Mamoru Maeda

Effect of fluorine in chemical‐vapor‐deposited tungsten silicide film on electrical breakdown of SiO2 film

The effect of fluorine in chemical‐vapor‐deposited tungsten silicide film on electrical breakdown of SiO2 film was investigated. Fluorine diffuses into the SiO2 film through the upper layer of poly Si above 800 °C. At 1000 °C, fluorine diffuses into the SiO2 film to a concentration on the order of 1020 cm−3. Electrical breakdown field of the SiO2 film degrades remarkably at 1000 °C. However, it was clear that the diffusion of fluorine was blocked by a thin chemical‐vapor‐deposited Si3N4 layer on the SiO2 film. In this case, the degradation of SiO2 film was not observed. From the above results, it is concluded that the diffusion of fluorine included in the chemical‐vapor‐deposited tungsten silicide film is one of the causes in degradation of electrical breakdown of the SiO2 film when the chemical‐vapor‐deposited tungsten silicide film was used as a gate electrode in metal oxide semiconductor integrated circuits.

Analysis of the effects of annealing on resistivity of chemical vapor deposition tungsten−silicide films

Chemical vapor deposition WSix films were formed with varying compositions and film thicknesses. The were films annealed in N2 for 30 min and their resistivities were measured. The maximum resistivity was obtained for annealing temperatures between 500 and 600 °C if the film was comparatively richer in tungsten and was thick. The film was analyzed by x‐ray diffraction and secondary ion mass spectrometry. The maximum resistivity is considered to be related to crystallization, which changes amorphous WSix into a mainly hexagonal structure, and to the difference in the conduction mechanism of electrons between amorphous, hexagonal, and tetragonal WSi2. It is clear that impurities in the film do not contribute to the maximum resistivity. There is also no relationship between resistivity and the number of grains of tetragonal WSi2 per unit length for annealing below 600 °C.

Analysis of stress in chemical vapor deposition tungsten silicide film

Stress in chemical vapor deposition (CVD) tungsten silicide film was studied by changing the flow ratio of SiH4 and WF6, deposition temperature, and annealing temperature. It was revealed that the stress is tensile on the order of 109 dyn/cm2. Stress is not significantly influenced by the flow rate of SiH4, but it is greatly influenced by the composition ratio of Si and W. As the film becomes Si rich, the stress is reduced remarkably, and after annealing its value peaks at 500 °C. This behavior was analyzed using an x‐ray diffractometer and secondary ion mass spectrometry (SIMS). Annealing behavior of the stress can not be explained solely by the Nakjima–Kinoshita model which indicates that increase of tensile stress is caused by grain size increase, but it can be fully explained by adding to the model an expression which takes the grain structure and orientation into account. A tensile stress increase accompanies reaction between the tungsten silicide film and Si substrate at a high temperature of 1100 °C.

Properties of Molybdenum Silicide Film Deposited by Chemical Vapor Deposition

films used as gate electrodes and interconnects were deposited on oxidized Si substrates via reactions between and . At high temperatures, generated by reactions involving , , and reacts with Si to form volatile , , , and ; thus Si is etched away and is not available to form , resulting in deposited film consisting of only metallic Mo. At low temperatures, the deposited film consists of which is thermodynamically stable. The deposited films show characteristics (resistivity, crystal structure, and so forth) similar to those of films deposited by sputtering. Chlorine in the deposited films has a gettering effect for mobile ions such as Na+.

