Engineered contacts and contact interfaces for stable and efficient DC electro-optic tuning

Abstract

An optical device with engineered contacts and contact interfaces for stable and efficient de electro-optic tuning. The device comprises a waveguide, comprising an electro¬ optic (EO) material having a first side and a second side, the first side of the EO material comprising at least one modified region; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on the first side of the substrate; and an electrode disposed in the at least one modified region of the EO material, wherein the at least one modified region is a chemically reduced region or a heightened conductivity region.

Description

HCU-07125 (HU9580, HU9581) ENGINEERED CONTACTS AND CONTACT INTERFACES FOR STABLE AND EFFICIENT DC ELECTRO-OPTIC TUNING

[0001] This invention was made with government support under 2137723 awarded by National Science Foundation (NSF) and under HR00112000137 awarded by U.S. Department of Defense / Defense Advanced Research Projects Agency (DOD / DARPA) and under W911N1F1810432 awarded by U.S. Army Research Office (ARO). The government has certain rights in this invention. CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of priority to U.S. Provisional App. No. 63 / 602,164, filed on November 22, 2023. The entire contents of which are hereby incorporated by reference. BACKGROUND

[0002] Embodiments of the present disclosure relate to engineered contacts and contact interfaces for stable and efficient dc electro-optic tuning. BRIEF SUMMARY

[0003] In a first example embodiment, the present invention is a device, comprising a waveguide, comprising an electro-optic (EO) material having a first side and a second side, the first side of the EO material comprising at least one modified region; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on the first side of the substrate; and an electrode disposed in the at least one modified region of theHCU-07125 (HU9580, HU9581) EO material, wherein the at least one modified region is a chemically reduced region or a heightened conductivity region.

[0004] In another example embodiment, the present invention is a device, comprising a waveguide, comprising an alkali metal ion-containing electro-optic (alkali EO) material having a first side and a second side; a layer of an alkali ion-blocking material disposed on at least one of the first side and the second side of the alkali EO material; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on or adjacent to the first side of the substrate.

[0005] In another example embodiment, the present invention is a method of manufacturing a device, comprising providing a waveguide comprising an electrooptical (EO) material having a first side and a second side, depositing a resist over a portion of the first side of the EO material; chemically reducing at least one region of the EO material, thereby producing at least one reduced region; optionally, removing the resist; and depositing at least one electrode on the at least one reduced region.

[0006] In another example embodiments, the present invention is a method of manufacturing a device, comprising providing a waveguide comprising an electrooptical (EO) material having a first side and a second side, subjecting at least one region of the EO material to a treatment that produces a heightened conductivity region, thereby producing at least one heightened conductivity region; and depositing at least one electrode on the at least one reduced region. BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0007] Fig.1A is a qualitative plot of an EO response versus time of a device according to the present disclosure.HCU-07125 (HU9580, HU9581)

[0008] Fig.1B is a qualitative plot of an EO reponse versus voltage of a device according to the present disclosure.

[0009] Fig.1C is a schematic view of a device annotated with phenomenological sources of dc bias drift according to the present disclosure.

[0010] Fig.1D is a schematic view of a double ring experiment used to monitor dc drift according to the present disclosure.

[0011] Fig.2A is a schematic view of a contact treatment procedure for a device.

[0012] Fig.2B is a plot of EO response versus time of an exemplary device according to the present disclosure.

[0013] Fig.2C is a plot of change in splitting versus voltage of an exemplary device according to the present disclosure with plasma treated contacts.

[0014] Fig.2D is a plot of current versus voltage of an exemplary device according to the present disclosure with plasma treated contacts.

[0015] Fig.2E is a plot of change in splitting versus voltage of an exemplary device according to the present disclosure without etched contacts.

[0016] Fig.2F is a plot of current versus voltage of an exemplary device according to the present disclosure without etched contacts.

[0017] Fig.3A is a plot of a raw transmission spectrum versus time of an exemplary device according to the present disclosure.

[0018] Fig.3B is a plot of splitting versus time of an exemplary device according to the present disclosure.

[0019] Fig.3C is a plot of representative optical mode splittings corresponding to a drive signal waveform used to measure EO voltage dependence according to the present disclosure.

[0020] Fig.3D is a plot of a representative drive signal waveform used to measure EO voltage dependence according to the present disclosure.HCU-07125 (HU9580, HU9581)

[0021] Fig.3E is a plot of splitting versus voltage for different EO voltage dependencies according to the present disclosure.

[0022] Fig.4A is a plot of splitting versus voltage of an exemplary device according to the present disclosure with annealing and air-cladding.

[0023] Fig.4B is a plot of splitting versus voltage of an exemplary device according to the present disclosure with an annealed, air-cladded, and C3F8-etched structure.

[0024] Fig.4C is a plot of splitting versus voltage of an exemplary device according to the present disclosure with LN and oxide annealed, oxide-cladded, and Ar-etched structure.

[0025] Fig.4D is a plot of EO response versus time of an exemplary device according to the present disclosure with annealing and air-cladding.

[0026] Fig.4E is a plot of EO response versus time of an exemplary device according to the present disclosure with an annealed, air-cladded, and C3F8-etched structure.

[0027] Fig.4F is a plot of EO response versus time of an exemplary device according to the present disclosure with LN and oxide annealed, oxide-cladded, and Ar-etched structure.

[0028] Fig.5 is a plot of EO response versus time of exemplary devices according to the present disclosure with annealing and air-cladding.

[0029] Fig.6 is a plot of change in splitting versus voltage of exemplary devices according to the present disclosure with annealing and air-cladding.

[0030] Fig.7A is a schematic view of a device cross-section with equivalent contact-limited circuit overlaid according to the present disclosure.

[0031] Fig.7B is a schematic view of interface band-bending with and without applied voltage according to the present disclosure.

[0032] Fig.7C is a plot of high-resolution XPS Nb 3d spectra of an annealed device according to the present disclosure.HCU-07125 (HU9580, HU9581)

[0033] Fig.7D is a plot of high-resolution XPS Nb 3d spectra of an Ar-etched TFLN device according to the present disclosure.

[0034] Fig.7E is a plot of high-resolution XPS O 1s spectra of an annealed device according to the present disclosure.

[0035] Fig.7F is a plot of high-resolution XPS O 1s spectra of an Ar-etched TFLN device according to the present disclosure.

[0036] Fig.8A is a plot of high-resolution Li 1s XPS spectra of an annealed device according to the present disclosure.

[0037] Fig.8B is a plot of high-resolution Li 1s XPS spectra of an Ar-etched TFLN device according to the present disclosure.

[0038] Fig.8C is a plot of intensity versus binding energy of an annealed device after the application of chemical cleaning solutions according to the present disclosure.

[0039] Fig.8D is a plot of the corresponding Li / Nb ratios of an annealed device after the application of chemical cleaning solutions according to the present disclosure.

[0040] Fig.8E-8G are plots of XPS spectra without background subtraction of the Li 1s / Nb 4s, Nb 3d, and O 1s lines in Ar-etched TFLN after annealing or cleaning in SC-1, respectively, according to the present disclosure.

[0041] Fig.8H is a plot of XPS spectra of the F 1s line after etching in a C3F8-containing reactive ion etch chemistry according to the present disclosure.

[0042] Fig.8I is a plot of XPS spectra of the F 1s line after exposure to hydrofluoric acid according to the present disclosure.

[0043] Fig.9A is a waterfall plot of Nb 3d XPS line with increasing etch time in a 4 keV single ion milling configuration according to the present disclosure.

[0044] Fig.9B is a plot of the corresponding reduced fraction of total niobium concentration [Nbx] / [Nb] as a function of etch time according to the present disclosure.HCU-07125 (HU9580, HU9581)

[0045] Figs.9C-9E are plots of Nb 3d XPS lines after in-situ cluster ion milling with 8 keV across 75 ions, 8 keV across 2000 ions, and 2 keV across 2000 ions, respectively, according to the present disclosure.

[0046] Fig.10A is a schematic view of a circuit for I-V measurements on blanket-processed TFLN, without any waveguide patterning by e-beam lithography according to the present disclosure.

[0047] Fig.10B is a plot of current versus voltage for several devices according to the present disclosure.

[0048] Fig.11A is a schematic view of nanofabrication flows for oxide-cladded TFLN EO modulators according to the present disclosure.

[0049] Figs.11B-11E is a plot of measured dc drift and multi-exponential fits for various devices according to the present disclosure.

[0050] Fig.11F is a plot of SIMS measurement of cladded LN samples according to the present disclosure.

[0051] Fig.12A is a schematic view of a circuit for a RC-ladder network supporting a single time constant according to the present disclosure.

[0052] Fig.12B is a schematic view of a circuit a RC-ladder network supporting two time constants according to the present disclosure.

[0053] Figs.13A-13D are plots of EO voltage dependence for TFLN modulators with SiO2 top cladding for various devices according to the present disclosure.

[0054] Fig.14A shows images of a cladded TFLN modulator cross-section that has experienced the oxide anneal according to the present disclosure.

[0055] Fig.14B is a plot of intensity profile vs depth extracted from the scanning transmission electron microscopy imaging according to the present disclosure.HCU-07125 (HU9580, HU9581)

[0056] Fig.15 is a plot of SIMS traces showing the Li distribution in oxide-cladded samples according to the present disclosure.

[0057] Fig.16A is a plot of frequency versus detuning of simulated anti-crossing in coupled ring resonators according to the present disclosure.

[0058] Fig.16B is a plot of splitting versus detuning in coupled ring resonators according to the present disclosure.

[0059] Fig.17 is a plot of measured current versus voltage of exemplary devices with electrodes of different materials according to the present disclosure.

[0060] Fig.18A is a fabrication process flow after waveguide formation, including cladding, metallization, thermal processing, and etching of the cladding oxide according to the present disclosure.

[0061] Fig.18B is a plot of EO response versus time for different electrode and processing conditions according to the present disclosure.

[0062] Fig.18C is a Bode plot showing the frequency dependence of the EO response for Mach Zehnder interferometers at their quadrature point according to the present disclosure.

[0063] Fig.19A is a plot of EO response versus time after various thermal treatment steps according to the present disclosure.

[0064] Fig.19B is a plot of the normalized EO response versus frequency for Mach Zehnder interferometers that have undergone various thermal treatments according to the present disclosure.

[0065] Fig.19C is a plot of normalized Li intensity versus sputter time for cladded lithium niobate samples according to the present disclosure.

[0066] Figs.20-33 are schematic views of exemplary devices according to the present disclosure.HCU-07125 (HU9580, HU9581) DETAILED DESCRIPTION

[0067] In certain materials, the linear electro-optic effect (often referred to as the Pockels effect) allows direct control of refractive index with an applied electric field. In practice, material inhomogeneities and defects can cause the bias field to drift over macroscopic time scales. Such phenomena can degrade the reliability and stability of integrated electro-optic devices. Systematic measurements of dc electro-optic response, electronic transport, and elemental composition were combined to examine the microscopic origins of dc instability in nanofabricated thin-film lithium niobate (TFLN) devices. Two key sources of instability were identified, including: the metal-LN (lithium niobate) contact interface where Schottky barriers impede charge injection, and the oxide-LN cladding interface where lithium diffusion occurs.

