Patterning of electrically tunable light-emitting photonic structures demonstrating bound states in the continuum

The ability to manipulate and confine light attracts considerable interest in both the scientific and the technological realms. By amplifying nonlinear phenomena, light confinement opens new possibilities for diverse applications, such as optical communications, lasers, biological sensing, energy conversion, and more. Typically, wave localization is achieved when appropriate outgoing waves either do not exist or are forbidden due to symmetry incompatibility. The first exception to this rule was proposed in 1929 by von Neumann and Wigner.¹ This early work postulated the existence of a quantum potential that could trap electrons with energy that would normally couple to outgoing waves. Since then the concept of bound states in the continuum (BICs)—states that lie inside the continuous spectral range spanned by the propagating waves but that remain perfectly confined, without any radiation—has been extended to a general wave phenomenon and has been identified in electromagnetic waves, acoustic waves, water waves, and elastic waves.^2,3 Unlike quantum systems, within photonic structures material and geometry are intrinsically decoupled, making them a flexible platform for BICs. While BICs have not yet been experimentally observed in quantum systems, various photonic implementations have been demonstrated, such as lasers,^4,5 chemical and biological sensors,^6,7 filters,⁸ and optofluidic setups.⁹ 

As recently shown, one effective way to achieve light confinement is through the patterning of a thin dielectric slab with a square array of cylindrical holes.¹⁰ Calculations predict that these photonic crystals (PhCs) will exhibit very high quality resonance bands, with modes approaching infinite lifetime, and up to 6 orders of magnitude field enhancement.¹¹ Therefore, these structures should demonstrate amplification capabilities comparable to confined plasmonic systems but without the drawbacks of metallic losses, e.g., high absorption at optical frequencies and heating effects. The BIC amplification would also be extended over the entire PhC surface rather than circumscribed to a small region.

One way to verify this hypothesis is to look at the nonlinear interaction between the BIC mode and a resonant emitting object. This measurement is complicated by the perturbation of the photonic structure, potentially capable of suppressing the BIC mode altogether, introduced by any entity in close proximity to the PhC. As an alternative, patterning of an optically active material allows us to study the impact of the BIC mode on the material emission with no distortion of the PhC properties.

In this work we developed a scalable process to pattern optically active and electrically conductive erbium-doped zinc oxide (Er:ZnO) deposited by magnetron sputtering.^12,13 The geometric parameters have been optimized through numerical simulations with the goal of obtaining a BIC mode with frequency matching the erbium (Er) emission. This material is of particular interest because it enables characterization of the interaction between the Er emission and the BIC mode, and, by virtue of being a wide band gap semiconductor, allows to study of the effects of polarization on the BIC. For all the advantages of Er:ZnO, one major limitation is the difficulty of patterning the photonic structure with conventional nanolithographic approaches. Electron beam lithography and plasma etching through a polymeric mask resulted in an unacceptable increase of the surface roughness, from an as-deposited root mean square (RMS) roughness of 0.5 nm to a postprocessing RMS roughness of 25 nm. Light scattering from such a rough surface converts the bound states into radiating resonant states with finite lifetimes.¹⁴ Exposure to UV radiation during plasma etching is the primary cause of the observed Er:ZnO roughening. The development of a patterning process involving a metallic mask successfully shields the material from UV light and yields patterned PhCs with a RMS roughness of 4.1 nm. Optical characterization of these structures demonstrates BIC resonance close to the expected frequency of 1540 nm.

Er:ZnO thin films were deposited using RF magnetron sputtering at room temperature in a Denton Discovery 18 confocal-target system with sputtering targets Er (99.99% purity) and ZnO (99.99% purity). Films were deposited on fused silica substrates (500 μm thickness) in argon atmosphere (Ar flow 12 sccm) at base pressure 1.0 × 10⁻⁷Torr. The sputtering powers for the Er and ZnO targets were 15 and 250 W, respectively. Postdeposition annealing was performed in a thermal furnace at 900 °C, for 60 min, in oxygen atmosphere. The final Er:ZnO film thickness was 175 ± 5 nm, measured by spectroscopic ellipsometry and surface profilometry.

