DOPED DIAMOND
BACKGROUND OF THE INVENTION
This invention relates to doped diamond
In order to create electronic devices in a material like silicon, the material needs to be doped in order to create the n- and/or p-type electrical characteristics This usually means that suitable impurity atoms such as B, P, or As have to be introduced to replace lattice atoms, e g Si atoms in silicon Whether it is possible to introduce such atoms, is subject to various considerations The reason why these defects are required, is that they create electronic levels in the band gap of the semi-conductor from where carriers can be excited into either the conduction band or the valence band If the impurity atom acts as a donor (typical for P dopant atoms), it creates an electronic level, containing an electron, near to the conduction band (CB) Such an electron can then be thermally excited or activated into the CB where it takes part in electrical conduction Such a material is called n-type If on the other hand, the impurity atom acts as an acceptor (typical for B dopant atoms), it creates an electronic level, which, in this case, does not contain an electron, near the valence band (VB) An electron can then be thermally activated from the VB into this level This process removes an electron from the VB This deficiency in the VB then acts as a positive carrier, called a hole, which can, like the electrons in the CB, conduct electricity Such a material is called a p-type material Mathematically, the holes can be treated as current carriers which are then activated from the "empty" acceptor level into the VB In other words, activation of a hole to a higher energy, means that it moves down in energy relative to the electron energy levels The donor and acceptor states are
defects Flaws can then be classified as all the defects that create extra electronic energy levels in the band gap of a semi-conducting material
Diamond promises to be a material which may have exotic applications which are not possible when using the standard semi-conducting materials available today Since the advent of plasma-assisted chemical vapour deposition (PACVD) this dream seems nearer than ever One of the main problems is to introduce shallow dopant flaws into the diamond structure Only two relatively shallow dopant atoms can be introduced into diamond during growth under high pressure conditions (at which diamond is the low energy form compared to graphite) These are nitrogen (N) and boron (B), and their introduction into the lattice under these growth conditions, show that their defect energies within the lattice must be lower than their energies would have been outside of the crystal Nitrogen forms donor states situated at ~ 1 ,7 eV below the CB This is extremely large (larger than the band gap of silicon), and accordingly renders this form of n-type diamond useless for most, if not all electronic applications Boron forms acceptor states situated at 0,37 eV above the VB This is much shallower, and such diamonds and diamond layers have been used to generate electronic devices even transistors However, even this activation energy is relatively high, and one would prefer shallower acceptor states Theoretical calculations invariably show that other dopant atoms like Al, P, As, O, F etc , should not form substitutional dopant flaw states, owing to the fact that the configurational energies of the states (also called formation energies), when these atoms are (theoretically) forced to occupy substitutional sites, are extremely high In fact, allowing for relaxation of the matrix atoms during theoretical modelling usually leads to a situation where the forced, theoretically created dopant state relaxes to a much lower configurational energy state, and even the latter state will, usually, be difficult to generate during growth As expected, such a configurational relaxation also affects the electron energy levels in the band gap They move away from the bands towards the middle of the gap Such deep states are, like the N-donor, not of much use for electronic
applications In fact, the nitrogen donor state can itself be considered as a relaxed state If the nitrogen could have been bonded within the lattice site to maintain its tetrahedral symmetry its donor level would probably be relatively shallow However, the nitrogen atom relaxes out of the symmetrical position, to form a lower energy state with a deeper donor level It is more than possible that the nitrogen atom also needs movements in the surrounding host atoms when it relaxes
SUMMARY OF THE INVENTION
According to a first aspect of the invention there is provided a method of forming flaws with high configurational energies in a diamond crystal lattice which includes the steps of