[0068] The ability to connect electronics and optics underpins optical communications and interconnects. Material phenomena that enable voltage-controlled optical modulation form a broad class of electro-refractive and electro-absorptive effects, including the Pockels, Franz- Keldysh, and free-carrier dispersion effects. Electronic and photonic properties may be considered together when designing such devices. This approach has been implemented in traditional semiconductor platforms such as silicon and III-V semiconductors, achieving low drive voltages and ultra-compact footprints. However, for long-haul communications requiring low-loss and high-bandwidth operation, Pockels electro-optic (EO) modulators based on wide-bandgap materials such as LN have long been the preferred choice. Because it is fundamentally rooted in the non-centrosymmetry of a crystal, the Pockels EO effect exhibits a linear, non-absorptive refractive index shift. Correspondingly, EO materials are generally understood not to suffer from many of the material limitations intrinsic to other modulation schemes, such as free-carrier absorption. Such claims have been bolstered by rapid advances in high-performance and compact Pockels modulators, largely driven byHCU-07125 (HU9580, HU9581) nanofabrication breakthroughs on the thin-film LN platform that have enabled electrodes to be placed closer to the optical waveguide, resulting in efficient overlap between the electric and optical fields. The regime of strong EO materials has also made accessible new functionalities, including low-power reconfigurable photonic circuits, broadband frequency combs, ultrafast pulse generation and manipulation, and on-chip spectrometers. Coupled with the wafer-scale availability of TFLN, entire electro-optical systems can be miniaturized at scale onto a wafer die using standard nanofabrication techniques.

[0069] Many of the above-mentioned applications rely on the ability to control the operating point of the EO device, which can be accomplished by applying bias dc fields. The EO response is expected to quickly settle to a constant value after application of a step voltage input, as well as exhibit high linearity with applied voltage (Fig.1A-1B). In particular, Fig. 1A is a schematic view showing a comparison of an ideal stable step response to a realistic response that decays. Fig.1B is a schematic view showing comparison of an ideal linear voltage dependence to a nonideal EO response with a threshold. In practice, drift of the dc operating point degrades the reliability of the EO devices over time and thresholding of the EO response prevents simple calibration of EO tuning, resulting in complex locking schemes. In conventional LN modulators based on low-confinement waveguides formed using proton exchange or Ti-indiffusion, drift rates can be minimized to feedback-acceptable levels. However, thin-film structures have received limited attention as a distinct format with unique material challenges, compounded by the expanded role of surfaces and interfaces. Indeed, leakage current is relegated to a nonideality, rather than an appreciable electronic response– an intrinsic consequence of charge carrier transport.

[0070] It is shown that the de facto simplification of TFLN as an ideal insulator during device design has critically limited the ability to rationally engineer devices and systems with minimized EO instability. Like any real crystal, TFLN has defects that can move, react, andHCU-07125 (HU9580, HU9581) trap charge carriers injected into the material—effects intrinsically temporal in nature (Fig. 1C). Referring to Fig.1C, a schematic view of an exemplary device 100 having a pair of electrodes 108a, 108b, an EO material 110, and a substrate 112 is shown with annotated phenomenological sources of dc bias drift. For example, charge (de)trapping may occur at interfaces 102, 104 with a characteristic timescale (^^, ^^), or ion migration can slowly screen applied fields (^^). Schottky barriers (SB) at contacts 106 (between electrodes 108a, 108b and EO material 110) can complicate charge redistribution by impeding charge injection and collection. No different from a semiconductor device explicitly using free carriers, the electronic and optical responses of an EO device must be designed for in tandem. The linearity and stability of the dc EO response was investigated in congruent TFLN nanophotonic modulators at telecommunication wavelengths, in response to a wide range of chemical and thermal processing conditions. Correlating device performance changes with measurements of the material structure and electronic properties, aa microscopic understanding of how the metal-LN contact interface influences thin-film electro-optic response was developed. It was found that while the EO performance of a pristine and untreated TFLN surface was highly sensitive to small variations in the nanofabrication process flow, plasma-induced damage arising during reactive ion etching can improve contact linearity. To identify different sources of dc drift throughout the entire device stack, this quasi-linear contact was applied to quantitatively study charge-trapping dynamics in oxide- cladded TFLN devices and investigate the influence of thermal annealing. While annealing is commonly used to improve device performance, it was shown such processing can also induce lithium diffusion at the oxide-LN cladding interface that degrades temporal flatness at long time scales (approximately hours). Specific electronic interfaces for targeted electro- optic engineering were identified.HCU-07125 (HU9580, HU9581)

[0071] Referring to Fig.1D, a schematic view of a double ring experiment used to monitor dc drift is shown. Air-cladded devices fabricated using standard nanofabrication processes for x-cut TFLN modulators were examined (shown in the inset of Fig.1D). An applied step voltage induced a phase shift, measured by the change in optical mode splitting. An exemplary device having a pair of electrodes 108a, 108b, an EO material 110, and a substrate 112 is shown in inset 114. The inset 114 shows an electrode cross-section for a baseline annealed, air-cladded device, where “BOX” refers to buried oxide. The baseline process has no intentional contact treatment beyond a brief O2plasma photoresist descum. Samples were fabricated from commercial ion-sliced x-cut thin-film lithium niobate on insulator wafers (NanoLN) with 600 nm LN, 4.7 µm thermally grown buried silicon dioxide, and 525 µm high resistivity silicon. All samples were cleaned with acetone and isopropanol prior to further processing. Electron-beam lithography with hydrogen silsesquioxane (HSQ) resist followed by Ar+-based reactive ion etching (300 nm etch depth) was used to pattern the waveguides. After patterning waveguides with electron-beam lithography and Ar-ion etching, redeposited material and residual resist arising from the dry etch process were removed with a series of wet chemical cleans. Silicon nitride deposited by PECVD was used as the HSQ adhesion layer to piranha-cleaned TFLN. The waveguide width tapered to 3 µm for ring resonators. After etching, a wet chemical process involving the RCA clean (SC-1—5:1:1 deionized water: 29 wt % NH4OH: 30 wt % H2O2 at 57 ºC), hydrofluoric acid, and piranha (3:1 H2SO4: H2O2) was used to clean the resist and redeposited material. As measured by reflectometry, this chemical cleaning process removed the top 6-10 nm of the TFLN surface. A two-hour anneal at 520 ºC in an O2 atmosphere may follow (“LN anneal”). Annealing in an oxygen environment at 520 ºC can reduce optical loss and photorefractive contributions associated with damage from ion slicing used to realize TFLN, as well as electron beam irradiation. For air-cladded devices, electrode deposition followed. Otherwise, the device was clad with 900HCU-07125 (HU9580, HU9581) nm of SiO2deposited using plasma-enhanced chemical vapor deposition (high-frequency PECVD, 13.56 MHz). A separate two-hour anneal (“oxide anneal”) with identical conditions to the LN anneal may then be included to improve oxide quality.

[0072] Photolithography, Ti / Au evaporation, and lift-off were then used to define contacts to TFLN. Electrode locations were defined using photolithography with a bilayer resist stack (LOR-3A, SPR-700-1.0) intended for metal lift-off in 80 ºC Remover PG. The electrode gap was 7.5 µm edge-to-edge for rings. For air-cladded devices, the contact region may undergo additional treatment with photoresist-compatible processes, using the bilayer resist stack as an etch mask. The baseline samples underwent a gentle O2 plasma cleaning (75W, 20 sccm) to clean off residual resist after developing, otherwise the contact regions were treated with a brief etch in either a C3F8+Ar+gas chemistry (20 seconds) or a pure Ar+etch (30 seconds). These processes etched away the top 20-25 nm of the TFLN contact region. The electrodes were then immediately deposited with electron-beam evaporation (15 nm Ti, 200 nm Au, base pressure <2E-6 torr).

[0073] If employing a top-cladded geometry, additional steps were taken prior to metallization to ensure intimate contact of the electrode with TFLN. Direct electrode contact has been used to greatly improve low-frequency roll-off by circumventing leakage through the oxide buffer. Reactive ion etching with a C3F8chemistry was used to create windows with near vertical sidewalls in the PECVD cladding that exposed the lithium niobate surface. If the etch terminated with the fluorine chemistry, the sample was referred to as “fluorine- terminated”, otherwise if the etch was switched to a purely physical Ar+plasma etch at the end it was referred to as “Ar-terminated.” This can create a SiO2 hard mask that allows for a wider range of possible chemical contact treatments not possible with photoresist, which may necessitate chemical stripping of the photoresist. In practice, precise alignment of a duplicateHCU-07125 (HU9580, HU9581) optical lithography step to redefine the electrode region for lift-off can be optimized, enabling more effective use of a self-aligned etch and lift-off mask.

[0074] Photolithography was used to define the electrodes (10 mm long, 150 µm wide) before 15 nm / 200nm of Ti / Au were deposited with electron-beam evaporation. Annealing and blanket surface treatments occured prior to electrode deposition. Etch chemistries (C3F8, Ar) were investigated with a 30 s blanket etch. RCA cleaning after Ar etching was done for 10 minutes. XPS and SIMS samples were produced in an identical manner to electrical test samples for the relevant thermal / chemical processing steps, without the addition of electrode patterning and deposition. The I-V characteristics in (Figs.2D, 2F) were measured on Mach- Zehnder modulators patterned using the corresponding full photonic device fabrication process. In this case, the waveguide width tapered to 1.2 µm in the electrode region, which has an edge-to-edge gap of 5.5 µm. I-V characteristics were taken using a Signatone probe station attached to a semiconductor parameter analyzer (Agilent B4156C). The delay time was set between 1 and 3 seconds to account for the settling time incurred by the high impedance device load. To counteract the large initial displacement current if the sweep were to start at a large negative(positive) voltage, the sweeps were instead conducted starting from 0 V, for example from 0 V to 15 V, then 0 V to -15 V.

[0075] Time-domain measurements of the electro-optic response were conducted by monitoring the transmission spectra of a double ring device with a tunable C-band laser (TSL-510) and 125 MHz photoreceiver (NewFocus 1811) and extracting the optical mode splitting as a function of time using a simple peak-finding algorithm (Figs.3A-3E). A laser light was vertically coupled into and out of the device using grating couplers with insertion loss approximately 5 dB per facet. In between transmission spectra acquisitions, the laser was maintained at a non-resonant wavelength. With a low probe laser power, (approximately 10 µW in the waveguide), photorefractive contributions may be negligible at telecommunicationHCU-07125 (HU9580, HU9581) wavelengths. This was verified experimentally by comparing the transmission spectra when the laser wavelength is swept, thereby confirming that the characteristic photorefractive resonance narrowing and broadening, respectively, does not occur.

[0076] Referring to Fig.3A, raw transmission spectra as a function of time for an examplary device is shown. Fig.3A shows resonance movement on the unbiased reference resonance(s), likely stemming from room temperature changes (thermo-optic drift).

