HTML
-
Our CHPW design effectively addresses the loss-confinement tradeoff by supporting supermodes formed through the superposition of an SPP mode and a hybrid plasmonic waveguide (HPW) coupled via a thin metal film. Light–matter interaction on two types of plasmonic interfaces—metal with high-index dielectric and metal with low-index dielectric—is suggested by combining dissimilar modes in one structure. The latter confines modal energy within the dielectric layer, allowing for more powerful linear and nonlinear processes. The modal analysis of the CHPW structure at $ \lambda = 1550 $ nm was performed using the Lumerical Finite-Difference Eigenmode (FDE) solver, with the perfectly matched layer boundary settings and mesh set to a maximum value of 0.001 μm. The FDE solver calculates the modal spatial profile and frequency dependence by solving Maxwell’s equations on a cross-section mesh of the waveguide. Table 1 summarizes the material parameters used in CHPW simulations.
Layer DC permittivity Work function (eV) Bandgap (eV) Electron mobility (cm2/(V·s)) Hole mobility (cm2/(V·s)) Refractive index Si 11.90 4.61 1.10 1471 470.50 3.48 ITO 3.90 5.40 2.60 12.10 – 2.14 ± j0.14 SiO2 3.90 – – – – 1.47 Al – 4.28 – – – 2.73 a-Si – – – – – 3.73 Table 1. Summary of material parameters used in CHPW simulations
The plasmonic mode profiles were calculated based on the carrier distribution in the ITO layer. The local permittivity in the ITO layer will determine the modal shape. However, permittivity is a function of carrier distribution. It can be calculated once the carrier distribution is determined. The optical properties of ITO, or more precisely, its permittivity, change as a function of the carrier distribution. The Drude model can describe the relation between the carrier distribution and permittivity as described by the following relation
$$ \begin{array}{l} \varepsilon_ {\rm{ITO}}(z,\omega) = \varepsilon_\infty - \dfrac{\omega^{2}_ {\rm{p}}(z)}{\omega^{2} - j\gamma \omega} \end{array} $$ (1) described in the Supplementary Information document, which sufficiently describes the relative permittivity at NIR wavelengths. The plasma frequency $ \omega_ {\rm{p}} $ is directly linked to the carrier density n via the following relation
$$ \begin{array}{l} \omega_ {\rm{p}} = \sqrt{q^{2}n/(\varepsilon_{0}m^{*})} \end{array} $$ (2) described in the Supplementary Information document. The parameters of the Drude model depend on the ITO material properties where $ \gamma = 1.8 \times 10^{14} $ rad/s at $ \lambda = 1550 $ nm and an effective electron mass $ m^{*} $ of 0.35$ m_0 $, where $ m_{0} \approx 9.109 \times 10^{-31} $ kg is the rest mass of an electron.
We used Lumerical’s Charge Transport (CHARGE) solver to calculate the spatial depletion and accumulation of electrons in the ITO layer across the SiO2/ITO interface when the bias voltage was varied between −10 V and 25 V. Theoretical information about the used algorithm is presented in the Supplementary Information document. The plots were obtained by calculating the carrier distributions at the SiO2/ITO interface at different voltage levels. These depletion and accumulation processes change the permittivity of ITO thin films in compliance with the Drude model, which opens gateways for realizing optical modulation. Once the change in permittivity reaches the ENZ regime from the modulation, it dramatically changes the absorption of ITO.
