Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Heterogeneously integrated InGaN-based green microdisk light-emitters on Si (100)

Open Access Open Access

Abstract

Heterogeneous integration of nitrides on Si (100) is expected to open the door to the new possibilities for this material system in the fields of high-speed integrated photonics and information processing. In this work, GaN epitaxial layer grown on the patterned sapphire substrate is transferred onto Si (100) by a combination of wafer bonding, laser lift-off and chemical mechanical polishing (CMP) processes. The GaN epilayer transferred is uniformly thinned down to 800 nm with a root mean square surface roughness as low as 2.33 Å. The residual stress within the InGaN quantum wells transferred is mitigated by 79.4% after the CMP process. Accordingly, its emission wavelength exhibits a blue shift of 8.8 nm, revealing an alleviated quantum-confined Stark effect. Based on this platform, an array of microcavities with diverse geometrics and sizes are fabricated, by which optically-pumped green lasing at ∼505.8 nm is achieved with a linewidth of ∼0.48 nm from ∼12 µm microdisks. A spontaneous emission coupling factor of around 10−4 is roughly estimated based on the light output characteristics with increasing the pumping densities. Lasing behaviors beyond the threshold suggest that the microdisk suffers less thermal effects as compared to its undercut counterparts. The electrically-injected microdisks are also fabricated, with a turn-on voltage of ∼2.0 V and a leakage current as low as ∼2.4 pA at -5 V. Being compatible with traditional semiconductor processing techniques, this work provides a feasible solution to fabricate large-area heterogeneously integrated optoelectronic devices based on nitrides.

© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

1. Introduction

Silicon photonics, compatible with large wafer-scale manufacturing foundries, have been considered a promising way to revolutionize the communication and computing technologies [14]. However, as an indirect bandgap semiconductor, Si itself cannot emit light efficiently [5]. As critical building blocks, III-nitrides have been demonstrated to have great potentials in realizing efficient light sources from deep ultraviolet to near infrared [6]. Recently, significant efforts have been made to integrate III-nitride photonic or electronic devices on Si substrates [79]. There is no doubt that hybrid integration of nitride-based photonic devices on a maturely-developed Si microelectronics platform can extend the applications and utility of the nitride-based material system [1012]. In this regard, direct growth of III-nitrides on Si substrate have enabled to demonstrate optically-pumped microcavity lasers in the wavebands covering the deep ultraviolet, blue, green, etc. [1321], while electrically-pumped microdisk or microring lasers from nitride-based materials have been further realized recently [2224]. However, at present, monolithically growth of nitrides is mostly limited on Si (111) substrates in the perspective of lattice-matching conditions with GaN [25]. Additionally, to integrate with conventional Si-based microelectronics, it requires to grow nitrides on Si (100), which brings great challenges to obtain III-nitrides with high quality [26]. On the other hand, epitaxial transfer of III-nitrides grown on sapphire allows the heterogeneous integration of nitrides onto Si (100) platform, giving a feasible technological solution. In this regard, multiple heterogeneously-integrated lasers have been demonstrated, including vertical-cavity surface-emitting lasers (VCSELs), Fabry-Perot cavity lasers and microdisk lasers [2729]. Nevertheless, one of the tough and urgent challenges for this epitaxial lift-off and transfer method is to fabricate large-area, smooth, flat GaN thin films on Si, providing a platform for versatile heterogeneously-integrated photonic devices. In particular, this challenge arises mainly from the fact that the exfoliated epitaxial nitrides are N-polar and highly susceptible to chemical corrosion, which usually ends up into pyramidal structures due to asymmetrical etching [30,31]. This is particularly true for the nitrides grown on a pattered sapphire substrate (PSS), which is the most commonly adopted method presently to reduce both the stress and defects for the epilayers [32,33]. The rough surface greatly reduces the optical confinement capability for the as-fabricated microcavities [30]. To obtain thin and smooth transferred nitrides, this problem can be solved by epitaxy of nitrides on planar substrates or by insertion of sacrificial layers, which in general deteriorates the quality of nitrides to a certain extent [3438]. Nevertheless, for long-wavelength InGaN materials, these technologies may still be applied to fabricate microcavities with smooth surfaces, at costs of a loss in the quantum efficiency potentially, as it is very sensitive to the defects and build-in strain. Therefore, for the epitaxial growth of nitrides on PSS, a new strategy needs to be introduced to reduce the surface roughness of the sample after laser lift-off (LLO), which achieves better optical confinement to reduce the lasing threshold.

In this work, we present a method for transferring ultrathin nitrides on the Si (100) to fabricate nitride microdisk light-emitters by wafer bonding, LLO, and chemical mechanical polishing (CMP) processes, which provides a feasible solution for large-area Si-based heterogeneous integration of nitrides. The method proposed has demonstrated its advantages of high compatibility with conventional semiconductor processes, mass production and excellent reproducibility, especially for the long-wavelength InGaN materials with high In composition. Moreover, various optical microcavities are fabricated on this platform, in which the green lasing from the as-transferred microdisks is realized by an optical pumping method and its lasing behaviors are investigated in detail. In addition, electrically-injected microdisks are also realized for exploring their electrical and optical performance.

