Low contact resistance ρ_s is a key requirement to the design of ohmic contacts, along with endurance and stability under severe operation conditions. However, an increase in the resistivity at the silicon/metal interface is unavoidable. The latter can be caused by either silicon surface treatment quality before metallization or the structure of the forming metallization layer and the charge properties at the Si/Me interface. Furthermore, the surface of silicon always has a natural oxide layer which directly affects the properties of the contact due to mechanical stress. Moreover, the formation of complex metallic phases whose resistivity is often higher than that of the raw components of the metallization increases the overall contact resistance.

Transition metal silicides, e.g. Me = Ti, Ni, Pd and Cr, have a low resistivity and thermal compatibility with silicon [1, 2], making them promising candidates for the fabrication of ohmic contacts. There is the assumption that rapid thermal annealing (RTA) in an H₂ atmosphere can fundamentally improve the properties of contacts to p⁺-Si due to the formation of the abovementioned silicides at the Si/Me interface. Moreover, this favors the purification of the silicon interface at an atomic level [3], thus further reducing the contribution to the contact resistance.

The reverse dark current is an important parameter of photodiode structures. The contribution of the surface component to the dark current of highly ohmic p-Si photodiodes comes mainly from the surface states at the interface of silicon oxide SiO₂, which is used as an insulating film, and from the semiconductor bulk itself. Thermally grown silicon oxide films are known to contain a positive built-in charge [4, 5]. Partial compensation of that positive charge stabilizes the charge properties at the SiO₂/p-Si interface. This can be achieved via hydrogen saturation of the dangling Si bonds [6], which reduces the density of the surface states at the SiO₂/p-Si interface. There is the assumption that RTA in an H₂ atmosphere can reduce the level of the reverse dark current of silicon photodiodes with silicon dioxide dielectric films due to a lower contribution of the surface component originating from the stabilization of the charge properties at the SiO₂/p-Si interface.

Presented below are data on the effect of RTA in an H₂ atmosphere on the properties of ohmic contacts for the example of a two-layered Ti/Au composition on р⁺-Si, as well as its effect on the reverse dark current of a p-Si photodiode with a SiO₂ dielectric film. Furthermore, the applicability of RTA in an H₂ atmosphere for silicon diode and photodiode structure technologies was evaluated.

The fabrication of ohmic contacts requires the capability to measure low specific resistances. There are a number of techniques and their modifications which are used for measuring the specific contact resistance of semiconductor structures. TLM (transmission line method) is one of the most widely used specific contact resistance measurement techniques which takes into account the spreading currents under the contact in the semiconductor bulk, due to the possibility of its adaptation to thin semiconductor films and diffusion layers. Generally, the basics of the technique are as follows. The specific contact resistance ρ_s is determined via probe measurements of the full contact resistance R_T between planar contact pads. The parameter R_T includes the contact resistance R_C and the semiconductor bulk resistance R_S [7]. Depending on the geometrical shape of the contact pads, the technique in question has two modifications: CTLM (circular transmission line method) and LTLM (linear transmission line method).

CTLM specific contact resistance measurement is described in detail in earlier works [8–11]. For example, the CTLM technique was used for studying the ohmic behavior of Ti/n⁺-ZnO/n-Ge and Ti/ITO/n-Ge MDS contacts [10]. Cr/Au, Cr/Au/Ag/Au, Ti/Pt/Au, Pt/Ti/Pt/Au, Pt/Au, Ti/Au, Ti/Pt/Ag and Ti/Pt/Ag/Au contact structures on p-GaSb were studied for the fabrication of high photocurrent density photocells (to 15 A/cm²) [11]. The LTLM technique was described elsewhere [12, 13]. Current propagation mechanisms in Pd/Ni ohmic contacts to p-GaN were studied and optimum metallization layer thickness ratio with minimum specific contact resistance were determined [12]. The contact properties of multilayered Ti (0.02 mm)/Au (0.06 mm)/Pd (0.04 mm)/Au (0.05 mm) metallization on n-GaN were studied [13].

The LTLM technique with linear semiconductor contacts was used in this work. LTLM measurements were conducted for test structures consisting of multiple similar rectangular contact pads with the length A and the width Z, having different distances between one another, i.e., L₁₂, L₁₃, L₂₃ etc. [14]. Minimization of the edge current spreading contribution requires the condition Z << A be met.

The specific contact resistance is determined via probe measurements of the resistance between the contact pads and plotting a resistance vs contact pad spacing graph (Fig. 1), whose tilt angle is R_S/A, with L_T being the transmission length.

Linear approximation of the measured resistance vs contact pad spacing curve allows determining the specific contact resistance using the following equations:

RT=2RC+RS(LZ) (1)

ρC=RC2Z2RS, (2)

where R_C is the contact resistance; R_S is the semiconductor bulk resistance; L is the contact pad spacing; Z is the contact pad width.

The studies were carried out for 60 mm n-Si wafers. The wafers were chemically pretreated in an ammonia-peroxide solution and boron doped by means of 100 keV accelerated ion implantation at a 6.0·10¹⁴ cm^-2 dose for producing a heavily doped p⁺ surface region.

The LTLM contacts were fabricated by vacuum deposition of a two-layered composition, i.e., gold with an adhesive titanium sublayer, followed by photolithographic stripping of similar rectangular pads with 10 to 550 mm distance from one another (Table 1). The contact pad sizes were 100×1000 mm².

