磷化铟晶片退火处理研究

1 Introduction

Semi-insulating (SI) InP is a promising material for Xray detectors, having large atomic number Zin, suitablevalue of the band-gap, high resistivity, large electron mo-bility and reasonably large hole mobility. Besides, highquality large single crystals of SI InP doped with iron canbe grown by Czochralski technique. While introduction ofdeep level iron impurity is necessary to obtain IP withsemi-insulating properties, the iron also acts as a trapping center of electrons and holes created by X-rays dur-ing their detection, which cuts back the charge collectionefficiency and resolution power of the detector. Therefore.there is a need to obtain semi-insulating InP either by dop-ing with smaller concentration of iron or with some moresuitable deep level impurity. Long-time high-temperatureannealing is a way to obtain SI InP with smaller concen-tration of trapping centers which should lead to a betterquality of InP X-ray detectors.

It was found that wafers of nominally undoped highpurity liquid-encapsulated Czochralski(LEC) InP becamesemi-insulating (SI) when annealed at high temperature ina phosphorus atmosphere 1. The result was interpretedas due to a compensation effect of a deep acceptor with themid-gap activation energy 0.63 0.65 eV and concentrationbelow 5 x 1015 cm-3.

Later, it was suggested that the mid-gap level could bedue to Fe2+ impurity in InP introduced by in-diffusion ofiron or by activation of Fe3+ ., since iron is well known to bean omnipresent contaminant in furnace annealing experiments 2]. Therefore, a comparative sensitive detection ofiron in undoped InP wafers before and after annealing was performed. It was concluded that iron was indeed incorporated and activated in InP wafers during annealing but its amount was not suffiffifficient to explain the semi-insulating properties of annealed InP.

Activation ofiron and reduction ofshallow donors werestudied in samples annealed for various periods of time 3The results showed that both the annihilation of shallowdonors and activation of Fe2+ deep acceptors were causedby fast diffusive transports, out and into the InP wafer.respectively.

In the first experiments, the iron was introduced intothe SI InP from an uncontrolled contamination during an-nealing which did not give fully reproducible results. Lateran effort was made to develop a controllable method of Fedoping on a comparably small level 4 . For this purposewas used a doping of InP wafers during annealing via thevapor phase by introducing a defined Fe source. An increase of the active Fe in InP was achieved, which was incontrast to older investigation showing no such increase 56]. It was suggested in 4] that the different behavior wascaused by different cooling rates, in particular that fastcooling rates (quenching) suppress the active Fe in InPThe same results were obtained independently by otherauthors (7 . This dependence on the cooling rates can beexplained by the hypothesis (7 that in quenched samplesmost Fe atoms are frozen in interstitial sites which are notelectrically active. On the contrary, a very slow cooling canfavor occupancy of the In sites by Fe2+.