WO1987000297A1

WO1987000297A1 - Infra-red lenses and methods of fabrication

Info

Publication number: WO1987000297A1
Application number: PCT/GB1986/000389
Authority: WO
Inventors: David John Pedder
Original assignee: Plessey Overseas Limited
Priority date: 1985-07-04
Filing date: 1986-07-04
Publication date: 1987-01-15
Also published as: ZA864959B; GB2181567A; GB2181567B; EP0227784A1; JPS63500826A

Abstract

In the prior art, infra-red lenses have been fabricated from germanium and zinc selenide, and these have shown to possess useful optical characteristics whilst also being robust. However, the commercial cost of design and fabrication of such lenses is considerable. To reduce the design and fabrication costs, Fresnel-type lenses of certain polymer materials have been made. Such lenses generally have unacceptable optical characteristics and lack thermal and mechanical stability. To overcome these problems, an infra-red lens having a Fresnel-type structure fabricated from an infra-red transparent elemental or compound semiconductor material such as silicon is proposed. A method of fabrication of such lens includes a step of impressing a desired lens structure into the surface of a wafer of the infra-red transparent material by means of a light controlled etching technique.

Description

INFRA-RED LENSES AND METHODS OF FABRICATION

The present invention relates to lenses and more particularly to a lens structure for an infra-red lens and to its method of fabrication.

Three classes of optical materials are commonly employed in windows and lenses for thermal detection and imaging devices. Thus the semiconductors germanium and silicon, II-IV materials such as MgF₂, CaF₂ MgO, ZnS and ZnSe, and the various chalcogenide glasses all find some application within the wide spectrum of an infra-red system (reference J.M. Lloyd, Thermal Imaging Systems, Plenum Press 1975, Chapter 6). However to date, out of this considerable list of materials, only germanium (in polycystalline or single cyrstal form) and zinc selenide (chemical vapour deposited) find real commercial exploitation in the fabrication of precision infra-red lenses. Both these materials possess the high refractive indices required to minimise lens curvatures and hence lens thickness, together with low infra-red absorption coefficients and adequate long wavelength cut-offs required of lenses that are typically between about 2 and 10mm in thickness. The cost of optical quality germanium is relatively high, and, being a strategic material of rather low natural abundance its price has risen significantly over recent years. This, together with the high costs of optical design and lens fabrication, means that germanium infra-red lenses are of considerable cost^'. Zinc selenide finds less application than germanium being mechanically less robust, but is employed where its very wide transmission range, low absorption and freedom from thermal runaway (a problem in relatively narrow bandgap semiconductor materials such as germanium) are demanded, for example in high power carbon dioxide infra-red laser work.

Some attention has been paid recently to methods to reduce the cost of infra-red optics, but has generally resulted in an unacceptable optical performance. Thus for example certain polymer materials, such as polythene, which possess acceptable transmission in the 8-14^xm infra-red band when used at low thicknesses, have been examined as alternatives to germanium. In order to achieve adequate transmission and yet obtain the relatively high lens curvatures demanded for short focal ^" length lenses in the low index polymer materials, a Fresnel lens structure has been employed. While the polymer material costs are very low and mass production moulding techniques can be employed to produce lenses at low unit cost, the optical performance and thermal and mechanical stability of these polymer Fresnel lenses are generally inadequate for specialist applications. The low refractive indices of polymer materials demands steep zone angles and large zone numbers in a Fresnel lens structure, thus limiting lens resolution and increasing zone edge losses_* Polymer materials also possess very high thermal expansion coefficients, and a large variation of refractive index with temperature thus giving poor stability of optical properties over a range of temperatures. An object of the present invention is to provide a lens structure, lens material and method of lens fabrication which has the potential to produce a very significantly lower cost lens for infra-red applications whilst retaining useful lens performance. According to the present invention there is provided an infra-red lens having a Fresnel type lens structure fabricated from an infra-red transparent elemental or compound semiconductor material.

According to another aspect of the present invention there is provided a method of fabricating an infra-red lens having a Fresnel type lens structure the method including the step of impressing the desired lens structure into the surface of a wafer of the infra-red transparent elemental or compound semiconductor material by means of a light controlled etching technique. In a preferred method of fabrication the light controlled etching technique includes a photoelectrochemical etching process•

Other suitable light controlled etching techniques suitable for performing the fabrication include uv light or excimer laser controlled photochemical or "photolytic" etching of the material in a reactive gas atmosphere.