The thickness of the Au layer was 0.5 mm, and that of the adhesive Ti sublayer, 0.02 mm. The metallization thickness was measured with a Bruker Dektak 3030 contact profilometer.

Then the test structures were subjected to RTA in an H₂ atmosphere. The heat treatment temperature should be far below the melting points of the metals used in the contact: t_{m Au} = 1337 °С and t_{m Ti} = 1668 °С [14, 15]. Thus, the following RTA temperatures were used: 300, 320, 340 and 360 °С. The RTA time also differed: 10, 20 and 30 s. The test structures after RTA at 360 °C for 10 s had initial signs of Au film exfoliation from the Ti sublayer, probably due to the difference in the coefficients of linear thermal expansion α (CLTA) of Au and Ti by almost 1.5 times: α_Au is (13.3–14.5)·10^-6 K^-1, while α_Ti is (7.23–8.82)·10^-6 K^-1 [15–17]. Nevertheless, the above test structures were also studied.

RTA was carried out in an RSO-650-200 NPO high-temperature vacuum melting furnace having 18 IR lamps 18 kW each and designed for the 50 to 650 °C range and a high heating rate (above 75 °С/s).

The electrophysical parameters of the test structures were probe-measured with a Hewlett Packard 4277ALCZ meter in the –50 to 50 V range. The test structure measurements were carried out twice: immediately after the formation of the two-layered Ti/Au metallization and after RTA. Comparison between the typical current-voltage characteristics of the test structures before and after RTA in an H₂ atmosphere at different temperatures for different time is shown in Fig. 2.

The data (see Fig. 2) suggest that immediately after the formation of the two-layered metallization, the current-voltage characteristics of all the test structures were nonlinear, even at low voltages, and RTA at 300 and 320 °C did not bring any significant improvement to the contact properties of the test structures. Linear and symmetrical current-voltage characteristics are achieved after RTA at 340 °C for 10 s or longer. Further increase in the heat treatment temperature to 360 °C provides for linear and symmetric current-voltage characteristics, despite Au film exfoliation from the adhesion Ti sublayer. Typical current-voltage characteristic before and after RTA in different modes are shown in Fig. 2. The optimum H₂ atmosphere RTA which is sufficient for the formation of an ohmic Ti/Au contact with p⁺ silicon was chosen to be 340 °C.

The time of RTA in an H₂ atmosphere at 340 °C providing for an ohmic contact with the lowest specific contact resistance was found using the LTLM technique. Test structure measurements with the above technique and calculations using Eqs (1) and (2) yielded the following average specific contact resistances:

– RTA for 10 s: 9·10^-3 Ohm·cm²;

– RTA for 20 s: 7·10^-3 Ohm·cm²;

– RTA for 30 s: 7·10^-3 Ohm·cm².

One can conclude that RTA in an H₂ atmosphere is an effective tool for fabricating ohmic contacts between two-layered Ti/Au compositions and p⁺-silicon. The mechanism of the observed process is as follows. Pulsed thermal annealing triggers the formation of titanium silicides at the Si/Ti interface, thus reducing the specific contact resistance. In this case, Ti can be replaced for other transition metals like Ni, Pd or Cr, since they also tend to form silicides upon interaction with silicon [1, 2].

The effect of rapid thermal annealing in an H₂ atmosphere on the ohmic properties of contacts were studied for the example of a two-layered Ti/Au composition on p⁺-Si. Test structures on n-Si with heavily doped p⁺ surface region were fabricated. Similar rectangular contact pads with 10 to 550 mm spacing were fabricated on the test structures for LTLM measurement with a linear semiconductor contact geometry.

The electrophysical parameters of the test structures were measured over a wide range of voltage (–50 to 50 V) immediately after the formation of metallization and after RTA at 300–360 °C with a 20 °C step for determining the temperature at which the current-voltage characteristics of the contacts become linear and symmetrical. The pulse treatment time providing for the lowest specific contact resistance was found via LTLM calculations of the specific contact resistances of the test structures after RTA at the selected temperature for 10, 20 and 30 s.

It was shown that RTA in an H₂ atmosphere at 340 °C for 20 s provides for the lowest specific contact resistance of Ti/Au on p⁺-Si, i.e., 7·10^-3 Ohm·cm². This is due to the formation of low-resistivity titanium silicides at the Si/Ti interface, which improves the contact properties. Ti can be replaced for other transition metals, such as Ni, Pd and Cr, since they also tend to form silicides after interaction with silicon.

The effect of RTA on the series resistance of diodes was studied for the example of p⁺–n clamping diodes. It was shown that pulse treatment in an H₂ atmosphere at 340 °C for 20 s improved the series resistance and, as a result, increased the yield of multi-pad diodes by 15 %. This confirmed the applicability of RTA for reducing the series resistance in silicon diode technology.

Furthermore, the effect of RTA on the reverse dark current was studied for the example of p–i–n photocells based on highly ohmic p-Si with SiO₂ insulating film. It was shown that pulsed treatment in an H₂ atmosphere at 450 °C for 5 s improved the dark currents of the PP and the guard ring of the photocells and, as a result, increased the photodiode yield by 15%. This confirms the applicability of RTA in an H₂ atmosphere for reducing the dark currents in the technology of highly ohmic p-Si photodiodes due to the hydrogen saturation of dangling Si bonds, providing for a decrease in the density of the surface states and stabilization of the charge properties at the SiO₂/p-Si interface.