2. Experiment

Figure 1(a) presents the flow chart illustrating the fabrication process for the heterogeneously integrated GaN-based microcavities on Si (100). The wafer for processing was grown on c-plane PSS by metal-organic chemical vapor deposition (MOCVD). In sequence, the epilayers consisted of a 2.5 µm undoped GaN layer, a 2.5 µm Si-doped n-GaN layer (∼2×1018 cm-3), 10 pairs of In0.2Ga0.8N/GaN-based MQWs, 10 pairs of Al0.2Ga0.8N/GaN superlattices (SLs)-based electron barrier layer (EBL) [39], and a 100 nm Mg-doped p-GaN (∼2×1019 cm-3) layer. The cross-sectional scanning electronic microscope (SEM) image for the as-grown wafer is shown in Fig. 1(b). The pyramid-shaped patterned substrate can be clearly seen and the total thickness for the overall epitaxial layer is approximately 5.6 µm. Before the wafer bonding, a multilayer metallic electrode consisted by Ni/Ag/Pt/Au (10/100/5/50 nm) was evaporated on the p-GaN layer by an electron beam (e-beam) evaporator. Then a rapid annealing treatment in O2 was carried out to form good ohmic contact to p-GaN. Next, the epitaxial wafer and p-Si (100) submounts were deposited on Ti/Au/In metals (5 nm/200 nm/1200 nm) as the bonding layer. A standard wafer-bonding process was then conducted in the vacuum chamber for 30 min at a temperature of ∼200 °C, which would allow the wafers to be bonded firmly. Afterwards, the backside of the sapphire was polished by a mechanical grinding and polishing process. Subsequently, a LLO process was carried out by using a 248 nm KrF excimer pulse laser as the irradiation source, in which the sapphire substrate was easily released when the GaN at the interface was decomposed into Ga droplets and gaseous N2. The cross-sectional view of the sample after the LLO process is shown in Fig. 1(c). From bottom to top, the Si substrate, alloyed bonding layer, metallic mirror, and the epitaxial layer can be clearly identified in sequence. The epilayer was further precisely polished to a few hundred nanometers by a CMP process. The SEM image illustrating the cross-section of the sample after the CMP is presented in Fig. 1(d), from which the thickness of the whole epitaxial layers is determined as only 800 nm. The material removal rate (MRR) was precisely controlled by the weight loss method in the polishing process so that a preset thickness could be achieved. Now the wafer was ready for the characterization and device processing. For the mesa definition, the polished sample was coated with a layer of 300 nm-thick silicon oxide (SiO2) using the inductively coupled plasma chemical vapor deposition (ICP-CVD), and the photoresist was used as another mask for shaping the SiO2 in a dry etching process. Using the SiO2 film as a hard mask, the GaN epilayer was etched away after an ICP etching process for 200 s using Cl2/Ar/BCl3 gases (40/5/5 sccm). A relatively slow etching rate (300 nm/min) was selected to achieve smooth and vertical sidewalls. The ICP radio-frequency (RF) coil power and platen power were set as 75 and 450 W, respectively. In this step, optical microcavities with diverse shapes and sizes were fabricated on the Si platform for the optical pumping test, including microrings, microdisks as well as hexagonal microcavities ranging from 5 µm to 20 µm. To demonstrate the current-injected devices, an array of microdisks with sizes from 40 µm to 200 µm were fabricated as well. It should be pointed out that, in addition to being used as a mask for etching GaN, the layer of SiO2 was slightly over etched in buffered oxide etchant (BOE) using the photoresist above as the protection mask, also acted as a current blocking layer (CBL) in the current-injected microdisk devices. Finally, Ti/Au (40/200 nm) were deposited on the n-GaN as the n-contacts.

 figure: Fig. 1.

Fig. 1. (a) Schematic diagrams illustrating the fabrication process of InGaN-based microdisk light-emitters heterogeneously integrated on Si (100). The SEM images illustrating the cross-sectional structures for the (b) as-grown samples, and samples (c) after LLO process, (d) after CMP process, and (e) as-fabricated electrically-injected microdisk devices on Si, respectively.

Download Full Size | PDF

The SEM image showing the cross-section of the current-injected devices is presented in Fig. 1(e). The surface morphology after the LLO and CMP process was checked by atomic force microscopy (AFM). The strain-relaxation behaviors were investigated by using the Raman spectroscopy. A 405 nm continuous-wave (CW) laser was used as the excitation source so as to check the Raman shift of GaN close to the InGaN QWs. The internal quantum efficiency (IQE) of the transferred InGaN-based thin-films was estimated by a temperature-dependent PL measurement from 10 K to 300 K using a 325 nm pulsed laser as the excitation source, and the excitation power density was fixed at ∼400 W/cm2. A homemade micro-photoluminescence (µ-PL) setup was used to measure emission characteristics of the InGaN microdisks on Si (100) at room temperature (RT) (T∼300 K), as shown in Fig. 2. An excimer laser operated at 355 nm with a pulse repetition rate of around 10 Hz and a pulse duration of ∼10 ns was used as the excitation source. The laser spot was around 60 µm. The PL signals were collected by a grating spectrometer (Princeton Instrument, ARC-SP-2356) and detected by a thermoelectrically-cooled charge coupled device (CCD) (Princeton Instruments, PIX-256E), which offers optical resolutions of ∼0.05 nm. The electrical properties of the transferred microdisk devices were probed in a Lake Shore TTPX probe station and analyzed by a Keithley 4200A-SCS semiconductor parameter analyzer. A pulsed electroluminescence (EL) measurement was further conducted by probing the microdisk devices as driven by a LDP 3830 pulsed laser diode driver.

 figure: Fig. 2.

Fig. 2. A homemade setup for power-dependent µ-PL test.

Download Full Size | PDF

3. Results and discussion

As shown in Fig. 3(a), the wafer with an area of 2 × 2 cm2 after the LLO process exhibits rainbow streaks due to the light dispersed by the periodical patterns on the wafer left by the PSS. The blurry areas on the wafer are the marks of laser scanning routinely appear in the LLO process, which are not the damages in epilayers. In contrast, a vivid reflection of an ISCAS badge from the mirror-like polished sample in Fig. 3(b) exhibits a very smooth surface. To check the smoothness of these samples, the surface roughness of both samples was probed by AFM over a randomly chosen area of 10 × 10 µm2. A close-up view on the surface morphology after LLO in Fig. 3(c) reveals an array of concave patterns with a period of 4 µm. On the contrary, the root mean square surface roughness (RMS) for the CMP sample is determined as low as ∼2.33 Å, as shown in Fig. 3(d). Several black dots in nanometer scale can be even found on the surface after the CMP process, corresponding to the threading dislocations which commonly exist in the nitride epilayers.

 figure: Fig. 3.

Fig. 3. The optical images for the wafers (a) before and (b) after the CMP process. Correspondingly, the AFM images showing the surface morphologies for the InGaN/GaN-on-Si wafer (c) before and (d) after the CMP process. (e) µ-PL spectra and (f) Raman spectra from the as-grown MQWs on sapphire, and wafers after LLO process, and after CMP process, respectively. (g) The statistical Raman E2(high) modes as measured across the wafer.

