The doping of crystalline substrates such as silicon carbide, diamond and other such materials to provide such substrates with semi-conducting properties and/or optically active centres, is well established in the art. Dopant atoms can be introduced into the substrate by using ion implantation techniques.

European Patent Publication No. 0 209 257 discloses a method of producing a doped diamond which includes the steps of bombarding the diamond with ions of a suitable energy to produce a damaged layer of point defects in the form of vacancies and interstitials within the crystal lattice. the bombardment being carried out at a temperature sufficiently low to inhibit diffusion of the point defects, followed by annealing the damaged substrate. Dopant atoms are introduced into the damaged layer by ion implantation during. before or after the damage-creating bombardment. The dopant atom implantation also takes place at a temperature sufficiently low to inhibit diffusion of the point defects in the damaged layer.

European Patent Publication No. 0 573 312 discloses a method of producing a doped diamond which includes the steps of creating a damaged layer of point defects in the form of vacancies and interstitial atoms within the crystal lattice of the diamond using low dose ion implantation at low temperature, introducing dopant atoms into the damaged layer using low dose ion implantation at low temperature, rapidly annealing the product to reduce lattice damage and to cause dopant interstitial atoms to diffuse into lattice positions, and repeating the doping and rapid annealing steps until a doped diamond having a desired amount of dopant is produced.

European Patent Publication No. 0 750 058 discloses a method of doping a crystalline substrate such as diamond which includes the steps of providing the substrate, creating a damaged layer containing vacancies and interstitial atoms in the crystal lattice, implanting dopant atoms under conditions to create a second damaged layer separate from the first damaged layer and containing the dopant atoms, and causing dopant atoms in the second layer to diffuse out of that layer and into vacancies in the first layer and thereby occupy substitutional positions in that layer.

In all the methods described above, effective doping of crystalline substrates. more particularly diamond. is achieved. However, as effective as the doping is. vacancies remain and these are electronically and optically active and can interfere or neutralise the electronic properties of the doped material. It is desirable to reduce the density of the vacancies to a very low level and preferably to zero.

SUMMARY OF THE INVENTION According to the present invention. a method of producing a doped crystalline substrate having a crystal lattice includes the steps of providing a crystalline substrate having a crystal lattice. implanting dopant atoms into the substrate to create a doped layer, implanting ions to create a damaged layer in the substrate which is separate from the doped layer and contains vacancies and interstitial atoms of the crystalline substrate. and causing interstitial atoms from the damaged layer to diffuse out of that layer and into vacancies in the doped layer.

The invention involves producing a doped layer and thereafter reducing the damage created in the doped layer by creating a damaged layer containing interstitial atoms and causing the interstitial atoms from this layer to diffuse into the doped layer and occupy or mop up vacancies in the doped layer.

The damaged layer is created separate from the doped layer. It may be deeper than the doped layer, it may be shallower than the doped layer, or it may be adjacent the doped layer.

The interstitial atoms of the crystalline substrate can be caused to diffuse into the doped layer at the same time as the damaged layer is created, for example by carrying out the ion implantation to create the damaged layer at temperatures at which the interstitial atoms can diffuse. Alternatively, if the ion implantation to create the damaged layer is carried out at low temperatures at which interstitial atoms cannot diffuse, then diffusion of the interstitials can take place in a subsequent annealing step.

BRIEF DESCRIPTION OF THE DRAWINGS Figures 1 to 3 illustrate graphicaily results produced in experiments using the method of the invention.

DESCRIPTION OF EMBODIMENTS The dopant atoms may be any known in the art such as boron. oxygen, nitrogen, phosphorus, arsenic, or any other atom which imparts electrical or optical properties to the substrate. A range of energies will preferably be produced creating a relatively wide doped layer. The dose will also typically be a relatively low dose thereby limiting the amount of damage created in the doped layer. The low dose dopant implantation will typically be such as to create a density of vacancies in the first layer of less than 5 x 102"cl-3 i. e. less than 0,3 atomic percent.