The Er:ZnO thin film was patterned with electron-beam lithography (Vistec VB300) and inductively coupled plasma (ICP) etching (Oxford Instruments PlasmaPro System 100). Two pattern transfer techniques were compared: a conventional approach using the e-beam resist as etching mask, and an alternative approach that involves the deposition of a metallic mask. As shown in Fig. 1, the sample was first covered with a 3 nm layer of SiO₂ deposited by plasma assisted atomic layer deposition (ALD) at 40 °C (Oxford Instruments FlexAL). This layer was necessary to protect the Er:ZnO from dissolution by amyl acetate, which is used to develop the e-beam resist required in subsequent processing steps, as described below. In the metallic mask process, a ∼150 nm poly(methyl methacrylate) (PMMA) layer was spin-coated on the ALD SiO₂ layer followed by a 10 nm Cr layer deposited by electron-beam evaporation (Semicore). A ∼500 nm layer of ZEP520A (Zeon) was spin-coated on the Cr and baked at 100 °C. In the conventional process the ZEP layer was deposited directly on top of the ALD SiO₂. Finally, the conductive polymer aquaSAVE (Mitsubishi Chemical Corporation) was spin-coated on the ZEP to serve as discharge layer during the e-beam exposure. The Cr mask was ICP etched using a mixture of Cl₂ and O₂ (process conditions: Cl₂ 35 sccm, O₂ 5 sccm, 5 mTorr, RF power 5 mW, ICP power 500 mW, and 20 °C). The PMMA was removed with O₂ reactive ion etching (RIE) (Oxford Instruments Plasmalab80Plus, O₂ 50 sccm, 50 mTorr, and 40 mW). The thin ALD SiO₂ was removed with CHF₃ and O₂ RIE (process conditions: CHF₃ 50 sccm, O₂ 5 sccm, 55 mTorr, 100 mW, and 20 °C). Er:ZnO was ICP etched using a mixture of CH₄, H₂, and Ar (process conditions: CH₄ 3 sccm, H₂ 8 sccm, Ar 5 sccm, 2 mTorr, RF power 200 mW, ICP power 500 mW, and 20 °C). The PMMA layer allowed for the Cr mask removal by lift-off in an ultrasonic bath of Remover PG (MicroChem) for 10 min. In the case of the conventional process, only the last two etching steps were performed: the thin ALD SiO₂ was removed with CHF₃ and O₂ RIE and the Er:ZnO was ICP etched using a mixture of CH₄, H₂, and Ar as described earlier. Any residual ZEP was removed in Remover PG.

Room-temperature photoluminescence (PL) measurements of annealed Er:ZnO thin films were performed using an Ar ion laser (Spectra Physics, 177–602) pumped at 488 nm. For the time-resolved PL measurements, the laser beam was modulated by a mechanical chopper at a frequency of 60 Hz. The PL signal was detected through a monochromator (Oriel Cornerstone 260) by an IR-sensitive photomultiplier tube (Hamamatsu R5509–73).

Surface/thin-film metrology was conducted on the following equipment. Scanning electron microscopy (SEM) was performed with a Zeiss Ultra 60 electron microscope. Etching rates were characterized with a Horiba UVISEL spectroscopic ellipsometer. Atomic force microscopy (AFM) was conducted with a Veeco Nanoscope 5.

PhCs were optically characterized in transmission. The sample was mounted on an automatic rotational stage that allowed for the careful adjustment of the angle of incidence. The sample was excited with a supercontinuum laser (SuperK Extreme, NKT Photonics), and the transmitted intensity was acquired by an optical spectrum analyzer (ANDO AQ6317B).

The numerical study of the excited modes of the photonic structure was performed with a fully tridimensional rigorous coupled wave approach (RCWA) based on a Fourier modal expansion.¹⁵ The simulated structure was an array of circular holes of radius r/a = 0.25 arranged in a square lattice, where a is the lattice constant. Figure 2 shows the band structure calculated around the Γ point. A band with vanishing width was identified at the normalized frequency ω_n = 0.687 (ω_n = a/λ) above the principal band of the PhC. This extremely narrow band corresponds to a BIC mode, which is characterized by an infinite lifetime.

In order to obtain a BIC mode with a resonant wavelength close to 1540 nm in our patterned Er:ZnO structure, the design parameters were set to lattice parameter a = 1035 nm and radius r = 259 nm.

Er:ZnO thin films were characterized after deposition, prior to patterning. As shown in Fig. 3, excitation at 488 nm produced room-temperature PL in the IR range with peak intensity at ca. 1550 nm. The PL lifetime was determined to be 3.8 ms by fitting the peak intensity as a function of time to an exponential function (inset in Fig. 3).

AFM imaging revealed a smooth surface [RMS roughness 0.5 nm, Fig. 4(a)].

Hollow cylindrical features arranged in a square lattice were patterned by e-beam lithography and plasma etching. Geometric parameters (radius = 260 nm and lattice constant = 1035 nm) were determined from our simulations to be optimized conditions to produce a BIC mode at 1540 nm at normal incidence (Fig. 2).

