Characterization of reacted ohmic contacts to GaAs

Characterization of reacted ohmic contacts to GaAs

Solid-Sam Printedin Electronics Vol. 29, No. 9, pp. 903-905, 1986 0038-I 101/86 $3.00 + 0.00 Great Britain PergamonJournalsLtd CHARACTERIZATION ...

287KB Sizes 1 Downloads 78 Views

Solid-Sam

Printedin

Electronics Vol. 29, No. 9, pp. 903-905, 1986

0038-I 101/86 $3.00 + 0.00

Great Britain

PergamonJournalsLtd

CHARACTERIZATION OF REACTED CONTACTS TO GaAs

OHMIC

H. P. KAITELUS~, J. L. TANDONS and M-A. NICOLET California Institute of Technology, Pasadena, CA 91125, U.S.A. (Received 17 September 1985)

Abstract-In the calculation of the true value of specific resistivity of reacted ohmic contacts, a modification in the conductivity of the semiconductor that may occur below the contact must be considered. Contacts to GaAs are discussed. A substantial decrease in the sheet resistance of p-type GaAs is measured for Pt-reacted contacts in which a single thin layer of Mg is interposed. It is pointed out that this lowering, which is attributed to the doping action of Mg, if not taken into account, can lead to serious errors in the estimation of specific contact resistivity.

Laterally uniform ohmic contacts formed to semiconductors can basically be classified into three categories, as shown schematically in Fig. 1. The idealized unreacted case (a) has a sharp interface between the metal M and the semiconductor S. In reality, such ideal contacts are rarely achievable, and most contacts involve reactions, as depicted in (b) and (c). Reactions of metal(s) with semiconductors to form ohmic contacts are often unavoidable, and in many instances desirable for stability. Additionally, the contacting metals could also contain elements, as in case (c), which upon reaction could enhance the doping of the semiconductor near the surface. Contacts of the type (c), comprising of Au-dopant-based eutectic compositions, are popularly used on GaAs[l,2]. However, because of lateral inhomogeneities introduced due to liquid-phase reactions involved in these types of contacts, the effects of the dopant elements in increasing the conductivity of GaAs below upon reaction are not fully understood[3]. This paper addresses laterally uniform contacts to p-type GaAs formed by the solid-phase reaction of Pt[4-71. The contacts were characterized by backscattering spectrometry, and in terms of their specific contact resistivity. In order to investigate the effects of dopants on contact resistivity, samples were also prepared with thin (m 100 A) layers of Mg interposed between Pt and GaAs. To facilitate contact resistivity measurements, thick (_ 2500 A) layers of Ag were deposited on the contacts which were separated from Pt films by TiN layers. The TiN layer acts as a diffusion barrier in confining the reaction of Pt with tPennanent address: Semiconductor Laboratory, Technical Research Centre of Finland, Otakaari 5A, SF-02150 Espoo 15, Finland. $Permanent address: Applied Solar Energy Corporation, City of Industry, CA 91749, U.S.A.

(a)

(bl

(c)

Fig. 1. Schematic diagram of possible metal (M) and semiconductor (S) ohmic contact configurations; (a) unreacted contacts, (b) reacted contacts, and (c) reacted contacts with a dopant element in the metallization system.

GaAs, and in preventing Ag from interfering with this reaction[l. Contact studies were carried out on p-type (Zndoped) epitaxial layers grown by metalorganic chemical vapor deposition (MGCVD) on (100) n-type GaAs substrates. The hole concentrations in the layers were -2 x lo’* cm-), and their thicknesses were l-2 pm. The films of Pt were evaporated by an electron beam, whereas Mg and Ag layers were deposited by RF magnetron sputtering in an Ar ambient. Films for TiN were prepared by reactively sputtering Ti in a premixed gas of 80% Ar and 20% N2. During TIN deposition, the substrates were negatively biased with respect to the sputtering chamber to improve the quality of the films[8]. The thicknesses of all layers were independently calibrated by backscattering spectrometry, assuming bulk density values[9]. For contact resistivity measurements, metallization patterns were defined photolithographically by liftoff, conforming to the circular transmission line model[lO]. Parallel samples were also prepared to monitor the reactions in the contact systems by backscattering spectrometry. The annealings of the 903

