Open Access
Add to BibSonomy
Abstract
Significant effort is being dedicated to developing alternative materials whose optical properties can be controllably and reversibly modified. Here, we experimentally demonstrate the reversible non-volatile molybdenum oxides MoO3-to-MoO2 transition associated to a change from a metallic to a dielectric behavior through cycles of thermal annealing in air and hydrogen (H2). A full cycle is demonstrated by characterizing structurally and optically the transition using Raman spectroscopy and spectroscopic ellipsometry. The potential applicability of the metal-to-insulator transition in MoOx is benchmarked through comparison with a canonical Mott insulator VO2 in a reconfigurable reflective configuration as well as in cladded waveguide schemes.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Optical devices with a reconfigurable response are in focus of ongoing photonic developments. At the basis of this research field is the study of novel materials whose optical properties can be modulated through external stimuli. In the last years, vanadium dioxide (VO2), which shows large modification in physical properties (e.g., refractive index, electrical resistivity) upon external excitation, has been of interest because of its insulator-to-metal transition (IMT) that can be induced by thermal heating (above Tc = 68°C), applied electric fields (E = 105 V/cm), injected carrier density (ne = 1018 cm−3) and optical pulses [1]. The IMT in VO2 leads to a volatile metallic state that, when external excitations are switched off, disappears bringing reversibly back VO2 to its insulating state. The large refractive modulation appearing during this process has been proven in applications related to reconfigurable antennas/metasurfaces [2,3], amplitude modulators integrated in waveguides [4–6], as well as in the dynamic modulation of spontaneous light emission [7–9], among others [1]. Although VO2 shows a large-refractive-index modulation, it also exhibits large extinction coefficient in the visible and near-IR, which directly correlates with undesirable optical losses in photonic devices.
Nevertheless, abrupt changes in electronic and optical properties can also be induced in transition metal oxides by tuning the oxidation states in the transition metal. For instance, molybdenum oxides, MoOx, can exist in a wide variety of different stoichiometric and non-stoichiometric phases due to the multiple valence states of molybdenum ranging from +3 to +6. Among all molybdenum oxides, MoO3 is of special interest due to its wide band gap semiconducting behavior with reported experimental value of the energy band gap of ≈ 3.0 eV [10]. MoO3 is a van der Waals semiconductor with an orthorhombic crystal lattice consisting of double layers of MoO6 octahedra as a framework, and recently is receiving significant attention due to its ability to support highly anisotropic phonon polaritons (PhPs)—infrared (IR) light coupled to lattice vibrations—offering an unprecedented platform for controlling the flow of energy at the nanoscale [11,12]. Conversely, MoO2 crystallizes in a distorted rutile structure and presents a metallic band structure. Its metallicity arises from the presence of Mo 4d orbitals near the Fermi level [13]. The metallicity of MoO2 has already been exploited to generate localized surface plasmon resonances (LSPRs) for surface-enhanced Raman spectroscopy (SERS) applications [13,14] as well as electrode material for applications such as dynamic random-access memory (DRAM) capacitors and lithium ion batteries [15–18]. Therefore, the MoO3-to-MoO2 transition enables an insulator-to-metal transition with the advantage of low-loss behavior of MoO3 in the visible and near-IR spectral region and a non-volatile character. This peculiarities are being exploited in the design of switchable reflective primary color subpixels for application in non-volatile displays based on the MoO3-to-MoO2 transition [19,20].
In this work, we experimentally demonstrate the reversible non-volatile MoO3-to-MoO2 transition reporting a change from a metallic to a dielectric behavior in the dielectric function. The MoO3-to-MoO2 transition has been achieved through cycles of thermal annealing in air and hydrogen (H2) and monitored structurally by Raman spectroscopy. The change of the dielectric function from insulating MoO3 to metallic MoO2 is probed by spectroscopic ellipsometry, from which we derived the refractive index contrast as key metric for the benchmarking of molybdenum oxide for reconfigurable photonics applications.
2. Methods
2.1 Sample fabrication
Amorphous MoO3 films were deposited at room temperature by e-beam evaporation using MoO3 pellets (Pi-KEM 99.99% purity) in a Leybold SYRUS pro 710 on 4-inch Si (100) and 40 nm Pt/5nm Ti/Si (100) substrates in the same run. The nominal thickness of the films was 150 nm and 170 nm. The Si (100) wafers were cleaned prior the deposition using a standard RCA cleaning procedure. The 40nm Pt/5nmTi metal stack was grown by sputtering in a TEMESCAL using Ti and Pt sputtering targets.
The crystallization of the samples was performed by thermal annealing of the films at T= 400°C in air or in H2 with a pressure of 100 mbar.
2.2 Structural and chemical characterization
Films, both as deposited and after annealing cycles, were characterized structurally by Raman spectroscopy (LabRam Horiba set up using a ×100 microscope objective (NA = 0.9) and excitation wavelength of 532 nm) as MoO2 and MoO3 assume different crystalline symmetry groups and Raman selection rules gives different Raman active phonon modes that are unique and specific for the MoO3 orthorhombic structure and MoO2 rutile structure
The stoichiometry of the as-deposited amorphous and crystallized films was checked by X-ray photoelectron spectroscopy (XPS) measurements carried out by a Scanning XPS Microprobe (PHI 5000 Versa Probe II, Physical Electronics) equipped with a monochromatic Al Kα x-ray source (1486.6 eV), with a spot size of 200 µm. Survey (0–1200 eV) and high-resolution spectra were acquired in fixed analyzer transmission (FTA) mode at a pass energy of 117.40 and 29.35 eV, respectively, and at a take-off angle of 45° with respect to the sample surface. Surface charging was compensated using a dual beam charge neutralization system, and the hydrocarbon component of C1s spectrum was used as internal standard for charging correction, and it was fixed at 284.5 eV.
