JPH0364473A - Vapor deposition of titanium nitride by cold wall cvd reactor - Google Patents

Abstract

PURPOSE: To execute vapor deposition without damaging the circuit elements on a cleaned wafer by passing a mixture composed of ammonia and titanium chloride heated to a prescribed temp. on the wafer, thereby forming the blanket layers of titanium nitride.

CONSTITUTION: The semiconductor wafer 24 is cleaned by exposing the wafer to ozone. Next, the cleaned wafer 24 is loaded into a cold wall CVD chamber 10. Atmosphere gases are sucked out of the chamber 10 via exhaust valves 34, 36, 40. The wafer 24 is then heated and treating gases 32 contg. the titanium chloride and the ammonia and a diluting agent selected from hydrogen and/or nitrogen are passed thereon at ≥250°C.

COPYRIGHT: (C)1991,JPO

Description

TECHNICAL FIELD OF THE INVENTION The present invention relates to a method of depositing a blanket layer of titanium nitride over a semiconductor wafer substrate.

[Background of the invention]

The oldest CVD technique involves the deposition of titanium nitride films, which were first reported in the 1920s. (
(See Paper Deposition, edited by C.F. Parael, New York Wiley, 1966, p. 378) Such films have been used commercially as jewelry or tool coatings for many years, but extremely high deposition temperatures are required. This has limited their interest as thin films for integrated circuits. More recently, reactive sputtering techniques have enabled lower temperature coatings, so such thin films are now used where tool coatings are sensitive to high humidity and as diffusion barriers in highly integrated circuits and solar cells. ing. (Geo E. Sandgren et al. Thin Solid Films 105°353 (198
3)). Particularly for integrated circuits, the poor process coverage of physical deposition techniques (ie, sputtering) is a problem. On the other hand, as CVD films appear to be compatible, an evolving low temperature CVD processing method may be of practical interest.

The first chemistry used to make CVD TiN was Tj
CQ4. It consists of passing a mixture of N2 and N2 at normal pressure over the heated sample. More recent studies have shown that CVD Tj
It has been shown that N can be deposited from TjCQ4 and NH3 at temperatures as low as 500°C compared to the approximately 1000°C required by older methods.

(Refer to Nis R Kara et al. Thin Solid Films 140.277 (1986)) Similarly, NH
It has also been shown that the 3° TiC114 process performed at low pressure in a hot wall tube reactor can deposit TiN films at 700°C, although an anneal of 900-1000'C is required.

As has already been shown in attempts to deposit TaSi2 by CVD, there are dramatic differences in the properties of the deposited film depending on whether the deposition is done in a hot wall or a cold wall.
Noise Publication 1987, page 100) Therefore, by carrying out the study in a low-pressure cold wall C■D reactor, NH3+Ti for the deposition of TjN.
CQ, it was decided to make further use of the law. The second, qt-wafer approach provides greater control over the process and may be a practical approach if high and sufficient deposition rates are achieved.

[Purpose of the invention]

It is therefore a primary object of the present invention to deposit a low reactivity blanket layer of titanium nitride onto a semiconductor wafer without damaging the circuitry on the wafer and without requiring high temperature annealing. An object of the present invention is to provide a method and apparatus for.

[Summary of the invention]

These and other objects of the present invention that will become apparent as the description progresses are such that, in short, after cleaning the wafer with ionized hydrogen gas, the NH
3 and T i CQ4 by depositing a Plancket I-layer of TjN on the heated wafer using a mixture heated to at least 250°C.

[Action of the invention]

An important advantage of the present invention is the ability to deposit a uniform blanket layer of TiN with low contact resistance to the substrate without the need for high temperature annealing.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings, and the symbols used in the drawings will be collectively explained as follows.