[0077] Referring to Fig.3B, the extracted EO step response showing this extrinsic common- mode drift can be cancelled out. For illustrative purposes the higher wavelength reference resonance was used; this makes no difference in the normalized EO response since reference resonances only differ by multiples of the free-spectral range. At each transition point, a few samples are averaged to extract the splitting. The initial samples before any voltage is applied are taken as the initial splitting.

[0078] Referring to Figs.3C-3D, the concept of measuring EO voltage dependence is shown. In particular, Fig.3D shows a representative drive signal waveform, showing interleaved periods of 0 V to attempt to compensate for memory effects. Fig.3C shows the corresponding optical mode splittings.

[0079] Referring to Fig. 3E, different EO voltage dependences are shown when sampling at the start or end of a voltage application period, using either a single point estimate or an average (for example, in Fig.3E averaging is over 5 points).

[0080] The step voltage was applied to a single ring with a ground-signal probe, sourced from a DC power supply (Keithley 2200). Immediately after the voltage was applied, the optical transmission was sampled every approximately 1.6 s for 60 samples to capture fast rise time dynamics. After this time, the sampling rate was decreased to once every approximately 10.6 s for the remainder of the experiment. When the experiment concluded, the device was held atHCU-07125 (HU9580, HU9581) 0 V to discharge the device. The temporal response changed dramatically after the first long measurement, and therefore the initial measurement of the given device are shown herein.

[0081] The short-time voltage dependence was measured using the same experimental setup. The sampling rate was once every approximately 1.6 s, and the applied voltage waveform followed a staircase pattern interleaved with periods of 0 V to discharge the device and more accurately capture the step response. Each period was 30 samples long. After extracting the splitting as a function of time, the voltage dependence was extracted by averaging a few samples at the start or end of the applied voltage period (Fig.3C). The initial samples before any voltage was applied was taken as the initial V=0 V splitting, due to level shifts that may occur after voltage application (Figs.4A-4C). Measurements were performed before and after each long-term drift measurement. Unless otherwise specified, EO voltage dependence was extracted using a 5-point average immediately after applying voltage to best correspond with I-V measurement delay times.

[0082] Referring to Figs.4A-4C, EO voltage dependence before and after dc drift measurement is shown for baseline air-cladded and annealed (Fig.4A); (E) air-cladded, annealed, and C3F8-etched (Fig.4B); and oxide-cladded, LN+oxide annealed, and Ar-etched (Fig.4C) samples. Characteristic phenomena include level shifting of the initial splitting and threshold shifting, possibly from stored charge. To summarize transients, Day 2 curves are plotted using both a single-point estimate immediately after voltage application (“day 2, start”), and a 5-point average after ~30s (“day 2, end”). Both baseline and C3F8-etched samples display unintuitive initial behavior that is non-monotonic with voltage and instead shifts the opposite direction predicted by the electro-optic effect.

[0083] Referring to Figs.4D-4F, consecutive dc drift measurements showing hysteresis occurring in the form of a stored decay across TFLN sample processing conditions, including baseline air-cladded and annealed (Fig.4D); air-cladded, annealed, and C3F8-etched (Fig.HCU-07125 (HU9580, HU9581) 4E); and oxide-cladded, LN+oxide annealed, and Ar-etched (Fig.4F) samples. Stored decay refers to dc drift that trends similar in magnitude to the final behavior of the previous drift, instead of reproducing the initial behavior. This occurs after an initial transient recovery period. All measurements take place <1 hour after the conclusion of the prior measurement.

[0084] The time-domain response was characterized using a variant of phase-tracking measurements, wherein the optical mode splitting of a double ring device was monitored as a function of time after a step voltage was applied (Fig.1D). The temporal EO response can be defined as the change in optical mode splitting, normalized to the applied voltage. Coupled ring devices can be both technologically relevant for emerging photonic devices such as microwave-to-optical transducers and frequency-domain beamsplitters and isolators, as well as useful sensors of the internal electric field distribution. These measurements can be conducted in an ambient environment: monitoring the splitting is a differential measurement first-order insensitive to common-mode noise (e.g., thermal shifts from either the room temperature or changes in the light in-coupling efficiency). The applied voltage for these measurements was 30 V. The EO voltage dependence was measured with the same experimental setup using a staircase waveform interleaved with periods of 0 V to discharge the device and more accurately capture the step response (Figs.3A-3E). This measurement was performed prior to any long- term dc drift measurement to avoid hysteresis.

[0085] For devices following the baseline fabrication flow without any intentional contact treatment, the step response was marked by a rapid initial transient that overshoots and then settles into the long-term drift response over the course of <0.1 hrs (shown in inset of Fig.5). Fig.5 shows the measured dc drift for two identically processed baseline devices. In addition to expected long-term drift, short-term transients cause initial response overshoot. The inset of Fig.5 shows a zoom-in of initial transients.HCU-07125 (HU9580, HU9581)

[0086] Fig.6 shows the measured EO-voltage curves. Despite nominally identical fabrication steps, devices with untreated contacts can show both linear and nonlinear voltage responses, highlighting the need to identify and minimize electrode process variations that critically influence dc performance. For some devices, the EO response showed negligible voltage dependence before surpassing a threshold voltage, at which point the resonance shift rapidly increased in magnitude (Baseline 2 of Fig.6). Both fast transients and thresholding marked significant departures from the ideal response, which may render dc EO biasing unreliable in practical applications. Both linear and nonlinear EO responses were observed across several devices following nominally identical fabrication flows, although initial transients were common regardless of the presence or absence of a voltage threshold. This variability suggested the pristine TFLN contact chemistry can be readily altered by nanofabrication processes, including those commonly considered benign such as organic residue cleans. Variations in EO responses across devices may be due to fluctuations in the concentration and temperature of the wet chemical cleans at the time of sample processing, leading to different levels of remaining redeposition. Indeed, x-ray photoelectron spectroscopy (XPS) of TFLN taken after wet cleaning indicated that the balance of Li, Nb, and O at the surface is highly sensitive to each of the chemicals employed (Figs.4A-4F).

[0087] To this end, device variability can be reduced by enhancing contact linearity through a controlled surface treatment prior to metal repositioned. The technique described herein is based on etching such that any surface contaminants are simultaneously removed while modifying the TFLN surface states (Fig.2A). Referring to Fig.2A, a contact treatment procedure 200 is shown in which an additional Ar-based etch (right) or C3F8-based etch (left) occurs before Ti / Au evaporation, after the optical layer has been fabricated by e-beam lithography, Ar-etching, redeposition cleaning, and thermal annealing in oxygen. Fig.2A shows an exemplary device having a substrate 202, an EO material 204, etched regions 206HCU-07125 (HU9580, HU9581) (e.g., Ar-based etch, C3F8-based etch), and electrodes 208a, 208b. The same bilayer photoresist mask is used for etch-masking and lift-off, self-aligning the treatment. Both a pure Ar and a F-based (Ar + C3F8) plasma were investigated for this step, as these gas chemistries are ubiquitous in nanofabrication and have been observed to influence low- frequency roll-off of the EO response. In contrast to the baseline samples, it was found that both the Ar and F-based plasma etched contacts suppressed initial transients and maintained linearity in their voltage response (Figs.2B-2C). Fig.2B shows the measured dc drift for different plasma etch chemistries. The inset of Fig.2B shows a zoom-in of the first 15 minutes, showing both plasma treatments suppress initial transient overshoot. Long-term drift persists, indicating that the underlying mechanisms are independent of the contact’s influence on short-term instability, and may have a different material origin entirely. This occurred despite ion-induced damage expected from the physical etching component, evident in XPS of etched samples showing significantly broadened Nb 4s and Li 1s peaks (Figs.4A-4F). Although the behavior at short time scales was qualitatively similar for Ar and F-based plasma, the long-term dc drift behavior was much richer for the F-exposed sample, displaying multiple enhancements and decays of the EO response compared to the monotonic decay of the pure Ar etch. This may be attributed to the specific chemical reactions associated with the F-containing etch, which can deposit undesired dielectric layers that provide further avenues for charge redistribution. Specifically, F can react with Li and other etching byproducts to produce nonvolatile fluorinated compounds such as LiFx salts or fluoropolymers, as indicated by XPS (Figs.4A-4F).

[0088] The EO voltage response was compared to the electronic characteristics of the device stack and was measured using simple two-probe current-voltage measurements (Figs.2C- 2F). Figs.2C-2D show a comparison of the EO voltage dependence and the electronic transport for a sample with plasma-treated contacts, showing that quasi-linearity isHCU-07125 (HU9580, HU9581) maintained in both types of measurement. The I-V more clearly shows the transition from rectifying to transmitting characteristics. Figs.2E-2F show a comparison of the EO voltage dependence and the electronic transport for a sample without etched contacts, showing rectification in both types of measurement. A clear resemblance is observed between EO-V and I-V, highlighting how electronic response controls EO response to a great degree in TFLN. A strong correspondence between the leakage current and the EO response was observed, indicating that electronic transport physics can be used to interpret and predict EO performance. The threshold phenomenon can be a consequence of rectifying contacts. This is clear from the I-V characteristic of the baseline fabrication flow (Fig.2F), which exhibits rectifying behavior that can be captured by a circuit model with opposing diodes at the contact interfaces (Fig.7A). Fig.7A shows a device 700, including a substrate 702, an EO material 704, and electrodes 706a,706b, with equivalent contact-limited circuit overlaid. Rectifying diodes indicate the potential barrier formed at the metal-LN contact interface due to work function mismatch, which may be further exacerbated by gap states and impurities. This model can be approximately solved to derive the analytical formulaEquation 1where ^^ is the saturation current, ^^^ is the thermal voltage ^^^ / ^ and ^ = 1 +^^^^^^^^^^ / 2^^^is a nonideality factor that accounts for bulk leakage pathways. Physically, this reflects the charge carrier transmissivity across the potential barrier created between the metal and LN. Because the current is clamped by the saturation current of the reverse-biased diode, in steady-state most of the voltage drop occurs at the interface instead of the central region where the optical mode resides (Fig.7B). Fig.7B shows a schematic of interface band-bending with and without applied voltage. Injected electrons diffuse as polaronsHCU-07125 (HU9580, HU9581) through LN. When thermionic emission (TE) dominates, electrons cannot surmount the barrier and rectification occurs. Barrier-tunneling (BT) can dominate instead if the barrier is sufficiently thinned, either by applying high voltage or highly doping the contact region. In EO devices, this manifests as an apparent lack of response when the current saturates. However, at sufficiently high voltages the reverse-biased diode can break down and start to conduct again due to barrier-lowering or tunneling processes induced by the steep band bending at the interface. Rapid initial transients can therefore arise as a consequence of the step voltage input when the conduction is contact-limited: the anti-series contact diodes act to suppress the voltage spike by rapidly increasing the reverse current, thereby clamping the overvoltage. The excess carriers injected during breakdown then diffuse into the optical mode region and induce the EO effect. This is a distinct phenomenon from simple displacement current contributions, which occur on the timescale of the voltage supply ramp time (approximately seconds). While diffusion can be slow in a polar oxide, the relevant carrier lifetimes (approximately microseconds) are orders of magnitude shorter than what is observed (approximately minutes). The abrupt hot injection of charges may induce local heating that adds additional index shifts through the thermo-optic and pyroelectric effect before thermalizing.

