Development of Micron Sized Photonic Devices Based on Deep Gan Etching

hv photonics

Communication Development of Micron Sized Photonic Devices Based on Deep GaN Etching

Karim Dogheche 1, Bandar Alshehri 2, Galles Patriache 3 and Elhadj Dogheche 1,*

1 Institute of Electronics, Université Polytechnique Hauts de France, Microelectronics & Nanotechnology IEMN CNRS UMR 8520, 59313 Valenciennes, France; [email protected] 2 TMO Transformation Management Ofﬁce, Riyad 11564, Saudi Arabia; [email protected] 3 Centre for Nanoscience and Nanotechnology (C2N), CNRS, UMR 9001, 91460 Marcoussis, France; [email protected] * Correspondence: [email protected]

Abstract: In order to design and development efﬁcient III-nitride based optoelectronic devices, technological processes require a major effort. We propose here a detailed review focussing on the etching procedure as a key step for enabling high date rate performances. In our reported research activity, dry etching of an InGaN/GaN heterogeneous structure was investigated by using an inductively coupled plasma reactive ion etching (ICP-RIE). We considered different combinations

of etch mask (Ni, SiO2, resist), focussing on the optimization of the deep etching process. A GaN mesa process with an etching depth up to 6 µm was performed in Cl2/Ar-based plasmas using ICP reactors for LEDs dimen sions ranging from 5 to 150 µm2. Our strategy was directed toward the mesa formation for vertical-type diode applications, where etch depths are relatively large. Etch characteristics were studied as a function of ICP parameters (RF power, chamber pressure, fixed total flow rate). Surface morphology, etch rates and sidewall profiles observed into InGaN/GaN structures were compared under different types of etching masks. For deep etching up to few microns into the GaN template, we state that a Ni or SiO2 mask is more suitable to obtain a good selectivity and Citation: Dogheche, K.; Alshehri, B.; vertical etch profiles. The optimized etch rate was about 200nm/min under moderate ICP conditions. Patriache, G.; Dogheche, E. We applied these conditions for the fabrication of micro/nano LEDs dedicated to LiFi applications. Development of Micron Sized Photonic Devices Based on Deep Keywords: InGaN/GaN; ICP; deep etching; mesa; photonic devices GaN Etching. Photonics 2021, 8, 68. https://doi.org/10.3390/ photonics8030068 1. Introduction Received: 27 December 2020 Accepted: 27 February 2021 GaN and related alloys are among the most attractive materials, due to their excel- Published: 2 March 2021 lent characteristics as well as large bandgap, high breakdown electrical field and high electron saturation velocity, for applications such light emitting diodes (LEDs), lasers, pho-

Publisher’s Note: MDPI stays neutral todetectors and high power electronics [1–6]. Mature material associated with optimized with regard to jurisdictional claims in technological processes is the key issue for the near future development of active/passive published maps and institutional afﬁl- RF-optoelectronic components. Dry etching processes dedicated to ternary structures (such iations. as InxGa1−xN and GaN) are of particular importance, since such devices generally include heterostructures. Hence, the fabrication of GaN-based optoelectronic devices depends on dry etching either partially or totally. Typically, the etch process should yield high etch rates, minimal surface roughening, good reproducibility and a high degree of anisotropy. All of these properties can be achieved with inductively coupled plasma (ICP) etching Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. Sidewall and etched surface morphology are critical parameters in the formation of mesas This article is an open access article or vertical devices with high ratio aspect. In the actual context of photonic devices, require- distributed under the terms and ments for GaN etching are ﬁxed in the order of few hundred nanometers [7–9], whereas conditions of the Creative Commons this is completely different for those expected for vertical power devices [10–12]. Deep Attribution (CC BY) license (https:// etching of GaN is presently needed for new applications such as photonic, power and creativecommons.org/licenses/by/ microelectromechanical system devices for device isolation, waveguide or vertical channel 4.0/).

Photonics 2021, 8, 68. https://doi.org/10.3390/photonics8030068 https://www.mdpi.com/journal/photonics Photonics 2021, 8, x FOR PEER REVIEW 2 of 7