2.3 Optical characterization
Optical properties, namely spectra of the complex pseudodielectric function, $\langle\epsilon\rangle=\langle\epsilon_1\rangle+i\langle\epsilon_2\rangle$, were measured by spectroscopic ellipsometry (UVISEL Horiba) in the photon energy range 0.75–6.5 eV with a resolution of 0.05 eV with an angle of incidence of 70°. The ellipsometry measurements were fit to a three-media substrate/film/air model, where the Si and Pt substrates were experimentally measured prior deposition of MoOx films. The MoO3 layer was parameterized using the Tauc–Lorentz oscillator model and MoO2 with a Tauc–Lorentz oscillator model plus a Drude term [21]. The experimental reflection spectra at 70° were also measured by spectroscopic ellipsometry.
2.3 Reflectance and colorimetry calculations
The reflectance of the multilayer structure was calculated using the Transfer Matrix Method (TMM). TMM allows to calculate the reflectance spectrum of an arbitrary system of homogeneous and non-magnetic multilayers by establishing conditions for the electric field along the boundary of two consecutive media [22].
Reflective color was calculated from experimental and theoretical reflectance spectra. The resulting color also depends on the illuminant and observer. All the color simulations presented here assume standard D65 illuminant and a CIE standard observer. Illuminant D65 corresponds to average daylight, and CIE standard observer represents mean human spectral sensitivity to visible spectrum range [23]. In this paper, CIE1931 space has been chosen because is one of the most accepted to study the resulting color by the reflective displays. The color difference (ΔE) has been quantified using the Lab color space as
2.5 Calculation of transmission through cladded Si3N4 waveguides
Finite-difference time-domain (FDTD) simulations were performed with Ansys Lumerical. The simulated geometry consists of a silicon nitride (Si3N4) waveguide (width = 800 nm and height = 300 nm) on a SiO2 substrate with a MoOx or VO2 patch on top of thickness 20 nm and variable length (see Fig. 5(a)). In the FDTD simulation region, a MODE source is used to inject a guided mode (wavelength λ = 1550 nm) into the waveguide inside the simulation region. From a list of possible modes, the TE mode is selected for injection.
3. Results
3.1 Reversible MoO3-to-MoO2 transition
Figure 1 summarizes the processing path to the reversible MoO3-to-MoO2 transition starting from MoO3-x films grown by e-beam deposition on Si and Pt substrates. After each processing step, the sample was structurally and optically characterized by Raman spectroscopy and spectroscopic ellipsometry. The steps followed to prove the reversible MoO3-to-MoO2 transition are as follows:
- 1. The starting point of experiments consist of 150 nm and 170 nm thick MoO3-x films deposited on Si and Pt substrates. The as deposited films are amorphous as revealed by the absence of peaks in the Raman spectrum as shown in Fig. 1(b). The peaks visible in the Raman spectra at 300 and 520 cm−1 are assigned to the Si substrate.
- 2. After annealing in air at 400°C, the film is crystallized MoO3 as it fully oxidizes following the chemical reaction MoOx + (3-x)/2 O2 → MoO3. This is corroborated by the Raman spectrum in Fig. 1(b), which shows Raman bands at 116, 128, 158, 198, 217, 245, 283, 337, 365, 378, 471, 666, 819 and 995 cm−1 ascribed to MoO3 [24]. The annealing process from amorphous MoOx to crystalline MoO3 entails a change in the optical properties of the film as revealed by the change in the imaginary part, $\left\langle\epsilon_{2}\right\rangle$, of the pseudo dielectric $\left(\langle\epsilon\rangle=\left\langle\epsilon_{1}\right\rangle+i\left\langle\epsilon_{2}\right\rangle\right)$ function in Fig. 1(d).
- 3. Subsequently, the sample was annealed in H2 at 400°C. Upon reduction in H2, the Raman spectrum changed, as shown in Fig. 1(c), revealing peaks at 201, 205, 227, 343, 350, 360, 422, 455, 466, 493, 567, 582 and 737 cm−1 consistent with MoO2 [25]. The thermal annealing in H2 involves the thermal activation of the oxygen desorption according to the reaction MoO3 + H2 → MoO2 + H2O. This chemical reduction of MoO3-to-MoO2 accompanies with a change in $\left\langle\epsilon_{2}\right\rangle$ as reported in Fig. 1(d). The change in $\left\langle\epsilon_{1}\right\rangle$ is shown in Figure S1. Furthermore, from ellipsometry analysis of a MoO3 film (initial thickness 170 nm) before and after the H2 annealing, a ≈ 10 nm (or 6%) expansion of the thickness of the film can be found.
- 4. In order to show a complete cycle and reversibility of the MoO3-to-MoO2 transition, the sample was again annealed in air. The Raman spectrum after this process is consistent with the complete transformation of MoO2 into MoO3 as shown in Fig. 1(c), according to the reaction MoO2 + O2 → MoO3. The reversibility of the process is also visible in the $\left\langle\epsilon_{2}\right\rangle$ spectrum of Fig. 1(d), which matches that of the MoO3 film before its reduction to MoO2 in step #3. Similar behavior is seen in $\left\langle\epsilon_{1}\right\rangle$ as shown in Figure S1.
