Interactions of deposited Ca with TiSe2 and TiTe2 surfaces

The interaction of evaporated Ca with TiSe2 and TiTe2 surfaces was studied using photoelectron spectroscopy. The surfaces were held at room temperature, and the measured spectra clearly show that the deposited Ca reacted with the substrates, without formation of any Ca overlayers. For depositions on TiSe2 corresponding to less than 2 nm of Ca there were indications of an intercalation reaction taking place, with charge transfer to the TiSe2 layers, but as more Ca was added a layer-breaking reaction dominated. For the case of TiTe2 only a large Ca deposition was studied, resulting in a layer-breaking reaction very similar to that found on the TiSe2 surface for similar amounts of deposited Ca.


Introduction
Layered transition metal dichalcogenides (TMDCs) are a group of highly anisotropic materials, which have been studied extensively for half a century [1][2][3][4]. They are described by the general formula TX 2 , where T is a transition metal and X stands for S, Se, or Te. Each layer consists of a hexagonal sheet of transition metal atoms, sandwiched between two hexagonal chalcogen sheets. The internal bonds in the layers are strong and of mixed ionic and covalent character, while the bonds between adjacent layers are weak and largely of van der Waals character. Due to this highly anisotropic bonding, one may obtain clean and flat cleavage in vacuo, which is a significant experimental advantage.
Depending on the valence band filling, the TMDCs can be either metallic or semiconducting. They have been used as model systems for studies of a wide range of interesting phenomena associated with their reduced dimensionality [5][6][7], and the possibility to produce single layers with truly two-dimensional properties has recently led to tremendous interest in these materials for nano-engineering and device application purposes [8,9]. Their properties can be altered by intercalation in situ with e.g. alkali metals, which has been extensively studied by photoelectron spectroscopy [10,11]. TiSe 2 and TiTe 2 both adapts the 1T structure with octahedral coordination of the Ti atoms. Regarding TiSe 2 there has been a longstanding debate whether there is a small band gap or not between the Ti 3d band and the Se 4p band, and this question is further complicated by a charge density wave (CDW) reconstruction occurring at low temperatures. Both the mechanism behind the CDW, and its connection to the gap/overlap issue, has been the subject of numerous studies [12][13][14][15][16][17]. In the otherwise isoelectronic TiTe 2 there is a significant overlap between the Ti 3d and Te 5p bands, which makes this compound an undisputed semi-metal [18,19]. In studies of in situ intercalation with alkali metals both compounds have behaved similarly, with rapid intercalation taking place at room temperature [20,21].
Intercalation of TMDCs in situ is possible also with other metals. Examples of such systens are Ag/TiS 2 [22], Pb/TiSe 2 [23] and Cu/TaS 2 [24], but many other combinations are unexplored. Intercalation with polyvalent metals may result in new interesting phenomena. If some valence electrons remain on the metal atoms, they may form localized or delocalized states, depending on the density and ordering of intercalated atoms. Given the potential interest in modification and interfacing of TMDCs by metal deposition, the aim of this study is to determine the behaviour of Ca atoms when deposited on TiSe 2 and TiTe 2 surfaces, and whether intercalation is achieved at room temperature in such systems. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Experimental details
Core level and valence band photoelectron spectra from the Ca/TiSe 2 and Ca/TiSe 2 systems were measured at the MAX-lab synchrotron radiation facility in Lund, Sweden in March 2009. The spectra were measured at the I311 beamline, which was equipped with a Scienta 200 electron energy analyser [25]. The beamline incorporated a modified SX-700 plane grating monochromator, and provided photons with energies in the range 30-1500 eV.
The TiSe 2 and TiTe 2 single crystals were attached to the sample holder by silver filled epoxy resin, and small metal handles were glued to the front faces of the samples. After curing of the resin, the samples were introduced into the spectrometer chamber, and ultra-high vacuum (UHV) was obtained after standard bake-out. Each sample was cleaved in vacuo by hitting its attached handle. This procedure resulted in mirrorlike (0001) surfaces, whose cleanliness were verified by photoelectron spectroscopy. The removed pieces remained in the measurement chamber but did not degrade the UHV, as the cured epoxy resin was UHV compatible. The thickness of the cleaved samples were several micrometers, corresponding to many thousands of layers. As the probing depth of photoelectron spectroscopy is a few nanometers, the sample thickness is not of critical importance in the present study.
