The utility of laser radiation to incorporate implanted dopants in semiconductors was identified in the late 1970s and remains an ongoing topic of research today. As in the semiconductor industry in general, most studies have focused on Si, where the incorporation of dopants like B, P, and As at concentrations in excess near 1021/cm3 has been realized. The formation of supersaturated substitutional alloys of Si and group III and V elements by ion implantation and pulsed-laser melting has also been studied.
Figure 1 shows a schematic time sequence of the stages of a typical II-PLM experiment: first the semiconductor wafer is implanted with the desired species and then the implanted sample is irradiated using a single pulse from an excimer laser. The laser photons are absorbed close to the surface and converted rapidly to heat which causes the implanted region to melt for times on the order of a few hundred ns. As the heat flows into the underlying substrate, epitaxial solidification begins from the underlying crystalline template and the liquid-solid interface. The solidification front moves at velocities of a few m/s and the quench rate can reach 109 K/s – parameters that are extraordinary in materials processing. This process results in an epitaxial film of the desired semiconductor alloy as the implanted atoms remain after solidification.
Figure 1. The steps involved in II-PLM processing (courtesy M. A. Scarpulla).
Both ion implantation and pulsed-laser melting occur far enough from thermodynamic equilibrium to be governed by kinetics (as opposed to thermodynamic equilibrium). Ion implantation produces non-uniform spatial distributions of implanted species which are quenched in due to near-negligible diffusion distances at room temperature. The subsequent pulsed-laser melting exploits the transient regime of heat flow following a near-impulse deposition of energy to drive a rapid liquid-phase epitaxial growth which maintains the supersaturation induced by ion implantation while reinstating crystalline order. These departures from thermodynamic equilibrium have also been exploited to form homogeneous dilute nitride semiconductors with N ~ 2 orders of magnitude higher than equilibrium solubility limits.
Recently synthesis of dilute ferromagnetic Ga1-xMnxAs and Ga1-xMnxP with Curie temperature as high as 130K and 60K, respectively using the PLM process has been demonstrated. In these studies, it was established that the incorporation of concentrations of impurities higher than the solubility limit at ambient temperatures without precipitation or the formation of secondary phasesis due to fast quenching during the laser-induced transient heat flow.
Figure 2. (a) N+ implantation followed by RTA only, (b) N+ implantation followed by RTA 0.34 J/cm2, (Images by J. Jasinski & Z. Liliental-Weber).
We have utilized the II-PLM process for the synthesis of highly mismatched alloys (HMAs), in particular the dilute III-V nitrides and II-VI oxides. Figure 2 shows TEM images of 2% N implanted GaAs samples (a) followed by RTA at 950°C for 10s and (b) after PLM at an energy fluence of 0.34 J/cm2. In addition to typical implantation induced defects, void-like defects were also present in the sample processed by RTA. Figure 1 (a) shows that the region extending from just below the surface and continued to a depth of about 0.6-0.7 mm contains a very high density of small N-related voids, with an average size of about 2-3 nm. Such N bubbles may account for the low N activation efficiency in these samples. In contrast to the RTA-only sample the TEM image of the PLM sample shows that the sub-surface layer is free from structural defects. A sharp interface at ~ 0.2 mm below the surface is observed indicating the orginal melt/crystalline interface.
Figure 3 shows that using the II-PLM method, the N incorporation efficiency varies with ximp ~ 60% for ximp < 0.02 and ~40% for ximp > 0.02. For high ximp, this represents a five times higher N activation than that observed in samples synthesized by RTA alone. This can be attributed to the extremely short melt duration (~200 nsec) and regrowth process that promotes N substitution in the As subslattice and inhibits the formation of nitrogen related voids, which have been observed in samples formed by N+-implantation followed by RTA only.
Group II-VI dilute oxide (II-O-VI) semiconductors with the anions partially replaced by highly electronegative isoelectronic O atoms are a direct analog of the III-V diluted nitrides. It has recently been demonstrated that group II-O-VI alloys in which highly electronegative O partially replaces the group VI element show behaviors that are similar to those of III-N-V alloys. The electronic band structure of the dilute oxides is determined by an anticrossing interaction between localized states of O and the extended states of the semiconductor matrix. For the CdTe sample implanted with 2% of O after PLM, the fundamental gap is 1.37 eV, corresponding to a band gap reduction DE = 140 meV. This large reduction in the band gap is a clear indication of O incorporation in the Te sublattice, forming CdOxTe1-x alloys. The amount of O incorporated in the Te sublattice in CdOxTe1-x HMAs can be determined using the BAC model. We found that the kinetic limit of solubility of O in CdTe for PLM at 40 mJ/cm2 is ~ 0.015. For a PLM fluence of 80 mJ/cm2 this limit decreases to ~0.01 because of the prolonged melt/crystallization duration.
For the cases of N in GaAs or O in CdTe, the localized states are located within the conduction band and consequently a relatively wide lower subband is formed through the anticrossing effect. The conduction-band states associated with the E- edge retain the extended EM-like character and those at the E+ edge have a more localized EN (or EO)-like character. A narrow lower band can be formed only if the localized states lie well below the conduction band edge. In this case the E- subband states are of highly localized character and the E+ subband states become more extended. This situation occurs when O is introduced into the II-VI semiconductors ZnTe, MnTe, and MgTe. Figure 4 shows PR spectra from a Zn0.88Mn0.12Te substrate and two Zn0.88Mn0.12Te samples implanted with 3.3% of O followed by PLM with laser energy fluence of 0.15 and 0.3 J/cm2. Two optical transitions in the vicinity of ~1.8 and 2.6 eV are clearly observed from the O+-implanted samples after PLM. These transitions occur at energies distinctly different from the fundamental band gap transition at EM = 2.32 eV of the Zn0.88Mn0.12Te matrix, and can be attributed to transitions from the valence band to the two conduction subbands, E+ (~2.6 eV) and E- (~1.8 eV) formed as a result of the hybridization of the localized O states and the extended conduction band states of ZnMnTe. Optical absorption shows that the intermediate band (E-) has a relatively large absorption coefficient (a ~ 105 cm‑1); similar to that obtained above the band gap in direct gap materials. The substitutional mole fractions of O for the Zn0.88Mn0.12OxTe1-x layers synthesized by II- PLM are calculated using the BAC model and estimated to be x~0.024.