5 zyxwvutsrqponmlkjihgfe 14 IEEE ELECTRON DEVICE LETTERS, VOL. EDL-5, KO. zyxw 12, DECEMBER 1984 Schottky gates,” zyxwvutsrqpon IEEE Electron Device Lett., vol. EDL-3, pp. [3] R.F.Leheny, R. E. Nahory, M. A. Pollack, and A. A. Ballman, 64-66, 1982. “Integrated Xno,53GQ 47A~ pin-FET photoreceiver,” Electron. Lett., zyxwvut [8] vol. 16, p. 153, 1980. [4] T. Y. Chang, R. F. Leheny, R. E. Nahory, E. Silberg, A. A. Ballman, E. A. Caridi, and C. J. Harrold, “Junction field-effect [9] transistors using Ino53G%.47As material grown by molecular beam epitaxy,” IEEE Electron Device Lett., vol. EDL-3, pp. 56-57, 1982. by MBE,” IEEE Electron Device Lett., vol. EDL-4, pp. 252-254, 1983. heterostructure GQ.4,1no 53A~ MESFETs with submicron gates,” IEEE Eiectron Device Lett., vol. EDL-1, pp, 174-176, 1980. [7] C. Y. Chen, A. Y. Cho, K. Alavi, and P. A. Garbinski, “Short [5] Y. G. Chai and R. Yeats, “Ino.s3Ga,-.4~As submicrometer FETs grown [IO] [6] J. Barnard, H. Ohno, C. E. C.Wood, and L. F. Eastman, “Double [ll] channel G~.4.11no.s3As/Alo,481no 5ZA~ selectively doped field-effect transistors,” IEEE Electron Device Lett., vol. EDL-3, pp. S. Bandy, C. Nishimoto, S. Hyder, and C. Hooper, “Saturation velocity determination for Ino.53Gao,47As field-effect transistors,” Appl. zyxwvu Phys. Lett., vol. 38, pp. 817-819, 1981. P. D. Gardner, S. Y. Narayan, Y.-H. Yun, S. Colvin, J. Paczkowski, B. Dornan, and R. E. Askew, “G%,471no,53As deep depletion and inversion mode MISFETs.” in Proc. Inst. Phys. Cony., ser. no. 65, 1983, ch. 5? pp. 399-406. H. H. Wieder, J. L. Veteran, A. R. Clawson,and D. P.Mullin, “Accumulation mode Ga&no.5sAs insulated gate field-effect tran- sistors,” Appl. Phys. Lett., vol. 43, pp. 287-289, 1983. H. Morkoq, S. G. Bandy, R. Sankaran, G. A. Antypas, and R. L. Bell, “A study of high-speed normally off and normally on A~O,~G%,~AS heterojunction gate GaAs FETs (HJFET),” IEEE Trans. Electron Devices, vol. ED-25, pp. 619-627, 1978. 205-208, 1982. Electrical En point Detection of VLSI Contact las tchin Abstract-A precise, in situ monitor of plasmacontactetching is described. This method of electrical endpoint detection provides a de electrical current which is proportional to the total open contacts and gives more than an order of magnitude improvement in signal-to-noise ratio over optical techniques. Endpoint detection is clearly deserved for total contact areas of less than 0.3 percent of a 3-in wafer. Using this method, contact resistances and leakage currents for both p I-n and n --p junctions,arecompatibleto that ofwet-etched or dry-etched contacts. I. INTRODUCTION LASMA ETCHING [1]-[3] is now widely used in the semiconductor industry to meet the demands for high resolution and tight control in the fabrication of VLSI circuits.Preciseendpoint detection is of great importance in controlling these processes, especially for the etching of contact holes. There are a number of endpoint detection methods which include the use of optical emission spectros- copy [4], mass spectroscopy [6], [7], laser interferometry [8], [9], and plasma impedance changes [IO], zyxwvuts [ 1 11, etc. [ 121. All of these methods have some drawbacks, Optical emission spectroscopy relies on the emission intensity of reactants or Manuscript received September 24, 1984. This work was supported in part by DARPA under Contract MD A903-79-C-0257 and by SRC under Contract 84-02-047. Stanford, CA 94305. The authors are with Integrated Circuit Laboratory, Stanford University, by-products in the plasma and suffers from small signal levels. Mass spectroscopy analyzes the residual gases and often misses active species in the glow discharge region which have short life times. Laser interferometry observes only a specific area on the wafer being etched and can enhance or affect the etching rate in the observed area [ 131, The plasma impedance method relies on the change in the plasma chemistry at the completion of etching and suffers from small signal level and sensitivity to the film type, pressure, and other processing parameters. For device fabrication,most etching steps involve material removal in large unmasked areas of the wafer, such as for polysilicon and aluminum etching. These steps have no problems for endpoint detection. Contact etching, however, involves oxide removal in only a small percentage of the wafer area (0.3 to 3 percent) and as such the endpoint can be extremely hard to detect by the bulk plasma monitoring techniques. In this letter, we describe an electrical endpoint detection technique for monitoring the etching of Si02 in contact vias. By eliminating alternative current paths, we monitor the induced dc current through the plasma as the oxide etching proceeds. The magnitude of this current will change signifi- cantly when contacts are opened. This electrical current has a large signal-to-noise ratio even when the total contact area is 0741-3106/84/1200-0514$01.00 zyxw 0 1984 IEEE