10...CVD reactor chamber 12.14...Turbo molecular pump 16...Load lock 18...Wax type valve 20...Airtight water-cooled light housing 22...Gas manifold 24...Wafer 26... ... Pyrometer 28 ... BaF2 window 30 ... Moving arm 32 ... Gas box 34, 36, 38, 40 ... Gate valve 42 ... Tungsten halogen lamp 50 ... Aluminum oxide cup 52 ... Experiments were conducted in a research CVD reactor as schematically shown in the stainless steel faceplate 54...tube cathode 56...cooling channel 58...exhaust diagram. The chamber 10 is made of stainless steel, and metal gas scalers 1 to 1 are used for all flange parts. Turbomolecular pumps 12,14 are used to evacuate main chamber 10 and loadlock 16, and roots blowers (not shown) are used to flow bulk reactant gases at appropriate pressures for deposition. Butterfly valve 18 allows independent control of pressure and gas flow. To enable well-characterized experiments,
Since the system was calcined, base pressures of 5 x 10-e Torr were easily obtained.

A tungsten halogen lamp (10 kW) 42 in an airtight water-cooled housing 20 is positioned inside the chamber 10. Thermal energy is radiated through the quartz window into the chamber where the wafer 24 (face up) is placed in the overlay of the lamp assembly.

Wafer loading is accomplished through a vacuum load lock using a bellows type insulated transfer arm 30. The chamber temperature is measured with a Linear Lab pyrometer 26 placed outside the chamber 10. Since the back side of the wafer is observed through the BaF2 window 28, 200
Temperatures as low as ℃ can be read. The lamp 24 is powered by a 5CR (not shown) and the temperature is controlled using the output of the pyrometer 26 with a Eurotherm controller 808 (not shown). With this system, temperatures as high as 900°C were obtained.

Gas is introduced from gas manifold 22 using a flow controller. TiCQ4 was directly vaporized at a temperature of 40℃,
Provide the necessary vapor pressure for this experiment.

The experiments reported here ranged from 450°C to 700°C.
, the pressure range is from 100mtorr to 300mtorr
r, the flow rate is >2 sec for T i CQ, N
Regarding H3> it was 40 sec. In the present invention, we choose to use N2 as a diluent rather than He, or Ar, or N2.

All experiments reported were performed at a ratio of NH3 to TicQ of 20. Also, since Ti(1, and NH3 react at room temperature to produce a solid product, N2 was used to isolate the TjCl, gas manifold from the NH3 manifold. The chosen flow rates were 2. 2s
ccm, 44 secm for NHa, and 20 secm for N2. Next, apply a thin film of 4" (approximately 1
0.16 cm) bare silicon wafers at various pressures and temperatures. The thickness was approximately 1,500 people.

For all depositions, the films were polycrystalline with columnar orientation. As seen in Figure 2, the diameter of the crystals was approximately equal to the film thickness. In addition, the surface is relatively smooth, and
I could barely see the conspicuous spots between 0 and 100 people.

One deposition was performed on a patterned wafer and a scanning electron micrograph of the film is shown in FIG. This film is approximately 1500 mm thick and appears to have a highly conformal coverage. The particles appearing on the thin film surface appear to be due to gas phase columnar nucleation. However, this effect is not observed in most inventive depositions.

All deposited films were tested for adhesion with the Scotch tape test and passed easily. Therefore, it seems that there is no problem with the adhesion of thin films with a thickness of 2000 mm or less.

The majority of deposited films exhibited the characteristic TiN gold color. Very thin films (=500 A) appeared silvery-gold, while thicker films (; 2000 A) exhibited a rose-gold color. In contrast, it has been reported that none of the thin films deposited by the Carrer and Gorton method exhibited the expected gold color.

Auger analysis and R
I3S analysis was performed. Auger analysis gave information about chlorine and oxygen in the thin film, but the Ti/N ratio could not be determined because the Ti and N lines overlapped. A typical Auger depth III line is shown in FIG.

Of interest is the high oxygen concentration near the surface. This layer is typically on the order of 60A thick and represents oxidation of the TiN at the surface. Similarly, Sl and T i
There appears to be a 200λ thick layer at the interface where N is mixed. However, the majority of the thin film appears to be TiN.

The Ti/N ratio could be determined by RBS analysis. T≧
This ratio was 1 for all films at -550°C. Deposition at T=450°C shows a nitrogen-rich film;
Ti/N was in the range of 0.84 to 0.92.

FIG. 5 shows the change in oxygen concentration due to pressure and temperature. Higher temperature deposition showed higher concentrations. Similarly, it was observed that the lower the pressure, the lower the oxygen concentration.

