The present application relates to a support for an optic and/or electronic chip, and in particular to a support by which the stress resulting from the intimate attachment of two dissimilar materials can be reduced.

Problems with stress are fundamental in many areas of mechanical engineering and arise when different materials are bonded together. This is a particularly concerning problem for many semiconductor devices which are of necessity bonded onto a substrate of a different material with different expansion coefficients, different thermal conductivity and different elastic moduli. It is widely known that sufficient stress applied to semiconductor devices can alter the electronic characteristics, electro-optic characteristics and is a significant source of reliability problems.

Figure 1 shows the result of bonding between two materials A &B where material A has a lower expansion coefficient than B. Heating the couple produces well known bending to minimise the total stress energy produced. There will be a shear stress at the interface.

One solution to the general problem of minimising stress for planar interfaces is to choose materials that are closely matched in expansion or have a bonding interface that contains a material which can plastically deform between the two materials. Lead-tin solder is such a material and is widely employed not only because it will wet to plated surfaces and provide good electrical contact, but its low melting point, low modulus and low elastic yield point allow it to absorb bonding stresses over a wide range of temperatures. If the interface bond is achieved by soldering for example, the shear stress may be sufficient to initiate creep in the solder. In this case the stress energy will be absorbed by producing irreversible structural changes in the solder. Another solution to this problem is to use a polymer or epoxy bonding fillet between the two bonded surfaces. A consequence of all of these is that some part of the bond material or structure undergoes plastic deformation in order to absorb the stress. As the temperature of the assembly changes, the material is cycled through continued plastic deformation. For most metals such as solders this means that in the plastic

regime the microstructure of the material changes. This is well known and the changes that arise are generally generation and movement of dislocations, growth and movement of grain boundaries and generation and movement of vacancies and interstitials. Generation of these entities leads to impaired mechanical and electrical properties and eventually to mechanical failure such as fracture. In addition, plastic deformation will also give rise to irrecoverable microscopic movement between the relevant materials. This is a major problem for some application such as micro-optical components where the above constraints are required and in addition the positions of the bonded components are extremely critical in order to satisfy the coupling of optical beams from one to another. For this situation, any plastic deformation will produce misplacement and misalignment of components such as semiconductor lasers or other electro-optic components and eventual loss or impairment of the electro- optical performance.

It is an aim of the present invention to provide a technique by which the above- discussed problems can be at least partially solved.

The present invention provides a support for an optic and/or electronic chip as defined in any of claims 1, 2, 15, 20, 24 and a method of controlling the movement of an optic and/or electronic chip according to claim 19. Embodiments of the invention are defined in the dependent claims.

Embodiments of the invention are described in detail hereunder, by way of example only, with reference to the accompanying drawings.

Embodiments of the invention solve the problem of limiting any distortion of material to the elastic regime and to spread the shear over a distance much larger than the conventional bonding interface. Figure 2 shows how this may be achieved by designing an interface in the form of a comb of two or more materials such that the properties of the comb structure are in some way combined to give a set of properties which can be engineered to meet several requirements. These would include; limits on the net strain without entering the plastic regime of the materials, limits on the stress which is transmitted to the sensitive component of the structure, limits on the resulting

electrical and/or thermal properties, and limits on the mechanical properties such as shear strength. The options for solving this problem by design are:-

1. The materials of the comb can be either one or other or both of the two materials (A, B) being bonded, or could be one of the materials (A or B) and air gaps, or could be one of the materials (A or B) and a filler of another material (C, or D) or could be two materials which are different from the two materials being joined, (C and D). (Figure 3)

2. The ratio of the widths of the alternating comb materials can be constant through the structure (Fig 3). The widths of the two materials comprising the comb can also be different; material C may have smaller finger width than material D (air) to give more compliance.

3. The separation between comb fingers and the width of the fingers through the structure may change. (Figure 4) This is a natural variation that would allow for greater strain between the two materials at the periphery of the bond compared to the centre region.

4. The length of the fingers may vary from centre to periphery to allow the nett compliance to be modified over the bond. This allows some freedom to make the stress evenly distributed across the structure. (Figure 5)

5. The fingers comprising the comb structure are shown aligned in one direction. This results in stress minimisation and modification in one direction perpendicular to the comb section. This may be sufficient for structures which have length considerably larger than the width. For structures where the dimensions of the interface are comparable in the two perpendicular directions, it may be advantageous to design a two dimensional comb structure such as an array of pins to give isotropic stress compensation. The arrangement of the pins may be subject to the considerations raised above for the one-dimensional cases. (Figure 6)

A practical example of this solution is in the attachment of lithium niobate optical waveguide devices to a substrate whereby the waveguide (located at the top of the

chip 1) is optically coupled to a semiconductor laser by means of micro-lenses. (Figure 7) The positional tolerance of the combined structure is less than one micron. The lithium niobate and laser devices are stress sensitive and have to operate over a range of temperatures. Since the lithium niobate device is several centimetres long, significant strain occurs between the lithium niobate and the mounting material which leads to continuous and unpredictable movement between the laser and lithium niobate over time and over temperature. Figure 8 shows the mechanical solution involving a combed structure which allows greater compliance at the end of the device to give low stress and mechanical positional stability for purposes of optical aligmnent to the laser.

Fabrication of a combed (interleaved) structure can be achieved by several means. The method chosen depends on the materials comprising the materials A & B and the comb materials C & D. In addition the dimension of the structure may be some tens of microns for example for bonding small chips or may be some millimeters or centimeters for larger mechanical structures. For these, precision mechanical milling, photolithographic metal etching, chemical etching, moulding, are all possible.

Due to the complexity of the problem of minimising stress for different geometrical and material structures it may be required to find the best solution by a design study using for example finite element analysis to determine the distortion, strain and stress values that result from specific designs and to verify that each material is operating within its elastic limits.