Download Full Size | PDF

Figure 3(e)–3(f) present the results from the µ-PL and Raman spectroscopic measurements. The dominant PL peaks for the three samples locate at 521.47 nm, 518.69 nm and 512.66 nm, respectively. As compared to the as-grown sample, the PL peak exhibits a blue-shift by 2.8 nm after the release of sapphire substrate in the LLO process, indicating a reduction in quantum-confined Stark effect caused by the strain-relaxation in MQWs [40]. The PL spectral peak continues to blueshift by 6.0 nm when the transferred epilayers are polished to 800 nm, suggesting the residual strain within the epilayers is able to be further mitigated. The in-plane biaxial strain relaxation can be also verified by the Raman shift in the GaN E2(high) mode, as illustrated in Fig. 3(f), which exhibits a continuous shift of E2(high). To avoid systematic errors, we randomly collected the Raman spectra at 10 different points over the sample for each case, and the E2(high) Raman mode statistics are given in the Fig. 3(g). In detail, the positions of the E2(high) modes for the as-grown, after LLO, and after CMP samples are 569.20, 568.33, and 567.94 cm-1, respectively, which reaffirms that the stress is greatly relieved after the epitaxial layers are transferred and thinned. The biaxial stress σxx can be estimated from the frequency deviation:

$$\Delta \mathrm{\omega =\ K}{\mathrm{\sigma }_{\textrm{xx}}}$$
where K is the linear stress factor (2.56 cm-1 GPa-1 for GaN [41]). Based on the calculation results from Eq. (1), it can be found that the biaxial stress in the as-grown epitaxial layers, the transferred film before and after polishing is 0.63, 0.29, and 0.13 GPa, respectively, indicating that the residual stress after the polishing has been reduced by 79.4%.

Optical photos of the as-fabricated microcavity array with diverse structures are presented in Fig. 4(a). In particular, an array of microrings (light blue), microdisks (light pink) as well as hexagonal microprisms (light grey) are fabricated. Here, as a demo, the lasing emission properties from the transferred InGaN/GaN microdisks are characterized as pumped by a pulse laser. Figure 4(b) presents the tilted-view SEM image of a typical as-transferred microdisk on Si (∼12.0 µm) for optical pumping tests. A magnified SEM image showing the zoom-in view on the microdisk sidewall is presented in Fig. 4(c). It can be observed that the disk sidewall is smooth and vertical, both of which are important for realizing low-threshold microdisk lasers via whispering-gallery modes (WGMs). Arrhenius plots of the temperature-dependent integrated PL intensities from the transferred thin films are presented in Fig. 4(d), and the PL spectra are shown in the inset of Fig. 4(d) accordingly. Here, the IQE at RT is calculated from the ratio of the PL intensities obtained at RT and ∼10 K, assuming that non-radiative recombinations at a low temperature of ∼10 K are totally eliminated and thus corresponding IQE reaches 100%. By this method, the RT IQEs are estimated to be 61.2% for the transferred materials. A periodical pattern emerges in the PL spectra, which should be attributed to the vertical standing waves existed in the 800 nm-thick thin films. Notably, in order to estimate the IQE more precisely, a low excitation power is used, leading to a much longer emission wavelength at low temperatures, which should be attributed to the recombinations of carriers from localized In clusters commonly exists in high-In QWs [42,43].

 figure: Fig. 4.

Fig. 4. (a) an array of as-transferred InGaN-based microcavities on Si (100) for optical pumping tests. (b) Tilted-view SEM image of a 12 µm microdisk on Si. (c) Magnified view on the microdisk sidewall. (d) The temperature-dependent integrated PL intensities for the transferred InGaN thin-film.

Download Full Size | PDF

The PL spectra under varied pumping power densities from an as-transferred microdisk (∼12 µm) on Si are plotted in Fig. 5(a). As increasing the excitation power densities from 75 to 199 kW/cm2, a series of WGMs with uniform spacings around ∼2 nm emerges in the emission spectra from the microdisk, in which three sharp spectral peaks located at 495.1 nm (Mode1), 501.9 nm (Mode2) and 505.8 nm (Mode3) become dominant, implying the onset of lasing established in the microdisk cavities. Notably, these WGM wavelengths are shorter than the QW emission as designed. This should be attributed to gain spectrum shift caused by a combined effect of the strain-relaxation and carrier band-filling in the InGaN QWs after the whole epilayers are transferred. Beyond the pumping power density of 106 kW/cm2, the dominance of Mode 1 in the PL spectra shifts to Mode2 and Mode3, suggesting a fierce mode competition for these WGMs. Mode hopping mainly comes from the shift of gain spectrum and the changes in the refractive index of gain medium at varying pumping conditions, particularly for the large cavities with dense optical modes. Potentially, it can be solved by the use of relaxed QWs with less built-in strain as the stable gain medium, and the reduction in the disk dimensions would be also helpful as well. Meanwhile, we can further observe the mode energies for the Mode 1, 2 and 3 hardly change with increasing the excitation densities, as shown in Fig. 5(b). With less red-shift in the emission wavelength, it implies that these transferred microdisks may suffer less heating effects as compared to the undercut ones, due to the relatively good thermal conductivity of the Si substrate [15,44]. Typically, the microdisk cavity with undercut architect uses air as the upper and lower optical confinement layers after partially removing the sacrificial layers, leaving a tiny post to uphold the upper microdisk, which seems evitable to suffer severe thermal effects, as only the supporting posts in the middle can effectively dissipate heat, even under pulsed optical pumping conditions [44]. Particularly, the three lasing modes emerge in the high energy side as compared to the emission background, which is controversial to the bulk GaN counterpart [35]. It verifies the optical gain of lasing is from the recombination of free carriers, which would blue shift due to the band-filling effects under high pumping conditions [45,46].

 figure: Fig. 5.

Fig. 5. (a) Power-dependent PL spectra from a 12 µm microdisk integrated on Si at RT. (b) Shift in WGM energies detected at around 495.1 nm, 501.9 nm, and 505.8 nm as increasing the pumping power densities.