Photonics 2021, 8, 68 2 of 7 and microelectromechanical system devices for device isolation, waveguide or vertical channel formation [13–15]. Low etch depth has been successfully employed for InGaN/GaN devices to create photonics crystal and LED mesas with, in general, etch depths below 1um,formation while deep [13–15 etch]. Low has etchbeen depth mainly has focused been successfully on GaN materials employed and for not InGaN/GaN de- actual devices. Hencevices to more create work photonics is required crystal to and apply LED and mesas optimize with, deep in general, etch to etch create depths below 1um, functional deviceswhile [16–18] deep The etch approach has been mainlyis to etch focused the epilayer on GaN from materials p-type and GaN not to actual sap- devices. Hence phire to form individualmore work or is interconnected required to apply microchips and optimize for a deepmicro-LED etch to createarray [19]). functional Our devices [16–18] purpose is to haveThe a approach chip-to-chip is to isolation etch the and epilayer to improve from p-type the electrical GaN to performance sapphire to formby individual or reducing the metalinterconnected surface and microchips global capa forcitance a micro-LED induced array by [19metallization]). Our purpose contacts is to have a chip-to- chip isolation and to improve the electrical performance by reducing the metal surface and (pads, connection line) on n-GaN. The deep etching process was performed to define in- global capacitance induced by metallization contacts (pads, connection line) on n-GaN. The dependent mesa. Different configurations of etching masks were defined. Photoresist deep etching process was performed to define independent mesa. Different configurations (AZ9260), metal (Ni) and silicon dioxide (SiO2) were used as etching masks. The objective of etching masks were defined. Photoresist (AZ9260), metal (Ni) and silicon dioxide (SiO ) was to assess surface morphology, mesa edges and the etching sidewall quality. Several 2 were used as etching masks. The objective was to assess surface morphology, mesa edges dimensions of etched mesas ranging from 5 × 5 to 150 × 150 µm² were evaluated as a func- and the etching sidewall quality. Several dimensions of etched mesas ranging from 5 × 5 to tion of etching profile. The effect of different masks on the etched mesa are discussed for 150 × 150 µm2 were evaluated as a function of etching profile. The effect of different masks depths varying from hundreds of nanometers up to a few micrometers. We also illustrated on the etched mesa are discussed for depths varying from hundreds of nanometers up to the effects of etched parameters on the resulting characteristics, such as etch rate, mask a few micrometers. We also illustrated the effects of etched parameters on the resulting erosion and, as a consequence, surface roughness. characteristics, such as etch rate, mask erosion and, as a consequence, surface roughness.

2. Experimental2. Details Experimental Details In our investigation,In our III-Nitride investigation, layers III-Nitride were grown layers by were NovaGaN grown Ltd by NovaGaN(Lausanne, Ltd. (Lausanne, Switzerland), usingSwitzerland), a metalorganic using achemical metalorganic vapour chemical deposition vapour system deposition (MOCVD) system on (MOCVD) c- on c-face face sapphire substrates.sapphire substrates.The epilayer The struct epilayerure consisted structure of consisted 2 µm thick of undoped 2 µm thick GaN, undoped 1 GaN, 1 µm µm thick n-GaN,thick ten n-GaN,periods tenof 10 periods nm thick of 10 GaN/3 nm thick nm GaN/3thick InGaN nm thick MQW InGaN and MQW100 nm and 100 nm thick thick p-GaN (as p-GaNshown in (as Figure shown 1a,b). in Figure Cross-1sectionsa,b). Cross-sections and plan-views and of plan-views the samples of were the samples were investigated by investigatedhigh-angle annular by high-angle dark fiel annulard scanning dark transmission ﬁeld scanning electron transmission microscopy electron microscopy (HAADF-STEM).(HAADF-STEM). Samples were prepared Samples by were tilting prepared along bythe tilting (1100) along and (1120) the (1100) axes andon (1120) axes on thin foils by Focusedthin foils Ion Beam by Focused (FIB) in Ion order Beam to (FIB)see the in dislocations, order to see the chemical dislocations, analysis chemical and analysis and stacking faults. Pinchingstacking faults.zones generate Pinching thre zonesading generate dislocations threading (yellow dislocations arrows) (yellowassociated arrows) associated with plastic relaxationwith plastic mechanism. relaxation They mechanism. also lead to They variation also lead of the to variationtop layer ofthickness the top layer thickness (p-GaN layer) above(p-GaN these layer) zones. above Our these invest zones.igations Our revealed investigations that InGaN/GaN revealed that samples InGaN/GaN samples had threading dislocationhad threading densities dislocation in the range densities of 3–5 in the× 10 range8 cm−2. of Dislocations 3–5 × 108 cmwere−2 .cre- Dislocations were ated during growthcreated due during to the lattice growth mismatch due to the between lattice the mismatch sapphire between substrate the and sapphire GaN substrate and [20–22]. Usually,GaN the [Cl202–/Ar22]. based Usually, plasma the Clshows2/Ar correct based plasmaresults showssuch as correct high etch results rates, such as high etch anisotropic profilerates, and anisotropic a smooth surf proﬁleace andof homogeneous a smooth surface GaN of [23,24]. homogeneous GaN [23,24].

(a) (b)

Figure 1. (Figurea) InGaN/GaN 1. (a) InGaN/GaN Multi-Quantum Multi-Quantum Well (MQW) Well structure (MQW) structure layers. (b layers.) Scanning (b) Scanning transmission transmission electron microscopy (STEM) crosselectron section microscopy along the (STEM) (1120) zone cross axis section showing along dislocations. the (1120) zone axis showing dislocations.