H. P.

KMTELUS

1 ........... 55Ok

30 min

rpt3401(

L 1.0

1.4

1.2

ENERGY

1.6

16

(MeV)

Fig. 2. 2MeV ‘He+ backscattering spectra of GaAs/Pt/ TiN/Ag contacts before and after annealing at 35o”C30 min and at 55O”C-30 min.

contacts were carried out in vacuum (N 7 x lo-’ torr) in the temperature range 350-55O”C, and for times up to 30min. Backscattering spectra obtained on GaAs/Pt/ TiN/Ag contacts, before and after heat treatments, are shown in Fig. 2. The spectra were taken using a 2 MeV ‘He+ beam. No change in the trailing edge of the Ti signal from TiN signifies the thermal stability of the TiN 6hn upon heat treatments up to 550°C for 30min. In addition, insignificant shift in the leading edge of the Pt signal shows that the reaction between Pt and GaAs is confined by the TiN diffusion barrier layer. The steps in the trailing edge of the Pt signal, and the leading edge of the GaAs signal can be attributed to the formation of compounds PtGa and PtAsz [4-7’J. Further TEM studies performed elsewhere[5] point to the fact that the solid-phase reaction of Pt with GaAs is laterally uniform. For the system considered here, the TiN film simply isolates the reaction of Pt with GaAs from interference by the top Ag layer[7j, and this reaction is also believed to be laterally uniform. Specific contact resistivity (p,) measurements were made on patterns conforming to the circular transmission line model[lO]. To eliminate errors due to excessive probe resistance, separate pairs of probes were used for the voltage and the current measurements. End resistance measurements were also made to calculate the modified sheet resistance below the reacted contacts[ 111. The ratio RJR,,, of the sheet resistance below the reacted contacts to the original sheet resistance of p-GaAs (N 180 n/Cl) is plotted in Fig. 3 as a function of annealing temperature. For the GaAs(p)/Pt/ TiN/Ag system the ratio RJR,,, is close to 1 and rises slightly upon annealing. This shows that the reaction of Pt with GaAs does not modify the conductivity of

et al.

the p-layer underneath significantly. The small increase in R,, can be attributed to the formation of intermetallic compounds (PtAs, and PtGa) upon reaction, resulting in a reduction in the effective thickness of the p-type GaAs layer. From backscattering measurements (Fig. 2), the thickness of the consumed GaAs upon reaction with Pt is estimated to be only about 800 A, which is much smaller than the original thickness of the p_GaAs layer (l-2 pm). Thus the increase in R,t is small. In contrast, with a thin (u 100 A) Mg layer interposed between GaAs and Pt in the GaAs(p)/Mg/Pt/TiN/Ag system, a dramatic drop in the KL/Rsh ratio is observed upon annealing. This lowering by more than a factor of 6 in %, presumably results from the doping action of Mg. The inclusion of Mg in the metallization system thus considerably increases the conductivity of the underlying p-type GaAs layer upon reaction. Corresponding contact resistivity (p,) values, for the contacts described in Fig. 3, are plotted in Fig. 4 as a function of annealing temperature. The solid circles and solid lines represent pc values calculated for the two contact systems when considering the modifications in the sheet resistances of the underlying p-type GaAs layer. These are the true values of pF. The fact that the contact resistivity is lower for contacts with interposed Mg layers is primarily due to the increased doping in the p-type GaAs layer below the contacts upon reaction with pt. Such a decrease in the contact resistivity of ohmic contacts