Download Full Size | PDF

Typical S-type increase in the integrated PL intensity (all modes) as a function of the excitation power density can be observed in Fig. 6(a), by deducting the background emission via a Gaussian fitting process. Such a nonlinear increase in the PL integrated intensity provides strong evidence of onset of lasing. The insets are the luminescence images captured by the CCD positioned on the PL setup below and above the Eth, respectively. Above the Eth, a periodical pattern along the microdisk boundary as well as bright speckles within the microdisk can be observed, both of which are signs of coherent light emission [15]. It is worthy to note that linewidth narrowing process in the WGMs were not always observed with our face-up collection setup, where the very strong emission background from the center of the microdisk may cover the signals of WGMs below the lasing threshold (Eth). Beyond the Eth, the transition process from the spontaneous emission to stimulated emission can be clearly identified. In particular, from the output characteristics, the Eth power densities for the Mode1, 2, 3 can be roughly determined as 86 kW/cm2(Eth1), 95 kW/cm2(Eth2), and 106 kW/cm2(Eth3), respectively. The threshold is relatively high, which is mainly due to the optical confinement of WGMs in the vertical direction is not satisfactory in the absence of designed waveguide layers. Additionally, the n-GaN layer on the top will absorb the incident power, leading to a severe increase in the Eth. The spontaneous emission coupling factor (β) can be roughly estimated to be ∼10−4, where the upper and lower black dashed lines in Fig. 6(a) represent spontaneous emission (bottom) and stimulated emission (top), respectively. The β factor represents the proportion of spontaneous emission coupling into stimulated emission [47]. Notably, as the extraction of the β factor requires the clear observation of the lasing mode below the Eth, it will be difficult to obtain it rigorously for larger microdisk diameters [17]. Therefore, to estimate the β factors for these transferred microcavities, the β factors of all microdisk lasers with different sizes are extracted based on their lasing characteristics, as summarized in Fig. 6(b). From a statistical analysis, it can be found that, for the microdisk cavities with diameters about 8 µm, the β factors are about 7 × 10−4. When the cavity sizes are larger than 12 µm, the β factors gradually decrease to a level of ∼10−5. It’s worth noting that β factors of the microring cavities (with widths of 2 µm) are generally larger than those of microdisks with the same sizes, which are around 5 × 10−5 when the sizes of microrings exceed 10 µm. In addition, when the cavity size gets smaller, the β factors for the microrings and the microdisks will get closer, at a level of ∼5 × 10−4, which approaches the previous results with similar sizes [16]. It should be pointed out that the accurate β factor can be calculated based on the carriers’ rate equations [16,17]. By suppressing the mode number, further reduction in the cavity size would help to improve the β factor, while it is still a challenge to maintain the good confinement for smaller cavities [15].

 figure: Fig. 6.

Fig. 6. (a) Integrated PL intensities of all modes as a function of the pumping densities from the InGaN/GaN microdisk integrated on Si. Insets are optical photos of the as-pumped microdisk devices below and above the Eth. (b) The extracted β factors from all microdisk and microring cavities based on their lasing output characteristics. (c) The WGM mode linewidth evolution as increasing the pumping power densities beyond the Eth. (d) Eth dependence of the microdisk diameter.

Download Full Size | PDF

Figure 6(c) shows the linewidth evolution for the WGMs above the Eth. As the power density increases, the WGMs linewidths broaden from ∼0.50, 0.45 and 0.48 nm to 0.87, 0.69 and 1.1 nm, respectively, due to the absorption of free carriers or plasma effect [46]. As the optical absorption of QWs is bleached beyond the Eth, the Q-factors of 990, 1115 and 1053 for Mode1, Mode2 and Mode3 can be roughly estimated by the relation λλ, where λ and Δλ represent the wavelength and spectral linewidth, respectively. Figure 6(d) presents the size-dependent variation in the Eth power densities for the as-transferred microdisk cavities. The Eth would gradually increase with reducing the disk sizes, which may be caused by the larger geometrical radiative loss for smaller microdisks. Concurrently, we also observe a splitting of lasing Eth as depicted by the dashed lines in Fig. 6(d). It is believed this divergence is very likely due to be variation in the optical scattering loss of WGMs for different modal families at the disk rim. The optical modes confined in the nitride microdisk cavities are mostly transverse-electric (TE)-dominant, so the disk sidewall surface roughness would greatly determine the scattering loss. Therefore, the positions of electromagnetic field as confined would play a critical role in determining the lasing Eth [32,48,49]. In this regard, it seems that the Eth for the second-order TE WGMs usually are lower than the first-order TE modes, as the first-order WGMs are closer to the cavity sidewalls so the scattering loss will be larger [15,32,49].

Finite-difference time-domain (FDTD) numerical simulations were carried out to clarify the optical modes in the transferred GaN microdisk and validate our discussions on the experimental results. In the simulations, the dipole wavelength range was set from 490 nm to 510 nm, the refractive index of the InGaN/GaN MQWs was 3.25 (larger than the actual InGaN/GaN MQWs [50]), and the diameter of the microdisk was ∼12.0 µm. The simulated optical resonances are presented in Fig. 7(a), which are in good agreement with the PL results. Notably, there are few modes from the simulations do not show up in the PL spectra. The mismatch with the PL spectral peaks is probably due to the fact that the optical loss for each specific optical mode is diverse, which is greatly determined by the scattering effects caused by the imperfections on the disk sidewalls after the ICP etching. This divergence would lead to the absence of the optical modes with higher optical loss. Additionally, for the lasing modes, they may also suffer intense mode competition due to the gain fluctuations caused by the spectral shift out of band filling or plasma effects. Generally, WGMs in the microdisk laser satisfy the following standing-wave conditions [51]:

$$\mathrm{m\lambda} /{n_{eff}} \mathrm{\ =\ 2\pi r}$$
where m presents the mode number in the azimuthal direction, r is the effective radius of the microdisk cavity and neff represents the group refractive index of the laser, respectively. Thereafter, the mode numbers ranging from 255 to 261 are calculated by Eq. (2) and labeled for comparison with the WGMs observed from the PL test. The free spectral range (FSR) for the WGMs is calculated as 1.8∼2 nm, by using the following Eq. (3):
$$\mathrm {FSR}\ =\ \lambda^2/2\pi{n_{eff}} \textrm{r\; }$$
which is well matched with 1.9 nm observed from the PL spectra [52]. The energy intensity patterns in the horizontal and vertical direction from the FDTD simulations at the wavelength of 505.8 nm (Mode3) are shown in the Fig. 7(b) and Fig. 7(c), respectively, suggesting this lasing mode can be attributed to be the first-order TE1,256. Moreover, the optical confinement factor is calculated to be 3.52%. A close-up view reveals that the confinement loss is caused by a severe electric field leakage out of MQWs into n-GaN layer, as shown in the inset of Fig. 7(c), due to a fact that the refractive index difference between MQWs and EBL (∼0.5) is larger than that between MQWs and n-GaN (∼0.25). This would lead to the WGMs energy being mainly concentrated in the MQWs and partially squeezed into the n-GaN layer. Further thinning down of n-GaN layer is an option to improve the light confinement in the vertical direction. In this case, light penetrating into the n-GaN layer could be confined by the interface between n-GaN and air. To achieve this goal, the physical thickness of the n-GaN needs to be thinned down about 100 nm.

 figure: Fig. 7.

