Enhanced optical four-wave-mixing in integrated ring resonators with graphene oxide ﬁlms

Layered two-dimensional (2D) graphene oxide (GO) ﬁlms are integrated with micro-ring resonators (MRRs) to experimentally demonstrate enhanced nonlinear optics in the form of four-wave mixing (FWM). Both uniformly coated and patterned GO ﬁlms are integrated on CMOS-compatible doped silica MRRs using a large-area, transfer-free, layer-by-layer GO coating method together with photolithography and lift-oﬀ processes, yielding precise control of the ﬁlm thickness, placement, and coating length. The high Kerr nonlinearity and low loss of the GO ﬁlms combined with the strong light-matter interaction within the MRRs results in a signiﬁcant improvement in the FWM eﬃciency in the hybrid MRRs. Detailed FWM measurements are performed at diﬀerent pump powers and resonant wavelengths for the uniformly coated MRRs with 1-5 layers of GO as well as the patterned devices with 10-50 layers of GO. The experimental results show good agreement with theory, achieving up to ˜7.6-dB enhancement in the FWM conversion eﬃciency (CE) for an MRR uniformly coated with 1 layer of GO and ˜10.3-dB for a patterned device with 50 layers of GO. By ﬁtting the measured CE as a function of pump power for devices with diﬀerent numbers of GO layers, we also extract the dependence of GO’s third-order nonlinearity on layer number and pump power, revealing interesting physical insights about the evolution of the layered GO ﬁlms from 2D monolayers to quasi bulk-like behavior. These results conﬁrm the high nonlinear optical performance of integrated photonic resonators incorporated with 2D layered GO ﬁlms.


Introduction
Nonlinear integrated photonic devices offer powerful solutions to generate and process signals all-optically, with far superior processing speed compared to electronic devices, as well as the added benefits of a compact footprint, low power consumption, high stability, and the potential to significantly reduce cost by mass production [1,2].Four-wave mixing (FWM), as a fundamental third-order (χ (3) ) nonlinear optical process [1,3], has been widely used for achieving all-optical signal generation and processing, such as wavelength conversion [4,5], optical frequency comb generation [6,7], optical sampling [8,9] , quantum entanglement [10,11], and many others [12][13][14].
Integrated micro-ring resonators (MRRs), with their strong light confinement in compact micro-scale resonant cavities, are key building blocks for photonic integrated circuits and play important roles in many applications from optical communications to optical interconnects, photonic processing, and biosensing [15][16][17].Compared with FWM in integrated waveguides, FWM in integrated MRRs can provide dramatically enhanced conversion efficiencies (CE) due to resonant enhancement of the optical field [18,19], thus significantly reducing the power requirements.FWM has been demonstrated in integrated MRRs fabricated on III-V platforms including GaAs and AlGaAs [19][20][21], and also complementary metal-oxide-semiconductor (CMOS) compatible platforms including silicon, silicon nitride, and high index doped silica glass [22][23][24].Although silicon has been a leading platform for integrated photonic devices, its strong two-photon absorption (TPA) at near-infrared wavelengths (TPA coefficient: ~0.9 cm/GW [2]) poses a fundamental limitation for FWM in the telecommunications band [1].
CMOS compatible platforms such as silicon nitride and high index doped silica glass have a much weaker TPA (no TPA observed in the telecommunications band even up to extremely high light intensities of ~25 GW/cm 2 [2]), although they face limitations in terms of FWM efficiency since their Kerr nonlinearity (n2) is over an order of magnitude smaller than that of silicon [2].
The quest for high-performance nonlinear integrated photonic devices has motivated the use of highly nonlinear materials on chips to overcome the limitations of existing platforms [8,25].
Owing to its ease of preparation as well as the tunability of its material properties, GO has become a highly promising member of the graphene family [27,[36][37][38][39].Previously, we reported GO films with a giant Kerr nonlinearity (n2) of about 4 orders of magnitude higher than that of silicon [27,40], and achieved enhanced FWM CE in GO-coated doped silica waveguides of up to 6.9 dB for a 1.5-cm-long waveguide uniformly coated with 2 layers of GO [29].Moreover, GO has a material absorption that is over 2 orders of magnitude lower than graphene [29] as well as a large bandgap (2.1−2.4 eV) that yields a low TPA in the telecommunications band [41,42].Recently, we achieved highly precise control of the placement, thickness, and length of the GO films coated on integrated photonic devices by using a large-area, transfer-free, layer-bylayer GO coating method together with standard photolithography and lift-off processes [43].This overcomes critical fabrication limitations in terms of layer transfer and precise patterning for on-chip integration of 2D materials and represents a significant advance towards manufacturing integrated photonic devices incorporated with 2D materials.
In this paper, we use our GO fabrication techniques to demonstrate enhanced FWM in MRRs integrated with 2D layered GO films.Owing to the strong light-matter interaction in the MRRs incorporating highly nonlinear GO films, the FWM efficiency in the hybrid MRRs is significantly improved.We perform FWM measurements at different pump powers and resonant wavelengths for CMOS-compatible doped silica MRRs uniformly coated with 1−5 layers of GO as well as devices patterned with 10−50 layers of GO, achieving up to ~7.6-dB enhancement in the FWM CE for an MRR uniformly coated with 1 layer of GO and ~10.3-dB for a device patterned with 50 layers of GO.Compared with the uniformly coated MRRs, the MRRs with patterned GO provide more flexibility to balance the trade-off between FWM enhancement and linear loss, which is critical for optimizing the CE and fully exploiting the high Kerr nonlinearity of 2D GO films.We also fit the measured CE based on FWM theory and obtain the dependence of the Kerr nonlinearity of the GO films on the number of layers and on the pump power.Our results reveal physical insights and trends of the layered GO films in evolving from 2D monolayers to quasi bulk-like behavior.These results confirm the effectiveness of introducing 2D layered GO films into integrated photonic resonators to improve the performance of nonlinear optical processes.