Ca was deposited onto the surfaces from a resistively heated evaporator, and a quartz crystal resonator was used to monitor the deposition rate. The cumulative deposited amounts of Ca are given in nanometers, not taking into account whether the Ca stays on the surface or not. Since there are no previous reports of Ca deposition on TMDC surfaces, the choice of deposition strategy is largely a matter of trial and error. The approach used for the TiSe 2 sample in this study involved cumulative small depositions, with recording of photoelectron spectra before, between and after the depositions. The alternative approach tried for the TiTe 2 sample instead consisted of one single large deposition, followed by a 17 h wait to look for delayed reactions. The samples were kept at room temperature throughout the experiments. A reference spectrum of Ca metal was Figure 1. Se 3d core level spectra from TiSe 2 obtained with hν = 130 eV. The uppermost spectrum was measured from the clean TiSe 2 surface, while the spectra below were obtained with gradually increased amounts of deposited Ca. Panel (a) show the primary spectra, while panel (b) show the same spectra after numerical removal of the Se 3d 3/2 component. measured from a thick film evaporated onto the sample holder, with the latter cooled by LN 2 . The pressure was kept in the low 10 −10 mbar range during the measurements.
The electron energy analyser was used in the angle integrating mode and the total energy resolution was typically 0.1 eV. The normalization of the spectra poses a problem due to the intensity changes of the synchrotron radiation. The radiation intensity on the sample is not strictly proportional to the storage ring current, as the width and position of the electron beam changes somewhat as the current falls off. It turned out to be useful to normalize the core level spectra to have the same background intensity at an energy just below the spectral peaks. This procedure resulted in reasonable spectral trends without erratic variations, although some uncertainty must be acknowledged before drawing quantitative conclusions. The background normalization procedure did not work well for the valence band spectra, which instead are presented normalized to equal maximum intensity.

Results and discussion
3.1. Ca/TiSe 2 Figure 1(a) shows Se 3d spectra during repeated Ca depositions. For pristine TiSe 2 the spectrum consists of two symmetric and relatively narrow peaks A and A', corresponding to the spin-orbit split levels 3d 5/2 and 3d 3/2 , respectively. After the first Ca deposition the peaks became broader and somewhat asymmetric. After further Ca depositions new peaks B and C appeared, together with spin-orbit shifted components B' and C', but the precise peak positions are problematic to evaluate due to the substantial overlapping of peaks. One possibility in order to clarify the spectral changes is to remove the 3d 3/2 components through a recursive procedure under the assumption that the 3d 3/2 contribution is just a spin-orbit shifted replica of the 3d 5/2 contribution. Figure 1(b) shows the same Se 3d spectra after such 3d 3/2 removal, obtained with a spin-orbit splitting of 0.85 eV and a branching ratio of 0.60. The oscillations seen on the low-energy side of the spectra are artefacts caused by the 3d 3/2 contributions not being perfect replicas of the corresponding 3d 5/2 features. The spin-orbit splitting and branching ratio were adjusted to minimize the oscillations. The presence of the artefacts calls for some caution in the interpretation of weak spectral features, but the conspicuous peaks can be identified more clearly than in the raw spectra. For pristine TiSe 2 peak A is the only conspicuous feature in figure 1(b), but after deposition of 0.6 nm Ca a shoulder is seen on the low-energy side of peak A. At 1.2 nm of Ca this shoulder has grown to yield peak B about 0.4 eV below peak A. In addition, a weak shoulder is suggested on the low-energy side of peak B. At 2.4 nm of Ca, this shoulder has developed into the broad and dominating peak C about 0.9 eV below peak A, which at this stage has been reduced to a shoulder on the high-energy side of peak C. After the final Ca deposition, corresponding