Since very little oxygen was introduced into the feed gas, this must have come from traces of water vapor and air in the chamber.

Figure 6 shows the change in chlorine concentration with respect to pressure and temperature. Contrary to the behavior of oxygen, the chlorine content becomes lower at higher deposition temperatures. Similarly, the high pressure cases above 450°C showed small amounts of chlorine. Although it has previously been reported that the chlorine content at the surface is seven times higher than in the thickness, no such surface peak was observed in our experiments.

About 40 years ago, TiN formed by CVD was shown to incorporate some hydrogen into it. (d)・
H. Bollard et al. Transactions of the Faraday Society 46.190 (1950)
) To examine this argument, Charles Evans
We chose the forward scattering technique developed by T.T. Pardein et al. Thin Solid Films 119.429 (1984) in this procedure, in which high-energy Heg particles strike a thin film at a shallow angle.

The He atoms and the ejected hydrogen atoms are scattered forward. After removing the He, the measured hydrogen flux provides information about the numerical density present in the thin film of hydrogen atoms.

The results of such measurements on the thin film of the present invention are shown in FIG. Here, increasing the deposition temperature increases the hydrogen content, but high pressure appears to suppress this.

As previously stated, high deposition rates would be advantageous if a single wafer reactor concept is pursued. In current experiments we have been able to achieve deposition rates as high as z700 people/min.

FIG. 8 shows changes in deposition rate depending on pressure and temperature.

Increasing pressure and temperature appears to lead to higher deposition rates. As expected from this, diffusion is limited at the highest temperatures, and since all our experiments were done with the same flow flux, we would not expect it to change significantly with pressure. Previous studies suggest that the deposition rate increases reported here should be without much difficulty.

It was possible to calculate the thin film resistance from Blomme 1 to Sea 1 resistance measurements with the Rix Omni-Map instrument and the film thickness [2] derived from the Auger curve. 9th result
As shown in the figure. Here the lowest values were obtained at higher pressures and temperatures. Compared to thin films deposited by reactive sputtering, values as low as only 100 .mu..OMEGA.-cln have been obtained with substantial ion bombardment of thin films. (See N.Searchieri Baka Solid State Technology, February 1988, p.75). N.H.
The lowest values in high-temperature hot tube experiments have been reported for thermal CVD thin films deposited using No. 3 to be 75-100 μΩcm for 700° C. deposition. The lowest value reported for atmospheric pressure cold wall equipment was 300 μΩ-cm for 650° C. deposition.

Based on the experimental results described herein, the inventors have shown that a low pressure cold wall CVD reactor can be used to deposit stoichiometric Tj,N thin films at moderate temperatures and high deposition rates. It was clearly stated here.

It was determined that the film contained small amounts of chlorine, oxygen and hydrogen. The thin film is polycrystalline and highly conformal (c
coat step (co formality)
at 5 tep). There appears to be no problem with adhesion, and the thin film exhibits the well-known golden color of TjN.

The as-deposited resistance was found to be as low as 100 μΩ-cm.

A shower head as shown in Figures 10 and 11.
A gas distributor/mixing chamber was placed an inch above the wafer through which the reactive gas was introduced. T j CQ, and NH3 react at room temperature to form a solid (powder) adduct TjCΩ, ·nNH3 (where 2 < n
< 8), it is difficult to introduce and mix these gases to maintain a particle-free chamber.

(Refer to Nis Al Kara et al. Thin Solid Films 140.277 (1986)) However, 250~
For temperatures in the range of 300° C. no powdery adducts are produced, but at the same time this temperature is too low to allow appreciable TiN deposition. As expected, we electrically heated our gas distributor/mixing chamber to 250
At ˜300° C., no deposits are observed either in the chamber or on the wafer undergoing deposition. In order to keep the internal surfaces of the metal CVD reactor free of deposits and particles, all of the surfaces can be maintained at 250-300°C.

This saves the time needed to clean the reactor.
Provides additional benefits that improve output.

In addition to acting as a diffusion barrier, CVD TiN
The thin film must have a low resistance ohmic contact to the silicon. Unfortunately, silicon is a very reactive material and forms very thick autogenous oxides (~25A) when exposed to the atmosphere.