The density of charge carriers at different voltage levels obtained from the CHARGE solver was transported to Lumerical FDE MODE solver through an index perturbation (np density) grid, which tuned the optical constant of ITO, leading to a change in the complex effective refractive index of the mode propagating in the CHPW structure. The plots exhibit changes in the real and imaginary parts of the effective refractive index with varying applied voltage. ER and the ON-state PL are parameters that can be used to quantify an optical modulator’s performance. An effective modulator has high ER and low PL levels. The complex effective refractive index resulting from the mode computations allowed us to describe the ER and PL for a given voltage bias. The PL was calculated using64
$$ \begin{array}{l} {\rm{PL\; [dB/ {\text μ} m]}} = 10 \log(e \cdot \alpha) \end{array} $$ (3) where $ e = \sum_{n = 0}^{\infty} {1}/{n!} $ is the base of natural logarithms (Euler’s number) and $ \alpha $ is the absorption coefficient, which can be determined as follows
$$ \alpha = \frac{4\pi k_ {\rm{eff}}}{\lambda} $$ (4) where $ k_ {\rm{eff}} $ is the effective extinction coefficient (i.e., the imaginary part of the effective refractive index). The ER in dB/μm is the difference between the absorption coefficient of the transmitting ON-state ($ \alpha_ {\rm{ON}} $) and absorbing OFF-state ($ \alpha_ {\rm{OFF}} $), i.e.,
$$ \begin{array}{l}{ \rm{ER}} = 10\log (e (\alpha_ {\rm{OFF}}-\alpha_ {\rm{ON}})) \end{array} $$ (5) Fig. 2 displays the observed changes in PL levels and the real and imaginary parts of the effective refractive index (Fig. 2a, b, respectively) of the CHPW waveguide varying applied voltage using the GIL model for an ITO with $ n = 6.3 \times 10^{20} $ cm−3. These plots demonstrate significantly lower variations in the optical constants given the nature of our ITO layer with a doping concentration $ n = 6.3\; \times 10^{20} $ cm−3 being within the ENZ regime.
Fig. 3 plots the observed changes in carrier density and electric field (Fig. 3a, b, respectively) levels across the CHPW waveguide with varying voltage using the GIL model. In Fig. 3a, we observe a depletion of free electrons after applying a negative bias at the insulator side. On the other hand, an accumulation of free electrons was achieved by applying a positive bias at the same terminal. This accumulation of electrons led to higher current flow levels that gradually increased with positive bias application. As one can observe in Fig. 3b, the magnitude of the electric field across the SiO2 layer is linearly dependent on the applied bias and is directly influenced by carrier accumulation/depletion at the SiO2/ITO interface, demonstrating that the ITO’s carrier dispersion properties are the key mechanism behind the operation of the CHPW modulator.
-
At the waveguide region, the Al/SiO2/ITO stack is equivalent to an MIS described by a capacitance $ C_ {\rm{MIS}} $ and a parallel (or shunt) resistance $ R_ {\rm{MIS}} $. Fig. 7 displays the capacitance-voltage ($ CV $) measurements that were performed on circular MISMs with a radius size of 100 μm. While the area-normalized capacitance values shown in Fig. 7 are reported for $ CV $ measurements performed with a small-signal frequency component of 1 MHz, the Supplementary Information document provides $ CV $ measurements performed with small-signal frequency components of 10 kHz and 100 kHz. The electrical probe station used to measure this structure is described in the materials and methods section under electronic properties. To have a properly functioning MISM device, the phase angle ($ \theta $) between the current through and voltage across the device must be around −90°C, implying that there is no significant influence of series resistance ($ R_ {\rm{s}} $) on the measured impedance ($ Z $) and hence it is entirely imaginary considering the following expression for a capacitor impedance
Fig. 7 MISM CV plots—Frequency response curves of fabricated SiO2/ITO MISMs with different nITO levels as functions of bias voltage.
$$ Z = R_ {\rm{s}} + \frac{1}{j\omega C} $$ (6) where $ j = \sqrt{-1} $ is the imaginary unit and C is the measured capacitance. In this case, the electrical reactance $ X_ {\rm{C}} $ is defined as $ X_ {\rm{C}} = -1/(\omega C) $ and the phase angle $ \theta_{Z} $ between the current and voltage can be calculated using
$$ \theta_{Z} = \arctan{\Bigg(\frac{X_ {\rm{C}}}{R_ {\rm{s}}}\Bigg)} $$ (7) The Supplementary Information document provides $ \theta_{Z} $ adherence curves for the poly-TiN/SiO2/ITO/Al MISM at 10 kHz, 100 kHz, and 1 MHz. By performing a small-signal analysis for the MISM capacitor structure, we calculated a negative bias resistance of 86.7 $ \mu\Omega $ with a capacitance value of 29 fF/μm2 at 20 GHz. These values, alongside the radius of the capacitor, yield an anticipated RC bandwidth of roughly 200 GHz, as obtained from calculated impedance parameters. It is worth noting that the large capacitor area relative to the thickness of the insulator and semiconductor layers yielded relatively small and large lumped series resistance and capacitance, respectively.
