic etching is exclusive to bilayer (2L) crystals ( Figure 3, main text) or applies generally to 2D MoS 2 , correlative Raman maps of the 2D MoS 2 crystal, imaged via optical microscopy, are acquired. The -frequency difference is calculated and its 1 1 2 variation across the crystal after irradiation to absorbed doses of 500 kGy and 1 MGy is shown in Figure   S7(e) and Figure S7(f), respectively. Note that, in order to aid interpretation of the Raman maps, 1L, 2L, 3L and 4L reference labels have been added to the -frequency difference z-axis scale as a guide 1 1 2 for the eye.
Pristine 2L MoS 2 exhibits an -frequency difference of ~ 21 cm -1 . 18 However, after irradiation to 1 1 2 an absorbed dose of 500 kGy, the frequency difference at the periphery of the 2L domain is just ~ 19 cm -1 ( Figure S6(e)). This is consistent with edge-selective radiolytic etching of 2L crystals to yield 1L domains and proceeds via the mechanism outlined in the main text.
Pristine 3L MoS 2 exhibits an -frequency difference of ~ 23 cm -1 . 18 It can be seen that the 3L 1 1 2 region present prior to irradiation is etched to yield 2L MoS 2 after irradiation to an absorbed dose of 500 kGy. This change in morphology is most easily visualised by noting how small the domain size exhibiting S14 a -frequency difference of ~ 23 cm -1 is after 500 kGy irradiation ( Figure S6(e)); relative to the 1 1 2 size of the 3L domain prior to irradiation ( Figure S6(d)).
Previous correlative atomic force microscopy measurements suggest that domains exhibiting an -1 1 2 frequency difference of ~ 23.8 cm -1 can be assigned as 4L MoS 2 . 4 The majority of the 4L regions present prior to irradiation ( Figure S6(d)) are etched to yield 3L MoS 2 after irradiation to an absorbed dose of 1 MGy ( Figure S6(f)). Moreover, at high absorbed doses, radiolytic etching is not limited to just the uppermost layer of 2D MoS 2 crystals. For example, at the periphery of the 4L and 3L domains, the crystal thickness is observed to reduce by more than one monolayer i.e. etching of 4L→2L and 3L→1L is observed ( Figure S6(f)). MoS 2 produced by edge-selective radiolytic etching of 2L crystals after 500 kGy irradiation, i.e. 1L border.
The spectra have been normalised to the intensity of the signal in pristine 1L MoS 2 .
′ 1 Numerous defect-activated modes are observed in the Raman spectrum of defective 1L MoS 2 . The most noteable of these is the mode observed at 227 cm -1 . 21 Figure S12(a) shows that the contribution ( ) of this signal towards the Raman spectrum of 1L MoS 2 crystals irradiated to an absorbed dose of 500 kGy is not significant. The scattering process responsible for the mode in 1L MoS 2 is described by the ( ) phonon-confinement model. [21][22][23] Consequently, the signal becomes observable only in MoS 2 ( ) crystals with an average inter-defect spacing of less than 10 nm, i.e. highly defective crystals produced by electron 24 and ion beam irradiation. 25 Therefore, Figure S12(a) suggests that the average inter-defect spacing within the basal plane of 1L MoS 2 crystals irradiated to 500 kGy must be greater than 10 nm.
The Raman spectrum of the 1L border region formed after 500 kGy irradiation is provided in Figure   S12(b). It is proposed that edge-selective radiolytic etching of the uppermost layer in a 2L crystal yields a border in which the surface of the 1L MoS 2 crystal is covered by oxidised Mo and S species. However, the Raman signals arising from Mo-O bonds 26 are not observed in the Raman spectrum of the 1L border ( Figure S12(b)). This is attributed to the amorphous morphology and < 1 nm thickness of the oxidised S20 layer. The fundamental Raman selection rule states that at least one component in the polarizability tensor of a vibrating molecule must change when interacting with the electric field of the incident photon.
However, as the polarizability tensor is anisotropic, the cross section for Raman scattering is decreases significantly during a crystalline-to-amorphous transition. 27 Therefore, Mo-O signals are not observed in the Raman spectrum of the 1L border (Figure S12(b)) due to the relatively small number and random orientation of Mo-O bonds within the atomically thin amorphous layer. However, the authors prove the existence of Mo-O bonds by using XPS and observing the Mo VI doublet (Figure 2, Main Text).   ( Figure S13(a)). However, the resolution of the spectrometer used in this study is ~ 1 cm -1 whilst the blueshift of the frequency distribution maximum is < 0.5 cm -1 . Moreover, as the kurtosis of the distribution increases upon irradiation, the slight blueshift of the mode cannot be considered statistically significant. irradiation to an absorbed dose of 40 kGy (Figure S13(b)). However, a significant increase in the kurtosis of the distribution is observed. This may suggest that the slight blueshift of the mode could be ′ 1 associated with Γ decreases and increases due to p-doping 25,30 and defect formation, 21 respectively.

Statistical Analysis of Raman/PL Maps of Polycrystalline Monolayer MoS 2 Films
However, akin to the frequency distribution, changes in the Γ of the mode cannot be considered ′ 1 ′ 1 statistically significant on account of the considerable overlap of the pristine and irradiated Γ distributions ( Figure S13(b)).
Conversely, the optical properties of the polycrystalline 1L MoS 2 film exhibit significant changes upon irradiation to an absorbed dose of 40 kGy. Figure S13(c) shows that the PL intensity of the film decreases considerably after exposure to gamma radiation. This is attributed to an increase in the number of nonradiative decay pathways on account of the introduction of defect states within the bandgap of irradiated 1L MoS 2 . 25 Moreover, an increase in the energy of the PL signal is observed after gamma irradiation ( Figure S13(d)). This is attributed to the dissociation of Atrions into neutral A excitons at higher energy.
These p-type changes in the optical properties of the irradiated film are facilitated by charge transfer from occupied conduction band states to adsorbed O 2 molecules, as described in the main text.

S22
However, as the increase in the energy of the PL signal is statistically significant upon irradiation to an absorbed dose of 40 kGy (Figure S13(d)), this suggests that the film becomes p-doped and should be accompanied by a significant blueshift and Γ decrease of the mode 25,30 which is not observed (Figures ′ 1 S13(a) and S13(b)). Therefore, the statistically significant shift in the energy of the PL signal could be attributed to the more numerous non-radiative decay pathways of trions. 31 For instance, in pristine materials, the PL intensity of trions is generally lower than excitons. Hence, they do not contribute significantly to the PL signal of 1L MoS 2 films after irradiation due to the overall reduction in PL intensity (Figure 13(c)).