For the formation of the etching mask, we considered three possible combinations: photoresist (4 µm thick AZ9260), SiO2 and Ni layers with a thickness of 600 nm and 520 nm, respectively. The latter were deposited by RF sputtering on p-GaN/InGaN/n-GaN. Patterning was realized by a lift-off process. Dry etching was performed using an Oxford

Photonics 2021, 8, x FOR PEER REVIEW 3 of 7

nm, respectively. The latter were deposited by RF sputtering on p-GaN/InGaN/n-GaN. Patterning was realized by a lift-off process. Dry etching was performed using an Oxford Photonics 2021, 8,plasmalab 68 100 plus system. Samples (1.5 × 2 cm) were mounted on a three inches Si wafer 3 of 7 using a thermally conductive paste and loaded on the wafer carrier. The temperature of the backside cooled sample chuck was kept at 10 ℃. Using Ni, SiOplasmalab2 and photoresist 100 plus system. masks, Samples the etching (1.5 × 2 conditions cm) were mounted were 10 on sccm a three of inchesCl2, 30 Si wafer sccm of Ar, 10 usingmTorr a thermallytotal pressure conductive and ICP paste power and loaded ranging on from the wafer 100 carrier.to 500 TheWatts. temperature The of ICP power and,the indirectly, backside cooledhigh-density sample chuckplasma was were kept limited at 10 ◦C. to 500W in order to avoid damage to the maskingUsing material Ni, SiO 2andand etching photoresist selectivity. masks, theEtch etching depths conditions were obtained were 10 from sccm of Cl2, surface profilometer30 sccm measurements of Ar, 10 mTorr after total mask pressure removal. and ICP The power surface ranging morphology from 100 to was 500 ex- Watts. The amined using SEM.ICP power and, indirectly, high-density plasma were limited to 500W in order to avoid damage to the masking material and etching selectivity. Etch depths were obtained from surface profilometer measurements after mask removal. The surface morphology was 3. Results and Discussionexamined using SEM. A first section of the p-GaN/InGaN/n-GaN MQW template was initially etched in order to attain the3. Results n-type and GaN Discussion layer. This resulted in a 3 µm depth mesa, as described in Figure 2a,b, under Athe first same section conditions of the p-GaN/InGaN/n-GaN described above. After MQW this, template we applied was initially Cl2/Ar etched in plasma to etch theorder last to section attain the of n-typethe templa GaNte, layer. the This100 nm resulted thick in p-GaN a 3 µm layer. depth During mesa, as chlo- described in rination of GaN,Figure chlorine2a,b, preferentially under the same reac conditionsts with the described gallium above. species After (Ga) this, surface we applied atoms Cl 2/Ar plasma to etch the last section of the template, the 100 nm thick p-GaN layer. During to form Ga chlorides. For etching of the InGaN/GaN MQW layered structure, the indium chlorination of GaN, chlorine preferentially reacts with the gallium species (Ga) surface species (In) is involvedatoms toform in the Ga chemical chlorides. reac For etchingtion and of thecontributes InGaN/GaN to the MQW formation layered structure,of In the chloride on surfaceindium [25], species which (In) is is not involved volatile in theat the chemical process reaction temperature. and contributes A hard to mask, the formation such as nickel (Ni),of In is chloride preferred on surfacefor photores [25], whichist because is not volatileetch selectivity at the process of GaN temperature. over pho- A hard toresist is moremask, advantageous such as nickel in (Ni),Cl2-based is preferred plasma for photoresistchemistry. becauseThe best etch etching selectivity surface of GaN over profile is also obtainedphotoresist using is more an Ni advantageous mask compared in Cl2 -basedto those plasma for SiO chemistry.2 or photoresist The best masks. etching surface GaN etch ratesprofile and the is alsoselectivity obtained over using etch an Nimask mask were compared evaluated to those from for SiOthe2 depthor photoresist of the masks. etched measuredGaN by etch the ratesprofilometry. and the selectivity We noti overced that etch maska selectivity were evaluated of 17 was from found the depth for of the etched measured by the profilometry. We noticed that a selectivity of 17 was found for the Ni mask, which is much higher than 5.3 and 1.5 for, respectively, SiO2 and photoresist the Ni mask, which is much higher than 5.3 and 1.5 for, respectively, SiO2 and photoresist masks as detailedmasks above. as detailed Using an above. ICP power Using anof ICP500 powerWatts ofand 500 RF Watts power and of RF 100 power Watts of for 100 Watts 2 each Ni, SiO andfor eachphotoresist Ni, SiO2 andmask, photoresist the resulting mask, theetched resulting surface etched profiles surface are profiles shown are in shown in Figure 2a,b. TheFigure surface2a,b. morphology The surface morphologyrevealed a more revealed homogenous a more homogenous and smoother and smoother aspect aspect when the Ni maskwhen was the selected. Ni mask was selected.

