9th International Conference on Nitride Semiconductors (ICNS-9) (ICNS 9)

Advances in Large Batch Gallium Nitride Etching Process for LED P Production d ti Solutions S l ti
Zhong g Ren* Ren*, , Stephanie p Baclet Baclet, , Andrew Newton Newton, , Andy y Goodyear Goodyear, y , Mark Dineen and Robert Gunn
Oxford Instruments Plasma Technology, Technology Yatton, Yatton Bristol, Bristol UK

1 Introduction 1. I t d ti
Gallium Galli m Nitride (GaN) is a wide ide band gap semicond semiconductor ctor for electronics and optoelectronics applications applications, i l di violet including i l t laser l diodes, di d detectors, d t t light-emitting li ht itti diodes di d (LEDs), (LED ) high hi h electron l t mobility bilit transistors t i t (HEMT) and bio-sensors. In particular, with high brightness LEDs (HB-LED) extensively used in large screen display p y technology gy and solid-state lighting, g g there is a drive for the fabrication roadmap p to give g dramatic acceleration in throughput. g p GaN etching is a key part of the HB-LED HB LED production process process. Plasma dry etching techniques are a mainstream solution to etch GaN epi layers in current manufacture. manufacture We report here the advances in large batch technique for GaN etching process, process based on leading-edge leading edge techniques which provides excellent uniformity niformit o over er large areas areas.

2 2 Based on a new technique using an active spacer 2.2

An special designed active spacer which sits between a ICP source and a main chamber. This is employed giving further control over plasma profile. profile Experimental results indicate that etch rate raises with increase of both RF and ICP powers and uniformity is successfully controlled at the range of 4~5%, 4 5% based on a variety of f chemistries, h i ti as shown h in i Figure Fi 8. Th That t means etch t h rate t i is able bl t to b be i independently d d tl adjusted while maintaining a good uniformity. As a result, we have achieved an etch rate of 160 nm/min with selectivity y of 0.9:1 to PR mask on a 27 2 inch batch process.

Uniformity profiles
25% 15%
Unifo ormit ty

BCl3/Cl2/Ar BCl3/Cl2 SiCl4/Cl2/Ar

2 E 2. Experimental i l and d di discussion i

21B 2.1 Based d on a high high density d ity ICP source
Our etching g experiments p were carried out in an OIPT Plasmalab133-ICP380 ( (inductively y coupled p plasma) p ) tool using BCl3 and Cl2 as process gases. gases The lower electrode is 330 mm in diameter and its temperature is able to be actively controlled controlled. A batch of 27 wafers was placed on a silicon carbide (SiC) carrier carrier-plate, plate as shown h in i Figure Fi 1. Th These wafers f were automatically t ti ll transferred t f d into i t process chamber h b to t perform f the th etch t h recipe. p A pre-etched pre etched wafer was cleaved and checked by field field-emission emission SEM (FE-SEM) (FE SEM) tools tools, as shown in Figure 2. A a result, As lt an etch t h rate t of f 120 nm/min / i with ith a good d selectivity l ti it of f 0.8:1 0 8 1 to t photoresist h t i t (PR) mask k has h been b achieved. SEM images reveal a near vertical profile and smooth etched surface, as shown in Figure 3.
Figure 8 Uniformity profiles of 27 2 inch GaN batch processes with a variety of chemistries, chemistries including BCl3/Cl2, BCl3/Cl2/Ar and SiCl4/Cl2/Ar, /Ar based on system133-ICP380 system133 ICP380 with an active spacer.

5% -5% -15% 15% -25% 0 20 40 60 80 100 120 140 160

Distance to the centre of electrode (mm)

2 3 Based on a new tool using VHF plasma source 2.3

A GaN G N batch b t h etching t hi process of f 55 x 2 i inch h wafer f h has b been d developed l d in i an OIPT NGP1000 tool which is shown in Figures 9 and 10. This new tool uses a very-high-frequency y g q y (VHF) ( ) plasma p source. By y delivering g power p in the VHF band this source provides a high plasma density comparable with conventional ICP sources and maintains the benefits of high etch rates but in a low damage environment environment. Operating at VHF requires careful design of the plasma source hardware and chamber components to minimise undesirable electromagnetic effects, effects asymmetries and nonnon uniformities that degrade the process across a batch of wafers wafers.

GaN Mask

Figure 9 (left), photograph of NGP1000 system.

Figure 1 27x2 inch GaN wafer were placed on a SiC carrier-plate p p for p process.