(1) subjecting the diamond crystal lattice to ion implantation to replace some of the diamond crystal lattice atoms with dopant atoms to produce a doped diamond, and
(2) annealing the doped diamond at a temperature which maximises (by "quenching" in) the density of high energy flaws which form shallow (smaller than 0,37 eV) dopant states
The annealing temperature must be low enough to maximise the density of high energy flaws which form shallow dopant states i e such that the host atoms of the diamond crystal lattice surrounding the inserted dopant atoms cannot relax to, in this way, reduce the configurational energy of the flaw
The annealing temperature will vary according to the nature of the dopant atoms which are implanted into the diamond crystal lattice
The annealing temperature is preferably below 600°C, more preferably about 300°C to 500°C inclusive
Preferably, steps (1 ) and (2) of the method of the invention are repeated one or more times, and more preferably, the diamond crystal lattice is subjected to multiple cold-implantation-rapid-annealing steps In this regard, step (1 ) is preferably carried out at a temperature which is around the temperature of liquid nitrogen
The dopant atoms are preferably Group VI dopant atoms, more preferably oxygen atoms, or the dopant atoms may be nitrogen atoms
According to a second aspect of the invention, there is provided a diamond having a crystal lattice and a plurality of Group VI dopant atoms within the diamond crystal lattice, and characterised by an effective number of the Group VI dopant atoms being in an energy level or state within the band gap of diamond and such as to provide the diamond with suitable n-type semiconducting properties Preferably the energy level or state within the band gap of diamond is 0,4 eV or less below the conduction band
The invention has particular application to diamond when doping this material with atoms which may form donor states Consider, for example, oxygen Although oxygen has been used in copious amounts during CVD growth, very little, if any, has been incorporated into the bulk of the diamond This means that any defect, within the bulk of diamond, which contains an O-atom must have a very high configurational energy However, by implanting oxygen ions, for example 0+, into diamond, these atoms end up within the bulk of this crystal, whether they like it or not The high formation energies for oxygen states within the bulk of the diamond crystal lattice are, in this way, breached by the ion energies The oxygen atoms find themselves within high energy states Provided the implantation is performed at a low temperature, e g that of liquid nitrogen, the oxygen atoms will find themselves in a variety of states that relate to interstitial and substitutional sites within the diamond crystal
of liquid nitrogen, the oxygen atoms will find themselves in a variety of states that relate to interstitial and substitutional sites within the diamond crystal lattice Subsequent annealing of the diamond to reduce or remove implantation damage, must be carried out at a suitably low temperature to ensure that the implanted oxygen atoms do not relax to an energy level below the band gap of diamond The oxygen atoms preferably occupy substitutional sites without relaxing into states that do not have tetrahedral symmetry
DESCRIPTION OF EMBODIMENTS
The diamond was subjected to multiple CIRA (cold-implantation-rapid- anneahng) steps using 0+-ιons The implantation was carried out at a temperature of liquid nitrogen and the annealing temperature was chosen to be 500°C As control experiments, two identical diamonds, were implanted with C+, and a combination of B+ and C+ ions The ion energies and doses (which ranged from 170 to 35 keV for the 0+ ions) were chosen such that in each diamond the amount and distribution of the radiation damage would (according to TRIM simulations) be the same Furthermore in the diamond co-implanted with B+ and C+ ions, the density and distribution of the boron atoms should be identical to that of the oxygen atoms in the first diamond Details thereof are set out in Table 1
TABLE 1
The resistance behaviour of the three diamonds after the first CIRA-step is compared in Figure 1. All three showed some conduction. The resistance of the C+- "doped" diamond is higher than for the other two diamonds. This proves that the lower resistances in the latter two diamonds are caused by the presence of the boron and oxygen atoms respectively. The resistance behaviour of the C+ -layer showed some hysteresis during heating and cooling. Thermal EMF measurements over the boron-implanted and oxygen-implanted layers, confirmed that they conducted p-type and n-type respectively.