Fig. 1. (a) Sketch summarizing the path to the reversible MoO3–to-MoO2 transition starting from a 150 nm MoO3-x films on a Si substrate. (b) Raman spectra of the as-deposited MoO3-x before (step #1) and after annealing in air (step #2). With vertical red lines are indicated the MoO3 Raman modes. (c) Raman spectra of the film annealed in H2 (step #3) and subsequent annealing in air (step #4). With vertical red and blue lines are indicated the MoO3 and MoO2 Raman modes. (d) Imaginary part, $\left\langle\epsilon_{2}\right\rangle$, of the pseudodielectric function $\left(\langle\epsilon\rangle=\left\langle\epsilon_{1}\right\rangle+i\left\langle\epsilon_{2}\right\rangle\right)$ at each step (from #1 to #4) of the annealing of the MoOx films of 150 nm deposited on Si. (e) XPS valence band spectra of the MoO3-x film after annealing in air (step #2) and of the film annealed in H2 (step #3).
The metal-to-insulator transition that entails the MoO3-to-MoO2 reversible transformation can also be inferred in the valence band analysis by XPS in Fig. 1(e), showing semiconducting high bandgap MoO3 after annealing in air and metallic states populating the region around the 0 eV (i.e. Fermi level) up to BE of 2 eV for the MoO2 obtained by annealing in H2. The peaks appearing in the XPS valence band spectra at 0.5 and 1.45 eV are in good agreement with values reported in literature (0.60 and 1.60 eV experimental values and 0.50 and 1.35 eV computed values) [26] and are assigned to Mo 4d bands. These peaks are associated with σ- and π-bonding states which are responsible for the metallic behavior of this oxide. In particular, the π bonding is not strong enough to split the t⊥ (π) bands off completely from the rest of the unoccupied Mo 4d bands to give an insulating character to the material [26,27]. In the case of MoO3, the Mo 4d contribution in the XPS valence band spectra around the Fermi level is negligible. Further analysis of the valence band spectrum shows for the case of MoO3 that the highest Mo 4d contribution to the occupied density of states are found in the regions labelled as I, II and III which overlap with the O 2p partial density of states seen as a broad feature at ≈ 6 eV. From a linear extrapolation of the valence band edge to zero, a value of = 2.9 ± 0.1 eV is calculated for the valence band onset. This value is consistent with other values reported in literature as determined by XPS and absorption measurements [26]. Further confirmation MoO3-to-MoO2 transformation is provided by the Mo 3d photoelectron core levels of the MoO3 and MoO2 films obtained after step #1 and step #3 and shown in Figure S2 and whose profiles are consistent with those provided by Scanlon et al [26].
These results are presented as proof-of-concept of the chemical mechanisms and the processing steps to be followed to attain the proposed reversible MoO3-to-MoO2 transformation. Further optimization of the annealing parameters is ongoing. Noteworthy, this innovative way of modulating the optical response of MoO3-to-MoO2 through annealing in air and H2 has similarly been used in VO2 as a new way to modulate its optical response [28].
3.2 Optical characterization of MoO3 and MoO2
Figure 2(a) shows the complex dielectric function of MoO3 and MoO2 extracted from spectroscopic ellipsometry measurements of the films after each step of the processing. MoO3 has been modeled using a Tauc-Lorentz dispersion that allow us to stablish its energy band gap at Eg = 2.7 ± 0.1 eV, in agreement with its semiconducting behavior. This value is consistent with the valence band onset calculated from the XPS measurements. In the case of MoO2, the dielectric function shows strong absorption below 3.0 eV consistent with measurements reported by Chase [29] and associated to quasiparticle absorption resulting from the alteration of the density of states by electronic correlation. The dielectric function has been modeled using a Tauc-Lorentz dispersion with a Drude component to model the metallic behavior at low photon energies. The Drude is justified by the appearance of metallic states around the Fermi level after annealing in H2 as shown in the XPS valence band spectra in Fig. 1(e).
Fig. 2. (a) Refractive index of MoO3 and MoO2 as extracted from the ellipsometry fitting. (b) Refractive index contrast between MoO3 and MoO2. As reference, it is also shown the refractive index contrast of VO2.
The MoO3-to-MoO2 refractive index contrast, defined as Δn = ninsulating – nmetallic and Δk = kinsulating - kmetallic, has been presented as a key metric to evaluate the performance of a material for its application in reconfigurable photonics [30]. The values of Δk are related with different light absorption between metallic and insulating phases (amplitude modulation), whereas Δn is related to the difference in phase shifts (phase modulation). The spectral value of Δn and Δk for MoOx is shown in Fig. 2(b) compared also to the value for VO2 as Ref. [2]. The spectral values of |Δk| are demonstrated to be higher for MoOx above 1 eV with the additional advantage that the insulating phase MoO3 is loss-less as opposed to the lossy behavior of both metallic and insulating VO2. Contrarily, for |Δn| the values of MoOx are almost constant (|Δn| ≈ 0.50-0-75) all over the visible range whereas for VO2 it takes higher and increasing values with decreasing photon energy (|Δn| ≈ 0.5-1.4). Therefore, considering these results, MoOx is a promising material to be integrated in photonic devices for amplitude modulation purposes in the visible and near IR considering its high values of Δk being one of the phases loss-less (i.e., MoO3). For VO2, a similar type of behavior is obtained in the infrared in 2 µm–10 µm range where kinsulating ≈ 0.07 and kmetallic= 5 - 10.