Comparison of phosphosilicate glass films deposited by three different chemical vapor deposition methods

Properties of phosphosilicate glass (PSG) films deposited by the plasma chemical vapor deposition (CVD), low‐pressure CVD, and atmospheric CVD methods are compared. Stress behavior of plasma PSG films is almost the same as that of low‐pressure PSG films. The stress values change from compressive to tensile with increasing phosphorus concentration. On the other hand, the stress of atmospheric PSG films is tensile, and decreases monotonously with increasing phosphorus concentration. One of the causes is thought to be that the densities of plasma PSG films and low‐pressure PSG films are larger than that of atmospheric PSG films. Infrared absorption (IR) at P=O bonds is greatest in the atmospheric PSG films, lower in plasma PSG films, and lowest in low‐pressure PSG films. This phenomenon is attributable to the quantity of oxygen gas present at reaction time. The surface morphologies of the plasma PSG and low‐pressure PSG after etching are rough, but the surface morphology of the atmospheric PSG is very smooth. Also, it was found that the change in infrared spectra after temperature‐humidity treatment is smallest for the atmospheric PSG films. But the crack in the atmospheric PSG films is generated more easily than for the plasma PSG films or low‐pressure PSG films.

High-temperature stress measurement on chemical-vapor-deposited tungsten silicide and tungsten films

Stresses in chemical‐vapor‐deposited tungsten silicide and tungsten films at high temperatures were measured. Tungsten silicide films were formed from WF6 and SiH4 or Si2H6. Tungsten films were formed from WF6 and H2. The stress in tungsten silicide films is tensile and in the order of 109–1010 dynes/cm2. For a composition ratio of Si/W≤2.6, the stress of a film of more than 1000 A has a maximum at about 500 °C. On the other hand, for a composition Si/W>2.9, the stress has no maximum. The maximum of the stress is caused by crystallization of the film. The stress has two components. One component is related to the difference of the thermal expansion coefficients between the film and the Si substrate. Another is related to the film crystallization. It was found that the stress concentrates in the portion of the film nearest the substrate. The stress in tungsten films also reaches a maximum at 550 °C, similar to the tungsten silicide films. However, the cause of this behavior is not clear.

Selective Growth of Polysilicon

A chemical vapor deposition (CVD) technique for selective growth of polysilicon has been developed. This technique makes it possible to grow polysilicon only on the exposed silicon without a Si nucleus on the or mask. Perfect selectivity over the whole wafer surface was obtained in a system in the temperature range of 900°–1000°C under a pressure of 100 Pa at a significantly high line velocity. These conditions provide a perfectly selective silicon epitaxy technique, while addition of trichloroethylene as a gas that causes extreme deterioration of the crystallinity produced the selective polysilicon technique without a degradation of selectivity. A SEM observation showed that the grain size of the grown film was about 0.3 μm. SIMS analysis indicated that the concentration of carbon atoms incorporated in the film was on the order of 1020 cm−3. X‐ray rocking curve analysis derived a smaller lattice parameter of 5.425A, due to incorporated carbon atoms.

Changes in Resistivity and Composition of Chemical Vapor Deposited Tungsten Silicide Films by Annealing