Fig. 7. (a) FDTD simulated resonances from a ∼12 µm microdisk and its comparison with PL spectra. The energy intensity patterns for the first-order TE WGMs in the (b) horizontal and (c) vertical direction at ∼505.8 nm. The insets show the enlarged view of the energy profiles in the red dashed boxes.

Download Full Size | PDF

Figure 8(a) shows the electrically-injected microdisk on Si probed for electrical performance characterizations. The CCD-captured luminescent images at injected currents of 0.1 mA, 1 mA and 10 mA are displayed below, respectively. The current-voltage characteristics are performed on a 160 µm microdisk, as voltage swept from -5 V to 5 V. A typical IV curve is presented in Fig. 8(b). As the drive current increases, the EL intensity gets stronger at the disk rim and becomes more bluish, again affirming the blueshift in the emission wavelength. The turn-on voltage for the transferred microdisks averages at ∼2.0 V (@1 µA), which is fair close to the typical green InGaN light-emitters, demonstrating good ohmic contacts to both p and n metal electrodes. The reverse IV curve of the device is replotted in the inset, which reveals a leakage current as low as only a few of pA. Meanwhile, the RT EL spectra from the transferred disk device at pulsed current density of 3.4 to 27.6 kA/cm2 are collected with a fixed optical fiber at a 30 degree, as shown in Fig. 8(c). The period and duration of the current pulse is set as 10 µs and 500 ns, respectively. The EL emission peak wavelength first blue-shifts and then red-shifts as increasing the current injection densities. The blue shift is due to band-filling effects and shielding of the electric field in the active regions, while the red shift is due to device self-heating effects. No signs of lasing are observed from these current-injected microdisk cavities transferred on Si. Two critical facts may hinder the device in achieving lasing. On one hand, although SiO2 CBL is utilized, the as-fabricated microdisks are still too large, which set up-limits for the current density injected. It further suggests that the highest current density achieved in the pulse mode at present is still far away from reaching the lasing Eth. As discussed above, to fabricate smaller microdisks is an option, yet it ultimately calls for a further reduction in the optical loss for the transferred devices. One the other hand, although the epilayer has been thinned down 800 nm in this work, it seems it’s still not enough to provide sufficient confinement in the vertical direction. A feasible solution is to transfer the laser epi-structure with a gain medium sandwiched by two waveguide layers, which would improve the both the light confinement and coupling for the WGMs [53]. Additionally, although many studies have reported the photonic integration of microdisk light emitters with optical components on Si platform, such as optical waveguides and gratings [5457], it is still a big challenge to achieve integration with Si-based microelectronic devices. All these problems may be solved in the future work.

 figure: Fig. 8.

Fig. 8. (a) As-fabricated 160 µm microdisk device for EL test and photos of the luminescence driven at currents of 0.1, 1, and 10 mA, respectively. (b) Typical IV curve of the transferred microdisk devices. The insets show the IV curve at reverse bias. (c) EL spectra under different injected current densities.

Download Full Size | PDF

4. Conclusions

In this work, GaN-based epilayers with an area of 2 × 2 cm2 natively grown on patterned sapphire substrate are successfully transferred onto Si (100) via a wafer-bonding process and a laser lift-off process. The transferred GaN film is further thinned down to 800 nm by a precise chemical mechanical polishing process, by which a smooth surface with an RMS roughness as low as 2.33 Å is obtained. Raman and µ-PL spectroscopy results reveal the as-built strain can be significantly mitigated after the GaN thin-film is released and polished. On this platform, an array of diverse microcavities in both shape and size are fabricated, by which optically-pumped multimode green lasing from these heterogeneously integrated InGaN/GaN-based microdisks is achieved on Si (100) at room temperature. The lasing wavelength is centered around 505.8 nm with the linewidth of ∼0.48 nm as the excitation exceeds the lasing threshold of ∼106 kW/cm2. Three-dimensional FDTD simulations have verified that the spectral lasing modes belong to the first-order TE whispering-gallery modes (WGMs) confined by the microdisk cavities. Current injection for the transferred microdisk cavities is further realized, which achieves a low leakage current of few pA at -5V. Being compatible with traditional semiconductor processing techniques, this technology offers a straightforward, yet simple, technique for reliable and high-yield hybrid integration of nitride microdisk lasers on Si (100) platform, which may find its applications in on-chip visible light communication and information processing as well as the development of high-density, monolithically integrated silicon-based optoelectronics.

Funding

National Key Research and Development Program of China ((2021YFB3601000, 2021YFB3601004)); Chinese Academy of Sciences.