3.3 Reflectance modulation in MoOx thin films
A thin-film material with adjustable complex permittivity can be considered as the simplest configuration to attain amplitude and phase modulators. Therefore, reflectance measurements taken at a 70° angle of incidence using unpolarized light on MoOx films deposited on Si and Pt substrates are shown in Figs. 3(a, c) at each step of the MoO3-to-MoO2 transition showing the good reversibility of the optical response of the films over a full cycle. For the film deposited on Si, Fig. 3(a) shows the initial reflectance spectrum in the MoO3 state shows a dip at 1.35 eV, which is well reversible and reproducible upon a full switching cycle, while the MoO2 reflectance at that photon energy is almost constant at approximately R = 0.38. Upon transformation from MoO3 to MoO2, the dip in the reflectance spectrum shifts to 2.25 eV where the reflectance of MoO3 takes a constant value of R = 0.37. Thus, through the MoO3-to-MoO2 transition, reflectance can be modulated at the two photon energies of 1.35 eV and 2.25 eV simultaneously, with ΔR = 0.35 and ΔR = 0.27 respectively. Figure 3(b) shows that the shift in the dip in the reflectance spectra is again well reproduced by simulated values of the reflectance using the transfer matrix method in a multilayer model that mimics the experimental conditions (see inset in Fig. 3(b)) and using the extracted dielectric functions in Fig. 2(a). Parallel measurements and simulations were run over a system consisting of a MoOx films on a Pt substrate. Figure 3(c) shows the MoO3 reflectance spectrum with two dips at 0.8 and 2.3 eV, which are also well reproducible upon a full transformation cycle. The line shape and dips of the experimental spectra are well reproduced by the simulated spectra in Fig. 3(d) using the transfer matrix method in a multilayer model that mimics the experimental conditions (see inset in Fig. 3(d)) and using the extracted dielectric functions in Fig. 2(a). The slight differences in the amplitude between the simulated and experimental reflectance spectra in Figs. 3(b, d) could be attributed to surface roughness effects of samples, which scatter light in a different direction from the specular one and that are not considered in the calculation.
Fig. 3. Reflectance (a,c) measurements and (b,d) simulation of a MoO3-to-MoO2 reversible cycle for the configurations sketched in (b) ad (d) and consisting of 150 nm and 170 nm MoOx thin films on Si and Pt, respectively, illuminated at 70° angle of incidence.
The metal (MoO2)-to-insulating (MoO3) behavior in molybdenum oxide has been recently proposed as the base for a new generation of reflective displays [19]. The reflective color can be calculated by converting reflectance spectra of the films to color coordinates using color matching functions. Figure 4 shows the color calculated using the experimental and simulated reflectance spectra in Fig. 3. In order to evaluate the reproducibility of the reflective color over the completion of a full cycle, the color difference, ΔE, between calculated colors from the experimental reflectance spectra at the beginning (step #2) and at the end (step #4) of the processing of the films were obtained following Eq. (1). This magnitude is defined as the Euclidian distance between the color coordinates of each state in the Lab color space. Only color differences ΔE > 3 are perceived by the human eye. The calculated color differences are very low, i.e., ΔE = 5.45 and 2.63 for the film on Pt and Si substrates respectively, demonstrating the good reproducibility of the reflective color over a full MoO3-to-MoO2 transformation cycle independently of the substrate. Noteworthy, there is also a good agreement between the color calculated from the experimental and simulated reflectance spectra. The color differences between experimental and simulated color are ΔE = 24.65 and 9.55 for MoO3 (step #2 and #4) on Si and Pt substrates. For MoO2 (step #3), the color differences are ΔE = 15.2 and 8.9 for Si and Pt substrates.
Fig. 4. Comparison between colors calculated from the experimental and simulated reflectance for the sample configuration sketched at the bottom and consisting of (a) a 150 nm MoOx thin film on Si and (b) a 170 nm MoOx thin film on Pt illuminated at the angle of incidence of 70°.
Specifically, in Ref. [19], the authors propose a system consisting of MoOx on a Pt mirror in which “on” CMY (cyan-magenta-yellow) subtractive vivid colors are achieved using insulating MoO3, while pale colors are obtained by metallic MoO2. This effect is herein demonstrated in Fig. 4(b), in which for MoO3 a pinkish vivid color is observed, while for MoO2 the reflective color turns grey. In the HSV (hue-saturation-value) color space, the saturation value drops from S = 22 for the “on” vivid color (i.e. MoO3) to S = 7 for the “off” dull pale color (i.e. MoO2). Interestingly, for both MoOx on Si and MoOx on Pt, after the completion of the cycle, we verified that the initial color of the films is recovered.
Other reconfigurable reflective displays have been proposed in literature based on the prototype phase-change material GST [31,32]. However, because both amorphous and crystalline phases present high losses optical frequencies, the change in the color present a lower contrast in luminosity and chromaticity [32]. Therefore, the advantage of MoOx with respect to GST is that, while for one of the states a vivid color can be achieved, for the other state a low luminosity dull color can be produced. In this way, high chroma and high luminosity “on” pixel states and low chroma and low luminosity “off” pixel states can be generated. This phenomenon relies on the fact that MoO3 is loss-less while MoO2 is highly absorbing.
Given the good agreement between the theoretical modeling and experimental data as well as the potential of MoOx for amplitude modulation photonic platforms, Figs. 5(a)-(c) shows the spectral value of reflectance of MoO2 and MoO3 as function of the molybdenum oxide layer thickness as well as the difference in reflectance, ΔR, to evaluate the potential of the insulator-to-metal transition (IMT) in light amplitude modulation schemes in the visible. In order to maximize the value of ΔR, the reflectance calculations have been performed in the configuration sketched in the inset, i.e., a MoOx layer of variable thickness on a Pt mirror illuminated at normal incidence, yielding values of ΔR in the range 0.40-0.05 in the visible range (1.5 to 3 eV). This configuration has been chosen as we have verified that is experimentally feasible. Nevertheless, higher ΔR could be achieved by choosing a substrate with higher reflectivity (e.g., replacing Pt with silver, Ag, values of ΔR in the range 0.70-0.95 can be achieved, see Figure S3).