If this oxide is left as is before TiN is deposited by CVD, a peak concentration (20%) will occur at the interface between TjN and Si.
) of oxygen, which causes an unacceptably high contact resistance.

To eliminate this problem, each wafer is rough-etched in-situ before each deposition. First, the wafer is exposed to ozone for 1 minute before it is placed in the chamber. This removes carbon on the wafer surface. This is important because even trace amounts of carbon in autogenous oxides make removal more difficult. An in situ hollow cathode discharge is then performed inside a small aluminum oxide cup (FIG. 12) while the wafer is in the reactor chamber 5.

Aluminum oxide cup 50 has a stainless steel faceplate 52. Tube 5 functioning as a cathode
4 (equipped with a water-cooled channel 56). Hydrogen gas is blown into the hollow cathode 54, from which the exhaust gas 58 is a mixture of H2, H1H' and e-. The ions quickly recombine with electrons, but the hydrogen atoms persist long enough to strike the surface of wafer 24. There it reacts with autogenous oxides to produce H20 vapor, which is pumped away. Auger scanning of the interfacial region of the TiN or Si thin films thus produced provided clear evidence of the effectiveness of this procedure. At this point, there is no evidence of an oxygen peak as there was for unprocessed wafers.

[Brief explanation of drawings]

Fig. 2 is a scanning electron micrograph showing the crystal size of titanium nitride; Fig. 3 is a scanning electron micrograph showing the step coating of titanium nitride on a patterned wafer. Micrograph, Figure 4 is a graph of a typical Auger depth curve, Figure 5 is a graph showing the relationship between oxygen concentration, pressure and temperature, and Figure 6 is a graph showing the relationship between chlorine concentration and pressure and temperature. , FIG. 7 is a graph of hydrogen versus pressure and temperature, FIG. 8 is a graph of deposition rate versus pressure and temperature, FIG. 9 is a graph of thin film resistance versus pressure and temperature, and FIG. 10 is a graph of used in the present invention. A bottom view showing the bottom of the shower head/gas distributor, Figure 1 is a partially cutaway side view of the shower head in Figure 10, and Figure 12 is a cross section of the hollow cathode hydrogen ion generator used in the present invention. It is a diagram. Main wing 10...CVD reactor chamber 12.14...Turbo molecular pump 16...Load lock 18/Butterfly valve 20...Airtight water-cooled light housing 22...Gas manifold 24...Wafer 26 Pyrometer 28 BaF2 window 30 Moving arm 32 Gas box 34, 36, 38, 40 Gate valve 42 Tungsten halogen lamp 50 Aluminum oxide cup 52... Stainless steel face plate 54...tube cathode 56/cooling channel 58...exhaust gas

Claims (6)

[Claims] 1. A method of forming a blanket layer of titanium nitride on a semiconductor wafer, comprising: (b) loading the semiconductor wafer into a cold wall chemical vapor deposition chamber; (c) evacuating atmospheric gas from the chamber; and (e) removing the wafer. (f) flowing a process gas containing titanium chloride and ammonia over said wafer. 2. The method of claim 1, wherein the process gas includes a diluent selected from the group consisting of hydrogen and nitrogen. 3. 3. The method of claim 2, wherein the process gas is mixed at a temperature above 250<0>C prior to flowing it over the wafer. 4. Further, before the step (b), (a) Process of exposing the wafer to ozone 4. The method according to claim 3, wherein: 5. After the step (c) and before the step (e), further (
4. The method of claim 3, further comprising the step of: d) cleaning the surface of the wafer with ionized hydrogen gas. 6. A method of forming a blanket layer of titanium nitride on an anticonductor wafer, comprising: (a) exposing the wafer to ozone; (b) loading the semiconductor wafer into a cold-wall chemical vapor deposition chamber; and (c) removing an atmosphere from the chamber. (d) cleaning the surface of the wafer with ionized hydrogen gas; (e) heating the wafer to a temperature of 450°C or higher; (f)
Processing gas containing titanium chloride and ammonia at 250℃
(g) flowing said processing gas over a heated wafer.