Device fabrication and characterization
Figure 1(a) shows a schematic of an integrated MRR incorporating a GO film.The MRR was fabricated on a high index doped silica glass platform using CMOS compatible fabrication processes [18,44] with chemical mechanical polishing (CMP) used as the last step to remove the upper cladding, so as to enable GO film coating on the top surface of the MRR.Benefiting from extraordinarily low linear and nonlinear loss, high index doped silica glass has been a successful integrated platform for nonlinear photonic devices [6,10,11,45,46].The n2 of the high index doped silica glass (~1.3×10 -19 m 2 /W) is lower than that of silicon (~4.5×10 -18 m 2 /W), while its negligible nonlinear loss even up to extremely high light intensities yields a nonlinear figure of merit (>>1) that is much higher than that of silicon (~0.3) [2].The CMOS compatible fabrication processes also make our devices comparable, in terms of fabrication maturity, to those implemented in silicon optoelectronic devices [2,18].
The coating of 2D layered GO films was achieved via a solution-based method that yields layer-by-layer GO film deposition on a dielectric substrate, as reported previously [29,41].
Four steps for the in-situ assembly of monolayer GO films were repeated to construct multilayer films on a target substrate.Figure 1(b) shows a scanning electron microscope (SEM) image of a 2D layered GO film on a silica substrate, with up to 5 layers of GO.Our GO coating approach, unlike the sophisticated transfer processes employed for coating other 2D materials such as graphene [47,48], enables large-area, transfer-free, and high-quality GO film coating on integrated photonic devices, with highly scalable fabrication processes and precise control of the number of GO layers (i.e., GO film thickness).Using this GO coating method, we achieved GO patterning on integrated photonic devices via photolithography and lift-off processes [43].Figure 1(e) shows microscopic images of an integrated MRR patterned with 50 layers of GO (~50 µm pattern length).Note that only the center ring of the 9 concentric rings (see inset) was coupled to through/drop bus waveguides to form an MRR − the rest were to aid in visual identification.We achieved GO film patterning with lengths down to ~150 nm via e-beam lithography and accurate placement on the MRR (deviation < 20 nm) by using gold alignment markers [43].The schematic illustration of the fabrication process flow is provided in Figure 1(f).The combination of GO coating with photolithography and lift-off allows precise control of the film placement, size, and thickness on integrated devices.This, in turn, allowed for investigating the layer dependent material properties of 2D layered GO films as well as optimizing the device performance such as FWM CE.Finally, along with large-area uniform coating capability, our fabrication techniques enable large-scale integrated devices incorporated with GO films.This is not only for nonlinear photonic devices but also for integrated devices based on other extraordinary material properties of GO such as broadband photoluminescence / light absorption [37,39], strong material anisotropy [43], and flexible bandgap engineering [27,42,49].
We fabricated and tested two types of GO-coated MRRs − uniformly coated with 1−5 layers of GO and patterned with 10−50 layers of GO (~50 µm pattern length).The radius of the MRR was ~592 μm, corresponding to a free spectral range (FSR) of ~0.4 nm (~49 GHz) for the uncoated MRR.Compared with the uniformly coated MRRs, the patterned MRRs enabled us to test the device performance with shorter GO film lengths but higher film thicknesses (up to 50 layers).Figures 2(a) and (b) show the measured transmission spectra of the MRR with uniformly coated and patterned GO films, all integrated with the same doped silica MRR and measured using a low-power (0 dBm) transverse electric (TE) polarized continuous-wave (CW) light.The coated