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

References

1. M. Hochberg and T. Baehr-Jones, “Towards fabless silicon photonics,” Nat. Photonics 4(8), 492–494 (2010). [CrossRef]  

2. A. Rickman, “The commercialization of silicon photonics,” Nat. Photonics 8(8), 579–582 (2014). [CrossRef]  

3. Y. Zeng, I. Roland, X. Checoury, Z. Han, M. E. Kurdi, S. Sauvage, B. Gayral, C. Brimont, T. Guillet, F. Semond, and P. Boucaud, “Imaging of photonic crystal localized modes through third-harmonic generation,” ACS Photonics 3(7), 1240–1247 (2016). [CrossRef]  

4. Y. Wang, X. Wang, B. Zhu, Z. Shi, J. Yuan, X. Gao, Y. Liu, X. Sun, D. Li, and H. Amano, “Full-duplex light communication with a monolithic multicomponent system,” Light: Sci. Appl. 7(1), 83 (2018). [CrossRef]  

5. E. M. T. Fadaly, A. Dijkstra, J. R. Suckert, D. Ziss, M. A. J. van Tilburg, C. Mao, Y. Ren, V. T. van Lange, K. Korzun, S. Kölling, M. A. Verheijen, D. Busse, C. Rödl, J. Furthmüller, F. Bechstedt, J. Stangl, J. J. Finley, S. Botti, J. E. M. Haverkort, and E. P. A. M. Bakkers, “Direct-bandgap emission from hexagonal Ge and SiGe alloys,” Nature 580(7802), 205–209 (2020). [CrossRef]  

6. S. Chang, N. B. Rex, R. K. Chang, G. Chong, and L. J. Guido, “Stimulated emission and lasing in whispering-gallery modes of GaN microdisk cavities,” Appl. Phys. Lett. 75(2), 166–168 (1999). [CrossRef]  

7. J. Justice, C. Bower, M. Meitl, M. B. Mooney, M. A. Gubbins, and B. Corbett, “Wafer-scale integration of group III-V lasers on silicon using transfer printing of epitaxial layers,” Nat. Photonics 6(9), 610–614 (2012). [CrossRef]  

8. K. H. Li, W. Y. Fu, Y. F. Cheung, K. K. Y. Wong, and H. W. Choi, “Monolithically integrated InGaN/GaN light-emitting diodes, photodetectors, and waveguides on Si substrate,” Optica 5(5), 564–569 (2018). [CrossRef]  

9. M. Feng, J. Liu, Q. Sun, and H. Yang, “III-Nitride Semiconductor Lasers Grown on Si,” Prog. Quantum Electron. 77, 100323 (2021). [CrossRef]  

10. Z. Shi, X. Li, G. Zhu, Z. Wang, P. Grünberg, H. Zhu, and Y. Wang, “Characteristics of GaN-based LED fabricated on a GaN-on-silicon platform,” Appl. Phys. Express 7(8), 082102 (2014). [CrossRef]  

11. Y. Zeng, I. Roland, X. Checoury, Z. Han, M. E. Kurdi, S. Sauvage, B. Gayral, C. Brimont, T. Guillet, and M. Mexis, “Resonant second harmonic generation in a gallium nitride two-dimensional photonic crystal on silicon,” Appl. Phys. Lett. 106(8), 081105 (2015). [CrossRef]  

12. M. Athanasiou, R. M. Smith, J. Pugh, Y. Gong, M. J. Cryan, and T. Wang, “Monolithically multi-color lasing from an InGaN microdisk on a Si substrate,” Sci. Rep. 7(1), 10086 (2017). [CrossRef]  

13. N. Niu, T. L. Liu, I. Aharonovich, K. J. Russell, A. Woolf, T. C. Sadler, H. El-Ella, M. J. Kappers, R. A. Oliver, and E. L. Hu, “A full free spectral range tuning of p-i-n doped gallium nitride microdisk cavity,” Appl. Phys. Lett. 101(16), 161105 (2012). [CrossRef]  

14. M. Athanasiou, R. Smith, B. Liu, and T. Wang, “Room temperature continuous–wave green lasing from an InGaN microdisk on silicon,” Sci. Rep. 4(1), 7250 (2015). [CrossRef]  

15. Y. Zhang, Z. Ma, X. Zhang, T. Wang, and H. W. Choi, “Optically pumped whispering-gallery mode lasing from 2-µm GaN micro-disks pivoted on Si,” Appl. Phys. Lett. 104(22), 221106 (2014). [CrossRef]  

16. J. Sellés, C. Brimont, G. Cassabois, P. Valvin, and B. Gayral, “Deep-UV nitride-on-silicon microdisk lasers,” Sci. Rep. 6(1), 21650 (2016). [CrossRef]  

17. J. Sellés, V. Crepel, I. Roland, M. El Kurdi, X. Checoury, P. Boucaud, M. Mexis, M. Leroux, B. Damilano, S. Rennesson, F. Semond, B. Gayral, C. Brimont, and T. Guillet, “III-Nitride-on-silicon microdisk lasers from the blue to the deep ultraviolet,” Appl. Phys. Lett. 109(23), 231101 (2016). [CrossRef]  

18. Y. Zhang, H. Li, P. Li, A. Dehzangi, L. Wang, X. Yi, and G. Wang, “Optically-pumped single-mode deep-ultraviolet microdisk lasers with AlGaN-based multiple quantum wells on Si substrate,” IEEE Photonics J. 9, 5 (2017). [CrossRef]  

19. D. Wang, T. Zhu, R. A. Oliver, and E. L. Hu, “Ultra-low-threshold InGaN/GaN quantum dot micro-ring lasers,” Opt. Lett. 43(4), 799–802 (2018). [CrossRef]  

20. G. Zhu, J. Li, J. Li, J. Guo, J. Dai, C. Xu, and Y. Wang, “Single-mode ultraviolet whispering gallery mode lasing from a floating GaN microdisk,” Opt. Lett. 43(4), 647–650 (2018). [CrossRef]  

21. C. Zhao, C. W. Tang, J. Wang, and K. M. Lau, “Ultra-low threshold green InGaN quantum dot microdisk lasers grown on silicon,” Appl. Phys. Lett. 117(3), 031104 (2020). [CrossRef]  

22. M. Kneissl, M. Teepe, N. Miyashita, N. M. Johnson, G. D. Chern, and R. K. Chang, “Current-injection spiral-shaped microcavity disk laser diodes with unidirectional emission,” Appl. Phys. Lett. 84(14), 2485–2487 (2004). [CrossRef]  

23. M. Feng, J. He, S. Qian, H. Gao, Z. Li, Z. Yu, J. Liu, S. Zhang, D. Li, and L. Zhang, “Room-temperature electrically pumped InGaN-based microdisk laser grown on Si,” Opt. Express 26(4), 5043–5051 (2018). [CrossRef]  

24. J. Wang, M. Feng, R. Zhou, Q. Sun, and H. Yang, “Continuous-wave electrically injected GaN-on-Si microdisk laser diodes,” Opt. Express 28(8), 12201–12208 (2020). [CrossRef]  