In a preferred embodiment of the present invention the infra-red transparent material is silicon. In other embodiments of the present invention the infra-red transparent material is germanium, gallium arsenide, gallium phosphide, indium phosphide and ternary or quaternary composed III-V semiconductors.

Preferably the infra-red transparent material is in single crystal form. This -enables the transfer of the desired Fresnel lens structure into the material by an etching process without the potential problems associated with preferential grain boundary etching and etching anisotropy related to the use of polycrystalline material.

High purity, single crystal silicon of well controlled impurity and defect content is readily available at reasonable cost, in the form of wafers up to 150mm diameter and typically 0.4 to 0.6mm thickness, as the starting material £or the silicon integrated circuit industry. The optical, thermal and mechanical properties of silicon are very significantly superior to those of the polymer materials. In particular the high refractive index of silicon (n * 3.42) means that shallow zone angles and modest zone numbers would be required in a Fresnel lens structure, thus enhancing the potential lens resolution and minimising zone edge effects. Silicon has not generally found application as an infra-red lens material to date because of its high absorption coefficient relative to germanium. However, if silicon were to be used in a Fresnel lens structure, then the low material thickness would ensure adequate infra-red transmission. Provided the silicon purity and resistivity were selected so that the absorption coefficient over the 8 to 14yu_n band did not exceed approximately 2cm~ , then absorption losses would be less than 10% for a typical silicon wafer thickness. Silicon materials that exceed this specification are available. Optical components in both silicon and germanium require single or multilayer antireflection/filter coatings to counteract the surface reflection losses inherent in the use of a high index material.

The basic physical and optical properties of germanium and silicon, but excluding the infra-red absorption characteristics which can vary considerably depending upon doping and impurity levels, are tabulated below

It can be seen in particular that silicon possesses a higher hardness, elastic moduli, melting point, specific heat, thermal conductivity and optical bandgap than germanium. On the other hand the density, thermal expansion coefficient, permittivity, refractive index, temperature coefficient of index and optical path are lower for silicon than for germanium. These property differences can be broadly understood in terms of the lower atomic mass of silicon and the stronger atomic bonding in the silicon lattice. The superior mechanical and thermal properties and the higher stablity of the optical properties of silicon are considered to be advantages in the present context. The infra-red refractive index of silicon, at 3.42, is still sufficiently high to provide low lens curvatures, as discussed later. The lower mass and stronger binding in silicon are also evident in the reduced infra-red transmission range, the long wavelength cut-off being defined by the lattice phonon frequencies, , which are, in the simplest case, given by an equation of the form: oCfF-M)* A.l where F is th atomic force constant and M is the atomic mass. The infra-red absorption properties of silicon and germanium in the infra-red transparent range, are significantly affected by free carrier absorption. The infra-red absorption properties of germanium have been discussed in detail (Wood et al, GEC Journal of Science and echnology, Vol. 48, No. 3, 1982, pp 141 to 151). The intrinsic carrier concentration, n^, in germanium at room temperature is relatively-high at 2.4 x lO^cπT^, as a consequence of the relatively narrow bandgap (0.67eV). This is equivalent to a resistivity of about 50 Λ.cra, and an intrinsic infra-red absorption coefficient at lO^uro of 0.03cm"¹. The intrinsic absorption coefficient is a sensitive function of temperature. The addition of, for example, gallium or antimony to dope the germanium either p or n-type, can be used to control and stabilise the absorption coefficient. The infra-red absorption crosssection for electrons is considerably smaller than that for holes lOjxm, and this means that deliberate n-type doping can _ie used to suppress the hole population and actually reduce the absorption coefficient, while at the same time providing a fixed carrier concentration (and hence a fixed absorption coefficient) over a range of temperature. This temperature range of stable absorption, known as the 'exhaustion range', increases with increasing dopant level, and a useful balance between absorption coefficient and exhaustion range is obtained at a resistivity of about 3_<A*.cm. This resistivity provides an absorption coefficient of 0.02cm"¹, remaining below 0.1cm"¹ at temperatures up to 77^βC (350K). The infra-red absorption properties of silicon have been widely studied, since the infra-red absorption spectrum is used to monitor doping and impurity levels in silicon wafers for the semiconductor industry. Commercially single crystal silicon for semiconductor use is grown by the Czokralski (CZ) or by the Float Zone (FZ) ,,- methods (T. Ravi, Imperfections and Impurities in

