Crosspolarisation characteristics of rectangular patch antennas

zyxwvutsrqponm zyxwvutsrqponm zyxwvutsrqp zyxwvutsrqponmlkjihg zyxwvutsrqponm over ordinary telephone lines'. Image Sequence Processing and Dynamic Scene Analysis in NATO AS1 series, 1983, pp. 314-336 3 ROSENFELD, A., and KAK, A. c.: 'Digital picture processing', vol. 2 (Academic Press, New York, 1982) 4 SHIOJAKI, A , : 'Edge extraction using entropy operator', Comput. Vision, Graphic. & Image Process., 1986, 36, pp. 1-9 CROSSPOLARISATION CHARACTERISTICS OF RECTANGULAR PATCH ANTENNAS Let the magnitude of the electric field corresponding to mode (m, n) be E,,,". If the patch is excited at the resonant frequency of the TM,, mode, the ratio of copolarisation to crosspolarisation in a particular direction is given by IE,,(B, 4)1/11 E,,(O, 4)l. Similarly, if the patch is excited at the resonant frequency of the TM,, mode, it is given by I E,,(& 4) I /IC E,,(@, 4) 1 . In the next Section, some numerical results showing the dependence of I E,,(& 4)I / I1E,,,,(O, 4) I on resonant frequency will be given for different substrate thicknesses and feed positions along the edges. For brevity, only the case corresponding to c, = 2.32 and a/h = 1.5 is presented. zyxwvutsr Indexing terms: Antennas, Microstrip, Antenna theory 1 Theoretical results describing the crosspolarisation characteristics of rectangular patch antennas are presented. Of particular interest is the result that the copolarisation radiation is not sensitive t o resonant frequency and substrate thickness while the crosspolarisation component increases with resonant frequency and/or substrate thickness. 50 x'ia = 0 4 zyxwvuts zyxwvuts Introduction ; Although the crosspolarisation characteristic is an important consideration in the design of microstrip antennas, few detailed theoretical results are available in the literature. Oberhart et al.' recently discussed the quantity 1 Ecopol I / I ExpolI of a rectangular patch excited in the TM,, mode and showed that it is dependent on the aspect ratio a/b. In this letter, we present detailed results showing the variation of this quantity for different feed positions, substrate thicknesses, and resonant frequencies. I I L.---A- I 0 2 4 6 resonant 8 1 0 0 frequency, GHz a US , 1 - 1 - . L - A 2 4 6 8 1 0 resonant frequency, G H z b zyxw 5ot Theory: Following Reference 1, we based our calculations on the cavity model. Let the substrate relative permittivity be E,, thickness t , and dimensions a and b, with a > b. The coaxial feed is located at (x', y') and is modelled by a current ribbon of width d. For a current of 1 A, the magnitudes of E , and E , in the far-zone consist of an infinite sum of modes: L A - 0 2 4 6 8 1 0 re s o nan t f re q u e n c y , GH z ,p;'i Fig. 1 x sin r 1 a 1 E,,, I / j E,, in broadside direction as a function of resonant frequency for different feed positions along side (1 ( t : , = 2.32. t = 0.159cm, a = 1.56, y'/b = 0) 6 I E,, 1 / 1 1 E,,,l in broadside direction as a function of resonant frequency for different feed positions along side h ( I , , = 2-32, t = 0.159cm, a = 1.56, x'/h = 0) c I E,, Ii 1 1E,, I in broadside direction as a function of resonant frequency for three substrate thicknesses ( E , = 2.32, LI = I.Sh, x ' / a = 0, y'/h = 0.2) zyxwvutsrq cos X - (Ed) mn)] .j , - 7 zyxwvutsrq (F)(F) cos 0: .10 G H ? 5 G H r ( k 2 - ki") IE, I = C sin 0 cos 0 1 E,,E,, sin [$(k, b$, - nn)] I%"=, x sin (Ed) [ & ( k , ~ l l/ ~mn)] .j o - ;m;!90 0 a -2 -20dB lE0l ' -20dB 0 0" -60" ' -20dB 1GHr -90"0 and,,,a 14th April 1988 -20dB 0 Fig. 2 a Copolarisation ( 1 E , , I ) patterns in = 0 plane for three resonant frequencies. Patterns are norrnalised with respect to 0 = 0 , f = IOGHz b Crosspolarisation ( 1 1E , 1 ) patterns in 4 = 0' plane for three resonant frequencies, Patterns are normalised wlth respect to H = O",f= IOGHZ E, = 2.32, t = 0.159cm, ai6 = 1.5, .Y','u = 0, y'/h = 0.2 is the effective loss tangent. ELECTRONICS LETTERS -20dB Vol. 24 No. 8 463 zyxwvutsrqponmlk zyxwvutsrqp zyxwvutsrqpo Numerical results: In our computation of he,, which appears in eqns. 1 and 2, the contribution due to surface waves is neglected. In the broadside direction (e = O O ) , the variation of I E,, I / 1 1E,, I as a function of resonant frequency for different feed positions along the edges of the patch are shown in Figs. l a and b. Fig. 2 shows the copolarisation 1 E,, I and crosspolarisation I E,, I patterns in the 4 = 0" plane for the case when thickness is fixed and resonant frequency varies. The case when resonant frequency is fixed and thickness is varied is shown in Fig. 3. Similar dependences are obtained for the 4 = 90" patterns. From the results, it is observed that: (i) I E , , I / I E,, -30" zyxwvutsrq zyxwvut zyxwvutsrqponm I decreases as the feed is moved along b, the 0" , , 7no t ~00795, 0 159.0 318cm \ 0 a 0 b -20dB lEOl ' -20dB 0 We have also studied the case when the antenna is excited at the resonant frequency of the TM,, mode. The dependences of I E,, I / I E,, I on resonant frequency and substrate thickness are qualitatively similar to I E,, I / I E,, I . However, I E,, I / 1 1E,, I is found to increase as the feed is moved along the shorter