For an ITO layer with a $ n_ {\rm{ITO}} $ = 6.05 × 1020 cm−3 at 0.50 V and a small-signal frequency of 1 MHz, the calculated capacitance value based on simulations ranged between 1.9 and 3.07 fF/μm2 depending on the ITO and SiO2 layer thicknesses. The measured capacitance value at 0.50 V and 1 MHz small-signal frequency is 2.19 fF/μm2 as can be seen in Fig. 7. This is within the calculated capacitance range for an MISM device with around a 2.5 nm average increase in ITO and SiO2 layer thicknesses. Therefore, we must note that variations in the actual structural layer thicknesses are not the only factor that affects the measured device parameters. Fabrication process imperfections and defects must have induced parasitic effects, which led to an increase in the “effective”' dielectric and semiconductor layers thicknesses (i.e., the approximately 2.5 nm increase in ITO and SiO2 layers), further contributing to lower measured capacitance levels.
As observed in Fig. 7, with a modest variation in negative bias levels, a significant drop in normalized capacitance levels occurred at higher positive bias levels, illustrating that higher current injection levels are achieved. This is a direct result of carrier accumulation at the SiO2/ITO interface. This mechanism behind the dispersion properties of charge carriers within the semiconductive ITO layer induces the necessary capacitance modulation upon which optical modulation is directly dependent. At all small-signal frequency levels (refer to the Supplementary Information document), capacitance levels are the highest across MISM devices with $ {F_{{{\rm{O}}_2}}} $ of 0.70 sccm (sample 3, $ n_ {\rm{ITO}} = 8.35 \times 10^{20} $ cm−3) and 0.80 sccm (sample 4, $ n_ {\rm{ITO}} = 6.05 \times 10^{20} $ cm−3), demonstrating relatively more robust capacitors when the device is operating close to or well within the ENZ regime. When comparing the $ CV $ measurements performed with a small-signal frequency component of 1 MHz (Fig. 7) with those with small-signal frequency components of 10 kHz and 100 kHz (refer to the Supplementary Information document), it is worth noting that the fabricated MISM devices display relatively high-performance levels when biased with small-signal frequency levels of 100 kHz and 1 MHz. At these small-signal frequency levels, lower noise levels (at 100 kHz and 1 MHz small-signal frequencies) and higher areal capacitance levels (at 1 MHz small-signal frequency) were achieved while maintaining considerable modulation in capacitance values.
MISM devices with ITO films deposited at $ {F_{{{\rm{O}}_2}}} $ = 0.70 and sccm have higher oxygen contents and therefore are less conductive. The ITO optical attributes described in Figure S1 of the Supplementary Information document illustrate how the dielectric constant, and hence the refractive index, varies with carrier concentration. As the carrier concentration increased through doping, there were points where the properties of ITO transitioned from those of a conductor to a semiconductor and then to an insulator66. Therefore, the $ CV $ measurements done on these ITO films deposited at different ${F_{{{\rm{O}}_2}}} $ levels are expected to reflect a trend of increasing refractive index for higher $ {F_{{{\rm{O}}_2}}} $67.