Semiconductor Silicon, John Wiley.& Sons, 1981). The former process provides silicon with resisti ities up to about 25*Λ_*.cm (carrier concentrations of 0.3 to 1 x 10 5cm~ ), _an<_ which also contains oxygen and carbon impurities. The inherent free carrier absorption coefficient for n-type CZ silicon at IOJΛΠI, for a carrier concentration of 1 x lO^cm^-3, is slight below 0.1cm"¹, with the phonon-related absorption coefficient contribution of about 0.4cm""¹. Interstitial oxygen in CZ silicon is of particular note since this gives rise to an absorption band at 9.1 um. The strength of this absorption band increases linearly with oxygen content, reaching a value of o = 4cm"¹ at the oxygen saturation content of silicon (1.8 x 10¹⁸cm~³). A typical CZ ingot will contain 6 to 8 x lO^cm"^ ₀f oxygen. equivalent to an absorption coefficient of 1.6cm"^*1. This value is below theβ ^s 2cm"¹ limit considered to be acceptable for a thin (0.5mm) silicon Fresnel lens. Thus, taking into account both the intrinsic phonon absorption and free carrier absorption characteristics over the 8-14t_,m band, lightly doped CZ silicon material proves acceptable for the proposed application.

As with germanium, the infra-red absorption cross- section in silicon for electrons is smaller than that for holes and thus n-type silicon is preferred for infra-red applications. This is fortunate, since n-type material is required to be able to conduct wafer surface profiling by the photoelectrochemical (PEC) or photolytic etching processes referred to later. Indeed at low carrier concentrations it has been found that the free carrier absorption coefficient for n-type CZ silicon decreases from 1 to 0.1cm^"1 as the carrier concentration falls from lO¹^^* to lO-^cm"³, whereas the coefficient for p-type CZ material plateaus at 1cm"¹ for carrier levels below 10¹⁶cm" . An infra-red transmission spectrum for a 4.8ιΛ».cm n-type CZ silicon wafer double polished to a thickness of 0.34mm is presented in Figure A.1. The average transmission level over the B-lA im band is about 95% of the theoretical figure, allowing for reflection losses. A shallow absorption bond at 9.1 u.m can be seen. The 9.1 urn oxygen absorption band can be suppressed by heat treatments which precipitate Siθ2 and hence reduce the interstitial oxygen level, although this is at the expense of producing dislocation loop defects that may affect the PEC etching process.

Higher resistivity silicon (up to 200Λcm, carrier concentration approximately 2 x 10¹³cm"³), which has very significantly reduced oxygen and carbon impurity content, may be produced by the FZ process. An n-type FZ silicon material with a carrier concentration of about^" 1 x lO^cm"³, is probably close to the ideal silicon material. As with germanium, the provision of a fixed dopant level provides a stable carrier concentration and hence a stable free carrier related absorption coefficient over a range of temperature. However, in the case of silicon, the significantly wider bandgap (l.leV) provides a^τ substantially greater exhaustion range than is the case for germanium (0.67eV). Thus an n-type FZ silicon material at a carrier concentration of 1 x 10¹5_CIn~3 should provide a free carrier absorption coefficient of about 0.1cm"¹ at 10^_m, which remains below 0.2cm"¹ for temperatures up to about 250^βC. The phonon contribution is about 0.4cm"¹ at 10 A.m, rising to about 0.6cm"¹ at 250°C. This is a very distinct advantage of silicon over germanium. It should be noted that the intrinsic carrier concentration for silicon at room temperature is 1.45 x lO-^cm"³; 3 orders below that of germanium. For this reason the initial suppression of the intrinsic absorption coefficient on low level n-type doping of germanium, is not found in the case of silicon, since the lowest practicable doping level in silicon will totally swamp the intrinsic carrier concentration and indeed at low dopant levels the phonon contribution to the absorption coefficient becomes dominant. Should uniformity σf- opant level prove crucial to the optical properties or PEC etching behaviour of the silicon, then neutron transmutation-doped (auto-doped) silicon could be employed.

