Enhanced epsilon-near-zero structures for photonics

. We present an experimental realization of a novel layered metamaterial we label enhanced epsilon-near-zero (eENZ). The structure is a stack of alternating thin films made of ENZ– and dielectric material and it can be designed for desired refractive/reflective properties by appropriately tuning the film thicknesses. The structure supports thin film resonances, guided modes and Ferrel-Berreman plasmon modes and the performance of the structure shows a large improvement to many currently available bulk ENZ materials. Additionally, we recently demonstrated the possible use of eENZ for coherence switching in lasers [1]. We demonstrate the design, fabrication and characterization of the optical properties of the eENZ stack and compare the measured transmission properties with transfer matrix method (TMM) simulations.


Introduction
ENZ-materials are materials in which the real part of the dielectric permittivity becomes zero at a certain frequency [2].This feature gives rise to unusual field dynamics, which can be utilized in engineering unconventional structures.The ENZ effect can occur in Drude-dispersive materials near the plasma frequency, which can have a wide range of values depending on the free carrier concentration of the material.For instance, some metals exhibit ENZ in the UV-region, transparent conducting oxides (TCO's) in the NIR-region and some polaritonic materials in the THz regime.On the other hand, ENZ can be realized by means of an artificial metamaterial, by creating composite materials, or engineering the structural parameters to yield an effective permittivity near zero [3].One of the main issues of bulk ENZ materials are the losses associated with the rather large imaginary part of the permittivity.Therefore, creating ENZ metamaterial devices with mitigated losses is important for realizing the full potential of the phenomenon.In this work, we fabricate and characterize a layered metamaterial that consists of alternating ENZ-and high index dielectric thin films.The structure was previously coined as enhanced ENZ (eENZ), and it may have interesting applications such as enhancing the directionality of light [4], or control of the coherence of light [1].

Design and fabrication
The sample schematic is depicted in Fig. 1.It is a bilayer thin film stack with alternating ENZ and dielectric films, with permittivities of ε1 and ε2, and film thicknesses of d1 and d2, respectively.The incident and transmission region permittivities are denoted as εi and εt, respectively.The ENZ material is indium tin oxide (ITO), which exhibits ENZ in the NIR-region and has been used in many experimental ENZ studies [5,6].The advantage of ITO is the easy tunability of the ENZ wavelength by changing the doping level or by altering the deposition process parameters such as substrate temperature, Ar and O2 gas pressures in sputtering and by annealing [7].The dielectric material is TiO2 which has a high refractive index and low dispersion in the NIR-region.Figs. 2 a) and b) display the measured real-and imaginary parts of the permittivity for 80 nm thick ITO films deposited by RF magnetron sputtering with different substrate temperatures.As the substrate temperature increases, the ENZ wavelength blueshifts.These wavelengths are 1454, 1382 and 1260 nm for 250, 300 and 350 ○ C, respectively.In addition, in Fig. 2 a), the real permittivity of a TiO2 film deposited by electron beam evaporation in 350 ○ C is displayed, demonstrating the negligible dispersion compared to the ITO films.

Characterization
The transmittance of the sample was measured for a wavelength range of 500 -1700 nm and varying angle of incidence in the range of 0-60 deg.The results were collected for both TE-and TM-polarized incident light and they are displayed in the Fig. 4 middle column for TE (upper) and TM (lower).The spectra show multiple resonances below 1200 nm and the amount of these resonances is proportional to the number of dielectric layers in the stack.In the NIR region the transmittance drops essentially to zero between 1200-1500 nm.The difference between TE- and TM-polarizations can be seen at around 1500 nm, where the transmittance drops to zero for TM-polarization at angles of incidence above 20 degrees.This is due to Ferrel-Berreman plasmonic modes in the structure getting excited by the TM-polarized input only.This polarization dependent response could be utilized for instance in coherence switching of lasers [4].The optical constants and film thicknesses of the structure were measured by variable angle spectroscopic ellipsometry (VASE) and they were used to simulate the transmission spectra with transfer matrix method (TMM).The simulation results are displayed in the left column of Fig.

Fig. 2 .
Fig. 2. a) Measured real permittivity of ITO for different substrate temperatures, indicated by the line style and of TiO2 deposited at 350 ○ C. b) Imaginary part of the permittivity of the deposited ITO films.The losses of the ITO films at the ENZ wavelength slightly decrease as the substrate temperature increases, from 0.52 for 250 ○ C to 0.46 for 350 ○ C. The eENZ sample had 15 layers with 8 ITO and 7 TiO2 films, the top and bottom film being ITO.The sample was fabricated on a single deposition run with Kurt J. Lesker CO.Lab 18 thin film deposition system at 350 ○ C substrate temperature.ITO films were sputtered at 3 mTorr Ar gas pressure and the TiO2 films were evaporated from TiO tablets at 0.2 mTorr partial O2 pressure.The designed film thicknesses were 65 nm for ITO and 225 nm for TiO2.SEM image and a photograph of the sample are displayed in the Fig.3.

Fig. 3 .
Fig. 3. Cross sectional SEM image and a photograph of the fabricated eENZ sample.

Fig. 4 .
Fig. 4. Simulated (left column) and measured (center column) transmittance of the eENZ stack as a function of angle of incidence.The error between simulated and experimental values (error = simulated -experimental) are shown in the rightmost column.Results for TE-and TM-polarized input are displayed in the top and bottom rows, respectively.
4. and they show excellent agreement with the measured data.The error between simulated and experimental values are shown in the right column of Fig. 4.