to 5 nm of Ca on the surface, only one broad peak, shifted further towards the lowenergy side, is seen. Figure 2 shows Ti 2p 3/2 spectra recorded during the series of Ca depositions. The pristine peak D was broadened but not markedly shifted as more Ca was added. Since peak D already from the beginning is much broader than peak A in the Se 3d spectra, it is more difficult to detect any shifted components, but it appears that a shoulder E is growing on the high-energy side of peak D as more Ca is added. The position of shoulder E relative to peak D can be clarified by examination of the second derivatives, and for the case of 2.4 nm of Ca it was established that the peak E is located about 1.3 eV above peak D. After the final Ca deposition the Ti 2p peaks are strongly suppressed, and peak E appears to dominate. Figure 3 shows corresponding valence band spectra. A detailed analysis of the valence band structure is not possible, since the analyser was operated in angle-integrating mode, but the appearance of peak F close to the Fermi level is very notable. It clearly shows that the narrow Ti 3d band, which is almost empty in pristine TiSe 2 , was gradually filled by charge transfer from Ca, as the latter was deposited on the surface. After deposition of 2.4 nm Ca peak F is found to be lower but broader, and after the final deposition it is almost gone, indicating a less metallic overlayer. Instead, the valence band emission is dominated by an intense and broad peak about 5 eV below the Fermi level, presumably of Se 4p character, although hybridized to some extent with Ti 3d and Ca 4s states. Figure 4 shows corresponding Ca 2p spectra, and for comparison also a spectrum measured from a thick Ca film obtained by massive Ca deposition on a LN 2 cooled sample holder. As more Ca was added to the TiSe 2 sample the peaks shifted slightly downward in energy and became somewhat broader, and some weak energyloss features became visible after the two last depositions. These features are shifted down in energy 6.6 eV relative to the primary Ca 2p peaks. For the thick Ca film the peaks are strongly asymmetric and exhibit plasmon Figure 3. Valence band spectra from TiSe 2 obtained with hν = 110 eV. The uppermost spectrum was measured from the clean TiSe 2 surface, while the spectra below were obtained with gradually increased amounts of deposited Ca. satellites in a fashion typical for metallic surfaces, which is in contrast to the more symmetric peak-shapes after the room temperature depositions on TiSe 2 . It is also notable that the intensity increase of the Ca 2p peaks is relatively modest when going from 0.6 to 1.2 nm of deposited Ca, compared to the much larger increase when going from 1.2 to 2.4 nm of Ca. It is also notable that the Ca 2p peaks after the room temperature depositions appears at lower energy than they do for the thick metallic Ca film.
The challenge is now to find a scenario for the interaction between the TiSe 2 host layers and the deposited Ca, which is consistent with the observed spectral changes. Three principally different outcomes can be imagined: (i) formation of a Ca overlayer on top of essentially unchanged host layers, (ii) intercalation of Ca between the host layers, and (iii) a layer-breaking chemical reaction between Ca and TiSe 2 , resulting e.g. in formation of CaSe and free Ti metal, or some non-layered compounds of Ca, Ti and Se.
The formation of a metallic Ca overlayer can be ruled out since the Ca 2p spectra obtained after the Ca depositions lack the typical metallicity signatures apparent in the spectrum from the thick Ca film. Also the spectral changes seen in figures 1-3 are inconsistent with formation of a metallic Ca overlayer, as one would then expect gradually damped but otherwise relatively unchanged peaks of TiSe 2 origin. Charge transfer from a Ca overlayer to the Ti 3d band of the uppermost TiSe 2 layer to produce peak F in figure 3 is conceivable, but the persistence of this peak when going up to 1.2 nm of Ca also makes the metallic overlayer picture improbable. It is of course likely that Ca is adsorbed on the surface in the very first stage of the deposition, but not in sufficient amounts to form a metallic overlayer.
