C. S. Fuller

Diffusion of Donor and Acceptor Elements in Silicon

The diffusion of the Group III (B, Al, Ga, In, and Tl) and Group V (P, As, Sb, and Bi) elements in silicon has been measured in the temperature range 1050–1350°C. A method based on change in conductivity through the penetration layer has been used for B and P. The p‐n junction method has been used for the other elements. Aside from B and P, which have similar diffusional properties, the acceptor elements diffuse more rapidly than the donor elements. Diffusion coefficients are given by DB, P=10.5 exp − (85 000/RT), DA1=8.0 exp − (80 000/RT), DGa=3.6 exp − (81 000/RT), DIn, T1=16.5 exp − (90 000/RT), DAs=0.32 ×exp − (82 000/RT), DSb=5.6 exp − (91 000/RT), DBi=1030 exp − (107 000/RT) with an average estimated error of about ±40%. This corresponds to an error in the activation energies of about ±5 kcal. Sources of error including the effects of impurities in the oxides are discussed. D0 values in most cases conform to the predictions of Zener for substitutional diffusion.

Effect of Heat Treatment upon the Electrical Properties of Silicon Crystals

Studies have been made of the process in which donors are introduced into silicon by heating in the temperature range 300°–500°C and are caused to disappear on heating at higher temperature. This phenomena is shown to depend on the conditions of growth and the heat‐treatment history of the crystal. Evidence is summarized which shows that oxygen is the impurity from which the donors are formed. The characteristics of the processes involved are described and possible mechanisms are discussed.

Photoluminescence of Cu‐Doped Gallium Arsenide

Photoluminescence measurements made at 20° and 77°K on Cu‐diffused GaAs are compared with Hall‐effect measurements on the same specimens. The level of the singly charged Cu acceptor lies 0.155 eV (20°K) above the valence band, compared to 0.145 eV (0°K) as determined by Hall effect. Intense phonon replicas indicate that this acceptor is strongly cojpled to the lattice; the observed discrepancy in the ionization energies may therefore be caused by a Stokes shift. Cu introduced into acceptor‐doped GaAs under intrinsic conditions shows the same acceptor level, but this level is absent when Cu is diffused into extrinsic p GaAs. No luminescence is found which can be assigned to the interstitial Cu donor, possibly because this donor level is paired with the acceptor is 77°K. Diffusion of Cu into Te‐doped GaAs causes the shift of a luminescence band at 1.2–1.33 eV; this is interpreted as a transformation of a donor‐vacancy center into a donor—Cu center. Shallow acceptor levels at 0.020 and 0.027 eV are found in ...

Changes in Electron Concentration of Donor‐Doped GaAs Crystals Caused by Annealing

GaAs doped with ∼1019 cm−3 Se or Te shows reversible temperature‐dependent changes in electron (donor) concentration over the temperature range 650 to 1100°C. Crystals containing between 2×1018 cm−3 to approximately 1019 cm−3 donors conform to the same equilibrium curve with an enthalpy of ∼0.5 eV. The behavior is that expected for a solubility equilibrium. Evidence against an equilibrium involving donor precipitation is presented, and the suggestion that the donors may interact to form molecules is discussed.

Effect of Structural Defects in Germanium on the Diffusion and Acceptor Behavior of Copper

The effect of dislocations in Ge upon the diffusion of Cu has been studied using Ge crystals having different etch‐pit counts as well as on Ge specimens bent on a (112) axis. The percentage of the total Cu introduced in a given time which attains the acceptor state was measured by comparing conductivity to radioactivity determinations employing Cu64. In addition, the detailed diffusion processes have been studied as a function of time and etch‐pit density by means of autoradiographs taken both parallel and perpendicular to the direction of diffusion and also by delineation of p‐n boundaries.The two most important findings are: (1) the slow attainment of acceptor equilibrium especially in Ge having low concentrations of dislocations and (2) the identification of dislocations as the initial loci of acceptor copper in Ge.The results call for a revision of our ideas on the mechanism of diffusion of Cu. The view is advanced that Cu atoms enter Ge interstitially at a high rate (>10−4 cm2/sec), but that their fl...