25. A. Dadgar, F. Schulze, M. Wienecke, A. Gadanecz, J. BläSing, P. Veit, T. Hempel, A. Diez, J. Christen, and A. Krost, “Epitaxy of GaN on silicon—impact of symmetry and surface reconstruction,” New J. Phys. 9(10), 389 (2007). [CrossRef]  

26. A. Mantarci and M. Kundaki, “Role of RF power on physical properties of RF magnetron sputtered GaN/p-Si(100) thin film,” Mater. Sci.-Pol. 37(3), 454–464 (2019). [CrossRef]  

27. D. Kasahara, D. Morita, T. Kosugi, K. Nakagawa, J. Kawamata, Y. Higuchi, H. Matsumura, and T. Mukai, “Demonstration of blue and green GaN-based vertical-cavity surface-emitting lasers by current injection at room temperature,” Appl. Phys. Express 4(7), 072103 (2011). [CrossRef]  

28. R. Zhou, M. Feng, J. Wang, Q. Sun, and H. Yang, “InGaN-based lasers with an inverted ridge waveguide heterogeneously integrated on Si(100),” ACS Photonics 7(10), 2636–2642 (2020). [CrossRef]  

29. K. H. Li, Y. F. Cheung, W. Y. Fu, and H. W. Choi, “Electrically injected whispering-gallery mode InGaN/GaN microdisks,” Appl. Phys. Lett. 119(10), 101106 (2021). [CrossRef]  

30. Y. Gao, M. Craven, J. S. Speck, S. P. Denbaars, and E. L. Hu, “Dislocation- and crystallographic-dependent photoelectrochemical wet etching of gallium nitride,” Appl. Phys. Lett. 84(17), 3322–3324 (2004). [CrossRef]  

31. C. X. Ren, T. J. Puchtler, T. Zhu, J. T. Griffiths, and R. A. Oliver, “Defects in III-nitride microdisk cavities,” Semicond. Sci. Technol. 32(3), 033002 (2017). [CrossRef]  

32. A. C. Tamboli, E. D. Haberer, R. Sharma, K. H. Lee, S. Nakamura, and E. L. Hu, “Room-temperature continuous-wave lasing in GaN/InGaN microdisks,” Nat. Photonics 1(1), 61–64 (2007). [CrossRef]  

33. I. Aharonovich, A. Woolf, K. J. Russell, T. Zhu, N. Nan, M. J. Kappers, R. A. Oliver, and E. L. Hu, “Low threshold, room-temperature microdisk lasers in the blue spectral range,” Appl. Phys. Lett. 103(2), 021112 (2013). [CrossRef]  

34. T. Kouno, K. Kishino, and M. Sakai, “Lasing action on whispering gallery mode of self-organized GaN hexagonal microdisk crystal fabricated by RF-plasma-assisted molecular beam epitaxy,” IEEE J. Quantum Electron. 47(12), 1565–1570 (2011). [CrossRef]  

35. H. Baek, C. H. Lee, K. Chung, and G. C. Yi, “Epitaxial GaN microdisk lasers grown on graphene microdots,” Nano Lett. 13(6), 2782–2785 (2013). [CrossRef]  

36. T. Kouno, M. Sakai, K. Kishino, A. Kikuchi, N. Umehara, and K. Hara, “Crystal structure and optical properties of a high-density InGaN nanoumbrella array as a white light source without phosphors,” NPG Asia Mater. 8(7), e289 (2016). [CrossRef]  

37. W. W. Wong, C. Jagadish, and H. H. Tan, “III-V Semiconductor whispering-gallery mode micro-cavity lasers: advances and prospects,” IEEE J. Quantum Electron. (2022).

38. H. Zi, Y. F. Cheung, B. Damilano, E. Frayssinet, B. Alloing, J.-Y. Duboz, P. Boucaud, F. Semond, and H. W. Choi, “Influence of surface roughness on the lasing characteristics of optically pumped thin-film GaN microdisks,” Opt. Lett. 47(6), 1521–1524 (2022). [CrossRef]  

39. J. H. Park, D. Y. Kim, S. Hwang, D. Meyaard, E. F. Schubert, Y. D. Han, J. W. Choi, J. Cho, and J. K. Kim, “Enhanced overall efficiency of GaInN-based light-emitting diodes with reduced efficiency droop by Al-composition-graded AlGaN/GaN superlattice electron blocking layer,” Appl. Phys. Lett. 103(6), 061104 (2013). [CrossRef]  

40. P. Mishra, B. Janjua, T. K. Ng, D. H. Anjum, R. T. Elafandy, A. Prabaswara, C. Shen, A. Salhi, A. Y. Alyamani, M. M. El-Desouki, and B. S. Ooi, “On the optical and microstrain analysis of graded InGaN/GaN MQWs based on plasma assisted molecular beam epitaxy,” Opt. Mater. Express 6(6), 2052–2062 (2016). [CrossRef]  

41. K. Kim, “Nanopatterned surface effect on the epitaxial growth of InGaN/GaN multi-quantum well light emitting diode structure,” Trans. Electr. Electron. Mater. 10(2), 40–43 (2009). [CrossRef]  

42. A. Woolf, T. Puchtler, I. Aharonovich, T. Zhu, N. Niu, D. Wang, R. Oliver, and E. L. Hu, “Distinctive signature of indium gallium nitride quantum dot lasing in microdisk cavities,” Proc. Natl. Acad. Sci. U.S.A. 111(39), 14042–14046 (2014). [CrossRef]  

43. Y. Liang, J. Liu, M. Ikeda, A. Tian, R. Zhou, S. Zhang, T. Liu, D. Li, L. Zhang, and H. Yang, “Effect of inhomogeneous broadening on threshold current of GaN-based green laser diodes,” J. Semicond. 40(5), 052802 (2019). [CrossRef]  

44. Y. Zhang, J.-A. Huang, K. H. Li, D. Bai, Y. Wang, T. Wang, and H. W. Choi, “Influence of strain on emission from GaN-on-Si microdisks,” J. Phys. D: Appl. Phys. 49(37), 375103 (2016). [CrossRef]  

45. M. Buerger, G. Callsen, T. Kure, A. Hoffmann, A. Pawlis, D. Reuter, and D. J. As, “Lasing properties of non-polar GaN quantum dots in cubic aluminum nitride microdisk cavities,” Appl. Phys. Lett. 103(2), 021107 (2013). [CrossRef]  