Vacancies can act as deep-lying donors as well as deep-lying acceptors. When they accept electrons, they form ND 1 centres that are situated at > ~ 3,15 eV below the conduction band (CB). When they donate electrons, they form positive vacancies situated at * 1 ,2 eV above the valence band (VB). They will, thus, compensate the dopant states in both p-type and n-type diamond layers. The boron-doped layer (B+ + C+. in Figure 1 ) behaves typically in the manner one will expect from a highly compensated p-type layer in diamond. At low temperatures it conducts with an activation energy of 0,37 eV, and as the temperature is increased, the activation energy increases to approach a value of « 0,8 eV. This is caused by the movement of the Fermi-
level to a position which lies roughly half-way between the boron acceptor states and the compensating, positive vacancy states
A similar behaviour in an n-type layer, where the donors are compensated by ND 1 centres is expected At low temperatures the layer will conduct with the activation energy of the donor states, while at higher temperatures the Fermi level will move towards the half-way position between the donor level and the ND 1 level or any other compensating level which is shallower This is, indeed, observed for the 0+ -CIRA layer At low temperatures it conducts with an activation energy of * 0,32 eV, and the slope starts to, increase at higher temperatures One should note that the deviation for the O-doped layer starts at a higher temperature than for the B-doped layer This is what one would expect from the fact that the compensating ND 1 -levels are situated much deeper below the CB It should, however, also be noted that each oxygen atom can donate two electrons The deviation may thus relate to a movement of the Fermi-level from the first, to the second lonisation levels
It is known from multiple-CIRA implantations, that with each CIRA-step the density of the activated dopant atoms will increase faster than the density of the compensating vacancies The resistance of the layer will then decrease, and the deviation to larger activation energies should move towards higher temperatures This can clearly be seen to be the case for the 0+-CIRA treated layer when one compares the resistance behaviour after the first and third steps (see Figure 1) It should be noted that the resistance of the O-doped layer decreased by nearly two orders of magnitude, while the activation energy at low temperatures still stayed « 0,32 eV
After the third step, Hall effect measurements confirmed n-type conduction, but the mobility was extremely low « 5 cm2/V-s This is not really surprising The layer is very shallow (~ 0,16μm) and many self-interstitials can escape before
annihilating vacancies during the annealing step If MeV implantations are used to create implanted layers with much larger depth-widths, the quality of these n-type layers will improve drastically
Nitrogen can be readily incorporated substitutionally during the growth of diamond, and it does form a donor state lying at ~ 1 ,7 eV below the conduction band As already mentioned above, this is very deep and not of much use in electronic applications Owing to the fact that it had been incorporated during growth, one may conclude that this donor state may be a low energy one The obvious question to ask in view of the results obtained on oxygen, is whether nitrogen may have higher configurational energy states with energy levels nearer to the conduction band edge A similar experiment was thus done using nitrogen ions
The energies and ion doses used for each CIRA step are 170, 155, 140, 125, 110, 95, 80, 65, 50 and 35 keV, and at each energy the N+-ιons were implanted to a dose of 5 x 1013 cm 2 (total ion dose = 5 x 1014 cm 2) For the same reasons as for oxygen, the annealing temperature was chosen as 500°C
As can be seen in Figure 2, already after the first step, the diamond conducted extremely well indeed, and showed an activation energy of « 0,23 eV appeared at low temperatures, and this type of activated conduction dominated up to higher temperatures after the third step When the dopant atoms are at a very high density, and are well compensated, conduction can occur by carrier tunneling from a "filled" dopant atom (having a carrier) to an ' empty" atom nearby In the case of the nitrogen donors the carrier will be an electron that tunnels from a neutral donor to a charged one This type of conduction is just called "hopping" or nearest neighbour hopping (NNH) In diamond NNH usually occurs with an activation energy of « 0,22 eV When one compares the nitrogen-doped resistance to the resistance obtained when doping with oxygen, it is clear that a much higher density of nitrogen became activated
Thus, the conditions for NNH are good Therefore, it seems highly probable that the conduction with an activation energy of 0,23 eV may be ascnbable to NNH
This means that the conduction with an activation energy of 0,29 eV must be band conduction Thermal EMF measurements confirmed n-type conduction Thus, the layer contains shallow donors which relate to nitrogen Owing to the fact that the activation energy is, in this case, 0,29 eV compared to 1 ,7 eV when nitrogen is introduced during growth, these donor states must be higher energy configurations of nitrogen atoms