The lens structure concerned is that of the Fresnel lens, in which a conventional lens is effectively divided into a series of thin zones and then reconstructed by displacing all the zones into one plane, on an appropriate thickness of supporting material. Several different designs of Fresnel lenses are possible in which the surfaces of each zone are plane or curved (spherical section and aspherical section curves), and with a variable number of zones. Fresnel zone surfaces may be defined on either or both surfaces of the lens material to produce the Fresnel equivalents of conventional plano-convex, plano-concave, convex, concave and meniscus lenses. The point of particular note in the context of the present invention is that such lenses are of low thickness.

The use of a material efficient lens structure, like the Fresnel lens, allows a thin lens to be achieved and thus allows materials to be employed that have higher infra-red absorption coefficients than would be required for more traditional lens thicknesses. Thus silicon, which can have an infra-red absorption coefficient that is largely below 2cm"¹ over the 8-14^».m band, provided the doping level and impurity content are suitably controlled, can be employed in a thin lens structure (below about 1mm thickness) without excessive loss of radiation through absorption. High Purity Germanium, with an infra-red absorption coefficient of about 0.03cm"¹, is essential for more conventional lens thicknesses. Thus the use of a Fresnel lens structure permits the use of a higher absorption material like silicon. Silicon is widely available in suitable wafer form with precisely controlled defect dopant and impurity content as the starting material for the semiconductor IC industry. Silicon is also very significantly cheaper than germanium.

What is required to complete this concept is a method of impressing the desired lens structure into the surface of the silicon wafer. Photoelectrochemical (PEC) etching. in which the depth of material removed by the etching process in any given location is controlled by the local visible or UV light intensity incident at that location, is a suitable technique. The PEC technique is applicable to n-type doped semiconductors, including Si, Ge, GaAs, InP and ternary and quaternary III-V compound semiconductors. The semiconductor, for example silicon, is immersed in an electrolyte and biase to a potential where the etch rate is proportional to the light intensity. The image of a photomask is projected onto the surface of the wafer to produce a spatial variation of light intensity to etch the desired shape. Such a mask may comprise for example very fine bars and spaces of^" varied mark-space ratios. The fine scale of the mask (submicron), combined with demagnification, slight defocus of the mask during projection onto the wafer or the deliberate exploitation of chromatic aberration with a white light source, effectively provides a smoothly varying light intensity at the wafer. The combination of the use of low cost silicon wafers, with a Fresnel lens structure and the PEC etching technique for defining the lenses provides a novel method for fabricating low cost infra-red lenses with a useful performance. Lenses for infra-red applications in the 8 to 14 um are band are produced in single and multielement form, using both spherical and aspherical surfaces. Aspheric and multielement lenses tend to be found where aperture ratios greater than about F/2 are required. Single element focussing lenses, of the meniscus or plano-convex form, are used for aperture ratios up to F/2. Popular lens diameters range from about 15 up to 50mm, with focal lengths from 25 up to 250mm. The Fresnel lens equivalents of all these single element lens types, meniscus, plano-convex and aspheric, can, in principal, be produced by PEC or photolytic etching, providing sufficient control of the illumination profile and the PEC process can be achieved. However a simple worked example, for a plano-convex lens, is sufficient to illustrate the principles involved.

To take for example a plano-convex lens which is required to give a focal length of 50mm J 2%, with a lens diameter of 50mm. Five such lenses could be obtained from a 150mm diameter silicon wafer, such as a wafer of lO l.cm double polished FZ material.

Using the standard thin lens formula

1/f = (n - 1) i^{1 + l} ) B.l ^rl ^r2 where f is the focal length, n is the refractive index, and ri and r2 are the radii of curvature of the lens surfaces, we have n _* 3.4177 for silicon at 10 m f « 50mm r2 ^s - ^βθ which gives r^ ^β 170.9mm for the radius of the convex surface. We now convert this plano-convex lens (minimum central thickness = 1.83mm) to a Fresnel lens, by dividing the lens into a series of slices of thickness t, and displacing these slices into a single plan on a supporting thickness of material, as illustrated in figure B.l. It may be shown that, for small values of t, the lens curvature, etched depth t and the outer radius of the nth zone, x_n, are related by the simple expression

we may calculate the Fresnel lens geometry for different etch depths, as tabulated below

Published work on the PEC etching of silicon has shown that etch depths of over 20 *m are possible with a smooth surface, free from pitting. A 20^m etch depth provides a reasonable number of zones, 91, with the narrowest zones still more than 10 wavelengths in width, the small step at each zone edge (^2 wavelengths) will minimise zone edge losses and obscuration for off-axis objects. With PEC etching the etch depth does not need te. remain constant across the lens. Variants where, for example, the depth increases towards the lens perimeter would be possible and would reduce zone numbers, the final PEC Fresnel lens thickness is now determined solely by the requirements for mechanical integrity. Thus, comparing the example lens, with a wafer thickness of say 0.6mm, with the original plano-convex lens (thickness 1.83mm), a material saving of at least a factor of 3 is obtained.