1.2

1.0

0.6

0.6

0.4

0.2

0

350

450

550

ANNALING TEWERATURE (‘C) ( 30 min in vawum 1

Fig. 3. The ratio of the modified sheet resistance below the contacts to the original sheet re&tance of pGaAs away from the contacts (RJ&) as a function of annealing temperature. A considerably low value of RJR,,, is measured for the contacts with an interposed Mg layer between Pt and pGaAs

Characterization

of reacted Ohmic contacts to GaAs

905

lated on the basis of models that assume unmodified sheet resistances of the underlying semiconductor layers. Such calculations can only be justified for unmatted contacts, and are therefore misleading. Since most ohmic contacts to GaAs are reacted contacts, it becomes imperative to consider the changes in the conducting properties of the underlying semiconductor layer in deriving the contact resistivity value from measured data. The discrepancy between the values calculated with or without considering sheet resistance modifications is expected to be substantial for metallization systems with dopant elements.

-I

IO 450

350 ANNALING (30min

550

Acknowledgements-This work was partially supported by Sandia National Laboratories (Len Beavis) under contract No. 47-3966, and, in its tinal phase, by the Army Research O&e under contract DAAG-29-85-K-0192. Dale Burger at Jet Propulsion Laboratories provided the computer program used in the calculations of specific contact resistivities. We thank Kevin Douglas at Applied Solar Energy Corporation (ASEC) for sample preparation. Encouragement provided by George Vendura at ASEC is also appreciated.

TEM’ERATURE ( ‘C ) invacuum)

Fig. 4. Contact resistivity of GaAs(p)/Pt/TiN/Ag and GaAs(p)/Mg/Pt/TiN/Ag systems as a function of annealing temperature. Solid circles and solid lines represent true values of contact resistivity. Open circles and dashed lines are the contact resistivity values for the GaAs(p)/ Mg/Pt/T’iN/Ag system calculated without taking the mod&xi sheet resistance value into account.

with an increase in the doping concentration of semiconductors is well documented[l]. Also plotted in Fig. 4 are contact rcsistivity values (open circles and dashed line) for the GaAs(p)/Mg/Pt/TiN/Ag system which were calculated assuming no change in the sheet resistance of the p-GaAs layer below the contacts. These values, which are substantially lower than those calculated with modified sheet resistance values, do not represent the true values of contact resistivity. It should be pointed out that most contact resistivity values reported in the literature are calcu-

REFERENCES

1. A. Piotrowska, A. Guivarc’h and G. Pelous, Solid-St. Electron. 26, 179 (1983). 2. V. L. Rideout, Solid-St. Electron. 18, 541 (1975). 3. T. S. Kuan. P. E. Batson. T. N. Jackson, H. Runurecht and E. L. Wilkie, .I. Ap&. Pkys. 54, 6952 (1983). 4. A. K. Sinha, T. E. Smith and J. J. Levinstein, IEEE Trans. Electron Devices 22, 218 (1975). 5. A. K. Sinha and J. M. Poate, in Thin FilmsInterdiffuion and Reactions (Edited bv J. M. Poate. K. N. Tu”and J. W. Mayer), p. 407. Wiley, New York (1978). 6. C. Fontaine, T. Okumura and K. N. Tu, J. Appl. Pkys. 54, 1404 (1983). 7. H. P. Kattelus, J. L. Tandon, A. H. Hamdi and M-A. Nicolet, Material Res. Sot. Symp. Proc., Pennsylvania (Edited by J. M. Gibson and L. Dawson) 37,589 (1985). 8. J.-E. Sundgren, B.-O. Johansson, H. T. G. Hentrell and S.-E. Karlsson, Thin Solid Films 105, 385 (1983). 9. W-K. Chu, J. W. Mayer and M-A. Nicolet, Backscaltering Spectrometry. Academic. New York (1978). 10. G. K. Reeves, Solid-St. Electron. 23, 487 (1980). 11. G. K. Reeves and H. B. Harrison, IEEE Electron Dev. Lerr. EDL3, 111 (1982).