GO films could be easily removed by plasma oxidation, and so the same uncoated MRR was reused for coating GO films with different numbers of GO layers.We used a 16-channel single-mode fiber (SMF) array to butt couple the CW light near 1550 nm into and out of the MRR chip.The mode coupling loss between the SMF array and the doped silica waveguide was ~8 dB/facet, which can readily be reduced to < 2.0 dB/facet with on-chip mode convertors [6,50].For the uniformly coated devices, the GO film thicknesses (1−5 layers, i.e., ~2−10 nm) were very small as compared with the geometry of the doped silica waveguide (~2 µm width ×1.5 µm height) and so their influence on the MRR's coupling condition was negligible.The difference between the coupling strength of the uncoated MRR and the uniformly coated MRR with 5 layers of GO was ~0.3%.For the patterned devices, on the other hand, the GO films were located outside the coupling area, and so they did not affect the MRR's coupling condition.We chose the TE polarization for the experiments because it supported the in-plane interaction between the evanescent field and the thin GO film, which is much stronger than the out-of-plane interaction due to the large optical anisotropy of 2D layered materials [43,51].We also measured the CW power-dependent transmission spectra of the GO-coated MRRs up to high powers (25 dBm) using a pump-probe method where we injected a high-power CW pump into a resonance around 1550.18 nm and a low-power probe (0 dBm, also used for measuring the spectra in Figures 2(a  excluding mode coupling loss between the SMF array and the input bus waveguide as well as GO-induced propagation loss of the input bus waveguide for the uniformly coated devices).
There were small but observable changes in the resonance wavelength and notch depth with pump power for the coated MRRs, reflecting a change in the GO material properties (e.g., refractive index and loss).For the uncoated MRR, on the other hand, we could not observe any obvious changes.The changes we observed were not permanent -the transmission spectra recovered to those in Figures 2(a  high Qs (~70,000) (although significantly less than those of doped silica MRRs with silica cladding [18,24]).As expected, both the ER and Q decreased with the number of GO layers, reflecting the additional loss induced by the GO film.As the input CW power was increased, the MRRs with both uniformly coated and patterned GO films exhibited a clear decrease in ER and Q, whereas the uncoated MRR did not.The propagation loss of the GO hybrid waveguides was obtained by using the scattering matrix method [52,53] to fit the ERs and Qs in Figures 3(a Note that the propagation loss obtained from the GO-coated MRRs is different from that obtained from the GO-coated waveguides [29,43] when considering the power-dependent loss changes of the GO films, since the light intensity in the MRRs is significantly higher due to resonant enhancements.The low-power (W/O in Figures 3(e)) propagation loss of the uncoated waveguide and the waveguide with a monolayer of GO was ~0.26 dB/cm and ~1.27 dB/cm, respectively, corresponding to an excess propagation loss of ~1 dB/cm induced by the GO film.This is over 2 orders of magnitude lower than that of integrated waveguides coated with graphene [54,55], indicating the low material absorption of GO and its strong potential for the realization of highperformance nonlinear photonic devices.The propagation loss increased with the GO layer number − a combined result of mode overlap and several other possible effects such as increased scattering loss and absorption induced by imperfect contact between the multiple GO layers as well as interaction between the GO layers, as reported previously [43].As the input CW power was increased, all the MRRs showed an increased propagation loss except for the uncoated MRR, which further confirms the change of GO material properties with light power in the GOcoated MRRs.