The creation of the damaged layer, separate from the doped layer, is achieved using ions of selected size and energy. The size and energy may be selected such that the damaged layer is deeper than the doped layer. For example, if light ions such as hydrogen or helium are used, then a deeper damaged layer can be produced at relatively low energies. If heavier ions such as oxygen, nitrogen or carbon are used, then higher energies are necessary to ensure that the damaged layer is deeper than the doped layer and there is minimum damage to the doped layer. High energy ions, by virtue of their speed, create little damage in the shallow doped layer. It is only when ions are slowed down to lower energies as they reach deeper levels of the substrate that they ballistically dislodge atoms in the substrate to create interstitial atoms and vacancies. It is preferable that damage to the doped layer be kept to a minimum.

The damaged layer can also be shallower than the doped layer. In this form of the invention, heavier atoms will generally be used. for example nitrogen, carbon or oxygen, at relatively low energies. After the interstitials in the damaged layer have been caused to diffuse into the doped layer. the damaged layer may be removed to expose the doped layer.

The damaged layer can also be created adjacent. for example to the side of, the doped layer.

Interstitials created in the damaged layer are caused to diffuse into the doped layer where they find and enter vacancies. This movement of interstitials may be caused by suitably annealing the ion implanted crystalline substrate. The anneal is preferably a rapid anneal. Rapid anneal means that the annealing temperature is reached in a time of less than two minutes or even as fast as 20 seconds and preferably in a time of less than five seconds. The annealing temperature will typically be in the range of 50°C to 900°C.

The interstitials may also be caused to diffuse into the doped layer during the creation of the damaged layer. This will happen at temperatures at which the interstitial atoms can diffuse, for example if room temperature or higher temperature conditions are used during the creation of the damaged layer.

The crystalline substrate will typically be a large band gap hard crystalline material such as diamond. cubic boron nitride, or silicon carbide. The invention has particular application to the doping of diamond.

The implantation with dopant atoms and the damage-creating ions can be carried out simultaneously or sequentially.

In a first experiment, H+-ions were co-implanted with P+ ion energies and doses used are given in Table 1: TABLE 1 Ion P+ Energy (keV) Energy(keV)(cm-2) Dose (cm-2) 170 2.16 701.29x10131013 145 1013451.03x1013x 120 1. 47 x 1013 30 0. 86 x 1013 95 1.47 x 10" This was followed by an implantation of H'-ions at an energy of 170 keV to a total ion dose of 4 x 10'6 cm-2. According to the Monte-Carlo computer program that simulates the damage created by means of ion implantation, the resultant vacancy and interstitial damage distribution will be as shown in Figure 1. The smaller damage peak near to the surface is caused mainly by the P+ ions and also contains these atoms. The larger, deeper peak has been created by the H+ implantation. Thus, on annealing, some of the interstitial atoms escaping from the deep damage peak can diffuse towards the surface where they then assist to mop up vacancies in the shallower P+-implanted layer.

Two experiments were done on the same diamond after it had been repolished each time to present an undamaged sub-surface substrate. In both cases, the diamond was implanted with the same P+-ion doses (see Table 1). However, in one case H+ -ions were co-implanted, and in the other case not. All implantations were done while maintaining the substrate temperature near liquid nitrogen temperatures, and this was followed by rapid thermal heating to 500°C, at which temperature the diamond was held for 30 minutes. The results were vastly different, and are shown and compared in Figure 2.

As can be seen, the resistance of the layer which had H'-ions co-implanted, is an order of magnitude lower than the resistance of the layer which was not treated with this co-implantation. This clearly shows that there are more uncompensated phosphorus donors in the layer with the lower resistance.

As pointed out above, it is not necessary to implant the damaged layer at low temperatures. If higher temperatures are used at which the interstitial atoms can diffuse, then interstitial atoms will diffuse into the doped layer during implantation.

An experiment was carried out to illustrate the point. In this experiment boron ions were used to dope a diamond p-type. The doping implantation was done cold (liquid nitrogen temperature) followed by rapid annealing to 500°C. It is known that such a layer will have a high resistance owing to the large number of residual vacancies present. To minimise these vacancies, the diamond was implanted with H'-ions of 170 keV to generate the deeper lying damaged layer. During implantation the diamond was maintained at 500°C such that the interstitials created in the damaged layer could diffuse. Those interstitials that then reached the shallower doped layer could then mop up vacancies. The results can be seen in the accompanying Figure 3. After doping, the resistivity of the doped layer was 4,38 x 10'° 52-cm. As the implantation proceeded, the resistance kept dropping and the last measurement was 3,12 x 1Os Q-cm, a decrease of over 5 orders in magnitude.