Fig. 5. Difference in reflectance, ΔR, between metallic and insulating phases of (a) MoOx and (d) VO2 on a Pt mirror at normal incidence as a function of the layer thickness. Reflectance, R, as a function of the layer thickness for the insulating phases (b) MoO3 and (e) insulating VO2. Reflectance, R, as a function of the layer thickness for the metallic phases (c) MoO2 and (f) metallic VO2.
The reflectance, R, for MoO2 decreases from 0.30 to 0.10 as the photon energy increases, while for MoO3 R goes from approximately 0.7-0.6 to 0.4-0.5. For comparison, Figs. 5(d)-(f) shows equivalent calculations of R and ΔR for the VO2 IMT, using the VO2 dielectric function from Ref. [2]. In the visible, VO2 shows values of ΔR in the 0.1-0.2 range, considerably lower than MoOx. VO2-based thin films configurations yields high reflectance modulations in the IR. For instance, Kats et al. [33] reported reflectivity modulation from 80% to 0.25% at λ = 11.6 µm for VO2 on sapphire substrate, while Butakov et al. [34], reported high reflectivity modulation in the IR by Ge/VO2/sapphire multilayer systems.
Thus, these results suggest that the MIT in MoOx shows a better performance for light modulation in the visible than VO2.
3.4 Amplitude modulation in cladded Si3N4 waveguides
Integration of materials with reconfigurable optical response such as VO2 or chalcogenide phase change materials (PCMs) such as GST or Sb2S3 in photonic integrated circuits (PICs) has led to a wide range of new functionalities ranging from programmable photonic processing, [35,36] optical memories, [37,38] optical switches [5,39] and modulators, [40–43] as well as neuromorphic computing [44]. Generally, VO2 as well as PCMs are integrated in the PICs by depositing small patches of these materials on top of the waveguides. By triggering the IMT of VO2 with an external excitation (i.e., heat, electric field, or light), the guided mode experiences a change in both optical phase and amplitude. For VO2 on Si waveguide modulators, it has been pointed out that insertion loss must be carefully considered due to the non-zero imaginary part (k ≠ 0) of the refractive index of insulating VO2, resulting in optical absorption in the passive state of the modulator [45]. In the case of MoOx, the insulating phase MoO3 presents an extinction coefficient equal to zero (k = 0), thus, resulting in a loss-less passive state. The possibility of integrating MoO3 into a silicon photonics platform has been recently demonstrated [46].
In order to explore the performance of MoOx as material to be integrated in PICs for optical switching applications, we have calculated the change in transmission, ΔT, through a waveguide as function of the length of a 20 nm thick MoOx patch in insulating (MoO3) and metallic states (MoO2) deposited on top of it as sketched in Fig. 6(a). The same simulations are performed for the two phases of VO2 (metallic and insulating). The waveguide (width = 800 nm and height = 300 nm) is made of silicon nitride (Si3N4) deposited on a silicon dioxide substrate (SiO2). Transmission calculations have been performed for a wavelength of λ = 1550 nm (telecom C-band). At this wavelength, the refractive index of the waveguide and the substrate are n =1.99 and n =1.44 respectively. In these simulations, the fundamental electrical mode is selected for the study. As shown in Fig. 6(b), the transmission through the waveguide as function of the length of the cladding layer strongly depends on the optical properties of the cladding material (insulating or metallic VO2 and MoO3 or MoO2).
Fig. 6. (a) Scheme of the hybrid waveguide. A 20 nm thick MoOx or VO2 deposited on top a silicon nitride waveguide over a silicon dioxide substrate. (b) Transmission through a waveguide as a function of the length for the different phases (metallic and insolating of MoOx and VO2). (c) Transmission contrast between the two phases of MoOx and VO2 for different lengths. In (b) and (c) is included equivalent calculations for amorphous and crystalline GST as reference. (d) Mode profile of the fundamental electric mode for the different cladding materials.
The refractive indices of the four phases considered in the calculations are: n = 2.00 + 2.83i (metallic VO2), n = 3.35 + 0.26i (insulating VO2), n = 2.26 + 1.59i (MoO2) and n = 2.19 + 0.00i (MoO3) for the wavelength of 1550 nm. The highest transmission is obtained for MoO3 as k = 0. Nevertheless, it presents low losses because its refractive index is slightly higher than the one of the waveguides (n = 1.99) and total internal reflection (TIR) inside the waveguide is not completely guaranteed. Insulating VO2 results in a lower transmission than MoO3 because of its extinction coefficient k = 0.26 and its real part much higher than the n = 1.99 of the Si3N4 waveguide. Metallic VO2 and MoO2 present low transmission due to its high value of the extinction coefficient, k, consistently with their metallic behavior. The transmission contrast ΔT between metallic and insulating phases of MoOx and VO2 is plotted in Fig. 5(c), showing a higher contrast for MoOx. Specifically, a contrast as a high as 0.8 can be obtained using MoOx as an active element in reconfigurable cladded waveguides. For comparison, in the case VO2, the highest value of ΔT is 0.2, i.e., four times lower than in the case of MoOx. The mode profile of the fundamental electric mode for the different phases is shown in Fig. 6(d).