Ni mask

Figure 2. Figure(a) Scanning 2. (a) electron Scanning microscope electron (SEM) microscope analysis (SEM) of mesa analysis sidewalls of after mesa inductively sidewalls coupled after inductively plasma (ICP) cou- etching with sputteredpled plasma nickel mask (ICP) (depth etching of 3 withµm for sputtered ICP power nickel of 500 mask Watts. (depth (b) Typical of 3 µm etching for ICP results power showing of 500 Ni Watts. mask. (b) Typical etching results showing Ni mask. We applied a deep etching procedure in order to analyze the maximal endurance and We appliedthe a deep toughness etching of the procedure SiO2 mask. in order As shown to analyze in Figure the3a, maximal the quality endurance of etching and mesa was satisfactory for mesa size larger than 20 × 20 µm2 but limited quality was obtained for the toughness of the SiO2 mask. As shown in Figure 3a, the quality of etching mesa was mesa below this dimension. Damaged areas were observed in the border to mesa indicating satisfactory for mesa size larger than 20 × 20 µm2 but limited quality was obtained for the limitation of the SiO etching mask. The formation of columnar defects also appeared mesa below this dimension. Damaged2 areas were observed in the border to mesa indicat- using the SiO2 cover-plate in this range of etching proﬁle. Columnar defects are generally ing the limitationdue of to the a micromasking SiO2 etching effect mask. [26 ].The They formation originate fromof columnar sputtering defects of hard also mask ap- materials, peared using thewhich SiO deposit2 cover-plate particles in onthis the range etched of surfaceetching and profile. create Columnar a local micromask: defects are Figure 3b generally due toshows a micromasking an etched GaN effect mesa [26]. with Th aney SiOoriginate2 cover from plate. sputtering The etched of surfacehard mask presented a

Photonics 2021, 8, x FOR PEER REVIEW 4 of 7

Photonics 2021, 8, 68 4 of 7 materials, which deposit particles on the etched surface and create a local micromask: Fig- ure 3b shows an etched GaN mesa with an SiO2 cover plate. The etched surface presented a regular morphologyregular but morphology with the presence but with theof GaN presence spikes of GaN with spikes diameters with diameters from 50 from to 200 50 to 200 nm, nm, and heights fromand heights1 to 2 µm, from at 1 tocertain 2 µm, places at certain of the places etched of the surface. etched surface.

(a) (b)

FigureFigure 3. (a) Cross 3. (a) section Cross SEMsection images SEM of images InGaN/GaN of InGaN/GaN MQW structure MQW after structure ICP etching after with ICP etching etching depth with of etch- 1 µm with