[0089] The plasma-treated samples can similarly be interpreted as engineering an electronic contact to quickly transition from rectifying to transmitting I-V, therefore appearing quasi- linear (Figs.2C-2D). The specific mechanism by which the potential barrier is modified depends on the charge carrier involved. As a polar oxide, charge conduction in LN lies intermediate between band conduction in crystalline semiconductors and hopping conduction in amorphous solids. Because of strong electrostatic interactions with the lattice, electrons in such materials self-localize to a single lattice site, forming small polarons. Electron transport in LN can be interpreted in terms of polaron motion, encompassing bound polarons trapped atHCU-07125 (HU9580, HU9581) niobium antisites NbLiand free polarons hopping between reduced niobium sites. Although ionic motion can also contribute to charge transport, all measurements were taken at room temperature, which is far below the transition temperatures required to thermally activate ions.

[0090] Referring to Figs.7C-7D, high-resolution XPS Nb 3d spectra of Ar-etched TFLN (Fig.7D) and an annealed baseline (Fig.7C) are shown. The mechanism of the plasma- treated contact was elucidated from XPS of the Nb 3d peak, showing clear emergence of the reduced Nb4+and Nb3+valence states at the surface after etching (Figs.7C-7D). Ar-etching effectively dopes the TFLN surface by increasing the local polaron hopping site density via chemical reduction. Surface reduction of metal oxides after Ar sputtering in vacuum was commonly observed and can be attributed to preferential sputtering of oxygen atoms, which can occur even with low energy (<1 keV) ion milling (Figs.8A-8I). In LN, this may be further compounded by loss of Li, leaving behind Li vacancy defects that Nb atoms can subsequently fill. Consequently, because both NbLi and reduced Nb serve as hopping sites for different species of polaron, chemical reduction can effectively modify the local conductivity of the LN in contact with the electrode. Due to the complex interplay between different hopping conduction mechanisms and their associated mobilities, LN conductivity may not in general increase monotonically with degree of reduction and can depend on the specific defect reactions incurred.

[0091] Referring to Figs.8A-8B, high-resolution Li 1s XPS spectra of Ar-etched TFLN (Fig. 8B) and an annealed baseline (Fig.8A) are shown. Figs.8A-8B show the clear emergence of a broad third peak attributed to lattice damage, labelled Nb 4s*. The Nb 4s line overlaps with the Li 1s line.

[0092] Referring to Fig.8C, the intensity versus binding energy is shown for a baseline device treated with piranha solution and SC-1. Piranha solution strips surface Li, evident from theHCU-07125 (HU9580, HU9581) reduced Li 1s peak intensity. Standard chemical cleans such as SC-1 can restore the TFLN surface Li / Nb ratio, indicating that the Li deficiency is localized to the top 6-10 nm etched away by the clean. Fig. 8D shows the corresponding Li / Nb ratios (after peak fitting) for a baseline device treated with piranha solution and SC-1.

[0093] Referring to Figs.8E-8G, XPS spectra without background subtraction of the Li 1s / Nb 4s (Fig.8E), Nb 3d (Fig.8F), and O 1s (Fig.8G) lines in Ar-etched TFLN after annealing or cleaning in SC-1 are shown. Figs.8E-8G show restoration of the surface chemistry.

[0094] Referring to Fig. 8H, XPS spectra of the F 1s line after etching in a C3F8-containing reactive ion etch chemistry is shown. Fig. 8H shows the emergence of a metal fluoride peak (LiFx) and C-F bonding from gas polymerization.

[0095] Referring to Fig.8I, XPS spectra of the F 1s line after exposure to hydrofluoric acid is shown. Fig.8I shows only the LiFx peak.

[0096] Referring to Figs.7E-7F, high-resolution XPS O 1s spectra of Ar-etched TFLN (Fig. 7F) and the annealed baseline (Fig. 7E) are shown. Figs. 7E-7F show reduction originates from oxygen deficiency after Ar-etching (area ratio [O 1s]Ar / [O 1s] = 0.61, normalized by the Nb 3d peak areas). In addition to the lattice oxygen (O1) peak, broadening is captured by the O2 peak attributed to adsorbed oxygen or oxygen vacancies. Although a clear deficiency of oxygen was detected in XPS post-etch (Figs. 7E-7F), it can be difficult to quantitatively analyze Li vacancy formation due to the emergence of the damage peak near Li 1s line, and therefore the relative contributions of bound and free polaron hopping (Figs.4A-4F). However, a simple I-V comparison between a pristine and blanket-etched TFLN sample showed that for the system described herein the contact resistance may decrease (Figs. 9A-9E). Such plasma etches have also been used to create ohmic contacts in ultra-wide bandgap semiconductors, ostensibly due to the high-density formation of donor-acting oxygen vacancies. In this view, the targeted surface reduction at the contact acts can act as a form of chemical doping for polarHCU-07125 (HU9580, HU9581) oxides in analogy to traditional impurity doping in semiconductors, effectively thinning the barrier or otherwise mediating the removal of dangling bonds and other trap states at the interface.

[0097] Fig.9A shows a waterfall plot of the Nb 3d XPS line with increasing etch time in a 4 keV single ion milling configuration, showing the emergence of side lobes characteristic of lower Nb oxidation states. Fig.9B shows corresponding reduced fraction of total niobium concentration [Nbx] / [Nb] as a function of etch time. Identical scattering cross-sections are assumed for all the Nb oxidation states. Figs.9C-9E show Nb 3d XPS lines after in-situ cluster ion milling with 8 keV across 75 ions (Fig.9C), 8 keV across 2000 ions (Fig.9D), and 2 keV across 2000 ions (Fig.9E). Slight reduction was seen for as low energies as 4 eV per ion (8keV / 2000 ions). Negligible etching was seen for both the 8keV / 2000 ions and 2keV / 2000 ions conditions.

[0098] DC stability is expected to not only be sensitive to local chemical modifications, but also to the presence of defects and impurities throughout the entire material stack. These inhomogeneities may trap (or detrap) charge carriers, or may be physically redistributed to screen the applied field. In this view, each independent drift process can be associated with a lifetime ^^, and arises in the dc drift curve as a linear combination of N exponential decays:Equation 2

[0099] The coefficients &^are weighting factors that can be positive or negative, and whose magnitude can be associated with the initial concentration of the participating defect or impurity. !- is the steady-state EO response after all the decays saturate. Such a model has also been understood from an RC-ladder network perspective, in which each RC block captures electronic inhomogeneity arising from interfaces or anisotropy (Figs.10A-10B).HCU-07125 (HU9580, HU9581) Fig.10A shows a device 1000, including a substrate 1002, an EO material 1004, and electrodes 1006a,1006b, with a circuit schematic overlaid for I-V measurements on blanket- processed TFLN, without any waveguide patterning by e-beam lithography. A surface shunt resistor has been added to model alternate conductivity paths that lead to deviation from perfect current saturation at high voltages. Fig.10B shows I-V curves for a baseline sample that has only undergone solvent cleaning in acetone and isopropanol, compared to an Ar-ion etched sample with and without additional cleaning in SC-1 to remove the surface damage layer. No samples have undergone thermal annealing. The saturation current is clearly higher in the Ar-etched sample, indicating the partial elimination of dangling bonds or other trap states that could be pinning the contact barrier. Interestingly, after cleaning off the top layer the saturation current is even lower than the baseline sample, suggesting that “cleaner” interfaces do not necessarily lead to greater electronic transmission.

[0100] In practical devices, the exact distribution of defects and impurities is modified by thermal cycling, for example, from annealing steps used to repair lattice damage, sinter contacts, and passivate defects. By isolating the effect of individual annealing steps on EO performance, one can probe material changes incurred by thermal annealing after characteristic stages in TFLN fabrication (Fig.11A). Fig.11A shows a nanofabrication flow 1100 for oxide-cladded TFLN EO modulators. Thermal annealing in oxygen may occur after the initial waveguide etch (LN anneal 1106) or after deposition of a SiO2 top cladding 1108 (oxide anneal 1110). After etching open electrode windows in the top cladding 1108, all devices undergo the Ar-plasma contact treatment (e.g., Ar-based etch 1114) to ensure linearity. In addition to post-etch annealing (herein termed the “LN anneal”), annealing after the deposition of a SiO2top-cladding (herein termed the “oxide anneal”) was investigated. Such an anneal can be used to improve the quality of oxide deposited by a low-temperature process such as plasma-enhanced chemical vapor deposition (PECVD). All device variationsHCU-07125 (HU9580, HU9581) utilized the Ar contact etch to ensure quasi-linearity, thereby enabling reliable probing of the exponential decays of interest without being obscured by the transient behavior of the contact (Figs.12A-12B). This need is evident in the RC-ladder network picture, which assumes that the connection between the power supply and the linear circuit modeling the trapping and screening processes is also linear (i.e. capacitive or resistive) (Figs.10A-10B). In this context, established defect spectroscopies, such as deep-level transient spectroscopy (DLTS), also necessitate the development of an ohmic contact to ensure the correct trap states are probed in the bulk instead of at the contact junctions.

[0101] Referring to Fig.12A, an exemplary device is shown, which includes a substrate 1202, an EO material 1204, and electrodes 1206a, 1206b, with a circuit schematic for a RC-ladder network supporting a single time constant overlaid. The RC-ladder network is characterized by N=2 unique blocks RC blocks. At t=0, the voltage is determined by capacitive voltage division (i.e. by the relative dielectric constants). Atthe voltage is determined by resistive voltage division (i.e. by the relative conductivities). This analysis is only valid if all elements in the circuit are linear, including the electrical connection between the drive and the RC-ladder network.

[0102] Referring to Fig.12B, an exemplary device is shown, which includes a substrate 1202, an EO material 1204, and electrodes 1206a, 1206b, with a circuit schematic for a RC- ladder network supporting two time constants is shown, characterized by N=3 unique blocks RC blocks. These trends hold for higher values of N, for example originating from contact impedance Rc-Cc or from a defect layer.