Hall‐Effect Levels Produced in Te‐Doped GaAs Crystals by Cu Diffusion

Hall‐effect measurements have been made on Te‐doped GaAs crystals diffused with known amounts of 64Cu sufficient to convert the crystals to p‐type. Energy levels are found above the valence band at 0.123±0.005 eV, 0.145±0.005 eV, 0.166±0.005 eV, 0.19±0.005 eV, and 0.43±0.01 eV. An analysis of the Hall data has been made by means of computer programs. The results may be summarized as follows: (1) Ga vacancies produced during the Cu diffusion associate with and neutralize Te as a compensating center. (2) The ionization energy of the TeCuGa pair is 0.19 eV rather than 0.166 eV as suggested previously.

Double Acceptor Behavior of Cu in Te‐Doped GaAs

Ionization energy levels introduced in Te‐doped GaAs after conversion to p type by diffusion of 64Cu have been measured by means of Hall effect. Levels at 0.145, 0.166, 0.20, and 0.44 eV are observed after various sequential heat treatments. The 0.145 and 0.44 eV levels are attributed to the two acceptor levels expected of Cu on a Ga site. The 0.20 eV level is assumed to arise from a Ga vacancy as previously reported, and the 0.166 eV level is interpreted as an ion pair between Cu on a Ga site and Te on an As site.

Investigation of quenched-in defects in Ge and Si by means of 64Cu

Abstract Quenched-in defects in Ge have been investigated by comparing the hole cones, produced by the acceptor defects with 64Cu cones, resulting when 64Cu is diffused into them at 500°C. It is found that for low defect cones. (~5 × 1014 cm−3) one 64Cu is added per defect hole whereas for higher cones. (>1015 cm−3) more than one 64Cu is added. This result is interpreted as support for the defect cluster model of Letaw. Further support for this model is provided by a study of the kinetics of loss of the defect acceptors on annealing. Ge, quenched in the presence of 3 × 1016 cm−3 64Cu, is found to add up to 5 × 1015 cm−3 64Cu of the same vintage at 500°C. This result is interpreted as supporting the clustering mechanism of vacancy precipitation. Peculiarities in the kinetics of precipitation of Cu from Ge are likewise able to be understood on this model. The cone, of ~4 × 1015 cm−3 vacancies found by tweet as present in as-grown Ge crystals at the m.p. has been verified. Results similar to those discussed above for Ge are reported for Si.

Defect Centers in GaAs Produced by Cu Diffusion

Evidence for defect centers having an ionization energy at Ev+0.10 eV in melt‐grown GaAs is given based on: (1) Hall‐effect measurements on Cu‐diffused crystals. (2) Residual 64Cu left in GaAs after extraction by indium. (3) Decrease in electron concentration of donor‐doped crystals after annealing. The failure to observe the defect level in photoluminescence is discussed. The results suggest that Ga vacancies are present in clusters in melt‐grown GaAs crystals.

Hall Effect Investigation on Lithium‐Diffused Gallium Arsenide

Hall effect results on GaAs crystals into which Li is introduced at 500°C, as well as at higher temperatures, and subsequently removed in Ga or in air at 500°C are reported. Four acceptor ionization energies: 0.023 and 0.11 eV (previously reported), 0.05 eV and 0.14 eV are found. The two levels 0.023 and 0.05 eV are attributed to Li self‐pairs. The 0.11‐ and 0.14‐eV levels are associated with Cu possibly already present in the GaAs crystals. The 0.023‐eV level is found in crystals produced by the floating‐zone process at all temperatures of the Li diffusion. The 0.023‐eV level is also found in the horizontal Bridgman crystals when diffusion is above ∼650°C. Below ∼500°C, however, the 0.05‐eV level predominates in Bridgman crystals and the 0.023‐eV level is not observed. The origins of the various levels are discussed.