SiO2 mask,ing depth and (b of) observation 1 µm with ofSiO mesa2 mask, sidewall and etching (b) observation with SiO2 maskof mesa (Etching sidewall depth etching of 4 µm with at 500 SiO Watts2 mask ICP power). (Etching depth of 4 µm at 500 Watts ICP power). In order to etch entirely the epilayer through the etch mask to expose the sapphire In order to etchsubstrate, entirely high the ion epilayer ﬂuxes were through required the toetch achieve mask high to expose etch rate the for sapphire the formation of a substrate, high ionthick fluxes mesa. were In addition, required high to achieve ICP power high (>500 etch Watts) rate couldfor the result formation in extensive of a roughness on the sidewall due to mask damage or erosion during etching. The use of an Ni mask thick mesa. In addition, high ICP power (>500 Watts) could result in extensive roughness allowed reduction of erosion of the mask edge and limited the formation of pillars observed on the sidewall due to mask damage or erosion during etching. The use of an Ni mask when using an oxide mask. The SiO2 mask erosion could be delayed by increasing the allowed reductionthickness of erosion of the of etchthe mask.mask Theedge presence and limited of nonuniformity the formation on the of SiO pillars2 mask ob- shape could served when usingcause an oxide damage mask or striations. The SiO to2 mask the sidewalls erosion ofcould the etched be delayed sample. by These increasing irregularities on the thickness of thethe etch mask mask. boundaries The presence could be of reduced nonuniformity by optimizing on the the SiO lithography2 mask shape process used to could cause damagepattern or striations the SiO2. to We the investigated sidewalls of the the experiment etched sample. with a SiOThese2 mask irregularities of 600 nm thickness, on the mask boundariesusing GaN could epilayers be reduced grown by thicker optimizing on a sapphire the lithograp substrate.hy process The SiO used2 mask to seemed to be a good choice for the mask because it offered sidewalls with reduced roughness and pattern the SiO2. We investigated the experiment with a SiO2 mask of 600 nm thickness, correct selectivity. The mesa sidewall as shown Figure3b, had a proﬁle with a lower using GaN epilayers grown thicker on a sapphire substrate. The SiO2 mask seemed to be slope, which was acceptable for a deep etch close to 4.5 µm. Compared to the Ni mask, a good choice for the mask because it offered sidewalls with reduced roughness and cor- higher ICP powers sputtered the SiO2 mask faster and caused the redeposition of residues rect selectivity. Theon mesa the surface sidewall which as createdshown aFigure more columnar3b, had a defectprofile due with to micromaskinga lower slope, during the which was acceptableetch. Duringfor a deep etching etch of close the InGaN/GaN to 4.5 µm. Compared by Cl2/Ar plasma, to the lowNi mask, volatile higher In chlorides can ICP powers sputteredbe formed the SiO on2 themask etched faster surface and caused [27]. In the could redeposition be involved of in residues the plasma on the reaction with surface which createdchlorine a more and favorablecolumnar to defect the pillar’s due formationto micromasking [28] by the during micro-masks the etch. on Dur- the etch surface. ing etching of theMoreover, InGaN/GaN part by of pillarsCl2/Ar results plasma, from low structural volatile defects In chlorides and dislocations can be formed in GaN. When the on the etched surfaceetch depth[27]. In increases, could be the involv densityed of in these the surfaceplasma features reaction was with found chlorine to increase. and When the etch depth was less than 2 µm, there was nearly no pillars observed on the etched GaN favorable to the pillar’s formation [28] by the micro-masks on the etch surface. Moreover, surface. Although a number of studies have been performed on the plasma etching of GaN, part of pillars resultsvery from little structural is known about defects the and mechanism dislocations of pillar in orGaN. pit formation When the on etch the depth etched surface of increases, the densitythe GaN/InGaN of these surface heterostructure features was [29 ].found to increase. When the etch depth was less than 2 µm, thereLet uswas consider nearly now no pi thellars third observed option with on applicationthe etched ofGaN a photoresist surface. Alt- mask. Etching hough a number ofconditions studies have were been 10 sccm performed of Cl2, 30 on sccm the plasma of Ar, 10 etching mTorr of total GaN, pressure very little and 500 Watts is known about theICP mechanism power. AZ9260 of pillar Photoresist or pit used formation as etching on maskthe etched exhibited surface poor persistenceof the to the GaN/InGaN heterostructureICP etching process.[29]. Photoresist etch masks have a low selectivity compared to GaN [30]. Let us considerA hardnow masksthe third such option as Ni with is preferred application for photoresist of a photoresist because mask. etch selectivityEtching of GaN over photoresist is more advantageous in Cl2-based plasma chemistry. We observed that conditions were 10 sccm of Cl2, 30 sccm of Ar, 10 mTorr total pressure and 500 Watts ICP GaN etching occurs without any pattern deformation at low ICP power (150 Watts) and power. AZ9260 Photoresistmoderate etch used rates as withinetching the mask same exhibited conditions. poor Up to persistence this power value,to the theICP photoresist etching process. Photoresistmask is drastically etch masks affected have by mask a low damage selectivity and edge compared degradations. to GaN Figure [30].4 aA shows the hard masks such as Ni is preferred for photoresist because etch selectivity of GaN over photoresist is more advantageous in Cl2-based plasma chemistry. We observed that GaN

Photonics 2021, 8, x FOR PEER REVIEW 5 of 7

Photonics 2021etching, 8, 68 occurs without any pattern deformation at low ICP power (150 Watts) and mod- 5 of 7 erate etch rates within the same conditions. Up to this power value, the photoresist mask is drastically affectedetching by maskoccurs damage without andany edgepattern degradations. deformation atFigure low ICP 4a showspower (150the etched Watts) and mod- mesa with an optimizederate etch depth rates obtainedwithin the at same 1.5 conditions.µm for larger Up todimensions this power value,(from the50 tophotoresist 200 mask isetched drastically mesa affected with an optimizedby mask damage depth and obtained edge degradations. at 1.5 µm for larger Figure dimensions 4a shows the (from etched 50 µm2). Concerning mesa sizes2 smaller than 10 µm², the etching process2 was limited in terms mesato 200 withµm ).an Concerning optimized mesadepth sizesobtained smaller at 1.5 than µm 10 forµm larger, the etchingdimensions process (from was 50 limited to 200 of quality and shape, showing defective areas in the mesa edges as illustrated in figure 4b. µmin terms2). Concerning of quality mesa and shape,sizes smaller showing than defective 10 µm², areasthe etching in the process mesa edges was limited as illustrated in terms in The photoresist mask was a material less robust and more affected with physical bom- ofFigure quality4b. Theand photoresistshape, showing mask defective was a material areas in less the robust mesa andedges more as illustrated affected with in figure physical 4b. bardment contribution.Thebombardment photoresist contribution. mask was a material less robust and more affected with physical bom- bardment contribution.