In Fig. 7, we can observe that at a negative bias around −1 V, there are dips in capacitance levels, where depletion occurs and $ {\rm {Re}}\{n\} $ drops, and hence the ITO electrical properties at the interface shift towards those of a conductive dielectric and thus start to exhibit smaller capacitance values. For small positive bias values (partial forward bias), we conjecture that the observed $ CV $ curves follow the $ {\rm{Re}}\{n\}{\text -}V $ curve exhibited in Fig. 2b. As the bias increases, accumulation effects occur, contributing to higher capacitance levels. At negative bias levels, the shape of the $ CV $ curves resembles an inversion-region operation, and hence the capacitance levels start to increase again. We also observe what resembles a roll-out at higher positive bias levels, which is possible cause by other competing effects. For instance, enhanced optical confinement is possible, implying an enhanced overlap with the electric field, which lowers capacitance levels. This phenomenon can be explained by the effects of plasma and the correlations between optical parameters and carrier density68. $ {\rm{Re}}\{\varepsilon\} $ is related to the energy stored within a medium and is affected by capacitance and $ {\rm{Im}}\{\varepsilon\} $ is associated with the loss or gain of energy within a medium. Capacitors with ITO films deposited at higher ${F_{{{\rm{O}}_2}}}$ are more resistive and therefore exhibit higher $ {\rm{Re}}\{\varepsilon\} $, which intuitively explains the greater capacitance levels displayed by ITO films deposited with ${F_{{{\rm{O}}_2}}}$ = 0.70 and 0.80 sccm. Finally, we note that as the operating frequency of the MISM devices gets higher, the $ CV $ curve gets smoother. This is probably caused by the noises being relatively lower in the driving frequency source. While for the 10 kHz operating frequency case, there are more noticeable slight dips in the capacitance levels around −1 V compared to higher operation frequencies. This may be caused by the higher modulation speed of carriers at the SiO2/ITO interface.
-
Fig. 8a displays the forward-biased $ IV $ response of the CHPW, manifesting an irreversible oxide breakdown around 35 V. This corresponds to a SiO2 dielectric strength of 17.5 MV/cm. The tunneling current up until 25 V bias was less than 2 nA, which indicates excellent capacitive coupling characteristics, and thus the shunt path bypassing the capacitor can be neglected (i.e., $ R_ {\rm{MIS}} \rightarrow \infty $). Moreover, the device’s capacitance was determined at an AC frequency and voltage of 1 MHz and 10 mV$ _ {\rm{rms}} $, respectively. Because of the compact device footprint nature and instrument resolution limitation, the high-frequency branch of the MIS $ CV $ curve could not be observed. $ C_ {\rm{MIS}} $ for the 5- and 10-μm long devices were measured to be about 3 fF and 5 fF, respectively, at a bias of 25 V. Fig. 8b displays the photocurrent characteristics. At 10 V, the MIS photodiode exhibited a photocurrent-to-dark ratio of 5.77.
Fig. 8 CHPW IV response—a Measured IV characteristics demonstrating a breakdown voltage around 35 V. b Measured dark- and photocurrent characteristics.
Several contact resistance terms are considered when describing the modulator’s electrical characteristics. The probe contact resistance is negligible as the coaxial high-frequency infinity waveguide probe used for device characterization has a typical contact resistance below 0.1 Ω to Al pads. The series resistance from the sputtered Al contact (around 5 Ω) is small relative to that of the other sputtered thin films, as their resistivities were measured to be between 8 × 10−7 Ω·cm and 2.9 × 10−4 Ω·cm. Therefore, the modulator resistance is dominated by the series resistance $ R_ {\rm{s}} $ of ITO and the contact resistance $ R_ {\rm{c}} $ between Al and ITO. Based on the dimensions of the fingers and modulators, $ R_ {\rm{s}} $ was determined to be 911.2 $ \Omega $ and 896.6 Ω for modulator lengths of 5 and 10 μm, respectively.