46. Y. Zhang, X. Zhang, K. H. Li, Y. F. Cheung, C. Feng, and H. W. Choi, “Advances in III-nitride semiconductor microdisk lasers,” Phys. Status Solidi A 212(5), 960–973 (2015). [CrossRef]  

47. H. Takashima, H. Fujiwara, S. Takeuchi, K. Sasaki, and M. Takahashi, “Control of spontaneous emission coupling factor β in fiber-coupled microsphere resonators,” Appl. Phys. Lett. 92(7), 071115 (2008). [CrossRef]  

48. F. Tabataba-Vakili, C. Brimont, B. Alloing, B. Damilano, and P. Boucaud, “Analysis of low-threshold optically pumped III-nitride microdisk lasers,” Appl. Phys. Lett. 117(12), 121103 (2020). [CrossRef]  

49. A. Raun and E. Hu, “Ultralow threshold blue quantum dot lasers: what's the true recipe for success?” Nanophotonics 10(1), 23–29 (2020). [CrossRef]  

50. A. L. Coldren and S. W. Corzine, “Diode lasers and photonic integrated circuits,” Opt. Eng. 36(2), 616 (1997). [CrossRef]  

51. X. Zhang, Y. F. Cheung, Y. Zhang, and H. W. Choi, “Whispering-gallery mode lasing from optically free-standing InGaN microdisks,” Opt. Lett. 39(19), 5614–5617 (2014). [CrossRef]  

52. F. F. Qin, G. Y. Zhu, J. B. Yang, L. Wei, Q. N. Cui, and Y. J. Wa Ng, “Unidirectional single-mode lasing realization and temperature-induced mode switching in asymmetric GaN coupled cavities,” Nanoscale 14(5), 1921–1928 (2022). [CrossRef]  

53. H. Zi, W. Y. Fu, H. Choi, F. Tabataba-Vakili, and F. Semond, “Whispering-gallery mode InGaN microdisks on GaN substrates,” Opt. Express 29(14), 21280–21289 (2021). [CrossRef]  

54. F. Tabataba-Vakili, S. Rennesson, B. Damilano, E. Frayssinet, and P. Boucaud, “III-nitride on silicon electrically injected microrings for nanophotonic circuits,” Opt. Express 27(8), 11800–11808 (2019). [CrossRef]  

55. I. Roland, M. Gromovyi, Y. Zeng, M. E. Kurdi, S. Sauvage, C. Brimont, T. Guillet, B. Gayral, F. Semond, and J. Y. Duboz, “Phase-matched second harmonic generation with on-chip GaN-on-Si microdisks,” Sci. Rep. 6(1), 34191 (2016). [CrossRef]  

56. F. Tabataba-Vakili, L. Doyennette, C. Brimont, T. Guillet, S. Rennesson, E. Frayssinet, B. Damilano, J. Y. Duboz, F. Semond, and I. Roland, “Blue microlasers integrated on a photonic platform on silicon,” ACS Photonics 5, 3643–3648 (2019). [CrossRef]  

57. Z. Shi, X. Gao, J. Yuan, S. Zhang, Y. Jiang, F. Zhang, Y. Jiang, H. Zhu, and Y. Wang, “Transferrable monolithic III-nitride photonic circuit for multifunctional optoelectronics,” Appl. Phys. Lett. 111(24), 241104 (2017). [CrossRef]  

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (8)

Fig. 1.
Fig. 1. (a) Schematic diagrams illustrating the fabrication process of InGaN-based microdisk light-emitters heterogeneously integrated on Si (100). The SEM images illustrating the cross-sectional structures for the (b) as-grown samples, and samples (c) after LLO process, (d) after CMP process, and (e) as-fabricated electrically-injected microdisk devices on Si, respectively.
Fig. 2.
Fig. 2. A homemade setup for power-dependent µ-PL test.
Fig. 3.
Fig. 3. The optical images for the wafers (a) before and (b) after the CMP process. Correspondingly, the AFM images showing the surface morphologies for the InGaN/GaN-on-Si wafer (c) before and (d) after the CMP process. (e) µ-PL spectra and (f) Raman spectra from the as-grown MQWs on sapphire, and wafers after LLO process, and after CMP process, respectively. (g) The statistical Raman E2(high) modes as measured across the wafer.
Fig. 4.
Fig. 4. (a) an array of as-transferred InGaN-based microcavities on Si (100) for optical pumping tests. (b) Tilted-view SEM image of a 12 µm microdisk on Si. (c) Magnified view on the microdisk sidewall. (d) The temperature-dependent integrated PL intensities for the transferred InGaN thin-film.
Fig. 5.
Fig. 5. (a) Power-dependent PL spectra from a 12 µm microdisk integrated on Si at RT. (b) Shift in WGM energies detected at around 495.1 nm, 501.9 nm, and 505.8 nm as increasing the pumping power densities.
Fig. 6.
Fig. 6. (a) Integrated PL intensities of all modes as a function of the pumping densities from the InGaN/GaN microdisk integrated on Si. Insets are optical photos of the as-pumped microdisk devices below and above the Eth. (b) The extracted β factors from all microdisk and microring cavities based on their lasing output characteristics. (c) The WGM mode linewidth evolution as increasing the pumping power densities beyond the Eth. (d) Eth dependence of the microdisk diameter.
Fig. 7.
Fig. 7. (a) FDTD simulated resonances from a ∼12 µm microdisk and its comparison with PL spectra. The energy intensity patterns for the first-order TE WGMs in the (b) horizontal and (c) vertical direction at ∼505.8 nm. The insets show the enlarged view of the energy profiles in the red dashed boxes.
Fig. 8.
Fig. 8. (a) As-fabricated 160 µm microdisk device for EL test and photos of the luminescence driven at currents of 0.1, 1, and 10 mA, respectively. (b) Typical IV curve of the transferred microdisk devices. The insets show the IV curve at reverse bias. (c) EL spectra under different injected current densities.

Equations (3)

Equations on this page are rendered with MathJax. Learn more.

Δ ω =   K σ xx
m λ / n e f f   =   2 π r
F S R   =   λ 2 / 2 π n e f f r\; 
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.