FWM experiment
Figure 4 shows the experimental setup used to measure FWM in the MRRs.Two CW tunable lasers separately amplified by erbium-doped fiber amplifiers (EDFAs) were used as the pump and signal sources, respectively.In each path, there was a polarization controller (PC) to ensure that the input light was TE-polarized.The pump and signals were combined with a 3-dB fiber coupler before being coupled into the MRR, which was mounted on a temperature control stage (TCS) to avoid thermal resonance drift and to maintain the wavelength alignment of the resonances to the CW pump and signal.An optical isolator was employed to prevent the reflected light from damaging the laser source.The signal output from the drop port of the MRR was sent to an optical spectrum analyzer (OSA) with a variable optical attenuator (VOA) being inserted before the OSA to prevent high-power damage.shows the opposite trend.This could reflect the trade-off between FWM enhancement (which dominates for the patterned MRRs with a short GO coating length) and loss (which dominates for the uniformly coated MRRs with a much longer GO coating length) in the GO-coated MRRs (see Section 4 for detailed discussion).

Figures 5(e) and (f)
show the measured CE versus pump power for the MRRs with uniformly coated and patterned GO films, respectively.We could not measure a CE lower than -66 dB since the generated idler was below the noise floor.For the uncoated MRR, the dependence of CE versus pump power shows a nearly linear relationship.In contrast, for the GO-coated MRRs, the measured CE shows a relatively obvious deviation from the linear relationship with pump power.Similar to the power dependent loss in Figures 3(e) and (f), this is also a reflection of the change in GO material properties with light power (see Section 4 for detailed discussion).
As the pump power increased, the measured CE increased with no obvious saturation for the uncoated MRR and the GO-coated MRRs, indicating the low TPA of both the high index doped silica glass and the GO films.Unlike graphene that has a metallic behavior and a zero bandgap, GO is a dielectric that has a large bandgap of 2.1−2.4 eV [41,42], which results in low linear and nonlinear light absorption in spectral regions below the bandgap, in particular, featuring greatly reduced TPA in the telecommunications band.This represents another important advantage of GO for implementing high-performance nonlinear photonic devices.In theory, the GO film with a bandgap > 2 eV should have no absorption for light at a wavelength near 1550 nm (~0.8 eV).We therefore infer that the loss of the coated GO films is mainly induced by light absorption from localized defects as well as scattering loss stemming from film unevenness and imperfect contact between the multilayers [29,43].The relatively low FWM CE is mainly induced by the low Kerr nonlinearity of doped silica.Although our GO-coated MRRs were based on a CMOS compatible doped silica platform, these GO films can readily be introduced into other integrated platforms (e.g., silicon and silicon nitride) offering significantly enhanced mode overlap between GO and waveguide with reduced waveguide dimensions.