[0103] As a baseline, a top-cladded device that has undergone no anneals was considered. Referring to Figs. 11B-11E, the measured dc drift and multi-exponential fits are shown for samples that have undergone no anneals (Fig.11B), only the LN anneal (Fig.11C), only the oxide anneal (Fig.11D), and both the LN and oxide anneals (Fig.11E). DC drift after a long-HCU-07125 (HU9580, HU9581) term measurement exhibits hysteresis that continues the decay of the previous drift measurement. This hysteresis is exploited in the fitting of Fig. 11D to stitch together two measurements and extrapolate the relevant lifetimes. Overall, annealing is observed to repairfast drift mechanisms (^ < 10^ s). However, oxide-annealed samples require additional long-lived decays (^ > 10^ s) in their fits, implying the emergence of a new slow-drift pathway. Thelack of annealing showed a fast rise time (^^ = 7"1$ × 10^ s) followed by a slow temporaldecay of the driving field, characterized by N=3 lifetimes total (Fig.11B). This response was then compared to devices that have undergone only the LN anneal, only the oxide anneal, or both. The phenomenological model underlying Eq.2 was used to infer material changes; the appearance or disappearance of a decay constant should correspond with the appearance or disappearance of some dielectric inhomogeneity. Characteristically, devices that only underwent the LN anneal lost the fast initial rise and exhibited dc drift well-fit by only N=2lifetimes (Fig. 11C). The fast drift process (^ ^^ < 10 s) vanished with any combination ofanneals (Figs.11C-11E), indicating that annealing in general removes a drift source intrinsic to nanofabricated TFLN-on-insulator. These initial defects may stem from the wafer-bonding process or initial etch, or from high energy electron irradiation during lithography. Any device that experienced an oxide anneal required at least N=3 lifetimes (Figs. 11D-11E). This was observed even though the oxide anneal ostensibly repaired fast-drift defects in the same manner as the LN anneal, thereby implying that the presence of oxide during annealing mediated the introduction of a different, unrelated relaxation pathway.

[0104] Secondary Ion Mass Spectrometry (SIMS) was used to obtain an understanding of the specific structural changes incurred by the oxide anneal (Fig.11F). Fig.11F shows Li diffusion after annealing in the presence of cladding. SIMS data was collected with Cs sputtering using a commercial time-of-flight SIMS (IONTOF M6). The Ar background pressure was set to a high 2E-6 torr to reduce charging, which was a significant for samples with unannealedHCU-07125 (HU9580, HU9581) cladding oxide. The total Li intensity was extracted by summing the contributions from all the Li-containing lines in the mass spectra, and then normalized to the total collected counts at each time point.

[0105] Samples that were annealed in the presence of cladding oxide had a measurable increased Li concentration at the cladding-LN interface, indicating the decomposition of LN at the interface and subsequent Li diffusion into the cladding. Negligible change to the Li signal was observed in the buried oxide layer, suggesting that the quality of the oxide can also play a role in the diffusion (see Figs.13A-13D for a comparison of the Li distribution in different cladding oxides). Figs.13A-13D show EO voltage dependence for TFLN modulators with SiO2 top cladding that have experienced no anneals (Fig.13A), only the LN anneal (Fig.13B), only the oxide anneal (Fig.13C), and both the LN and oxide anneals (Fig. 13D), prior to contact formation. Regardless of bulk conductivity changes incurred by different combinations of thermal annealing, (quasi)linearity is maintained in the EO response. The contributions to drift processes may be multifold. The main temporal contribution may stem from Li vacancies in the TFLN, Li ion contamination in the cladding, or a combination of both. Li vacancies were associated with a variety of point defects in LN, including NbLi defects serving as preferential trapping sites for bound polarons. As such, an increased antisite concentration at the interface may impact overall charge carrier dynamics.

[0106] Referring to Fig.14A, a scanning electron microscopy image of a cladded TFLN modulator cross-section that has experienced the oxide anneal is shown. The inset of Fig. 14A shows a bright-field scanning transmission electron microscopy (STEM) zoomed into the boxed region, showing a change in contrast at the interface between the cladding silica and the TFLN top surface. An intensity profile versus depth was extracted from the STEM imaging (Fig.14B). Greater intensity (darker region in image) corresponds to higher mass density, which in TFLN is associated with Li vacancies. Scanning transmission electronHCU-07125 (HU9580, HU9581) microscopy (STEM) imaging indicated that the LN is denser within 10-15 nm of the interface (Figs.14A-14B), a signature of NbLi formation. Alternatively, slow ion migration may result in an irreversible screening process that may also help explain observations of hysteresis, where the dc drift after a long-term measurement was not reproducible and instead continued the decay of the previous drift measurement (Fig.15). However, given that hysteresis occurred even for samples that only experienced the LN anneal (Fig.11C), it is shown that Li ion migration in the top cladding can only partially explain memory effects. Other possible hysteretic contributions inherent to the TFLN-on-insulator platform include ion migration in the TFLN or the buried oxide, or ferroelectric domain inversion under the hot electrode.

[0107] Referring to Fig.15, SIMS traces are shown of the Li distribution in oxide-cladded samples (top) before annealing in the presence of cladding oxide, and (bottom) after annealing. The oxides examined were deposited by PECVD using either high frequency (13.56 MHz) or low-frequency (380 kHz) power supplies. Low-frequency PECVD oxide is less dense and exhibits a lithium distribution markedly different from high-frequency oxide, with an overall higher concentration throughout.

[0108] The effects of surface chemistry and interfaces on device performance in general, and on reliable EO in particular, only become exacerbated as materials scale down in dimensions and surface-volume ratios increase. An understanding of the interfacial conditions that enable stable and linear EO operation in TFLN nanophotonic devices was developed. While thermal annealing has been generally assumed to improve device performance, it is shown that TFLN electro-optic devices must be carefully engineered to prevent concurrent Li diffusion into the cladding, in the same way that silicon foundries are constrained by a thermal budget to prevent diffusion of dopants. Moreover, small modifications to the contact interface can drastically affect the evolution of dc response, highlighting the need to consider electronic properties of TFLN beyond an ideal defect-free dielectric in order to optimize itsHCU-07125 (HU9580, HU9581) performance for electro-optic devices. To this end, the polaronic nature of conduction in LN was exploited to develop a chemical reduction protocol that extends contact linearity, despite etch damage. Contact engineering is well-understood in integrated circuits and silicon photonics, but uncommonly used in electro-optic devices that do not explicitly rely on carrier injection for index modulation. The utility of linear contacts for efficient and stable dc EO coupling is simple and can be generalized to other EO platforms such as polymer modulators, barium titanate, and lithium tantalate. Devices described herein can utilize a low-frequency feedback control loop to lock bias points. Free-running operation and therefore greater system complexity can be possible with device processing informed by the material understanding described.

[0109] For a diode circuit mode, one can assume that the diode is exponential, in other wordsEquation 3 where I0is the saturation current, q is the magnitude of the electron charge, kBis the Boltzmann constant, T is the temperature, and VD is the voltage across the diode. One can abbreviate the value kBT / q as the thermal voltage Vth, which is approximately 25 mV at room temperature. At this point one makes no assumptions about where the exponential behavior arises from or the origin of the saturation current.

[0110] Taking the circuit model of Fig. 10A, if the bulk leakage resistance Rleakageis set to 0 then the circuit can be solved analytically. The solution is tanh^2^^^Equation 4 where I is the current drawn by the voltage source V. If the bulk resistance is included, the voltage as a function of current is given byHCU-07125 (HU9580, HU9581)Equation 5 which is not well defined if I=I0and therefore not invertible, thereby complicating the determination of the I-V relation. Note that one can neglect the contribution of the surface conductivity at this point since its only role is to add a linear current I=V / Rsurfon top of the final expression. However, one can Taylor expand in the I ≪ I0 regime to obtainEquation 6

[0111] Equation 6 can be inverted,Equation 7 with n=1+I0Rbulk / 2Vth.

[0112] The desire is to find a function that can be linearly expanded for small V and saturates to 1 for large V. Taking inspiration from the analytical form for the Rbulk=0 limit, this is obeyed by the hyperbolic tangent. Thus, one can obtain an approximate I-V relation as followsEquation 8 which behaves as expected in the small V, large V, and n=1 (Rbulk=0) limits.

[0113] Some remarks on the physical justification for this circuit model are made. Rectification in lithium niobate, and more broadly leakage conduction phenomena, can occur through ionic or electronic means. In either case, the potential barrier occurs at the contact interface, hence the modeling of the metal-LN junctions as diodes. To elicit more insight, it isHCU-07125 (HU9580, HU9581) noted that at low temperatures electronic conduction dominates via polaron hopping, while at elevated temperatures ions are the primary charge carrier. At room temperature (approximately 293 K) conduction is expected to be electronic in nature. The transmission of electrons across the metal-LN junction can be viewed as a Schottky diode with Schottky barrier height ^^ (6H^ ∝ ^ ' 9 / ^9:Equation 9 The proportionality is carried by geometric factors. Indeed, Schottky diodes follow an exponential I-V as expected by Eq.3.

[0114] Linear deviations past saturation can be captured by a surface shunt conductance (Fig. 10B). Surface conductivity may partially account for electro-optic relaxation phenomena observed in frequency-domain measurements of Mach-Zehnder modulators. Several other electronic processes may also cause deviations from saturation, including Schottky emission (image force barrier lowering), Fowler-Nordheim tunneling emission, and Poole-Frenkel conduction. These typically occur at field strengths comparable to the breakdown field of LN (Ecoerc ~21 V / µm, Ebreakdown ~75-80 V / µm), much higher than applied in measurements (Eext ~4 V / µm). One may consider their contribution since the exact distribution of the electric field can depend on the particular device geometry. For example, if the conduction is contact-limited then most of the voltage is dropped along the interface, possibly leading to large fields across a very thin region which may be surmised for Fig.6. Moreover, these effects are expected to become more salient as electro-optic devices scale down, trading off device length for larger bias fields.

[0115] I-V data can be interpreted using this model through the following conclusions: (1) For a fixed temperature, the Schottky barrier height can determine the saturation current.HCU-07125 (HU9580, HU9581) (2) Bulk conductivity relative to the saturation current can determine the onset of saturation. (3) Surface conductivity can determine the slope past saturation. In any case, proper preparation of electrode material and TFLN surface chemistry are imperative for efficient transmission of the relevant charge carrier.

[0116] For illustrative purposes, in Fig.12A the simplest possible mode, a series RC ladder with only two unique blocks (Rz-Cz, Rx-Cx), is considered. The labelling used in the literature is used, which attributes the different impedance elements to the material anisotropy in the x- axis vs z-axis (c-axis) directions. In that case, the relevant electric field that couples to the optical mode is Ez(E in the main text). For a ridge waveguide device geometry, it is noted that most of the relevant electric field coupling to the optical mode may go through the LN slab, providing some justification for a purely series RC-ladder network.

[0117] The transient response can be readily solved with the boundary conditions that at time t→0 the circuit is a capacitive voltage divider, and at time t→∞ the circuit is a resistive voltage divider.Equation 11 This network is supported by a single time constant that depends on the interplay between the individual RC blocks and their respective time constants.

[0118] If an additional RC block is added in series, for example, stemming from the contact resistance, then the exact expressions can become far more cumbersome. Although an analytical solution can be found using Laplace methods, the main result is reproduced here:HCU-07125 (HU9580, HU9581)Equation 12 A full treatment including the functional forms of A, B, C, τ1, and τ2can be found. Suffice to say, it is difficult to identify any individual term with a particular RC block. Rather, the lifetime formulas reflect interactions between the different electronic regions abstracted by the RC blocks – in other words, interfaces.