One particular advantage of using a high index material in a Fresnel lens structure arises from the (n-1) factor in equation B.l. This factor is 3 for germanium and 2.4 for silicon, compared with 0.5 for a typical polymer material. For a given Fresnel zone thickness or depth, t, a polymer lens would therefore require 5 to 6 times the number of zones of a silicon or germanium lens. Conversely, for a given number of zones, the polymer - IB -

material would require a similar increase on zone depth. In either case a significant increase in zone edge losses and obscuration results for the polymer material.

The tolerance on focal length in the example lens, at + 2%, is equivalent to a similar percentage tolerance on etch depth, i.e. +0.4 mm in 20 ι*.m. It is suggested that optical interference techniques, using, for example, a laser interferometer, could well be used to provide in-situ monitoring of the development of the Fresnel lens profile. The PEC process is, in effect, a 'non-contact optical machining' process which would allow such monitoring , through a beam splitter, by appropriate periodic interruption of the etch process. Control to better than one half wavelength of visible light (i.e. ._- 0.3 jam) , should be possible.

For a PEC etch that is simply linearly related to light intensity we may show that the illumination profile in the nth zone should follow an equation of the form

T(x) = (0.1 + ° 2 [ ^ - (n-l)t]] B.3 t 2R where x lies between x_n_ι and x_n (i.e. in the nth zone), t, n and R are as defined earlier, the maximum transmission is taken as unity, and the minimum mask transmission is taken as 0.1 The spherical aberration of the example lens is poor. but this is dominated by the choice of a plano-convex lens of large aperture. This aberration is largely related to lens form and aperture, and is relatively insensitive to material optical properties. A memiscus lens, with R2 * 1.5 Rχ_» would give a more acceptable spherical aberration and coma figure for apertures up to F/2. The chromatic aberrations on the other hand are controlled by the material's optical properties. Thus a lens fabricated in silicon will show chromatic aberration of a factor of 3 lower than an equivalent lens in germanium, since the refractive index of silicon varies more slowly with wavelength over the 8-11^ n band.

Lenses with curvature on both faces can in principal be fabricated by the PEC technique, provided a suitable front-to-back alignment technique can be established. A potentially suitable alignment technique would involve the etching of alignment holes right through the silicon slice prior to PEC etching of the two surfaces, using anisotropic etching processes based on KOH or ethylene dia ine-pyrocatechol-water (EDP) etchant solutions. These selective, anisotropic etching techniques, and appropriate masking techniques, are well established. Again the PEC process should also be suitable for defining aspheric Fresnel lens surfaces, with attendant gains in aberration control, providing the mask definition and etching processes are sufficiently controllable. The general principles of PEC etching of n-type semiconductors have been outlined above. The purpose of the description below is to briefly review some of the relevant work on the PEC etching of silicon and germanium, and to indicate other relevant literature.

The Anodic dissolution of n-type silicon under illumination has been studied by Sharpe and Lilley (J. Electrochem. Soc. Vol. 127 No. 9, September 1980, p. 1918). Their work was essentially concerned with developing a method for measuring carrier concentration profiles of epitaxial silicon structures by a combination of anodic dissolution and in-situ C-V measurements. This involved establishing an electrolyte and suitable operating potentials for (1) smooth and flat dissolution to depths in excess of 20 uun to allo'_*^ controlled profiling and (ii) interface conditions approximating a Schottky barrier so that meaningful C-V measurments could be made. In the present context the former criterion is of particular interest. The electrolyte system studied by Sharpe and Lilley was the NaF: H₂Sθ4:H₂0 system.