As reference in Fig. 6(b), (c) is shown the data corresponding to the amorphous-to-crystalline transition of GST, as prototypical system (namorph + ikamorph = 4.352 + 0.060i and ncryst + ikcryst= 7.912 + 1.030i). This material shows a higher ΔT than VO2, but lower than MoOx. Nevertheless, it should be noted that because the switching speed for MoOx is lower and its actuation in this particular configuration is probably more complicated than for GST (annealing in different atmosphere versus only annealing/melt quench), MoOx may not competitive for its integration in reconfigurable PIC architectures as the ones proposed for GST related to programmable photonics [35,47], neuromorphic computing [44,48], non-volatile and rewritable data storage [37,49].
Conclusions
We have demonstrated the reversible non-volatile MoO3-to-MoO2 transition associated to change from a metallic (MoO2) to a dielectric (MoO3) behavior in the dielectric function. Amorphous MoOx-3 on 4 inches silicon wafers were fabricated by e-beam deposition and subsequently crystallized into MoO3 by annealing in air. MoO3-to-MoO2 non-volatile transition has been achieved reversibly through cycles of annealing in hydrogen and air. The analysis of the refractive index contrast between both phases and its comparison with analogous values for canonical insulating-to-metal transition material VO2, reveals the superior performance of MoOx for amplitude modulation photonic platforms at the C telecom band (1550 nm) as well as for reflectivity modulators in the visible range.
Funding
European Union's Horizon 2020 research and innovation program (No 899598 – PHEMTRONICS).
Disclosures
The authors declare no conflict of interests.
Data availability
The refractive index of MoO2 and MoO3 are available at [50]
Supplemental document
See Supplement 1 for supporting content.
References
1. S. Cueff, J. John, Z. Zhang, J. Parra, J. Sun, R. Orobtchouk, S. Ramanathan, and P. Sanchis, “VO2 nanophotonics,” APL Photonics 5(11), 110901 (2020). [CrossRef]
2. J. John, Y. Gutierrez, Z. Zhang, H. Karl, S. Ramanathan, R. Orobtchouk, F. Moreno, and S. Cueff, “Multipolar Resonances with Designer Tunability Using VO2 Phase-Change Materials,” Phys. Rev. Appl. 13(4), 044053 (2020). [CrossRef]
3. A. Tripathi, J. John, S. Kruk, Z. Zhang, H. S. Nguyen, L. Berguiga, P. R. Romeo, R. Orobtchouk, S. Ramanathan, Y. Kivshar, and S. Cueff, “Tunable Mie-Resonant Dielectric Metasurfaces Based on VO2 Phase-Transition Materials,” ACS Photonics 8(4), 1206–1213 (2021). [CrossRef]
4. R. M. Briggs, I. M. Pryce, and H. A. Atwater, “Compact silicon photonic waveguide modulator based on the vanadium dioxide metal-insulator phase transition,” Opt. Express 18(11), 11192 (2010). [CrossRef]
5. K. J. Miller, K. A. Hallman, R. F. Haglund, and S. M. Weiss, “Silicon waveguide optical switch with embedded phase change material,” Opt. Express 25(22), 26527 (2017). [CrossRef]
6. I. Olivares, L. Sánchez, J. Parra, R. Larrea, A. Griol, M. Menghini, P. Homm, L.-W. Jang, B. van Bilzen, J. W. Seo, J.-P. Locquet, and P. Sanchis, “Optical switching in hybrid VO2/Si waveguides thermally triggered by lateral microheaters,” Opt. Express 26(10), 12387 (2018). [CrossRef]
7. S. Cueff, D. Li, Y. Zhou, F. J. Wong, J. A. Kurvits, S. Ramanathan, and R. Zia, “Dynamic control of light emission faster than the lifetime limit using VO2 phase-change,” Nat. Commun. 6(1), 8636 (2015). [CrossRef]
8. E. Petronijevic, M. Centini, T. Cesca, G. Mattei, F. A. Bovino, and C. Sibilia, “Control of Au nanoantenna emission enhancement of magnetic dipolar emitters by means of VO2 phase change layers,” Opt. Express 27(17), 24260 (2019). [CrossRef]
9. P. K. Jha, H. Akbari, Y. Kim, S. Biswas, and H. A. Atwater, “Nanoscale axial position and orientation measurement of hexagonal boron nitride quantum emitters using a tunable nanophotonic environment,” Nanotechnology 33(1), 015001 (2022). [CrossRef]
10. T. He and J. Yao, “Photochromism of molybdenum oxide,” J. Photochem. Photobiol., C 4(2), 125–143 (2003). [CrossRef]
11. G. Álvarez-Pérez, T. G. Folland, I. Errea, J. Taboada-Gutiérrez, J. Duan, J. Martín-Sánchez, A. I. F. Tresguerres-Mata, J. R. Matson, A. Bylinkin, M. He, W. Ma, Q. Bao, J. I. Martín, J. D. Caldwell, A. Y. Nikitin, and P. Alonso-González, “Infrared Permittivity of the Biaxial van der Waals Semiconductor α-MoO3 from Near- and Far-Field Correlative Studies,” Adv. Mater. 32(29), 1908176 (2020). [CrossRef]