Theoretical analysis and discussion
We used the theory from Ref. [19] to model FWM in the hybrid integrated MRRs.Assuming negligible depletion of the pump and signal powers due to the generation of the idler, the FWM CE of the GO-coated MRR is given by where Pidler, out and Psignal, in are the output power of the idler and input power of the signal, respectively.CEWG is the FWM CE in a straight waveguide with the same length as the circumference of the MRR, which can be expressed as where Ai(L) and As(0) are the amplitudes of the idler and signal waves along the light propagation direction, respectively, and L is the waveguide length (i.e., circumference of the MRR).For the waveguide with uniformly coated GO, CEWG was calculated by solving coupled differential equations for FWM processes based on the theory in Ref. [29].For the waveguide with patterned GO, CEWG was calculated by dividing the waveguide into coated and uncoated parts and solving the FWM differential equations in each part.In our simulation, the uncoated doped silica waveguide has an anomalous dispersion (group velocity dispersion β2 ≈ -3.9×10 -26 s 2 /m), and the hybrid waveguides with GO films show slightly enhanced anomalous dispersion (e.g., β2 ≈ -3.92×10 -26 s 2 /m for the hybrid waveguide with 5 layers of GO) for better phase matching.It is also worth mentioning that the effect of phase matching is negligible in our case since Lβ2Δω 2 << 1 [28,29], where L is the circumference of the MRR, and Δω < 1.6 × 10 13 rad/s (i.e., ~20 nm wavelength spacing) is the angular frequency detuning range.
FEp,s,i in Eq. ( 1) are resonant field enhancement factors for the pump, signal, and idler, respectively, which can be written as where κr and tr are the field coupling and transmission coefficients between the MRR and the bus waveguides, respectively.Ar is the round-trip field transmission factor and ϕ p, s, i are the round-trip phase shift of the pump, signal, and idler, respectively, which can be further expressed as where αc, u and Lc, u are the loss factors (including both linear and nonlinear loss) and lengths of the coated and uncoated waveguide segments, respectively, and kpc, sc, ic and kpu, su, iu are the wavenumbers of the pump, signal, and idler for the coated and uncoated segments, respectively.
For αc, we used the power dependent loss for each layer number obtained from Figures 3(e) and (f).Note that 15−25 dBm of power for a single CW pump in Figures 3(e) and (f) corresponds to 12−22 dBm of pump power in Figures 5(e) and (f) since we used the same power for the pump and signal in the FWM measurements and so there were the same overall CW powers in the MRRs (the idler power could be neglected).Eq. ( 3) is equivalent to that reported in our previous work [18] and is slightly different from that in Refs.[19,31] due to the use of a 4-port MRR in our case.
Based on Eqs.(1)−( 5), we fit the experimentally measured CE in Figures 5(e) and (f) to obtain the nonlinear parameter γ of the GO hybrid waveguide.We then obtained the Kerr coefficient n2 of the layered GO film using the following equation [29] where λ is the pump wavelength, D is the integral domain over the material regions with the fields, and Sz is the time-averaged Poynting vector calculated using Lumerical FDTD commercial mode solving software.n0 (x, y) and n2 (x, y) are the linear refractive index and n2 profiles over the waveguide cross section, respectively.In our FDTD simulation, we used the material refractive index of doped silica and GO measured via spectral ellipsometry and neglected the small modal distribution difference induced by the slight change of GO's refractive index with power during FWM.This work was performed in the regime close to degeneracy where the three FWM frequencies (pump, signal, idler) were close together (compared with any variation in n2 arising from its dispersion [29]).We therefore used n2 instead of χ (3) in our subsequent analysis and discussion of the material third-order nonlinearity.The values of n2 for pure silica and doped silica used in our calculation were 2.60 × 10 -20 m 2 /W [2] and 1.28 × 10 -19 m 2 /W, respectively, the latter obtained by fitting the experimental results from the uncoated MRR.Note that γ in Eq. ( 6) is an effective nonlinear parameter weighted by not only n2 (x, y) but also n0 (x, y) in different material regions, which is more accurate for highindex-contrast hybrid waveguides studied here as compared with the theory in Ref. [31].The FWM CE can be further optimized by redesigning the waveguide cross section to improve mode overlap with the GO film as well as changing the coupling strength to achieve a better field enhancement.a) The loss is the linear propagation loss of the hybrid WG after excluding the propagation loss of the bare WG. b) n2 of silicon = 4.5×10 -18 m 2 /W (Ref.[2]).c) The information is not provided.