[0119] General rules of thumb may be ascertained: (1) Every additional electrical inhomogeneity in series with the sensed layer (e.g., from defects, material anisotropy, surfaces and interfaces) can impose an additional time constant. (2) The short-time behavior can be determined by capacitive voltage division. (3) The steady-state behavior can be determined by resistive voltage division. This allows one to adopt a more phenomenological interpretation of measured dc drift that does not necessarily seek to identify every time constant with a specific microscopic origin, but rather highlights changes in the functional form. That is, the inclusion or omission of time constants can be corresponded with measured changes in the material structure. The multiexponential behavior and subsequent interpretations all arise from a linear system analysis. If any nonlinear elements are included, for example, a diode arising from a contact potential barrier, then the observed behavior may be less intuitive.

[0120] For simplicity, in the treatment here the top cladding or layers below the LN stack (i.e. the BOX and Si handle) have not been included. Parallel elements to the sensed layer Ez can simply be lumped into an equivalent leakage impedance in a circuit model, and do not affect the number of time constants. They can, however, affect the duration of the time constants.HCU-07125 (HU9580, HU9581)

[0121] Technically, the EO response is not linear over the free spectral range (FSR) of the reference (unbiased) ring due to the anti-crossing that arises. In the following, the regime in which this nonlinearity is significant is explored.

[0122] The Hamiltonian for a coupled ring system is given byEquation 13 where ωjis the cavity resonance frequency of ring j, ajis the corresponding annihilationoperator, and µ is the coupling strength between the two optical cavities. Diagonalizing toobtain the hybridized energies, the optical mode splitting Δ is given by U= V4T^ + X^Equation 14 where δ=ω1-ω2is the splitting if the modes were uncoupled (µ=0). This treatment of course assumes operation in the limit of a two-level system. One may account for the fact that ring resonators have a FSR by truncating the domain to 0 ≤ δ ≤ FSR / 2 and assuming the behavior follows symmetrically for FSR / 2 ≤ δ ≤ FSR. Taylor expanding about δ=FSR / 2, fractional splitting error is defined asEquation 15 For the parameters of ring resonators (FSR=40 GHz, 2µ=6 GHz), the error is only 11.7% ifthe resonances are kept 2µ away.

[0123] It can be seen graphically that the behavior is essentially linear mid-FSR (Figs.16A- 16B). Figs.16A-16B show simulated anti-crossing and optical mode splitting, respectively for characteristic parameters of coupled ring resonators used (FSR=40 GHz, 2µ=6 GHz). Grey regions indicate the operating regions of experiments, where the behavior is wellHCU-07125 (HU9580, HU9581) matched by a linear approximation. Indeed, devices with small initial splittings are measuredsuch that it is expected that the transients stay ±2µ away from the reference ring resonances.In the case that the EO response is stronger than expected, the applied voltage was lowered to stay mid-FSR. This was the case for the main text LN+oxide anneal sample, which was measured under 22.5 V bias (Fig.11E). Given that the measured data was well-fit by the provided exponential models without obvious distortion, it was concluded that the avoided crossing had a negligible effect on measurements.

[0124] The electro-optic response and its long-term stability are important parameters in circuit, device, and system design for integrated devices. Overcoming long-term drift and reduction in response from these devices is paramount to making them deployable in commercial settings. A thin film coating can be applied to integrated lithium niobate photonic structures to reduce diffusion of Lithium during nanofabrication to improve the DC stability response of integrated devices. It is shown that the long-term stability of these devices is dependent on the thermal budget of the nanofabrication process. In particular, correlations are shown between the long-term DC stability of coated devices, the number of times the sample is treated at moderate (>450 C) heating treatments, the amount of Lithium that diffuses out of the surface of the lithium niobate and into the oxide film. Lithium diffusion barriers may enhance the electro-optic stability of these devices, and provide candidate oxide and nitride materials that can form stable diffusion barriers compatible with scalable semiconductor manufacturing.

[0125] Referring to Fig.18A, a fabrication process flow 1800 after waveguide formation on a substrate 1802 is shown. The fabrication process flow 1800 includes cladding (e.g., depositing a top cladding 1806 on the EO-material 1804 containing waveguide), metallization (e.g., application of electrodes 1808), thermal processing 1810, 1812, and etching of the cladding oxide is shown. After creating a mask on the lithium niobate, the sample is etched.HCU-07125 (HU9580, HU9581) The sample may go through an annealing process (here, described as “LN anneal” 1810) to improve the optical losses of the device. Then, an oxide top cladding is deposited on the lithium niobate, and the sample may be annealed after deposition (referred to as “Oxide Anneal” 1812). Lithography is then used to define a mask for metal contacts. The oxide layer is removed through etching. At this point, the sample may go through a chemical treatment or may go through an additional annealing step (referred to as “contact anneal”); after any additional steps, contacts are deposited within the oxide openings.

[0126] Referring to Fig.18B, the DC response for ring resonators is shown. DC electro-optic response curves are shown over long time scales for three different electrode and processing conditions, yielding a 5 order of magnitude difference in response timescale. Here, 30 V is applied in a ground-signal-ground configuration in a ring resonator and the shift in resonant frequency is recorded over time. For samples with electrodes placed above the cladding oxide, the response decays to steady state with a characteristic timescale of 100 ms. Placing the electrodes on the lithium niobate makes the stability better, while annealing the samples following the bottom process flow path in Fig.18A yields a response that is stronger, but varies substantially over time.

[0127] Referring to Fig.18C, a Bode plot is shown displaying the frequency dependence of the response for Mach Zehnder interferometers at their quadrature point. Flatness is increased when the contacts are placed directly on the lithium niobate, and the response is further enhanced when the samples are annealed. At low frequencies, the top contact sample (i.e. on the silicon dioxide) is nearly 50 dB (~300x) lower in response than the annealed sample with contacts applied directly to the lithium niobate. Notably, the flatness is improved and the act of annealing appears to monotonically increase the overall device response.

[0128] To better understand the role that thermal annealing has on the low frequency response and DC stability, devices that have seen variants of the thermal treatments in theHCU-07125 (HU9580, HU9581) presented standard nanofabrication were characterized. The difference between no thermal processing, a single anneal step after waveguide formation, and a single anneal step after cladding oxide deposition was characterized.

[0129] Figs.19A-9C show the impact of the annealing steps on the low frequency response. Fig.19A shows the DC electro-optic response for samples without any anneals 1902, annealing just the lithium niobate 1904, and annealing only after the oxide is deposited.The response at long timescales appears higher when the samples go through at least one anneal, but the temporal response is markedly different. Here, the response starts at a low value when just the lithium niobate is annealed, whereas the response starts at its maximum and decays to its steady state when annealed after the oxide is deposited. The Bode diagram shows the electro-optic response of a Mach-Zehnder interferometer (MZI) as a function of modulation frequency for a small signal (V-pp = 100 mV) in Fig.19B. Fig.19B shows a low frequency response of MZI devices that have undergone the same treatments as in Fig.19A. Notably, the low-frequency response generally increases with the number of anneals. The frequency-dependent behavior for the annealed samples features regions with enhanced response, indicating that a combination of mechanisms are at play with the low-frequency response.

[0130] Secondary Ion Mass Spectrometry (SIMS) was performed on two lithium niobate samples that underwent the lithium niobate anneal and PECVD silicon dioxide cladding step; one sample was also annealed after the cladding deposition process (“after anneal”), while the other sample saw no additional annealing (“before anneal”). Fig.19C shows SIMS data for cladded lithium niobate samples, showing that Lithium diffuses from the LiNbO3 into the cladding oxide under moderate thermal treatment. As shown in Fig.19C, samples that are annealed have a measurable and increased Li concentration at the oxide-LiNbO3 interface, indicating the decomposition of LN at the interface and subsequent lithium diffusion into theHCU-07125 (HU9580, HU9581) cladding. While the conductivity of all materials within the sample may be different after annealing, it is also clear that compositional variations may also impact the low frequency response.

[0131] Lithium vacancies are associated with a variety of optically-relevant defects such as antisite defect NbLi . The compositional variations, certainly at the surface, could be related to enhanced surface conductivities that give rise to the enhanced response at kHz frequencies observed in the frequency dependent response curves. Further, Lithium is a fast diffuser in silica, making the material non-CMOS compatible. Introducing diffusion barriers above and below the lithium niobate layer may alleviate these concerns. This would require low-loss thermal or electrochemical diffusion barriers for Lithium.

[0132] Lithium diffusion barriers can be formed of ZrN, ZrO2, and HfO2. ZrO2 and HfO2 are also used in the semiconductor industry as high-k dielectrics and can be deposited using scalable technologies like Atomic Layer Deposition. Including these in the process flow may improve both the DC stability of these devices and the low frequency response of these materials.

[0133] The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustration, and combinations ofHCU-07125 (HU9580, HU9581) blocks in the block diagrams and / or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.

[0134] Referring to Figs.20-33, schematic views of various embodiments of exemplary devices that can be used to practice the present invention are shown.

[0135] Fig.20 is a schematic view of a device 2000 comprising a substrate 2002 having a first side 2002a and a second side 2002b, an EO material 2004 having a first side 2004a and a second side 2004b, a modified region 2006, an electrode 2008 disposed on the modified region 2006, and ridge 2012. Optical mode 2010 is shown.

[0136] Fig.21 is a schematic view of a device 2100 comprising a substrate 2102 having a first side 2102a and a second side 2102b, an EO material 2104 having a first side 2104a and a second side 2104b, and a ridge 2112. An optical mode 2110 is shown.

[0137] Fig.22 is a schematic view of a device 2200 comprising a substrate 2202 having a first side 2202a and a second side 2202b, an EO material 2204 having a first side 2204a and a second side 2204b, and a ridge 2212. An optical mode 2210 is shown.

[0138] Fig.23 is a schematic view of a device 2300 comprising a substrate 2302 having a first side 2302a and a second side 2302b, an EO material 2304 having a first side 2304a and a second side 2304b, and a doped volume 2312. An optical mode 2310 is shown.

[0139] Fig.24 is a schematic view of a device 2400 comprising a substrate 2402 having a first side 2402a and a second side 2402b, an EO material 2404 having a first side 2404a and a second side 2404b, modified regions 2406a, 2406b, electrodes 2408a, 2408b disposed in the modified regions 2406a, 2406b, and a doped volume 2412. An optical mode 2410 is shown.

[0140] Fig.25 is a schematic view of a device 2500 comprising a substrate 2502 having a first side 2502a and a second side 2502b, an EO material 2504 having a first side 2504a andHCU-07125 (HU9580, HU9581) a second side 2504b, modified regions 2506a, 2506b, electrodes 2508a, 2508b disposed in the modified regions 2506a, 2506b, and a doped volume 2512. An optical mode 2510 is shown.

[0141] Fig.26 is a schematic view of a device 2600 comprising a substrate 2602 having a first side 2602a and a second side 2602b, an EO material 2604 having a first side 2604a and a second side 2604b, modified regions 2606a, 2606b, electrodes 2608a, 2608b disposed in the modified regions 2606a, 2606b, and a ridge 2612. An optical mode 2610 is shown.

