1. Introduction

Being indirect bandgap semiconductors, bulk transition metal dichalcogenide (TMD) crystals [1,2], in contrast to monolayers [3–5], do not possess luminescent properties. However, the situation changes qualitatively when halogen molecules intercalate into the van der Waals (vdW) gap of these layered compounds [6,7]. These diatomic molecules are unambiguously embedded in the quasi-tetrahedral voids of the vdW gap, forming neutral centres, with properties similar to isoelectronic impurities that provide efficient radiative recombination in some indirect-gap semiconductors [8,9]. Previously, we studied the luminescence of bound excitons in molybdenum and tungsten disulfide crystals intercalated by halogens (Cl₂, Br₂ and I₂) [7,10–12]. In this paper, we present the results of studying the steady–state luminescence spectra of bulk 2H-MoSe₂:I₂ crystals intercalated with iodine. In order to interpret the phonon replicas that form the spectra of the observed photoluminescence (PL), we also studied the Raman scattering spectra of this compound.

2. Experimental

The 2H-MoSe₂:I₂ single crystals were grown from primary components by means of chemical vapour transport (CVT) in a two-zone tube furnace, using iodine as a transport agent. The crystallization chamber temperature at the growth zone was maintained at a level of 1050°C. The ampoules were held inside the furnace for a period of up to 10 days, after which they were slowly cooled to room temperature. The grown layered bulk crystals were then exfoliated to obtain smooth van der Waals surfaces. The thickness of each crystalline sample was of an order of a few tens of microns with diameters varying from 3 to 5 mm.

The photoluminescence (PL) measurements in the temperature range T = 11-100 K were performed with a grating monochromator using an IR Hamamatsu PMT and a standard lock-in detection technique. The steady-state PL was excited using a CW laser with wavelength 447 nm (2.77 eV). The spectral resolution during measurements was 0.17 meV (1.4 cm⁻¹). The intensity of the exciting radiation did not exceed a few mW/cm² to avoid local heating of the sample, which strongly affects the temperature dependence of the observed spectra.

Raman spectra of the same 2H-MoSe₂:I₂ samples, on which the luminescent properties have been studied, were measured at room temperature, using two wavelengths for excitation (532 nm and 785 nm).

3. Results and discussion

Figure 1 shows the evolution of the PL spectral shape of the MoSe₂:I₂ crystal as a function of temperature in the range T = 11-55 K. The structure and temperature behaviour of the spectra is similar to the characteristics of the recombination emission spectra of bound excitons observed earlier for MoS₂:Cl₂ [7,11] and WS₂:Br₂ [10,11]. However, since the dimensions of the iodine molecule somewhat exceed the dimensions of the quasi-tetrahedral voids in the van der Waals gap [6,11], where halogen molecules are embedded during the growth of the MoSe₂ crystal, the observed exciton lines are much wider than in the case of excitons bound to Cl₂ or Br₂ in layered MoS₂ or WS₂ crystals. At low temperatures there are two zero-phonon excitonic lines (ZPLs) labelled as A and B, that dominate the spectra, followed by their phonon replicas (Fig. 1). The energies of the spectral maxima of the ZPLs are E_A = 1.0360 eV and E_B = 1.0416 eV, which is approximately by 0.1 eV less than the width of the indirect band gap of molybdenum diselenide $E_g^{ind} = $ 1.10 eV [13]. The energies of observed ZPL and phonon replica spectral peaks are presented in Table 1.

 

Fig. 1. Steady-state luminescence spectra of the 2H-MoSe₂:I₂ layered crystals at different temperatures.

Download Full Size | PDF

Table 1. Energies of observed exciton lines and of phonon replica peaks and their interpretation.

View Table

With the increase of temperature, the intensities of A and B lines (${I_A}\; $ and ${I_B}$), separated by 5.6 meV (45 cm⁻¹) and located at E_A = 1.0360 eV and E_B = 1.0416 eV, are redistributed in favour of the short wavelength line B. A similar redistribution of intensities was also observed for 2H-MoS₂:Cl₂ [7] and 2H-WS₂:Br₂ [10], and is due to the thermal population of the excitonic state with a higher energy B, the rate of radiative recombination of which is much faster than that of the state A. For two excited levels, the energy gap between which is ${\mathrm{\Delta }_{AB}}$, in thermodynamic equilibrium the temperature dependence of the intensity ratio ${I_B}(T )/{I_A}(T )$ can be described by the formula

In Fig. 2 along with the experimental temperature dependence of the ratio $[{I_B}(T )/{I_A}(T )]$^exp, there is a curve calculated by Eq. (1), using the value of the multiplier $({{g_B}/{g_A}} )({{\tau_A}/{\tau_B}} )= 76$ calculated for ${\mathrm{\Delta }_{AB}}$=5.6 meV and the experimental value $[{I_B}/{I_A}]$^exp = 0.172 at T = 11 K. Thus, as in the case of MoS₂ and WS₂ intercalated by halogen molecules, the lifetime of the short-wavelength B-state of the bound exciton is much shorter than that of the A-state. And just like for the above-mentioned TMD compounds, an increase in temperature above 60 K leads to the very fast thermal quenching of the luminescence with an activation energy of E_qx = 0.12 eV, which occurs due to the extrinsic self-trapping of the electron component of the bound exciton [12,14–16].