The $ R_ {\rm{c}} $ for the CHPW modulator was determined via circular transmission line measurement (CTLM). CTLM is preferred over standard transmission line measurement for devices with micron-scale areas as current crowding can be avoided, and electrical isolation is automatically established69. As shown in the inset in Fig. 9, the CTLM test structure consisted of a conducting circular inner region of radius L, an air gap of width d, and a conducting outer part70, 71. Specifically, the total resistance $ R_ {\rm{T}} $ between the internal and external contacts can be expressed as
Fig. 9 Contact resistance characterization—a A plot summarizing CTLM results for the Al (250 nm)/ITO (10 nm) contacts with L = 75 μm. b IV curves for a CTLM structure with d = 10 μm and L = 75 μm.
$$ R_ {\rm{T}} = \frac{R_ {\rm{sh}}}{2\pi L}(d + 2L_ {\rm{T}})\mathbb{C}, $$ (8) where $ R_ {\rm{sh}} $ is the sheet resistance, $ \mathbb{C} $ is a correction factor, and $ L_ {\rm{T}} $ is the transfer length describing the average distance traveled by carriers within the semiconductor region. These two parameters were calculated as $ \mathbb{C}=L/d\ln(1+d/L) $ and $ L_ {\rm{T}} = \sqrt{\rho_ {\rm{c}}/R_ {\rm{sh}}} $, where $ \rho_ {\rm{c}} = R_ {\rm{c}}(\pi L^{2}_ {\rm{T}}) $ is the specific contact resistivity that takes into account all interfacial effects and effects from regions close to the interface71. CTLM results from the ITO/Al interface are depicted in Fig. 9. The conducting regions consisted of 250-nm-thick Al on 10-nm-thick ITO on SOI. Because ITO is a highly degenerate semiconductor, the barrier height and the width of the depletion region at the ITO/metal interface are small. Therefore, ITO forms an ohmic contact with metals, allowing for smooth carrier injection. Using the measured $ R_ {\rm{sh}} $ of 290 $ \Omega/\square $, the $ \rho_ {\rm{c}} $ of the CTLM structures is extracted to be 7.2 × 10−3 Ω·cm2. This is significantly lower than the 3 × 10−1 Ω·cm2 previously reported for ITO/Al contacts with similar $ R_ {\rm{sh}} $ values72, but higher than 10−6Ω·cm2 as estimated by the standard metal/semiconductor contact theory 73. Such deviation from theoretical value is common for ITO devices, as the interaction between oxygen and metal has been observed to alter ITO’s carrier density at the interface and thus leads to an increase in $ \rho_ {\rm{c}} $. With contact area of 82.72 μm2, the $ R_ {\rm{c}} $ for the fabricated CHPW modulators was measured to be 8.7 kΩ.
-
Lower carrier concentrations lead to lower optical losses, which result in better modulator’s IL but at the expense of lower conductivity, implying lower $ RC $ time constants. Therefore, we require the carrier mobility to be as high as possible to increase the conductivity without increasing the carrier concentration, leading to lower insertion and optical losses. The sheet resistivity of ITO is strongly correlated with the O2 and Sn contents. The density of free electrons in ITO is directly influenced by O2 vacancies, with an inverse proportionality to the O2 content. On the other hand, the electron mobility in ITO is heavily influenced by the O2 content, as higher O2 contents enhance the crystallization of ITO thin films74.
-
Unbiased-state CHPW modulator characterization has been performed using the cut-back method. As observed in Fig. 10, the IL and CE at $ \lambda = 1550 $ nm were extracted to be around 0.128 dB/μm and 52.50%, respectively. The loss value is higher than the simulated value of 0.03 dB/μm (simulations are not shown here). The higher experimental IL level is attributed to the additional free carrier absorption caused by elevated ITO carrier density and scattering losses from the contact region. Effects from deposited layer roughnesses and various intrinsic defects, such as surface states and oxygen vacancies, could have played an additional role in this higher measured IL level75-77. Nonetheless, IL and CE values exhibit low variations between $ \lambda = 1520 $ nm and 1600 nm, indicating a potential broadband modulator operation.