Table I compares the FWM performance of integrated photonic devices incorporating graphene and GO.Here, we focus on integrated devices, although we note that strong FWM has also been demonstrated for optical fibers and spatial-light systems incorporating 2D materials [58][59][60].As can be seen, although the n2 of GO is about 1 order of magnitude lower than graphene, the loss induced by GO is over two orders of magnitude lower.Further, the n2 of GO is still over 3 orders of magnitude higher than silicon, suggesting that the GO films can also be introduced into other integrated platforms (e.g., silicon and silicon nitride) to improve the FWM performance.4 orders of magnitude higher than that of silicon and agree reasonably well with our previous waveguide FWM experiments [29] and Z-scan measurments [40].Table II compares the measured n2 of GO in our work with previous reports on solid GO films.Note that our work is the first study of the dependence of n2 on the number of layers for 2D GO films, which is challenging for Z-scan measurements because of the weak response of extremely thin 2D films [27,40].We also note that the Kerr nonlinearity of GO and GO nanocomposite has been studied in Refs.[61][62][63] via the Z-scan method, with the measured n2 being lower than those in Table II, possibly due to their samples being dispersed in solutions in contrast to solid films in our case.The high n2 of GO films highlights their strong Kerr nonlinearity for not only FWM but also other third-order (χ (3) ) nonlinear processes such as third harmonic generation, self-phase modulation, and cross-phase modulation [12, 64 -68], an n2 that is over a thousand times higher even than the highest nonlinear chalcogenide glasses [69][70][71][72][73][74][75] .showing a similar trend to WS2 measured by a spatial-light system [77].This is probably due to increased inhomogeneous defects within the GO layers as well as imperfect contact between the multiple GO layers.We also note that the decrease in n2 with GO layer number becomes more gradual for thicker GO films.This could reflect the transition of the GO film properties towards the bulk material (with a thickness independent n2) for thick films.At 22 dBm, n2 is higher than at 12 dBm, indicating a more significant change of the GO optical properties with inceasing power.The rate of decrease in n2 with GO layer number at 22 dBm is lower than that at 12 dBm, possibly reflecting the fact that thicker GO films are more easily affected by light power.GO layer number, n2 varies non-monotonically with pump power, perhaps with slight oscillations.This, along with the power-dependent loss in Figures 3(e) and (f), can be attributed to power-sensitive photo-thermal changes of GO in the MRRs at high power, as noted previously [43,78], as well as self-heating and thermal dissipation in the multilayer GO films.
These effects could lead to a dynamic change in the GO material properties (e.g., linear refractive index, Kerr nonlinearity, linear/nonlinear absorption, and dispersion) with power, although in practice it is difficult to distinguish this from the effects of changes in loss and CE.
In our FWM experiments, we slightly tuned the wavelength of the input CW light from blue to red around the resonances of the MRRs until it reached a steady thermal equilibrium state for FWM, and a temperature controller stage was used to maintain the steady state for long periods.
In the steady state, the FWM CE did not show any time-dependent variation for fixed input pump power, indicating that any power-dependent change in the GO film's n2 occurred very quickly and so any effects of this were not significant during the FWM experiments.We also note that the CW power in the MRRs (< 0.5 W) was not high enough to excite obvious satuable absorption of the GO films [27,40].Since it was difficult to accurately measure the slight change in film thickness and dispersion of GO with the pump power during FWM in the GOcoated MRRs, we neglected any change in these parameters.In principle, this approximation could lead to slight deviations in n2, possibly explaining the non-monotonic relationship between n2 and pump power.Despite this, the fit n2 can still be regarded as a parameter reflecting the over-all FWM performance at different pump powers.
To compare the photo-thermal changes of the GO films during FWM, we also tested the performance of FWM in doped silica waveguides uniformly coated with 1 layer of GO or patterned with 50 layers of GO.The waveguides had the same geometry as those of the MRR, but with a longer length (~1.5 cm).This resulted in the generated idlers being above the noise floor.The GO coating length of the patterned waveguide (50 µm) was the same as that of the patterned MRR.The n2 of GO films versus pump power obtained from FWM measurements using the GO-coated waveguides are plotted in Figures 7(d) and (e), together with those obtained from FWM measurements using the GO-coated MRRs.As can be seen, n2 shows a much greater variation with pump power in the MRRs than the waveguides, reflecting a more dramatic change of GO material properties in the MRRs.This is not surprising since the power is significantly higher in the MRRs.were not permanent.Note that a variation in bandgap and material properties of GO can be obtained by changing GO's chemical structure and we previously showed [27,37,49] that the material properties of GO can also be permanently changed by femtosecond laser pulses with significantly higher power levels than those used here.This is distinct from the photo-thermal changes observed here.Finally, FWM in MRRs depends on many factors in terms of material properties, including the third-order nonlinearity, the linear and nonlinear loss, dispersion, etc.For the GO-coated MRRs in our case, the physics of FWM are more complex due to the change of GO material properties with layer number and light power.Despite this, the layer-number and powerdependent material properties of the layered GO film yield many new device properties that are difficult to achieve for typical integrated photonic devices.We believe this could enable one to tailor the device performance well beyond simply enhancing the FWM CE as reported here.