 

Fig. 2. Experimental values of the intensity of the spectral lines ratio ${I_B}(T )/{I_A}(T )$. On the insert – two levels of the exciton bound on iodine molecule embedded in the MoSe₂ wan der Waals gap.

Download Full Size | PDF

Along with the observed dramatic change in the ${I_B}(T )/{I_A}(T )$ intensity ratio with increasing temperature, the intensities of the phonon replica peaks are also redistributed, which made it possible to identify the origin of these replicas. Figure 3 shows expanded PL spectra at two temperatures (11 K and 35 K), with the beginning of the frequency scale in cm⁻¹ aligned with the position of the maximum of the short-wavelength zero-phonon line B. Below, on the same frequency scale, the Raman spectra of the same sample obtained at excitation by radiation with wavelengths of 532 nm and 785 nm are also shown. The interpretation of the spectra, which follows from the comparative analysis of the positions of the phonon replica’s maxima at two temperatures, is presented in Table 1.

 

Fig. 3. PL spectra at temperatures of 11 K (a) and 35 K (b), and Raman spectra at room temperature and two excitation wavelengths of 532 nm (c) and 785 nm (d) of a MoSe₂:I₂ single crystal. In figures (a) and (b), along with the frequencies identifying the peaks of phonon replicas, the ZPL, from which this replica originates, is indicated in parentheses.

Download Full Size | PDF

From the data given in the table and Fig. 3(a,b) it follows that the spectrum of phonon repetitions is formed due to two phonons with frequencies ν_ph1 = 144 cm⁻¹, ν_ph2 = 190 cm⁻¹ and vibrational modes with the sum frequencies of these phonons. Due to the fact that the distance between the lines of the A-B exciton doublet is close to the frequency difference of these phonons (ν_ph2–ν_ph1 = 46 cm⁻¹), peaks with frequencies ν₂ = 190 cm⁻¹ and ν₅ = 336 cm⁻¹ correspond to superpositions of replicas of both ZPLs.

The Raman spectra measured at room temperature on the same sample as the PL spectra (Fig. 3(c, d)) are similar to the spectra obtained with excitation at the same two wavelengths in [17], as well as in [18], where the excitation was carried out using infrared radiation, resonant with MoSe₂ indirect band gap. It can be seen, that the frequency ν_ph1 coincides with the frequency of the second-order Raman mode attributed, according to [17], to the sum of E_1g and LA branches at the M point of the Brillouin zone. As for the second phonon frequency ν_ph2, in the previously studied Raman spectra of 2H-MoSe₂ crystals [17–19], no lines at the frequency 190 cm⁻¹ were observed.

Thus, it can be assumed that a phonon with a frequency of ν_ph2 = 190 cm⁻¹ corresponds to a local mode caused by vibrations of iodine molecules embedded in the van der Waals gap of a layered host crystal.

4. Conclusion

As in the case of other layered indirect-gap TMD semiconductor compounds, iodine intercalated into MoSe₂ crystals imparts them luminescence properties at low temperature, arising from the recombination of excitons bound on neutral halogen molecules. It is noteworthy that, in all TMD compounds, the spectra of this luminescence always consist of narrow lines, which indicates the unambiguous arrangement of diatomic molecules in the van der Waals gap of the crystals. It should be noted that the authors of [20] recently observed a similar exciton luminescence in the HfS₂ crystals intercalated with iodine. This means that in this representative of layered TMD materials there is also a size matching between halogen molecules and corresponding interstitial sites, which provides an ordered intercalation of iodine into the host lattice.

Another noteworthy circumstance is that the dimensions of the vdW quasi-tetrahedral cavities, where diatomic halogen molecules can be intercalated, are nearly identical for all (!) TMD compounds (WS₂, WSe₂, MoS₂, MoSe₂ etc.). In fact, this provides an opportunity to embed halogen not only between monolayers of one compound but also in the interface between the layers of different TMD compounds forming, for example, van der Waals heterojunction. By selecting a halogen molecule during the synthesis of these heterojunctions, one can practically eliminate the distortion of the layered structure, thus creating an opportunity to control and modify the electronic properties of the heterojunction. Technologically, such a procedure can be done by Chemical Vapour Deposition or Atomic Layer Deposition methods.