-
Fig. 11a plots the transmission of 10-μm-long CHPW modulators under forward bias. The maximum voltage was kept below 26 V to avoid oxide breakdown, and the results have been normalized to optical transmission at zero bias. We note that the observed optical transmission level decreases marginally to a small extent as the voltage is increased up to 20 V. After the applied voltage exceeds 20 V, a dramatic increase in optical absorption takes place, and transmission is reduced to −10 dB at 25 V, which corresponds to strong ER of approximately 1 dB/μm, one of the highest measured to-date to our knowledge6. The reason for resorting to applying a high voltage level of 20 V is that a 20-nm-thick SiO2 (rather than a high-κ dielectric material) layer was incorporated in the MOS region of the modulator for simplicity of design, hence the facile integration of SiO2/ITO. In other words, our CHPW design necessitates applying a relatively high electric field, given the relatively low dielectric constant of SiO2, which serves as a reference to other field-confining materials. The voltage can be reduced to improve the modulation efficiency by incorporating a high-κ dielectric material instead, such as hafnium oxide (HfO2).
Fig. 11 CHPW DC/AC response—a Normalized transmission curves of a 10-μm-long SiO2/ITO-based CHPW modulator as functions of bias voltage. The behavior of the modulator without an ITO layer is also displayed. b Normalized transmission of 5- and 15-μm-long SiO2/ITO-based CHPW modulators as functions of bias voltage. c Normalized transmission curves of a 10-μm-long SiO2/ITO-based CHPW modulator as functions of bias voltage and wavelength. d Optoelectronic bandwidth measurements of the SiO2/ITO-based CHPW modulator.
The behavior of the same modulator without an ITO layer is also provided for comparison. As no modulation can be observed under both scenarios, the measured modulation response can be attributed to field-induced carrier accumulation rather than drift in the set-up, optothermal effects at high bias voltages, or optoelectronic events in materials other than ITO. Moreover, modulation depth varies with device length, and the normalized transmission for a 15-μm-long modulator can reach −12 dB at 26 V as can be seen in Fig. 11c, where the broadband response of the CHPW modulator is shown. It was verified by measuring the optical transmission as a function of voltage at different wavelengths (from $ \lambda = 1.53 $–1.6 μm), as shown in the figure. Beyond $ \lambda = 1550 $ nm, the modulation depth starts to decrease as the permittivity of ITO moves away from the ENZ regime. However, an ER of about 1 dB/μm was maintained between $ \lambda = 1.53 $ μm and 1.56 μm.
-
Fig. 11d depicts the 10-μm-long modulator’s frequency response. One can observe that the frequency response is flat up to about 26 GHz, and the calculated 3-dB cutoff frequency (i.e., anticipated $ RC $ bandwidth), given a parasitic capacitance of 5 fF, is approximately 636 GHz for 50 Ω load resistance. The frequency response measurement, including the substrate characteristics and effects from the RF probe, was normalized to the response peak point. Using small-signal analysis, we calculated the junction current at 0 V. We extracted the resistance from the real part of the resulting impedance for operating frequencies up to 636 GHz at the CHPW’s SiO2/ITO heterojunction. At static and up to 26 GHz, the resistance was found to be around 75.6 Ω at 0 V, while the junction capacitance was calculated numerically using the finite difference method (i.e., $ dv/dq $ with infinitesimal voltage steps) and was estimated to be around 0.21 fF/μm (normalized to the propagation distance). For operating frequencies from 26 to above 636 GHz (up to 700 GHz), the heterojunction resistance and capacitance levels dropped dismally to approximately 73.6 Ω and 0.18 fF/μm, respectively.