12. M. Chen, X. Lin, T. H. Dinh, Z. Zheng, J. Shen, Q. Ma, H. Chen, P. Jarillo-Herrero, and S. Dai, “Configurable phonon polaritons in twisted α-MoO3,” Nat. Mater. 19(12), 1307–1311 (2020). [CrossRef]
13. Q. Zhang, X. Li, Q. Ma, Q. Zhang, H. Bai, W. Yi, J. Liu, J. Han, and G. Xi, “A metallic molybdenum dioxide with high stability for surface enhanced Raman spectroscopy,” Nat. Commun. 8(1), 14903 (2017). [CrossRef]
14. Y. Zhan, Y. Liu, H. Zu, Y. Guo, S. Wu, H. Yang, Z. Liu, B. Lei, J. Zhuang, X. Zhang, D. Huang, and C. Hu, “Phase-controlled synthesis of molybdenum oxide nanoparticles for surface enhanced Raman scattering and photothermal therapy,” Nanoscale 10(13), 5997–6004 (2018). [CrossRef]
15. W. Lee, C. J. Cho, W. C. Lee, C. S. Hwang, R. P. H. Chang, and S. K. Kim, “MoO2 as a thermally stable oxide electrode for dynamic random-access memory capacitors,” J. Mater. Chem. C 6(48), 13250–13256 (2018). [CrossRef]
16. Y. Shi, B. Guo, S. A. Corr, Q. Shi, Y.-S. Hu, K. R. Heier, L. Chen, R. Seshadri, and G. D. Stucky, “Ordered Mesoporous Metallic MoO2 Materials with Highly Reversible Lithium Storage Capacity,” Nano Lett. 9(12), 4215–4220 (2009). [CrossRef]
17. B. Guo, X. Fang, B. Li, Y. Shi, C. Ouyang, Y.-S. Hu, Z. Wang, G. D. Stucky, and L. Chen, “Synthesis and Lithium Storage Mechanism of Ultrafine MoO2 Nanorods,” Chem. Mater. 24(3), 457–463 (2012). [CrossRef]
18. L. C. Yang, Q. S. Gao, Y. Tang, Y. P. Wu, and R. Holze, “MoO2 synthesized by reduction of MoO3 with ethanol vapor as an anode material with good rate capability for the lithium ion battery,” J. Power Sources 179(1), 357–360 (2008). [CrossRef]
19. G. Santos, F. Gonzalez, D. Ortiz, J. M. Saiz, M. Losurdo, F. Moreno, and Y. Gutierrez, “Dynamic reflective color pixels based on molybdenum oxide,” Opt. Express 29(13), 19417 (2021). [CrossRef]
20. G. Santos, F. González, D. Ortiz, J. M. Saiz, M. Losurdo, Y. Gutiérrez, and F. Moreno, “Design of Switchable On/Off Subpixels for Primary Color Generation Based on Molybdenum Oxide Gratings,” Physics 3(3), 655–663 (2021). [CrossRef]
21. G. E. Jellison and F. A. Modine, “Parameterization of the optical functions of amorphous materials in the interband region,” Appl. Phys. Lett. 69(3), 371–373 (1996). [CrossRef]
22. M. Born, E. Wolf, A. B. Bhatia, P. C. Clemmow, D. Gabor, A. R. Stokes, A. M. Taylor, P. A. Wayman, and W. L. Wilcock, Principles of Optics (Cambridge University Press, 1999).
23. J. Schanda, Colorimetry: Understanding the CIE System (John Wiley & Sons, Inc., 2007).
24. M. Dieterle and G. Mestl, “Raman spectroscopy of molybdenum oxides,” Phys. Chem. Chem. Phys. 4(5), 822–826 (2002). [CrossRef]
25. M. A. Camacho-López, L. Escobar-Alarcón, M. Picquart, R. Arroyo, G. Córdoba, and E. Haro-Poniatowski, “Micro-Raman study of the m-MoO2 to α- MoO3 transformation induced by cw-laser irradiation,” Opt. Mater. 33(3), 480–484 (2011). [CrossRef]
26. D. O. Scanlon, G. W. Watson, D. J. Payne, G. R. Atkinson, R. G. Egdell, and D. S. L. Law, “Theoretical and Experimental Study of the Electronic Structures of MoO3 and MoO2,” J. Phys. Chem. C 114(10), 4636–4645 (2010). [CrossRef]
27. V. Eyert, R. Horny, K.-H. Höck, and S. Horn, “Embedded Peierls instability and the electronic structure of MoO2,” J. Phys.: Condens. Matter 12(23), 4923–4946 (2000). [CrossRef]
28. X. Duan, S. T. White, Y. Cui, F. Neubrech, Y. Gao, R. F. Haglund, and N. Liu, “Reconfigurable Multistate Optical Systems Enabled by VO2 Phase Transitions,” ACS Photonics 7(11), 2958–2965 (2020). [CrossRef]
29. L. L. Chase, “Optical properties of CrO2 and MoO2 from 0.1 to 6 eV,” Phys. Rev. B 10(6), 2226–2231 (1974). [CrossRef]
30. S. Abdollahramezani, O. Hemmatyar, H. Taghinejad, A. Krasnok, Y. Kiarashinejad, M. Zandehshahvar, A. Alu, and A. Adibi, “Tunable nanophotonics enabled by chalcogenide phase-change materials,” Nanophotonics 9(5), 1189–1241 (2020). [CrossRef]
31. P. Hosseini, C. D. Wright, and H. Bhaskaran, “An optoelectronic framework enabled by low-dimensional phase-change films,” Nature 511(7508), 206–211 (2014). [CrossRef]
32. B. Broughton, L. Bandhu, C. Talagrand, S. Garcia-Castillo, M. Yang, H. Bhaskaran, and P. Hosseini, “38-4: Solid-State Reflective Displays (SRD ®) Utilizing Ultrathin Phase-Change Materials,” SID Symp. Dig. Tech. Pap. 48(1), 546–549 (2017). [CrossRef]