[0142] Fig.27 is a schematic view of a device 2700 comprising a substrate 2702 having a first side 2702a and a second side 2702b, an EO material 2704 having a first side 2704a and a second side 2704b, modified regions 2706a, 2706b, electrodes 2708a, 2708b disposed in the modified regions 2706a, 2706b, and comprising a ridge 2712. An optical mode 2710 is shown.

[0143] Fig.28 is a schematic view of a device 2800 comprising a substrate 2802 having a first side 2802a and a second side 2802b, an EO material 2804 having a first side 2804a and a second side 2804b, a modified region 2806, an insulator 2814, an electrode 2808a disposed on the modified region 2806, an electrode 2808b disposed on the insulator 2814, and a doped volume 2812. An optical mode 2810 is shown.

[0144] Fig.29 is a schematic view of a device 2900 comprising a substrate 2902 having a first side 2902a and a second side 2902b, an EO material 2904 having a first side 2904a and a second side 2904b, a modified region 2906, electrodes 2908a, 2908b, a doped volume 2912, and an insulator 2914. An optical mode 2910 is shown.

[0145] Fig.30 is a schematic view of a device 3000 comprising a substrate 3002 having a first side 3002a and a second side 3002b, an EO material 3004 having a first side 3004a and a second side 3004b, a modified region 3006, electrodes 3008a, 3008b, a ridge 3012, and an insulator 3014. An optical mode 3010 is shown.HCU-07125 (HU9580, HU9581)

[0146] Fig.31 is a schematic view of a device 3100 comprising a substrate 3102 having a first side 3102a and a second side 3102b, an EO material 3104 having a first side 3104a and a second side 3104b, at least one modified region 3106b, and electrode 3108b disposed on the modified region 3106b. Alternative electrode arrangements for a second electrode are also shown: (1) a second modified region 3106a and an electrode 3108a, disposed in the second modified region 3106a; or (2) a second electrode 3108c disposed on the second side 3102b of the substrate 3102. An optical mode 3110 is also shown.

[0147] Fig.32 is a schematic view of a device 3200 comprising a substrate 3202 having a first side 3202a and a second side 3202b, an EO material 3204 having a first side 3204a and a second side 3204b, chemically reduced regions 3206a, 3206b, and electrodes 3208a, 3208b disposed in the chemically reduced regions 3206a, 3206b. The EO material 3204 comprises a ridge 3210 and two planes 3212a, 3212b. The device 3200 further comprises a cladding 3214.

[0148] Fig.33 is a schematic view of a device 3300 comprising a substrate 3302 having a first side 3302a and a second side 3302b, an alkali metal ion-containing EO material 3304 having a first side 3304a and a second side 3304b, and layers of an alkali ion-blocking material 3306a, 3306b. An optical mode 3310 is also shown.

[0149] Any of the devices shown in Figs.20-33 can further comprise a support having a first side and a second side and / or a cladding. The support can be adjacent to or disposed on the second side of the substrate in any of the devices illustrated in Figs.20-33. The cladding can be adjacent to or disposed on or over the first side of the EO material in any of the devices illustrated in Figs.20-33.

[0150] Accordingly, in a first example embodiment, the present invention is a device. In a 1staspect of the 1stexample embodiment, the device comprises a waveguide, comprising an electro-optic (EO) material having a first side and a second side, the first side of the EOHCU-07125 (HU9580, HU9581) material comprising at least one modified region; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on the first side of the substrate; and an electrode disposed in the at least one modified region of the EO material, wherein the at least one modified region is a chemically reduced region or a heightened conductivity region.

[0151] As used herein, a “substrate” refers to a material having a refractive index that is less than that of the EO material at the operating wavelength. For example, if the EO material is LN operated at 1550 nm, nsub< 2.14.

[0152] As used herein, a “chemically reduced region” results from a gain in electrons or a loss of oxygen.

[0153] As used herein, a “region of heightened conductivity” refers to a region having a higher conductivity higher than the bulk EO material. In certain embodiments, a region of heightened conductivity is a region of reduced contact resistance. In alternative embodiments, a region of heightened conductivity is a region of quasi-linear conductivity.

[0154] In a 2ndaspect of the 1stexample embodiment, the waveguide comprises a ridge disposed on the first side of the EO material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1staspect.

[0155] In a 3rdaspect of the 1stexample embodiment, the waveguide comprises a ridge disposed on or adjacent to the second side of the EO material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 2ndaspects.

[0156] As used herein, “adjacent to” refers to a geometric arrangement that permits a spatial gap between two objects that are positioned next to each other.HCU-07125 (HU9580, HU9581)

[0157] In a 4thaspect of the 1stexample embodiment, the ridge comprises the EO material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 3rdaspects.

[0158] In a 5thaspect of the 1stexample embodiment, the ridge comprises a ridge material, different from the EO material. A ridge material can be Si, Si3N4, SiON (silicon oxynitride), or hydrogen silsesquioxane (HSQ). The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 4thaspects.

[0159] In a 6thaspect of the 1stexample embodiment, the waveguide comprises a doped volume disposed within the EO material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 5thaspects.

[0160] Ti-indiffusion can form a doped volume by introducing titanium atoms into the EO material. Proton-exchanged lithium niobate can form a doped volume by replacing lithium ions with protons.

[0161] In a 7thaspect of the 1stexample embodiment, the EO material comprises at least two modified regions, and further wherein the device comprises at least two electrodes, each electrode disposed in the modified regions. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 6thaspects.

[0162] In an 8thaspect of the 1stexample embodiment, the device comprises at least one electrode disposed in the at least one modified region; an insulator disposed on the EO material; and at least one electrode disposed on the insulator. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 7thaspects.HCU-07125 (HU9580, HU9581)

[0163] As used herein, an “insulator” refers to an electrically insulating material such as SiO2, Al2O3.

[0164] In a 9thaspect of the 1stexample embodiment, the at least one modified region is recessed in the EO material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 8thaspects.

[0165] In a 10thaspect of the 1stexample embodiment, the device further comprises a support disposed on the second side of the substrate. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 9thaspects.

[0166] In an 11thaspect of the 1stexample embodiment, the device further comprises at least one electrode disposed in the at least one modified region; a support having a first side and a second side, the support disposed on the second side of the substrate; and at least one electrode disposed either on the second side of the substrate or the second side of the support. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 10thaspects.

[0167] In an 12thaspect of the 1stexample embodiment, the support comprises Si, lithium niobate, SiO2, or Al2O3. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 11thaspects.

[0168] In a 13thaspect of the 1stexample embodiment, the device further comprises a cladding disposed over the first side of the EO material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 12thaspects.

[0169] As used herein, the term “cladding” refers to a material disposed on the EO material and having a refractive index that is less than the refractive index of the EO material.HCU-07125 (HU9580, HU9581)

[0170] In a 14thaspect of the 1stexample embodiment, the cladding comprises SiO2. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 13thaspects.

[0171] In a 15thaspect of the 1stexample embodiment, the substrate comprises a material selected from SiO2, Al2O3, SiON, HfO2, or ZrO2. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 14thaspects.

[0172] In a 16thaspect of the 1stexample embodiment, the EO material is selected from a metal oxide, metal nitride or a ferroelectric material. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 15thaspects.

[0173] In a 17thaspect of the 1stexample embodiment, the EO material is selected from LiNbO3, LiTaO3, BaTiO3, KNbO3, ScxAl(1-x)N, ScN, AlScN, or AlN. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 16thaspects.

[0174] As used herein, the notation ScxAl(1-x)N denotes non-stoichiometric ratios of theelements Sc and Al, where x is in the range of 0 ≤ c ≤ 1. In an example embodiment, x <0.5 (such a material is ferroelectric).

[0175] In an 18thaspect of the 1stexample embodiment, the EO material is selected from potassium titanyl phosphate (KTP) or potassium dihydrogen phosphate (KDP). The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 18thaspects.

[0176] In an 19thaspect of the 1stexample embodiment, at least one electrode comprises Ti / Au, indium tin oxide (ITO), Al, Cr, Ni, NiCr, MoOx, LiFx, CsFx, or KFx.HCU-07125 (HU9580, HU9581)

[0177] As used herein, a notation ABxdenotes an optionally non-stoichiometric ratios of elements A and B. In other words, the value of x can vary from any value that is greater than zero up to the maximum value permitted by the valency.

[0178] In a 20thaspect of the 1stexample embodiment, the first side of the EO material comprises a ridge and two planes, the ridge comprising the EO material; the two planes are separated by the ridge, each plane extending along the ridge on the first side of the substrate, each plane comprising a chemically reduced region recessed in the EO material; the device comprises two electrodes disposed on the chemically reduced regions such that the ridge extends between the two electrodes; and further wherein: the EO material is LiNbO3; the substrate comprises SiO2; the electrodes comprise Ti / Au; the device further comprises a support disposed on the second side of the substrate, the support comprising Si; the device further comprises a cladding disposed over the first side of the EO material, the cladding comprising SiO2. The remainder of the features and example features of the 1stexample embodiment are as described above with respect to the 1stthrough 19thaspects.

[0179] In a second example embodiment, the present invention is a device. In a 1staspect of the 2ndexample embodiment, the device comprises comprising a waveguide, comprising an alkali metal ion-containing electro-optic (alkali EO) material having a first side and a second side; a layer of an alkali ion-blocking material disposed on at least one of the first side and the second side of the alkali EO material; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on or adjacent to the first side of the substrate.

[0180] As used herein, the terms “alkali metal” or “alkali ion” refer to metals and their ions listed in the first group of the periodic table, such as Li, K, Na, etc.

[0181] In a 2ndaspect of the 2ndexample embodiment, the waveguide comprises a ridge disposed on or adjacent to the first side of the alkali EO material. The remainder of theHCU-07125 (HU9580, HU9581) features and example features of the 2ndexample embodiment are as described above with respect to the 1staspect.

[0182] In a 3rdaspect of the 2ndexample embodiment, the waveguide comprises a ridge disposed on or adjacent to the second side of the alkali EO material. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 2ndaspects.

[0183] In a 4thaspect of the 2ndexample embodiment, the ridge comprises the alkali EO material. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 3rdaspects.

[0184] In a 5thaspect of the 2ndexample embodiment, the ridge comprises a ridge material, different from the alkali EO material. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 4thaspects.

[0185] In a 6thaspect of the 2ndexample embodiment, the waveguide comprises a doped volume disposed within the alkali EO material. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 5thaspects.

[0186] In a 7thaspect of the 2ndexample embodiment, the device further comprises a support 3102 disposed on the second side of the substrate. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 6thaspects.

[0187] In an 8thaspect of the 2ndexample embodiment, the support comprises Si, lithium niobate, SiO2, or Al2O3. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 7thaspects.

[0188] In a 9thaspect of the 2ndexample embodiment, the device further comprises a cladding disposed over the first side of the alkali EO material. The remainder of the featuresHCU-07125 (HU9580, HU9581) and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 8thaspects.