The formation of an intercalation compound Ca x TiSe 2 should in principle be possible, but it is unknown whether this actually happens under the present circumstances. It is well known that alkali metals readily form intercalation compounds when deposited in situ onto clean TMDC cleavage surfaces [11], and since Ca (and other alkaline earth metals) have many similarities with the alkali metals, it may be expected that in situ intercalation with Ca is feasible too, but this has not previously been experimentally verified. It should be noted that intercalation with Ca, in analogy with alkali metal intercalation, should not require formation of strong chemical bonds between Ca atoms and the host layers, but rather be associated with a certain amount of charge transfer. A reliable signature of in situ intercalation with alkali metals is the appearance of distinct chemically shifted peaks in the alkali metal core level spectra. These intercalation peaks typically appears at higher energies than the corresponding peaks originating from adsorbed alkali atoms, and they are narrower. The higher energies (i.e. lower binding energies) of the intercalation peaks has tentatively been attributed to more efficient final state screening of intercalated alkali atoms, and the narrow peak-shapes are likely due to more uniform surroundings of intercalated atoms. However, there are no analogous intercalation peaks to be seen in the Ca 2p spectra in figure 4. Still, this does not exclude that intercalation has occurred. Instead, the explanation could simply be that the chemical shift of core level peaks from intercalated Ca atoms, relative to adsorbed species, is just too small to be resolved. One observation in favour of intercalation is the modest increase of Ca 2p emission in figure 4 when going from 0.6nm of Ca to 1.2 nm. If most of the Ca added in that step intercalated, it would be buried under at least one host layer and therefore not contribute as much to the spectral intensity as the Ca adsorbed on the surface. The corresponding valence band spectra in figure 3 shows an increased filling of the Ti 3d band just below the Fermi level, together with dramatic changes in the emission from the lower bands of predominantly Se 4p character. Also these changes are consistent with intercalation of Ca. Looking at the corresponding Se 3d spectra in figure 1, the most striking change seen here is the appearance and growth of peak B. It is therefore reasonable to suggest that peak B is a feature of the intercalation compound Ca x TiSe 2 . The changes in the corresponding Ti 2p spectra are less dramatic, but a moderate increase of asymmetry of both Se 3d and Ti 2p peaks can be attributed to metallic screening by the partially filled Ti 3d band. The third option stated above, a layer-breaking chemical reaction between Ca and TiSe 2 is possibly becoming prominent during the final depositions, going from 1.2 nm of Ca to 2.4 nm and further on to 5 nm. Contrary to the intercalation reaction, this would not remove Ca from the surface, but rather result in the buildup of reaction products, e.g. CaSe and Ti metal, on top of the host material. As such a reaction would break up the TiSe 2 layers, the resulting overlayer would probably become polycrystalline with a large degree of disorder. A strong indication that such a layer-breaking chemical reaction takes place during the final deposition is the occurrence of peak C in figure 1. This peak, at lower energi than peaks A and B, clearly signals the presence of a Se compound different from those giving rise to peaks A and B. The large width of peak C is likely due to the inhomogeneity of such a disordered overlayer. A general broadening of spectral peaks is observed also in the Ti 2p, Ca 2p and valence band spectra, which also can be associated with increased disorder and inhomogeneity at the surface. Finally, there is a shoulder E appearing in the Ti 2p spectra, which possibly could be due to Ti metal or some Ti compound produced in the layer-breaking reaction. It is notable that the Ti 2p intensity at the later stages is diminished more than the Se 3d intensity, which implicates that most of the released Ti atoms are trapped deeper down in the reacted overlayer.
Considering the spectral changes seen in figures 1-4, one arrives at the following scenario: Initially Ca atoms are adsorbed on the TiSe 2 surface, but soon there is an onset of intercalation. As the deposited amount of Ca increases further, the intercalation reaction is replaced by layer-breaking chemical reactions giving rise to a disordered and inhomogeneous overlayer of reaction products. The switch from intercalation to layer-breaking reactions appears to be gradual, as peaks C and E are seen as shoulders also at the earlier stages when intercalation seems to be dominant. The relative importance of the different reactions is likely to depend on both temperature and deposition rate. Figure 5 shows Te 4d spectra for the clean TiTe 2 surface, after one deposition of 5 nm of Ca, and repeated 17 h later, without any further Ca deposition. The obvious effects seen immediately after the deposition are a shift down in energy by 1.4 eV, together with a strong broadening. After the 17 h wait the energy shift relative to the clean surface was reduced to 1.0 eV, but the peaks became even broader. An attempt was made to remove the Te Figure 6. Ti 2p 3/2 core level spectra from TiTe 2 obtained with hν = 530 eV. The uppermost spectrum was measured from the clean TiTe 2 surface, the middle one immediately after a 5 nm Ca deposition, and the bottom one 17 h later. 