(a) (a) (b) (b)

FigureFigure 4. SEM 4. images SEM images of mesaFigure of sidewalls mesa 4. SEM sidewalls usingimages a photoresistof using mesa a sidewalls photor mask:esist using total mask: etchinga photor total depthesist etching mask: of 1.5 depthtotalµm, withetching of 1.5 a mesa depthµm, width of 1.5 of µm, (a) 100withµm anda mesa (b) 10 widthµm -ICP ofwith ( powera) a 100 mesa ofµm 150width and Watts. of(b ()a 10) 100 µm µm -ICP and power (b) 10 µmof 150 -ICP Watts. power of 150 Watts.

The Ni mask representsThe Ni maskthe best represents choice thefor best obtaining choice forfoa rhigh-selectivity obtainingobtaining aa high-selectivityhigh-selectivity and vertical and verticalvertical etch profiles for deep etched structures. Throughout the experiment, RF power, ratio of etch profiles for deepetch profilesetched forstructures. deep etched Thro structures.ughout the Throughout experiment, the experiment,RF power, ratio RF power, of ratio of mixed gas and chamber pressure were conserved, which seems to be a good compromise mixed gas and chamber pressure were conserved, which seems to be a good compromise for high etching rate and to investigate the impact of ICP power on the etched profileprofile and for high etching ratenature and of to masks. investigate The temperature the impact of of the ICP ba backsideckside power cooled on the sample etched chuck profile was and maintained at nature of masks. The10 ◦℃ C.temperature. We We therefore of studied the ba ckside the etch cooled rate as sample a function chuck of ICP was power. maintained Further at experiments 10 ℃. We thereforeusing studied different the gasetchgas combinationscombinations rate as a function will will be be investigatedof investigated ICP power. for for improvements,Further improvements, experiments while while focusing focusing on using different gasonone combinations one mask. mask. will be investigated for improvements, while focusing on one mask. Ni masks permitted permitted us us to to study study the the ICP ICP power power parameter, parameter, demonstrating demonstrating a akey key ele- el- Ni masks permittedmentement in in usthe the toetching etching study rate. rate.the ICPThe The differentpower different parameter, values values used used demonstrating in in this experiment a key led ele- to extractingextracting ment in the etchingdifferent rate. etchingThe different conditions. values Increasing used in the this ICP experiment power led to led increasing to extracting chemical compo- nent of the etch flow at identical conditions. Figure5 shows the linear influence on the different etching conditions.nent of the etchIncreasing flow at theidentical ICP power conditions. led to Figure increasing 5 shows chemical the linear compo- influence on the etch rate and the DC bias bias when when increasing increasing the ICP ICP power. power. We We observed that a compromise nent of the etch flow at identical conditions. Figure 5 shows the linear influence on the was found for a very low inclination and smoothsmooth sidewall mesa for the Ni mask when ICP etch rate and the DC bias when increasing the ICP power. We observed that a compromise power was 500 Watts withwith anan etchetch raterate ofof 198198 nm/min.nm/min. was found for a very low inclination and smooth sidewall mesa for the Ni mask when ICP power was 500 Watts with an etch rate of 198 nm/min.

Figure 5. Etch rate of GaN and corresponding DC bias versus ICP power.

Photonics 2021, 8, 68 6 of 7

4. Conclusions

Dry etching of p-GaN/InGaN/n-GaN structures was carried out in Cl2/Ar inductively coupled plasmas. The objective was directed toward mesa formation for vertical-type devices, where the etch depth is up to 3 µm. Through a combination of etch masks and the plasma etch process, mesa structures with smooth sidewalls and reduced post-etch residues were achieved. In our investigations, dry etch development was focused on deep mesa into GaN materials. The etch parameters were adjusted to obtain high etch depth with low defects and good verticality. The experiment indicated that the Ni mask is more appropriate for the deep etch process in comparison to SiO2 and photoresist masks in the case of micro-mesa etching. Such a mask is more suitable to obtain a good selectivity and vertical etch proﬁles for an etch depth exceeding 6µm. The highest etch rate was 198 nm/min at ICP power of 500 Watts. Here we found that the most vertical and smooth sidewall mesa were produced for the higher ICP power. We observed an increase in the etch rate of GaN with increasing ICP power attributed to increase in Cl radical and ion density in the plasma. The pattern transfer during etching was intimately limited by the quality of the etch mask and an inappropriate mask could result in surface morphologies that include pits and/or pillars. ICP etching can potentially provide relatively low damage while maintaining fast etching rates and superior uniformity with an Ni mask. The nature of etching masks also plays an important role in device performance due their junction and possible contamination of sidewalls during etching process. Subsequently, the evaluation of electrical response is largely impacted by the nature and the quality of the etching process.