Figure 2 SEM image of pre pre-etched etched GaN samples with PR masks.

Figure 3 SEM images show a near vertical profile fil and d smooth th surface f on processed d samples Insert: a top view of LED Cells. samples. Cells

Figure gu e 10 0( (right), g t), 55x2 55 inch c Ga GaN wafer were placed on a SiC carrier-plate p for p process.

At the 13.56 MHz frequency q y used, , it is g generally y considered that the ion energy gy will mainly y depend p on the potential difference between the plasma and the lower electrode, which in turn mainly comprises of the easyeasy to-measure to measure DC self-bias self bias voltage. voltage After etching, etching the test GaN wafers were cleaved and examined by FE FE-SEM SEM in order to measure step height and s surface rface q quality. alit Etch depths were ere meas measured red across the whole hole test batch gi giving ing the f full ll batch uniformity. niformit A di discussion i was performed f d to t analyze l the th effect ff t of f key k processing i parameters t on uniformity. if it These Th parameters were: Process p pressure, , as shown in Figure g 4 RF power, as shown in Figure 5 ICP power power, as shown in Figure 6 Gas ratio ratio, as shown in Figure 7

Effect of variable process p parameters p on uniformity y has been studied in order to get g an accurate control of uniformity y across platen. p Influence of p parameters like p pressure, , VHF power and flow have been observed and the results from these parameters is described in Figure 11. This work has demonstrated that pressure and flow are key parameters for uniformity control across platen platen. Influence Infl ence on etch rate and selecti selectivity it has also been obser observed ed and a balance h b has been f found dt to allow ll th the b best t process performances. f A An etch t h rate t of f 140 nm/min / i with ith selectivity of 1:1 to PR masks and uniformity of <5% have been achieved.

Uniformity change with pressure

Uniformit ty(+/-% %)

Uniformity change with flow

Un niform mity (+/-%) ( )
6 5 4 3 2 1 0
30 45

Uniformity change with VHF power

5 4 3 2 1 0
400 500

Uniform mity (+/-% %)

Initial experiments e periments were ere performed without itho t an acti active e spacer spacer; f further rther processes were ere then performed with ith the spacer.

8 7 6 5 4 3 2 1 0
4.4 10

Pressure increase

Flow increase

VHF Power increase

(a)

(b)

(c)

Fi Figure 11 Variation V i ti of f uniformity if it with ith pressure ( (a), ) flow fl (b) and d VHF power (c). ( )

3 Conclusions 3.
Large a ge ba batch c e etching c g techniques ec ques have a e been bee further u e developed de e oped for o LED manufacture, a u ac u e, based on plasma etching techniques techniques. An etch rate of 160 nm/min with selectivity of 0 0.9:1 9:1 to PR mask and uniformity of <5% has been successfully achieved on a 27 2 inch GaN batch process process, using an active spacer. Th The process performed f d an excellent ll t repeatability t bilit i in LED chip hi f fabrication. b i ti Based on a new tool of NGP1000, we have also achieved an etch rate of 140 nm/min with selectivity y of 1:1 to PR masks and uniformity y of <5% on a 55 2 inch GaN batch ba c p process. ocess This se enables ab es throughput oug pu to o be remarkably e a ab y and a d stably s ab y increased. c eased

Figure 4 Variation of uniformity with process pressures under a constant BCl3/Cl2 chemical ratio of 1:3: at RF power of 400W and ICP power p of 400W.

Figure 5 Variation Fi V i ti of f uniformity if it with ith RF power at t a constant t t process pressure of 1.5mtorr: 1 5mtorr: at ICP power of 400W under a constant BCl3/Cl2 chemical ratio of 1:6 1:6, and at ICP power of 300W under a constant BCl3/Cl2 chemical ratio of 3:5.

Acknowledgement:
Thi work This k was supported t db by FP7 project j t SMASH contract t t CP CP-IP IP 228999-2 228999 2.

Figure 6 Variation of uniformity with ICP power at a constant process pressure of 4mtorr: at RF power of 200W under a constant t t BCl3/Cl2 chemical h i l ratio ti of f 1:6; 16

Figure g 7 Variation of uniformity y and etch rate with the ratio of BCl3 to Cl2: at a constant pressure of 1.5mtorr, RF power of 400W and ICP power of 400W.

*E-mail: E-mail: zhong.ren@oxinst.com zhong ren@oxinst com

www oxford instruments com www.oxford-instruments.com