The etching process was adapted from previously reported ICP recipes for ZnO.^16–18 The etch rate was measured with ellipsometry and found to be 25 nm/min.^16–18 

Etching through a polymer mask (the e-beam resist ZEP) resulted in a surface roughness increase to a RMS value of 25 nm (Fig. 5). Light scattering at the surface resulted in blurring of the PhC resonances and converted the bound states into radiating resonant states with finite lifetimes.¹⁴ The reason for this change in the surface morphology was found to be the exposure to UV radiation. This effect was previously observed for ZnO, and was confirmed by exposing the as-deposited material to UV-ozone radiation (in a conventional table top tool).¹⁶ 

This initial etching attempt also showed a further problem with Er:ZnO processing; the slow dissolution of Er:ZnO in ZEP developer (amyl acetate) resulted in the ZEP mask peeling off during development. In order to circumvent this issue without considerably altering the Er:ZnO optical properties, a thin layer of SiO₂ (about 3 nm thick) was deposited by plasma assisted ALD. The ALD process involved short oxygen plasma pulses. The effect of the plasma pulses on surface roughness was assessed by AFM. The roughness was found to be fairly constant with a post-ALD RMS value of 2.2 nm [Fig. 4(b)]. Use of a thin SiO₂ layer is thus an effective method for protecting the Er:ZnO surface during ZEP development.

A chromium mask was used to avert surface roughening during plasma etching. Direct Cr deposition on the Er:ZnO would preclude its removal after, as ZnO has been demonstrated to be soluble in acidic solutions.^19,20 The protective SiO₂ layer could not prevent the Cr etchant from completely dissolving the Er:ZnO, so a sacrificial PMMA layer was deposited underneath the Cr mask (Fig. 1). The PMMA layer allowed for the lift-off of the Cr mask in a Remover PG ultrasonic bath without any damage to the patterned layer.

Use of this PMMA-Cr stack required modification of the conventional procedure for ZEP deposition. The multilayer cracked if baked at 180 °C resulting in a highly defective PhC. To prevent this problem, post-spin-coating ZEP baking was done at 100 °C. No cracking was observed, and ZEP patterning proceeded without any change in resolution (the process was tested on a Si substrate). Following this strategy, after e-beam exposure and development, the pattern was transferred to the e-beam evaporated Cr layer with an ICP etching process using Cl₂ (etch rate 20 nm/min).²¹ The underlying PMMA layer was removed with O₂ RIE and the 3 nm SiO₂ layer was removed with CHF₃-O₂ RIE. The pattern transfer was then completed with 7 min of CH₄-H₂-Ar ICP etching. The resulting film was characterized by SEM and AFM. As shown in Fig. 6, a modest roughness increase was observed (RMS value 4.1 nm), considerably smaller than surface roughness resulting from etching through a polymer mask (RMS roughness 25 nm).

Optical characterization of the Er:ZnO PhC patterned with a Cr mask demonstrated a very sharp peak (FWHM ∼ 0.8 nm) at 1496 nm at normal incidence (Fig. 7) corresponding to the BIC mode predicted by our simulations (Fig. 2). The peak width is affected by light scattering due to surface roughness. This result constitutes a considerable improvement over the sample patterned with a polymer mask, which did not demonstrate any narrow BIC peak because of excessive light scattering. Any further reduction of the sample surface roughness, and patterning imperfections in general, would result in a reduced FWHM of the BIC peak.

We have demonstrated a scalable process to pattern optically active and electrically conductive thin films of Er:ZnO deposited by magnetron sputtering. The process is based on e-beam lithography and ICP etching. Roughening of the Er:ZnO surface upon exposure to the UV radiation generated during plasma etching required the use of a Cr mask for pattern transfer and its removal with a lift-off process in organic solvents that do not attack the Er:ZnO film. This approach allowed for the successful patterning of PhCs made of Er:ZnO with a modest roughening of the surface, from a RMS value of 0.5 nm for the as-deposited film, to a RMS value of 4.1 nm after patterning. Samples patterned with the metallic mask demonstrated the expected photonic behavior, displaying a narrow resonance at 1495 nm, which, according to our simulations, corresponds to a BIC mode.

The authors thank the Molecular Foundry clean room staff Simone Sassolini and Michael Elowson for technical support. Thanks to Deirdre Olynick for advice on plasma etching and to Adam Schwartzberg for advice on ALD. Work at the Molecular Foundry was supported by the Office of Science, Office of Basic Energy Sciences of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. L.D.N. would like to thank the support of the NSF-EAGER program under Award No. ECCS 1643118.