Author Contributions: K.D. and B.A. designed the masks for clean room experiment and performed the etching process. G.P. performed the TEM characterizations. E.D. wrote the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This work was funded by the Bureau Culturel Saoudien-Ambassade d’Arabie Saoudite, Paris, France. Data Availability Statement: The data presented in this study are available on request from the corresponding author. Acknowledgments: We acknowledge the support from the Saudi Cultural Attaché, Bureau Culturel Saoudien, Ambassade d’Arabie Saoudite à Paris, France. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References 1. Li, X.; Jiang, Y.; Li, J.; Shi, Z.; Zhu, G.; Wang, Y. Integrated photonics chip with InGaN/GaN light-emitting diode and bended waveguide for visible-light communicacations. Opt. Laser Technol. 2019, 114, 103–109. [CrossRef] 2. Zhou, S.; Wang, S.; Liu, S.; Ding, H. High power GaN-based LEDs with low optical loss electrode structure. Opt. Laser Technol. 2013, 54, 321–325. [CrossRef] 3. Li, K.H.; Fu, W.Y.; Cheung, Y.F.; Wong, K.K.Y.; Wang, Y.; Lau, K.M.; Choi, H.W. Monolithically integrated InGaN/GaN light- emitting diodes, photodetectors, and waveguides on Si substrate. Optica 2018, 5, 564. [CrossRef] 4. Nakamura, S. InGaN-Based laser diodes. Annu. Rev. Mater. Sci. 1998, 28, 125–152. [CrossRef] 5. Omnès, F.; Monroy, E.; Muñoz, E.; Reverchon, J.-L. Wide bandgap UV photodetectors: A short review of devices and applications. SPIE 2007, 6473, 64730E. 6. Lee, K.H.; Joo, D.H.; Kim, M.S.; Yu, J.S. Improved light extraction of InGaN/GaN blue LEDs by GaOOH NRAs using a thin ATO seed layer. Nanoscale Res. Lett. 2012, 7, 458. [CrossRef] 7. Khokhar, A.Z.; Parsons, K.; Hubbard, G.; Rahman, F.; Macintyre, D.S.; Xiong, C.; Massoubre, D.; Gong, Z.; Johnson, N.P.; De La Rue, R.M.; et al. Nanofabrication of gallium nitride photonic crystal light-emitting diodes. Microelectron. Eng. 2010, 87, 2200–2207. [CrossRef] 8. Hong, S.; Cho, C.; Lee, S.; Yim, S.; Lim, W.; Kim, S.; Park, S. Localized surface plasmon-enhanced near-ultraviolet emission from InGaN/GaN light-emitting diodes using silver and platinum nanoparticles. Opt. Express 2013, 21, 3138–3144. [CrossRef] [PubMed] 9. Baik, K.H.; Pearton, S.J. Dry etching characteristics of GaN for blue/green light-emitting diode fabrication. Appl. Surf. Sci. 2009, 255, 5948–5951. [CrossRef] 10. Zhang, Y.; Sun, M.; Piedra, D.; Azize, M.; Zhang, X.; Fujishima, T.; Palacios, T. GaN-on-Si vertical schottky and p-n diodes. IEEE Electron. Dev. Lett. 2014, 35, 618–620. Photonics 2021, 8, 68 7 of 7