4d 3/2 contribution in a similar manner as was done for the Se 3d 3/2 contribution in figure 1, but this did not reveal any additional spectral features. Figure 6 shows the corresponding Ti 2p 3/2 spectra. The Ca deposition strongly suppressed the spectral peak, and the maximum of the peak shifted 1.3 eV towards higher energy. This is very similar to what was found in figure 2 after the final Ca deposition on TiSe 2 . The spectrum obtained 17 h later showed only minor changes. Figure 7 shows corresponding valence band spectra. An intense and narrow peak just below the Fermi energy is seen already for the pristine TiTe 2 , because the Ti 3d band is partially filled due to energy overlap with bands of Te 5p character. After the fairly massive Ca deposition the sharp peak at the Fermi level is wiped out in a manner similar to that observed in figure 3 after the final Ca depostion on TiSe 2 . The spectrum obtained 17 h later, shows even less intensity close to the Fermi level, together with some changes in the rather strong emission from the deeper valence bands. Figure 8 shows corresponding Ca 2p spectra, and for comparison again the spectrum measured from a thick Ca film. The Ca 2p peaks found after the deposition are strong, rather symmetric, and have associated energyloss features shifted down in energy 6.0 eV relative to the primary Ca 2p peaks. The spectrum is quite similar to that seen in figure 4 after the final deposition on TiSe 2 , and differs markedly from the spectrum from the thick Ca film. After the 17 h wait, the peaks were shifted upwards by 0.5 eV but otherwise almost unchanged.

Ca/TiTe 2
Since the Ca was deposited on TiTe 2 in one fairly large step, rather than in several smaller steps as for the TiSe 2 case, there is less data to draw detailed conclusions from. This deposition strategy was chosen in order to test the hypothesis that a large deposition possibly could produce a metallic Ca overlayer which would slowly intercalate, or react in some other fashion, in the course of time. Such a behaviour has been reported after deposition of Cu on 1T-TaS 2 [24], but in the present study it is clear that a reacted overlayer is formed quite immediately with only minor changes during the 17 h waiting period. It may still be the case that intercalation is a possible outcome if the Ca deposition is done differently, e.g. in smaller doses or at higher temperatures.

Conclusions
It is quite clear from the above results that Ca reacts with the substrates when deposited on the clevage planes of TiSe 2 and TiTe 2 held at room temperature. One may imagine two fundamentally different types of reactions, either intercalation of Ca between essentially unchanged substrate layers, or layer-breaking reactions in which strong bonds are formed between Ca and Se (or Te) atoms. For depositions corresponding to more than 2 nm of Ca, it appears that the layer-breaking reactions dominated, to produce a disordered overlayer of Ca x Se (or Ca x Te) of unknown stoichiometry with intermixing of Ti atoms. In the Se 3d (or Te 4d) spectra this disordered overlayer was manifested by a broad peak shifted down in energy by about 1.5 eV. The weakness of the Ti 2p emission at this stage indicates that most of the Ti atoms were trapped deeper into the overlayer.
However, in the case of depositions onto TiSe 2 corresponding to less than 2 nm of Ca, there were indications of an intercalation reaction. This supposed intercalation reaction was associated with a Se 3d peak shifted down in energy by 0.4 eV, and charge transfer from Ca to the empty Ti 3d band of pristine TiSe 2 , resulting in an intense peak just below the Fermi level in the valence band spectrum. It is quite possible that smaller Ca depositions onto TiTe 2 also would have resulted in intercalation, given the very similar behaviour of TiSe 2 and TiTe 2 for larger depositions, but this remains to be proven.
From the Ca 2p spectra, it is very clear that Ca deposition on the cleavage planes of TiSe 2 and TiTe 2 at room temperature does not result in formation of metallic Ca overlayers, at least not for the amounts of Ca deposited in this study. Distinct energy-loss features were seen in the Ca 2p spectra about 6 eV below the primary peaks, but a more detailed knowledge of the overlayer composition and structure would be required to fully understand these. Figure 8. Ca 2p core level spectra from TiTe 2 obtained with hν = 430 eV. At the bottom is a background spectrum measured from the clean TiTe 2 surface, while the next two spectra were obtained immediately after a 5 nm Ca deposition, and 17 h later. The spectrum at the very top was recorded from a thick film of Ca metal.