Conclusion
We demonstrate enhanced FWM in MRRs integrated with layered GO films.CMOScompatible doped silica MRRs with both uniformly coated and patterned GO films are fabricated based on a large-area, transfer-free, layer-by-layer GO coating method together with photolithography and lift-off processes, which yield precise control of the film thickness, placement, and coating length.We perform FWM measurements for MRRs uniformly coated

Figure 1 .
Figure 1.(a) Schematic illustration of GO-coated integrated MRR.Inset shows schematic atomic structure of GO.(b) Scanning electron microscope (SEM) image of 2D layered GO film.The numbers 1−5 refer to the number of GO layers for that part of the image.Inset shows 5 layers of GO on a silica substrate.(c) Raman spectra of an integrated chip without GO and with 2 layers of GO.Insets show the corresponding microscope images.(d) Measured GO film thickness versus GO layer number.Insets show the images of a silica circular substrate uniformly coated with 0 (uncoated), 10, and 20 layers of GO.(e) Microscopic image of an integrated MRR

Figure 1 (
Figure 1(c) shows the measured Raman spectra of an integrated chip (including doped silica

Figure 1 (
d) shows the thickness of GO films versus the number of layers, characterized by atomic force microscopy (AFM).The plots show the average of measurements on three samples and the error bars reflect the variations.The GO film thickness shows a nearly linear relationship with the layer number, with a thickness of ~2.25 nm on average for each layer.The insets present images of a silica substrate uniformly coated with 0 (uncoated), 10, and 20 layers of GO, showing large-area GO film coating with high uniformity.

Figure 2 .
Figure 2. (a)−(b) Transmission spectra of an integrated MRR uniformly coated with 1−5 layers of GO and patterned with 10−50 layers of GO measured using a low-power CW light, respectively.The measured transmission spectrum of the uncoated MRR (GO-0) is also shown for comparison.(c)−(d) Transmission spectra of the MRRs with 1 layer of uniformly coated and 50 layers of patterned GO measured using a low-power CW probe when another high-power CW pump was injected into a resonance around 1550.18 nm, respectively.The values of 15−25 dBm represent the incident pump powers.The transmission spectra measured using the low-power CW probe without the high-power CW pump (W/O) are also shown for comparison.
) and (b)) to scan the spectra around another resonance.