[0189] In a 10thaspect of the 2ndexample embodiment, the cladding comprises SiO2. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 9thaspects.

[0190] In an 11thaspect of the 2ndexample embodiment, the substrate comprises a material selected from SiO2, Al2O3, SiON, HfO2, or ZrO2. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 10thaspects.

[0191] In a 12thaspect of the 2ndexample embodiment, the alkali EO material is LiNbO3 or LiTaO3. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 11thaspects.

[0192] In a 13thaspect of the 2ndexample embodiment, the alkali ion-blocking material is selected from Al2O3, ZrO2, HfO2, SrF2, or CaO. The remainder of the features and example features of the 2ndexample embodiment are as described above with respect to the 1stthrough 12thaspects.

[0193] In a 3rdexample embodiment, the present invention is a method of manufacturing a device. In a 1staspect of the 3rdexample embodiment, the method comprises providing a waveguide comprising an electrooptical (EO) material having a first side and a second side, depositing a resist over a portion of the first side of the EO material; chemically reducing at least one region of the EO material, thereby producing at least one reduced region; optionally, removing the resist; and depositing at least one electrode on the at least one reduced region.

[0194] In a 2ndaspect of the 3rdexample embodiment, reducing the at least one region comprises bombarding the at least one region with ions. The remainder of the features andHCU-07125 (HU9580, HU9581) example features of the 3rdexample embodiment are as described above with respect to the 1staspect.

[0195] In a 3rdaspect of the 3rdexample embodiment, the ions are selected hydrogen, helium, argon, xenon or krypton. The remainder of the features and example features of the 3rdexample embodiment are as described above with respect to the 1stthrough 2ndaspects.

[0196] In a 4thaspect of the 3rdexample embodiment, reducing the at least one region comprises contacting the at least one region with a reducing agent. The remainder of the features and example features of the 3rdexample embodiment are as described above with respect to the 1stthrough 3rdaspects.

[0197] In a 5thaspect of the 3rdexample embodiment, the reducing agent is Ti, trimethylaluminum (TMA), CaH2, or NaBH4. The remainder of the features and example features of the 3rdexample embodiment are as described above with respect to the 1stthrough 4thaspects.

[0198] In a 6thaspect of the 3rdexample embodiment, reducing the at least one region comprises exposing the at least one region to a thermal treatment in a reducing environment (e.g., oxygen deficient environment, nitrogen-rich environment). The remainder of the features and example features of the 3rdexample embodiment are as described above with respect to the 1stthrough 5thaspects.

[0199] As used herein, the phrase “thermal treatment” refers to a step of exposing a material to a temperature above 300 C for a time period required to achieve the desired result (see Zhang, Y., L. Guilbert, and P. Bourson. "Characterization of Ti: LiNbO 3 waveguides by micro-raman and luminescence spectroscopy." Applied Physics B 78 (2004): 355-361.)

[0200] In a 4thexample embodiment, the present invention is a method of manufacturing a device. In a 1staspect of the 4thexample embodiment, the method comprises providing a waveguide comprising an electrooptical (EO) material having a first side and a second side,HCU-07125 (HU9580, HU9581) subjecting at least one region of the EO material to a treatment that produces a heightened conductivity region, thereby producing at least one heightened conductivity region; and depositing at least one electrode on the at least one reduced region.

[0201] In a 2ndaspect of the 4thexample embodiment, subjecting at least one region of the EO material to a treatment that produces a heightened conductivity region comprises ion implantation, doping, or alloying.

[0202] Ion implantation can include accelerating and directing ions (e.g., iron ions, lithium ions) into the at least one region of the EO material.

[0203] Doping can increase conductivity of the at least on region of the EO material for high temperature operation (e.g., temperature above room temperature). Doping by indiffusion can include depositing ion-rich materials onto the at least one region of the EO material and heating the region. Heating of the material allows the ions to diffuse into the material, altering conductivity of the EO material, depending on the temperature and duration of the heating process.

[0204] Doping can include using Li glass, LiF, or Li pentoxides to produce a heightened conductivity region.

[0205] Alloying includes heating a metal to an elevated temperature (e.g., greater than 300 C) for a duration inversely related to the temperature. Metals for alloying can include Al, Ti, Cr, Ni, Cu, or their respective alloys.

[0206] The remainder of the features and example features of the 4thexample embodiment are as described above with respect to the 1staspect.

[0207] The descriptions of the various embodiments of the present disclosure have been presented for purposes of illustration, but are not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the describedHCU-07125 (HU9580, HU9581) embodiments. The terminology used herein was chosen to best explain the principles of the embodiments, the practical application or technical improvement over technologies found in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

HCU-07125 (HU9580, HU9581) CLAIMS 1. A device, comprising: a waveguide, comprising an electro-optic (EO) material having a first side and a second side, the first side of the EO material comprising at least one modified region; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on the first side of the substrate; and an electrode disposed in the at least one modified region of the EO material, wherein the at least one modified region is a chemically reduced region or a heightened conductivity region.

2. The device of Claim 1, wherein the waveguide comprises a ridge disposed on the first side of the EO material.

3. The device of Claim 1, wherein the waveguide comprises a ridge disposed on or adjacent to the second side of the EO material.

4. The device of Claim 2, wherein the ridge comprises the EO material.

5. The device of Claim 2 or Claim 3, wherein the ridge comprises a ridge material, different from the EO material.

6. The device of Claim 1, wherein the waveguide comprises a doped volume disposed within the EO material.HCU-07125 (HU9580, HU9581) 7. The device of Claim 1, wherein the EO material comprises at least two modified regions, and further wherein the device comprises at least two electrodes, each electrode disposed in the modified regions.

8. The device of Claim 1, wherein the device comprises: at least one electrode disposed in the at least one modified region; an insulator disposed on the EO material; and at least one electrode disposed on the insulator.

9. The device of any one of Claims 1-8, wherein the at least one modified region is recessed in the EO material.

10. The device of any one of Claims 1-9, further comprising a support disposed on the second side of the substrate.

11. The device of Claim 1, wherein the device further comprises: at least one electrode disposed in the at least one modified region; a support having a first side and a second side, the support disposed on the second side of the substrate; and at least one electrode disposed either on the second side of the substrate or the second side of the support.

12. The device of Claim 10 or 811 wherein the support comprises Si, lithium niobate, SiO2, or Al2O3.HCU-07125 (HU9580, HU9581) 13. The device of any one of Claims 1-12, further comprising a cladding disposed over the first side of the EO material.

14. The device of Claim 13, wherein the cladding comprises SiO2.

15. The device of any of Claims 1-14, wherein the substrate comprises a material selected from SiO2, Al2O3, SiON, HfO2, or ZrO2.

16. The device of any of Claims 1-15, wherein the EO material is selected from a metal oxide, metal nitride or a ferroelectric material.

17. The device of Claim 16, wherein the EO material is selected from LiNbO3, LiTaO3, BaTiO3, KNbO3, ScxAl(1-x)N, ScN, AlScN, or AlN.

18. The device of Claim 15, wherein the EO material is selected from potassium titanyl phosphate (KTP) or potassium dihydrogen phosphate (KDP).

19. The device of any one of Claims 1-18, wherein at least one electrode comprises Ti / Au, indium tin oxide (ITO), Al, Cr, Ni, NiCr, MoOx, LiFx, CsFx, or KFx.

20. The device of Claim 1, wherein: the first side of the EO material comprises a ridge and two planes, the ridge comprising the EO material;HCU-07125 (HU9580, HU9581) the two planes are separated by the ridge, each planes extending along the ridge on the first side of the substrate, each plane comprising a chemically reduced region recessed in the EO material; the device comprises two electrodes disposed on the chemically reduced regions such that the ridge extends between the two electrodes; and further wherein: the EO material is LiNbO3; the substrate comprises SiO2; the electrodes comprise Ti / Au; the device further comprises a support disposed on the second side of the substrate, the support comprising Si; the device further comprises a cladding disposed over the first side of the EO material, the cladding comprising SiO2.

21. A device, comprising: a waveguide, comprising an alkali metal ion-containing electro-optic (alkali EO) material having a first side and a second side; a layer of an alkali ion-blocking material disposed on at least one of the first side and the second side of the alkali EO material; a substrate having a first side and a second side, wherein the second side of the EO material is disposed on or adjacent to the first side of the substrate.

22. The device of Claim 21, wherein the waveguide comprises a ridge disposed on or adjacent to the first side of the alkali EO material.HCU-07125 (HU9580, HU9581) 23. The device of Claim 21, wherein the waveguide comprises a ridge disposed on or adjacent to the second side of the alkali EO material.

24. The device of Claim 22, wherein the ridge comprises the alkali EO material.

25. The device of Claim 22 or Claim 23, wherein the ridge comprises a ridge material, different from the alkali EO material.

26. The device of Claim 21, wherein the waveguide comprises a doped volume disposed within the alkali EO material.

27. The device of any one of Claims 21-26, further comprising a support disposed on the second side of the substrate.

28. The device of Claim 27, wherein the support comprises Si, lithium niobate, SiO2, or Al2O3.

29. The device of any one of Claims 21-28, further comprising a cladding disposed over the first side of the alkali EO material.

30. The device of Claim 29, wherein the cladding comprises SiO2.

31. The device of any of Claims 21-30, wherein the substrate comprises a material selected from SiO2, Al2O3, SiON, HfO2, or ZrO2.HCU-07125 (HU9580, HU9581) 32. The device of any one of Claims 21-31, wherein the alkali EO material is LiNbO3or LiTaO3.

33. The device of any one of Claims 21-32, wherein the alkali ion-blocking material is selected from Al2O3, ZrO2, HfO2, SrF2, or CaO.

34. A method of manufacturing a device, comprising: providing a waveguide comprising an electrooptical (EO) material having a first side and a second side, depositing a resist over a portion of the first side of the EO material; chemically reducing at least one region of the EO material, thereby producing at least one reduced region; optionally, removing the resist; and depositing at least one electrode on the at least one reduced region.

35. The method of Claim 34, wherein reducing the at least one region comprises bombarding the at least one region with ions.

36. The method of Claim 35, wherein the ions are selected hydrogen, helium, argon, xenon or krypton.

37. The method of Claim 34, wherein reducing the at least one region comprises contacting the at least one region with a reducing agent.HCU-07125 (HU9580, HU9581) 38. The method of Claim 37, wherein the reducing agent is Ti, trimethylaluminum (TMA), CaH2, or NaBH4.

39. The method of Claim 34, wherein reducing the at least one region comprises exposing the at least one region to a thermal treatment in a reducing environment.

40. A method of manufacturing a device, comprising: providing a waveguide comprising an electrooptical (EO) material having a first side and a second side, subjecting at least one region of the EO material to a treatment that produces a heightened conductivity region, thereby producing at least one heightened conductivity region; and depositing at least one electrode on the at least one reduced region.

41. The method of Claim 40, wherein subjecting at least one region of the EO material to a treatment that produces a heightened conductivity region comprises ion implantation, doping, or alloying.

Images