11. Hu, J.; Zhang, Y.; Sun, M.; Piedra, D.; Chowdhury, N.; Palacios, T. Materials and processing issues in vertical GaN power electronics. Mater. Sci. Semicond. Proc. 2018, 78, 75–84. [CrossRef] 12. Tohru, O. Recent development of vertical GaN power devices: Japanese. J. Appl. Phys. 2019, 58, SB0805. 13. Xiao, M.; Yan, X.; Xie, J.; Beam, E.; Cao, Y.; Wang, H.; Zhang, Y. Origin of leakage current in vertical GaN devices with nonplanar regrown p-GaN. Appl. Phys. Lett. 2020, 117, 183502. [CrossRef] 14. Sun, Y.; Kang, X.; Zheng, Y.; Lu, J.; Tian, X.; Wei, K.; Wu, H.; Wang, W.; Liu, X.; Zhang, G. Review of the Recent Progress on GaN-Based Vertical Power Schottky Barrier Diodes (SBDs). Electronics 2019, 8, 575. [CrossRef] 15. Hsieh, Y.; Chen, W.; Chang, L.; Chow, L.; Borges, S.; Schulte, J.A.; Huang, S.; Jeng, M.; Yu, C. Etched Gallium Nitride Waveguide for RamanSpectroscopic Applications. Crystals 2019, 9, 176. [CrossRef] 16. Qiu, R.; Lu, H.; Chen, D.; Zhang, R.; Zheng, Y. Optimization of inductively coupled plasma deep etching of GaN and etching damage analysis. Appl. Surf. Sci. 2011, 257, 2700. [CrossRef] 17. le Boulbar, E.D.; Lewins, C.J.; Allsopp, D.W.E.; Bowen, C.R.; Shields, P.A. Fabrication of high-aspect ratio GaN nanostructures nanostructures for advanced photonic devices. Microelectron. Eng. 2016, 153, 132–136. [CrossRef] 18. Okada, N.; Nojima, K.; Ishibashi, N.; Nagatoshi, K.; Itagaki, N.; Inomoto, R.; Motoyama, S.; Kobayashi, T.; Tadatomo, K. Formation of distinctive structures of GaN by inductively-coupled-plasma and reactive ion etching under optimized chemical etching conditions. AIP Adv. 2017, 7, 065111. [CrossRef] 19. Jeon, C.W.; Choi, H.W.; Dawson, M.D. Fabrication of Matrix-Addressable InGaN-Based Microdisplays of High Array Density. IEEE Photonics Technol. Lett. 2003, 15, 1516–1518. [CrossRef] 20. Wu, X.H.; Elsass, C.R.; Abare, A.; MacK, M.; Keller, S.; Petroff, P.M.; Denbaars, S.P.; Speck, J.S.; Rosner, S.J. Structural origin of V-defects and correlation with localized excitonic centers in InGaN/GaN multiple quantum wells. Appl. Phys. Lett. 1998, 72, 692–694. [CrossRef] 21. Lester, S.D.; Ponce, F.A.; Craford, M.G.; Steigerwald, D.A. High dislocation densities in high efficiency GaN-based light-emitting diodes. Appl. Phys. Lett. 1995, 66, 1249–1251. [CrossRef] 22. Kim, I.H.; Park, H.S.; Park, Y.J.; Kim, T. Formation of V-shaped pits in InGaN/GaN multiquantum wells and bulk InGaN films. Appl. Phys. Lett. 1998, 73, 1634–1636. [CrossRef] 23. Yang, G.F.; Chen, P.; Wu, Z.L.; Yu, Z.G.; Zhao, H.; Liu, B.; Hua, X.M.; Xie, Z.L.; Xiu, X.Q.; Han, P. Characteristics of GaN thin films by inductively coupled plasma etching with Cl2/BCl3 and Cl2/Ar. J. Mater. Sci. Mater. Electron. 2012, 23, 1224–1228. [CrossRef] 24. Cho, H.; Hahn, Y.B.; Hays, D.C.; Jung, K.B.; Donovan, S.M.; Abernathy, C.R.; Pearton, S.J.; Shul, R.J. Inductively Coupled Plasma Etching of III-Nitrides in Cl2/Xe, Cl2/Ar and Cl2/He. Mater. Res. Soc. Internet J. Nitride Semicond. Res. 1999, 4, 763–768. [CrossRef] 25. Rawal, D.; Arora, H.; Sehgal, B.; Muralidharan, R. Comparative study of GaN mesa etch characteristics in Cl2 based inductively coupled plasma with Ar and BCl3 as additive gases. J. Vacuum Sci. Technol. A 2014, 32, 031301. [CrossRef] 26. Kim, H.K.; Lin, H.; Ra, Y. Etching mechanism of a GaN/InGaN/GaN heterostructure in Cl2 and CH4 based inductively coupled plasmas. J. Vac. Sci. Technol. A 2004, 22, 598–601. [CrossRef] 27. Lai, Y.; Yeh, C.; Wang, J.; Wang, H.H.; Chen, C.; Hung, W. Sputtering and Etching of GaN Surfaces. J. Phys. Chem. B 2001, 105, 10029–10036. [CrossRef] 28. Ladroue, J.; Meritan, A.; Boufnichel, M.; Lefaucheux, P.; Ranson, P.; Dussart, R. Deep GaN etching by inductively coupled plasma and induced surface defects. J. Vac. Sci. Technol. A 2010, 28, 1226. [CrossRef] 29. Evgeny, Z.; Sergei, S.; Alan, G.; Wang, W.N.; Shreter, Y.G.; Tarkhin, D.V.; Bochkareva, N.I. ICP etching of III-nitride based laser structure with Cl2–Ar plasma assisted by Si coverplate material. J. Vac. Sci. Technol. A 2005, 23, 687–692. 30. Lee, H.; Harris, J. Iron nitride mask and reactive ion etching of GaN films. J. Electron. Mater. 1998, 27, 185–189. [CrossRef]