edge of the patch and decreases as the feed is moved along the longer edge. For brevity, these results are not shown here. In conclusion, theoretical results for the crosspolarisation characteristics of rectangular patch antennas have been presented. Of particular interest is the result that the copolarisation radiation changes only slightly with resonant frequency and substrate thickness while the crosspolarisation component increases with resonant frequency and/or substrate thickness. A c k n o w l e d g m e n t : This work is supported by the National Aeronautics and Space Administration, Lewis Research Centre, Cleveland, Ohio. \/A -20dB zyxwvutsrqp zyxwvutsrqponml zyxwvuts -20dB IPEmol 0 Fig. 3 a Copolarisation ( I E,, I ) patterns in 9 = 0" plane for three substrate thicknesses. Patterns are normalised with respect to H = OL, t = 0.318~111 b Crosspolarisation ( 11 E,, I ) patterns in 4 = 0 plane for three substrate thicknesses. Patterns are normalised with respect to 0 = o",t = 0.318cm E, = 2-32,fool= 6GHz, ajh = 1.5, x'ja = 0, y'lh = 0.2 HEAVY METAL GETTERING I N BURIED NITRIDE SILICON-ON-INSULATOR STRUCTURES Indexing terms: Semiconductor devices and materials, Ion implantation, Silicon Using Rutherford backscattering spectrometry the redistribution and gettering of implanted gold was investigated in silicon-on-insulator structures produced by high-dose nitrogen implantation. The gettering efficiency of a structure annealed at 1ooo"C containing a highly damaged silicon region near to the buried compound layer is more pronounced than that of a 1200°C-annealed structure with a lightly damaged silicon top layer. Gettering takes place at the interregion between silicon and silicon nitride. Introduction: Recently, the investigation of silicon-on-insulator (SOI) structures, formed by ion beam synthesis of buried dielectric layers of Si3N, or SiO,, has received considerable attention.' Despite considerable research activity in the field of buried oxide formation, the use of silicon nitride layers offers several advantages: lower doses for stoichiometric layers, diffusion inhibition against impurities,' higher radiation hardness and high minority carrier generation lifetime in epitaxial layers deposited on such SOI-stru~tures.~ We have reported on the latter aspect in previous paper^.^-^ It was concluded from electrical measurements that heavy metal gettering could be the cause of this effect.' In this letter results are reported concerning the gold redistribution in, and the gettering efficiency of, buried nitride SOIstructures. Gold was used for two reasons. It is a typical heavy 464 (ii) For frequencies above 1 GHz, 1 E,, I / 1 1E,, I decreases as the resonant frequency and/or the substrate thickness increases. The copolarisation radiation remains essentially unchanged while the crosspolarisation radiation increases as the resonant frequency and/or the substrate thickness increases. zyxwvutsrqpo On -60" shorter edge of the patch, and increases as the feed is moved along a, the longer edge of the patch. T. HUYNH K. F. LEE Department of Electrical Engineering University of Akron Akron, O H 44325, U S A 1st March 1988 R. Q. LEE N A S A Lewis Research Centre 21000 Brookpark Road Cleveland, OH 4 4 / 3 5 , U S A Reference 1 and LEE, R . Q. H.: 'New simple feed network for an array module of four microstrip elements', Electron. Lett., 1987, 23, (9), pp. 436437 OBERHART, M. L., LO, Y. T., metal impurity in silicon which diffuses very fast. On the other hand it can easily be detected by Rutherford backscattering spectrometry (RBS) owing to its high atomic mass. E x p e r i m e n t a l : I4N+ ions were implanted with an energy of 330keV into the central 2 x 2cm2 part of 50.8mm silicon wafers (n-type, 3-5Rcm, (100)). The implanted dose was 1.2 x 10I8cm-' leading to a nearly stoichiometric silicon nitride layer.6 Target heating to a temperature of 500°C was realised by the ion beam itself. Post-implantation annealing was done by furnace in dry nitrogen. Firstly, a 1000°C, 2 h anneal step was used to form substrates with a buried amorphous silicon nitride with relatively highly damaged silicon regions above and below it.' Previously, such substrates were used to test the carrier lifetime of epitaxial layers.3p5 Secondly, SOI-structures with silicon layers characterised by a low defect density throughout the film were produced by a 1200"C, 5 h anneal step. These substrates are further characterised by narrow Si/Si,N, interregions and a polycrystalline silicon nitride layer.6 Then, I9'Au+ ions were implanted with an energy of 5OkeV and a dose of 1 x 10I6cm-' serving as a diffusion source at the silicon surface. The gold redistribution was performed at 1O O O T for 15 min. RBS and channelling techniques with 1.7MeV, ,He+ ions were applied to assess the atomic profile of nitrogen and gold and the radiation damage of the silicon top layer, respectively. R e s u l t s : In Fig. 1 the random and channelled spectra taken after the full preparation are shown. In the channel region (CHR) lock265 the buried layer system is visible. The lowered yield in the CHR 1W190 of the random spectra is caused by ELECTRONICS LETTERS 14th April 1988 Vol. 24 No. 8