Inductively coupled plasma etching of deep photonic crystal holes in InP using Cl/sub 2/

Abstract

To]", R. Notzel P.J. van Veldhoven"' and H.W.M. Saleminka.b7" b We have investigated ICP-etching of deep photonic crystal holes in InP using solely Clz as supplied etching gas. The influence of process parameters on hole geometry is discussed and optical test results are reported.InP based two-dimensional (2D) deeply etched holetype photonic crystals are likely to be present in many of the hture optical devices involving the telecommunication wavelength of 1550 nm. For operation around 1550 nm a triangular lattice of etched holes with a lattice constant a of -400 nm and a diameter d of-250 nm can be used. To fulfill the requirement of low optical loss, the holes should be etched through an InP/lnGaAsP/InP planar waveguide structure with a depth of -2.5 pm and with smooth and vertical sidewalls. (1) Deep photonic crystal holes in InP have been successfully etched using chemically assisted ion beam etching and electron cyclotron resonance reactive ion etching, all based on Cl?-chemistry. (2,3,4) A versatile etching technique for largescale fabrication is inductively coupled plasma (ICP) etching, however, so far only one ICP process has been reported for deep hole etching, based on SIC4. ( 5 ) ICP etching of InP using Clz as main etch gas has been used in other photonic crystal applications such as pillars and shallow holes for membranes. (6, 7) In this work we investigate ICP-etching for fabrication of deeply etched hole-type photonic crystals in the InP-based material system using gas flow consisting solely of Ch.All experiments were performed on (100) n-type Sndoped InP substrates with a size of approximately 8x8 mm'. For optical experiments a waveguide structure consisting of 500 nm InGaAsP (h=1.25 pm) with a 500 nm InP upper cladding on top, grown lattice matched to the KnP by metal organic vapour phase epitaxy, was used. The photonic crystal pattern is defined into a layer of ZEP520A (positive e-beam resist) with e-beam lithography. This pattern is then transferred into a 400 nm thick, PECVD deposited Si,N,-masking layer with a CHF3-based RIE process. The ICP etch experiments were carried out in a load-locked Oxford Plasmalab IO0 system. Since the main etch-product, Inch, is not enough volatile at room temperature the etching was performed at elevated temperature. Sample temperature control in ICP etching can be difficult due to significant heating by the ion bombardment. (8) To keep the sample temperature as close to the preset value as possible the 0-7&03-8891-7105/$20.00 02005 lEEE following measures were taken. The samples were glued with heat conducting paste onto a 4 in. silicon carrier wafer. The table temperature was regulated by resistive heating in combination with short sequenced processing (etch steps of 30 s). The stage temperature was measured with a thermocouple. The ion energy was controlled by the negative DC-bias voltage, which is induced by capacitive coupling of additional rf-power to the plasma. Hereby, the relatively low plasma potential is neglected. The chlorine flow was kept const...