The modulation speed of a CHPW is determined by two factors: (1) the time the accumulation layer inside the semiconductor takes to form, and (2) the delay time arising from parasitic $ RC $ effects. Electron accumulation is a relatively fast process dictated by majority carrier dynamics. The formation time can be estimated using $ \tau = t/v_ {\rm{drift}} $, where, in our case, $ t = t_ {\rm{ITO}} = 10 $ nm and $ v_ {\rm{drift}} $ are the ITO film thickness and the drift velocity of carriers in ITO, respectively. Here, $ v_ {\rm{drift}} $ can be estimated using $ v_ {\rm{drift}} = \mu E $, where μ is the carrier mobility in ITO and $ E $ is the applied electric field. The latter is proportional to the magnitude of applied gate voltage divided by the dielectric film thickness ($ t_ {\rm{SiO _{2}}} = 20 $ nm). Using $ \mu = 33 $ cm2/(V·s) as obtained from Hall measurements, the resulting modulation speed exceeds 300 THz. Similarly, the electron scattering rate of ITO is around 28.65 THz as derived from the collision frequency ($ \gamma = 1.8 \times 10^{14} $ rad/s); hence, the modulator speed is also not limited by the transient response of the accumulation layer when the gate is turned off78. Thus, similar to other MIS-based modulators, our CHPW modulator’s speed is dictated by the $ RC $ time constant79. While the $ RC $ response of the 10-μm-long CHPW modulator has an anticipated 3-dB cutoff frequency above 600 GHz, the bandwidth of the testing instrument used to characterize the modulator is limited to 26 GHz. Because the response of light to the plasma dispersion effect is significantly faster than the $ RC $ limit, delays from strong local field decays can be disregarded, and, as a result, the theoretical electrical modulation speed of 636 GHz is directly relevant to the optical domain.
-
The performance of the 10-μm-long CHPW modulator was characterized over broad wavelength and temperature ranges. The evolution in the modulator’s ER level at elevated stage temperatures (T = 35–100 °C) relative to the modulator’s response at room temperature (ΔERRT, calculated as an absolute numerical difference) is shown Fig. 12 as a function of voltage. Because of power output limitations in the thermoelectric cooler (TEC) source meter, the Peltier heat pump requires a voltage-limited mode of operation, and, therefore, the stage temperature exhibited a ±0.2 °C fluctuation. Regardless of the noise caused by temperature fluctuations, the modulator’s ER performance did not degrade; instead, it increased with increasing stage temperature. For instance, an approximately 0.74 dB increase in ER was detected at 50 °C and 100 °C at 25 V.
MISM $ CV $ characterization
CHPW electrical characterization
CHPW optical modulator characterization
CHPW unbiased response
CHPW DC response
CHPW AC response
CHPW ER analysis
-
We have successfully demonstrated the facile growth, integration, and characterization of Al/ITO/SiO2 MISM heterojunctions on a poly-TiN/Si platform, facilitating the design and fabrication of practical CHPW devices. The research focus on oxide materials for plasmon-electronics has significantly surged, given its potential to achieve tunable material characteristics. However, achieving optimal device performance requires extended optimization of material quality, device design, and process technology. Herein, we focused on developing insulator–semiconductor device topologies by integrating an electro-optic SiO2/ITO interface in MIS devices housed on bulk Si and SOI platforms. The advantage of this integration scheme is rooted in the use of CMOS-compatible materials, along with breaking the loss-confinement barrier in plasmon-electronic devices by tuning the waveguide architectural characteristics and hence achieving control over supermode attributes. XRD, $ IV $, $ CV $, and parameter analyzer measurements were used to characterize the grown materials and fabricated devices. The fabricated CHPW modulator shows ER and IL levels of around 1 dB/μm and 0.128 dB/μm, respectively, for a 10-μm-long waveguide length. While our initial device performance seems promising, challenges still exist, and these are the development of the ability to produce large-area optoelectronic devices using current material synthesis and device fabrication methodologies. Table 2 summarizes our CHPW modulator performance and provides a performance comparison with select modulator configurations as reported in the literature.
Modulator device structure Footprint (μm2) Speed (GHz) Operation wavelength (nm) Driving voltage (V) ER (dB) Ref. MIS–Schottky CHPW 10 × 0.2 >600 1550 25 10.63 This work MISM capacitor 3 × 104 >200 1550 15 – This work Plasmonic Si Mach–Zehnder 10 × 1 >70 1550 3 6 Haffner et al. 80 Graphene-based Si waveguide 50 × 10 5.9 1550 2.5 5.2 Hu et al. 81 Electro-optic graphene 1.5 × 30 >150 1570 10 28 Phare et al. 82 Table 2. Summary of fabricated CHPW modulator performance parameters, along with performance comparison with select modulator configurations.