Chemical vapor deposited (CVD) tungsten silicide films were formed by a cold wall reactor. These films were annealed in N2 to investigate changes in resistivity, composition, thickness, and impurity. The change in resistivity after 1000~ annealing becomes larger as the film reaches the stoichiometric value. A composition change occurs in a film whose composit ion Si/W is more than 2.6. Excess Si in the WSi, films (x > 2.6) is segregated in the boundary between WSi~. and poly-Si. A thickness change of about 15% occurs after 1000~ annealing at WSi~.4 on SiO~; this value is smaller than the calculated value. F and H, which are impurities in WSi, films decrease gradually and diffuse into gate SiO~ after 1000~ annealing. Progress of metal oxide semiconductor (MOS) large scale integrated circuits (LSI) is remarkably fast. Since the one kilo bit dynamic random access memory (1 Kbit DRAM) was developed in 1970, integration has advanced 4 times every 3 years. Now, one mega bit (1 Mbit) DRAM has been manufactured as a trial. The design rule for 1 Mbit DRAM is 1.2-1.3 #m, and the cell area has become very small: ranging from 20 to 35 /~m ~. For this reason, new capacitor s t ructures such as trench and stacked capacitors are used (1-4). TiSi~, TaSi~, and other refractory metal silicides with a base layer of poly-Si and refractory metals such as W are being used as interconnection materials. The reason for employing refractory metals or refractory metal silicides is that as the l inewidth becomes narrower and line length longer, resulting in high densification of devices, the signal propagation delay times become larger with the usually used poly-Si interconnection. Currently, there are two methods for forming these films: the physical vapor deposition (PVD) method, and the chemical vapor deposition (CVD) method. Among these methods, the CVD method is frequently used because of good step coverage. For example, we refer to the studies on deposition of WSi2 films by plasma CVD (5, 6). The low pressure chemical vapor deposited WSi, film, developed by Brors et al., however, is beginning to be widely used because of lower contamination and resistivity (7). Detailed reports on resistivity and capacitance-voltage characteris t ics have already been written. Generally, electrical characteristics degrade in the reaction between poly-Si and SiO~ during high temperature annealing (8). Consequently, it is necessary to study reactions and composit ion changes which include changes of F and H in the WSi, film by annealing; the mechanism of change also needs to be studied. Hara, et at. reported that reaction between WSi, film and the reaction between poly-Si and SiO.2 begins at 1000~ (9). We repor t on the changes in resis t ivi ty, composi t ion , and behav ior of F and H as well as the decreas ing fi lm th ickness after annealing. LOW PRESSURE WSi x CVD SYSTEM MASS FLOW F ~ CONTROLLER L WF6 He S i l l 4 t EACTORIuLF He VALVEI PUMP V E N T ~ RF Fig. 1. Schematic diagram of the cold wall CVD equipment Experimental WSi~ films were formed by cold wall CVD (see Fig. 1). WF6 and Sill4 were used as reaction gases. The flow rate of WF6 was fixed to 2 cmrVmin, but the flow rate of Sill4 was varied from 30 to 120 cm3/min. Helium was used as a dilution gas. The substrate temperature was changed from 325 ~ to 425~ The pressure was 40 Pa. The WSix films were deposited on the (100) plane, the poly-Si plane, and the oxidized plane of a 4 in. Si wafer. Film composition was controlled by changing the substrate temperature. Samples were annealed for 30 min in N2, therefore preventing 02 contamination. Film thickness w a s measured by Talystep after etching with HNO3:HF (ratio was 50:2). The resistivity was measured by a 4 point probe. Film composition was analyzed by Rutherford backscattering (RBS) method. The energy of the He ion was 2.275 MeV. The annealing behavior of impurities in these films was analyzed by secondary ion mass spectrometry (SIMS).

Properties of chemical vapor deposited tungsten silicide films using reaction of WF/sub 6/ and Si/sub 2/H/sub 6/

Tungsten silicide films were formed by the chemical vapor deposition method using the reaction WF/sub 6/ and Si/sub 2/H/sub 6/. The deposition rate, resistivity, composition, stress, crystal structure, and content of impurities were studied and compared with tungsten silicide films deposited by reaction of WF/sub 6/ and SiH/sub 4/. The tungsten silicide films made using Si/sub 2/H/sub 6/ have a higher deposition rate and higher Si concentration than those made by using SiH/sub 4/ at the same substrate temperature. For these reasons, the tungsten silicide films made by using Si/sub 2/H/sub 6/ were found to have a resistivity that is a little higher and, after annealing, a stress that is smaller than that made by SiH/sub 4/.

Noise reduction and recording density increase in magnetic‐coated disks

We studied the effects of several conditions on noise of medium of magnetic‐coated disks and succeeded in reducing the noise significantly. In our study, we focused on nonmagnetic particles and surface roughness and found that: (1) Nonmagnetic particles, like alumina, affect noise unless they are small and as well‐dispersed as the magnetic particles. (2) ’’As‐coated’’ roughness, meaning the surface roughness before polishing, causes the noise. By eliminating nonmagnetic particles and improving the coating method to reduce the ’’as‐coated’’ roughness, we produced extremely low‐noise magnetic‐coated disks. Using Co adsorbed iron oxide powder (Hc:650 Oe), we obtained high signal to noise ratio (SNR) (40 dB) magnetic‐coated disks at a high linear density of 20 000 FRPI.