The anodic V-I characteristics showed a saturation current for both p and n-type silicon, the saturation current increasing with increasing NaF concentration and with illumination level for the n-type material. Currents below the saturation level gave a pitted etch, together with traces of a dark film. However samples etched in the saturation region were found to be considerably smoother, particuarly under strong illumination, with electrolytes with the higher acid content giving the smoothest etches with few pits or spikes on the surfaces. It was also apparent that electrolytes with a high sodium fluoride content gave a flatter etched surface in terms of gross uniformity, and an electrolyte comprising 1M NaF: 0.05M ^H2^So4 ^Was selected as a reasonable comprise. Uniform, deep etches (_ 10^um) were possible, provided the electrolyte was pulsated during the etching process.

The dissolution depth, w, in the PEC process is given by _{w =} M ( I.dt C.l

DNAF where M and D are the molecular weight and density of the semicondcutor, A is the interface area, N is the dissolution valcence, F is the Faraday and I is the current passed in the interval dt. Since in the case of a PEC etched Fresnel lens the wafer is uniformly doped throughout its thickness, this equation becomes ^w = ^M . Ξ . t C.2

D.N.F. A The dissolution valence for silicon was estimated to be 3.50 + 0.3 for standard cell conditions (cell potential 1.0V, 0.05M H₂S0₄: 1M NaF). Under suitably intense illumination, current densities of between 1 and lOmA/cm² can be readily obtained, corresponding to silicon dissolution rates of between 1.3 and 13 nm/hr. Since in principal the PEC etching process can be made fully automatic, as have PEC profilers for the assessment of epitaxial layers, this is considered a useful etch rate. The work by Sharpe and Lilley indicates that a PEC etching technology for fabricating Fresnel lenses in silicon is possible.

Similar work has been reported in the PEC etching of n-type germanium, although without details of the etched surface topography (Ashok K. Vijh, Electrochemistry of Metals and Semiconductors, Marcel Dekker Inc. New York, 1973, Chapter, p.64). The material employed was 3 Λcm n-type germanium, which was anodically dissolved in a 0.1N NaOH electrolyte under 0.4 to 0.7 i m visible illumination. The characteristic saturation current was observed for increasing anodic bias, this current being proportional to illumination level. A maximum current density of 10mA/cm² was achieved. It was reported that the dissolution valency of the germanium was a function of the electrolyte _

and varied between 2 and 4, indicating that both holes and electrons are involved in the overall dissolution process.

Some of the original and pioneering work on PEC etching of semiconductors was conducted by Ambridge,

Elliot and Faktor in 1973 (Journal Applied Electrochem. 3, 1973, pp 14 to 15). These authors were concerned with the electrochemical characterisation of n-type GaAs and, in addition to much useful experimental data, published a useful model of the PEC phenomena from which the photo- current density can be related to the illumination level, the depletion region width, the minority, carrier diffusion length, the surface potential barrier, and the optical absorption coefficient at the visible wavelength of interest. The closest agreement of theory and experiment, the widest range of voltages over which a well defined saturation current plateau was observed, and the best resulting etched surface finish, was obtained at the lowest dopant levels. These observations, which should also apply for the PEC etching of silicon, are encouraging since low dopant levels, correspon ing to a low infra-red absorption coefficient, are required for the present application. An example of experimental data from the Ambridge work, showing excellent linearity of saturation current with illumination level, is presented in Figure C.l.

Claims

CLAIMS :

1. An infra-red lens having a Fresnel type lens structure fabricated from an infra-red transparent elemental or compound semiconductor material.

2. An infra-red lens according to claim^; 1, wherein the infra-red transparent material comprises:silicon.

3. An infra-red lens according to claim 1, wherein the infra-red transparent material comprises one or more of the elements/compounds germanium, gallium arsenide, gallium phosphide, indium phosphide and ternary or quarternary composed III-V semiconductors.

4. An infra-red lens according to claim 1, claim 2, or claim 3, wherein at least part of the infra-red transparent material is in single crystal form.

5. A method of fabricating an infra-red lens according to any one of the preceding claims, wherein the method includes impressing a desired lens structure into the surface of a wafer of the infra-red transparent elemental or compound semiconductor material by means of a light controlled etching technique.

6. A method according to claim 5, wherein the light controlled etching technique includes a photoelectrochemical etching process.

7. A method according to claim 5, wherein the light controlled etching technique includes uv light or excimer laser controlled photochemical or "photolytic" etching of the infra-red transparent material in a reactive gas atmosphere.