Figures 2 (
Figures 2(c) and (d) show the transmission spectra of the MRRs with 1 layer of uniformly ) and (b) when we turned off the high-power CW pump, with the spectra in Figures2(c) and (d) being repeatable (see Section 4 for detailed discussion).

Figure 3 .
Figure 3. Power dependent (a)−(b) extinction ratios (ERs) and (c)−(d) quality factors (Qs) for the MRRs with 1−5 layers of uniformly coated and 10−50 layers of patterned GO films, respectively.The ERs and Qs of the uncoated MRR (GO-0) are also shown for comparison.The Qs are not shown when the ERs are < 5 dB.(e)−(f) Fit propagation loss obtained from (a)−(d), respectively.The values of 15−25 dBm represent the incident pump powers.The data points for W/O correspond to the values measured using a low-power CW probe without the high-power CW pump.

Figure 5 (
Figure 5(a) shows the FWM spectra of the MRRs uniformly coated with 1−5 layers of GO,

Figure 5 .
Figure 5. (a)−(b) Optical spectra of FWM at a pump power of 22 dBm for the MRRs with 1−5 layers of uniformly coated and 10−50 layers of patterned GO films, respectively.(c)−(d) CE enhancement and additional insertion loss extracted from (a) and (b), respectively.(e)−(f) CE versus pump power for the MRRs with 1−5 layers of uniformly coated and 10−50 layers of patterned GO films, respectively.In (a), (b), (e), and (f), the results for uncoated MRR (GO-0) are also shown for comparison.

Figure 6 .
Figure 6.(a)−(c) Optical spectra of FWM at different resonant wavelengths for the uncoated MRR, the MRR uniformly coated with 1 layer of GO, and the MRR patterned with 50 layers of GO, respectively.Δλ and FSR represent the wavelength spacing between pump and signal and the free spectral ranges of the MRRs, respectively.(d) Measured CE versus Δλ/FSR for the MRRs in (a)−(c).The pump power in (a)−(d) was 22 dBm.

Figure 7 (
Figure 7(a) shows n2 of GO films versus layer number obtained from the measured CEs in

Figures 7 (
Figures 7(b) and (c) show n2 of GO films versus pump power obtained from the measured

Figures 7 (
Figures 7(e) and (f) present the transmission spectra of GO-coated MRRs before and after

Figure 7 .
Figure 7. Analyses of the change of GO material properties during the FWM process.(a) n 2 of GO versus layer number at fixed pump powers of 12 dBm and 22 dBm obtained from the MRR FWM experiment.(b)−(c) n 2 of GO versus pump power for 1−5 layers and 10−50 layers of GO obtained from the MRR FWM experiment, respectively.(d)−(e) n 2 of GO versus pump power (1 layer of GO in (d) and 50 layers of GO in (e)) obtained from the MRR and waveguide (WG) FWM experiments.(f)−(g) Measured transmission spectra of GO-coated MRRs (uniformly coated with 1 and 5 layers of GO in (f) and patterned with 10 and 50 layers of GO in (g)) before and after FWM with a pump power of 22 dBm.

with 1 −
5 layers of GO and patterned with 10−50 layers of GO, achieving up to ~7.6-dB and ~10.3-dB enhancement in the FWM CE for the MRRs uniformly coated with 1 layer of GO and patterned with 50 layers of GO, respectively.We also fit the measured CE to theory and obtain the change in the third-order nonlinearity of GO films with layer number and pump power.The high nonlinear performance of the GO-coated MRRs confirms the effectiveness of introducing 2D layered GO films into integrated photonic resonators to improve the performance of nonlinear optical processes.by the Canada Research Chair Program.He also acknowledges additional support by the Professorship Program (grant 074-U 01) and by the 1000 Talents Sichuan Program in China.

Table II . Comparison of n2 of GO film. WG: waveguide. MRR: microring resonator.
some n2 values at 12 dBm are not shown since the generated idlers were below the noise floor.n2 (both at 12 dBm and 22 dBm) decreases with GO layer number,