33. M. A. Kats, D. Sharma, J. Lin, P. Genevet, R. Blanchard, Z. Yang, M. M. Qazilbash, D. N. Basov, S. Ramanathan, and F. Capasso, “Ultra-thin perfect absorber employing a tunable phase change material,” Appl. Phys. Lett. 101(22), 221101 (2012). [CrossRef]
34. N. A. Butakov, M. W. Knight, T. Lewi, P. P. Iyer, D. Higgs, H. T. Chorsi, J. Trastoy, J. Del Valle Granda, I. Valmianski, C. Urban, Y. Kalcheim, P. Y. Wang, P. W. C. Hon, I. K. Schuller, and J. A. Schuller, “Broadband Electrically Tunable Dielectric Resonators Using Metal–Insulator Transitions,” ACS Photonics 5(10), 4056–4060 (2018). [CrossRef]
35. M. Stegmaier, C. Ríos, H. Bhaskaran, C. D. Wright, and W. H. P. Pernice, “Nonvolatile All-Optical 1 × 2 Switch for Chipscale Photonic Networks,” Adv. Opt. Mater. 5(1), 1600346 (2017). [CrossRef]
36. M. Delaney, I. Zeimpekis, D. Lawson, D. W. Hewak, and O. L. Muskens, “A New Family of Ultralow Loss Reversible Phase-Change Materials for Photonic Integrated Circuits: Sb2S3 and Sb2Se3,” Adv. Funct. Mater. 30(36), 2002447 (2020). [CrossRef]
37. C. Ríos, M. Stegmaier, P. Hosseini, D. Wang, T. Scherer, C. D. Wright, H. Bhaskaran, and W. H. P. Pernice, “Integrated all-photonic non-volatile multi-level memory,” Nat. Photonics 9(11), 725–732 (2015). [CrossRef]
38. Y. Jung, H. Han, A. Sharma, J. Jeong, S. S. P. Parkin, and J. K. S. Poon, “Integrated Hybrid VO2 –Silicon Optical Memory,” ACS Photonics 9(1), 217–223 (2022). [CrossRef]
39. A. Joushaghani, J. Jeong, S. Paradis, D. Alain, J. Stewart Aitchison, and J. K. S. Poon, “Wavelength-size hybrid Si-VO2 waveguide electroabsorption optical switches and photodetectors,” Opt. Express 23(3), 3657 (2015). [CrossRef]
40. H. M. K. Wong, Z. Yan, K. A. Hallman, R. E. Marvel, R. P. Prasankumar, R. F. Haglund, and A. S. Helmy, “Broadband, Integrated, Micron-Scale, All-Optical Si3N4/VO2 Modulators with pJ Switching Energy,” ACS Photonics 6(11), 2734–2740 (2019). [CrossRef]
41. V. Jeyaselvan, A. Pal, P. S. Anil Kumar, and S. K. Selvaraja, “Thermally-induced optical modulation in a vanadium dioxide-on-silicon waveguide,” OSA Continuum 3(1), 132 (2020). [CrossRef]
42. Y. Gutiérrez, A. P. Ovvyan, G. Santos, D. Juan, S. A. Rosales, J. Junquera, P. García-Fernández, S. Dicorato, M. M. Giangregorio, E. Dilonardo, F. Palumbo, M. Modreanu, J. Resl, O. Ishchenko, G. Garry, T. Jonuzi, M. Georghe, C. Cobianu, K. Hingerl, C. Cobet, F. Moreno, W. H. P. Pernice, and M. Losurdo, “Interlaboratory study on Sb2S3 interplay between structure, dielectric function, and amorphous-to-crystalline phase change for photonics,” iScience 25(6), 104377 (2022). [CrossRef]
43. J. Faneca, I. Zeimpekis, S. T. Ilie, T. D. Bucio, K. Grabska, D. W. Hewak, and F. Y. Gardes, “Towards low loss non-volatile phase change materials in mid index waveguides,” Neuromorph. Comput. Eng. 1(1), 014004 (2021). [CrossRef]
44. J. Feldmann, N. Youngblood, C. D. Wright, H. Bhaskaran, and W. H. P. Pernice, “All-optical spiking neurosynaptic networks with self-learning capabilities,” Nature 569(7755), 208–214 (2019). [CrossRef]
45. K. J. Miller, R. F. Haglund, and S. M. Weiss, “Optical phase change materials in integrated silicon photonic devices: review,” Opt. Mater. Express 8(8), 2415 (2018). [CrossRef]
46. G. Ren, B. Y. Zhang, A. Mitchell, and J. Z. Ou, “A Novel 2D Plasmonic MoO3 Driven pH Sensor on Silicon Photonics Platform,” in 2019 Photonics & Electromagnetics Research Symposium - Fall (PIERS - Fall) (IEEE, 2019), pp. 2544–2555.
47. M. Delaney, I. Zeimpekis, H. Du, X. Yan, M. Banakar, D. J. Thomson, D. W. Hewak, and O. L. Muskens, “Nonvolatile programmable silicon photonics using an ultralow-loss Sb2Se3, phase change material,” Sci. Adv. 7(25), 1 (2021). [CrossRef]
48. X. Chen, Y. Xue, Y. Sun, J. Shen, S. Song, M. Zhu, Z. Song, Z. Cheng, and P. Zhou, “Neuromorphic Photonic Memory Devices Using Ultrafast, Non-volatile Phase-change Materials,” Adv. Mater. 2203909 (2022). [CrossRef]
49. M. Wuttig and N. Yamada, “Phase-change materials for rewriteable data storage,” Nat. Mater. 6(11), 824–832 (2007). [CrossRef]
50. Y. Gutierrez, G. Santos, F. Palumbo, M. Modreanu, F. Moreno, and M. Losurdo, “Data Set-Optical Properties of MoO3 and MoO2,” Zenodo, 2022, https://zenodo.org/record/6592425.
