Ion implantation is a process designed to create favorable electrical characteristics in a semiconductor. In gallium arsenide, silicon (to create n-type material) and magnesium (to create p-type) are used as 'dopant' ions, changing the conductivity of the semiconductor. These ions are implanted to a certain depth in the gallium arsenide lattice by means of implanter apparatus employing acceleration energies of 3 to 500 kilo electron-Volts.
For gallium arsenide, implantation is the means used to dope the material - diffusion of ions is a common technique in other materials, but gallium arsenide evaporates at the temperatures necessary for diffusion. Therefore, ion implantation is currently the only means by which to place desired impurities into the semiconducting material. In order to implant the ions, they are accelerated to high velocity and sent as a focussed beam onto the target semiconductor surface. This method has several advantageous features over diffusion:
GaAsTEK employs two Varian IIS ion implanters, a 400-10 and a 200-III-V. The 400-10 was
made in 1983 and has maximum acceleration voltage of 400 keV; using doubly charged beam it
can provide up to 800 keV of energy. It has a cold cathode source which is capable of no
more than a few micro-amps beam current. The 400-10 is an older model which is no longer
supported by Varian.
The 200-III-V is an unusual version of the standard model 350D. It has 200 keV
acceleration voltage and a hot filament source which can generate mA's of current. It also
has a vaporizer for implanting from solid sources.
In general, Varian's 200-III-V implanter is a medium current ion implanter used on gallium
arsenide, with ion energies betwen 10 and 200 keV. As with the 400-10, use of
doubly-charged ions allows doubled energies, in this case up to 400 keV. The 200-III-V is
capable of implanting wafers from 2 to 6 inches, at a rate of 200 wafers implanted per
hour.
The ion implantation process generally takes place in a three chamber implanting apparatus. Ions to be implanted are separated from a gaseous mixture, then accelerated by a voltage towards their desired target. The energy the ions acquire determines the depth that they are implanted to in the target wafer.
The ion implantation process occurs in several stages (numbers refer to the numbers on the diagram):
When the beam of ions hits the wafer, the concentration of ions in the substrate
becomes a function of the depth of the ions. This function is a Gaussian relationship of
the form
N(x)=Np*exp((-(x-Rp)2)/(2*DRp2))
where Np is the maximum (desired) concentration, Rp is the range or
desired depth, and DRp is the straggle or standard
deviation.
The implantation process will have as its target either a crystalline or an amorphous target. In the latter case, the Gaussian distribution described will be expected. The same applies in the case of the crystalline target, with an important exception: in a crystalline target, the orientation of the target crystal becomes very important. When oriented in certain ways, there is a path for the implanted ions to travel much deeper into the substrate than desired, an effect called 'channeling'. This condition is generally handled easily by orienting the target crystal in a direction that will not lead to channeling.
The range and straggle are related to the acceleration energy in an almost linear
fashion:
Ideally, the ion concentration relationship should be a step function rather than a
Gaussian distribution. While it is impossible to get a pure step function through ion
implantation, it is possible to get close by implanting ions with different energies and
summing their individual Gaussian distributions.
During the implantation of crystalline materials, target atoms may be displaced from their lattice sites as a result of collisions with incident ions and, if the energy and mass of the ion beam species is great enough, cascades of displacements and hence zones of amorphization may be produced. If the doping of semiconductors by implantation is to be a successful method of device fabrication, the lattice damage has to be removed, and this is achieved by thermal annealing. For GaAs it is found that annealing at temperatures up to and above 900oC is necessary to minimize sufficiently the number of atoms, although it has been found that an annealing stage exists at around 150-200oC. It has been observed that, by implanting at temperatures above 150oC, no amorphous layers are formed even for high doses. The high-temperature annealing of GaAs to remove radiation damage has been one of the main obstacles in advancing the use of ion implantation technology. The difficulty with GaAs lies in the fact that above about 600oC the surface of GaAs crystals dissociates rapidly with As evaporating at a far greater rate than Ga. Three techniques have been developed to preserve the GaAs surface during the annealing process:
The most widely used encapsulant has been Si3N4 which, when plasma deposited, was found to be superior to SiO2 for the passivation of Te-implanted GaAs. Si3N4 encapsulation has permitted the successful annealing of implanted GaAs to temperatures as high as 950oC in a conventional furnace and 1000oC for transient annealing. Another encapsulant that has been found to be suitable for the passivation of GaAs surfaces is RF sputter deposited oxygen-rich AlN. Better activation of high doses of n-type dopants have been observed when this encapsulant was used than when Si3N4 deposited by the same technique was employed. The difference in behaviour of the two caps may be associated with the magnitudes of the expansion coefficients; there is a large thermal mismatch between Si3N4 and GaAs but very little difference between the thermal expansion coefficients of AlN and GaAs.
A technique that has recently been found to be suitable for GaAs at temperatures as high as 900oC is that of annealing in an arsenic vapor overpressure sufficiently great to prevent As evaporation. The attraction of this method is that since there is no encapsulation involved, no indiffusion of unwanted impurities takes place, and the whole annealing process is considerably simplified. In practice, this technique yields variable results owing to the lack of control of the actual overpressure near the sample surface.
The transient process involves raising the sample to a higher temperature (up to 1000oC) for a very short time. Incoherent light beams, scanned electron beams, and graphite strip heaters have been used to carry out transient annealing. The technique of scanned electron beam is very successful in silicon and is now beginning to yield good results for GaAs.
N-type GaAs is produced by implanting column VI elements (Se, Te, or S) or column IV elements (Si or Sn). For doses greater than or equal to 1014 ions/cm2, the implantation of the column VI ions must be carried out in samples heated to 500 degrees Celsius or more. Implantation of these ions at such temperatures results in a higher doping efficiency, as is evident from the graph below:
Column IV must engage Ga sites in order to behave like donors. Implantation of Si at
high temperatures will show a small increase in the doping efficiency of the atoms.
However, increase in doping efficiency due to the sample heating during the implantation
is substantial for Sn atoms. Anneal temperature is another important factor. An anneal
temperature is chosen between 850 and 9500 C. Pulsed laser beam annealing has been
demonstrated to remove the radiation damage from high dose implanted layers. It has also
been applied to high-dose p-type implants in GaAs.
P-type doping of GaAs requires the implantation of group III ions (Zn, Be, Cd, Mg, and C).
From the graph below, we can see that at low doses (up to 1014 cm-2)
the doping efficiency is close to 100%. For larger doses, a saturation effect was observed
with a decrease in doping efficiency. Pulsed laser beam annealing has been applied to
high-dose p-type implants in GaAs. Peak carrier concentrations of 3x1019 to
7x1019 cm-3 and corresponding mobilities of 40-80 cm-2 V-1
s-1 are typical.
