二氧化硅薄膜的化学蚀刻

Some important factors that affect the dimensional  control of oxide films on silicon were studied. Both Nand P -type silicon with resistivities in the range of  0.014 to 200 ohm -cm and a (111) surface orientation  were employed in this experiment. The etching rates of  silicon dioxide in hydrofluoric acid (IIF) were studied  as a function of the concentration of HF, temperature,  and stirring speed. The experimental results show that  the etching rates varied directly with these variables,  but no difference in etching rate was found due to concentration or type of impurity in the silicon substrate  over the range studied.

The oxide layers on silicon used in this experiment were prepared by five different oxidation methods.  They are: wet oxygen, dry oxygen, steam, wet nitrogen  during diffusion of boron, and dry oxygen during diffusion of phosphorus. The etching rates of the oxide  layer grown by the above methods have the same average value except for the oxide layer grown in dry oxygen  during a phosphorus diffusion which has a much faster  etching rate.

The thickness of the oxide layers employed in this  experiment was determined by a multiple -beam interference  method. Comparisons of this method to other optical  interference methods were made. It was found that the  multiple -beam method was the most accurate of the four  interference techniques.

The oxide film on a silicon surface has the property  of masking against certain elements from diffusing into  the silicon. Therefore, it is widely employed in the  fabrication of semiconductor devices. With the development of integrated circuits, the precise geometric  control of the mask is becoming more important. The  dimensional control of the oxide film involves how the  oxide film is grown, and the method of sectioning.

In order to understand the important factors that  affect the dimensional control of chemical etching, it  is necessary to investigate the important variables involved and their effects on the kinetics of the etching  rate of the silicon dioxide layer. The results of this  study are concerned with how the concentration of the  etchant, the temperature, and the stirring speed affect  the etching rate. Also comparisons were made on the etching rates of silicon dioxide layers which were  prepared in different thermal oxidation atmospheres.  These are: wet oxygen, dry oxygen, steam, wet nitrogen  during a boron diffusion, and dry oxygen during a phosphorus diffusion. These studies are designed to clarify  some of the surface problems that are associated with  semiconductor device fabrication.

Both N -and P-type silicon with resistivities in the  range of 0.014 to 200 ohm -cm were cut into circular  slices about 10 mm in diameter, 0.60 mm in thickness,  and in (111) surface orientation. These samples were  mechanically lapped with 600 grit silicon carbide on a Beuhler polishing machine, and then polished to a mirror  finish with 1 micron alumina powder as a final abrasive.  Then they were degreased in hot acetone and rinsed first  in 48% hydrofluoric acid and then in deionized water.  Some of the samples were further chemically polished in  a solution consisting of 10 parts of concentrated nitric  acid to 1 part of 48% hydrofluoric acid to 6 parts of  glacial acetic acid. Thermally grown layers were prepared on the mechanically polished samples as well as  the chemically polished samples.

Fig1

For the thickness of the oxide films in the range of  about 2000 to 12,000 Á, the 95% confidence interval for  these four methods was evaluated from a set of six  samples by taking five readings from each sample for  each method. The 95% confidence intervals of the standard deviation of a normal distribution (2, p. 225) were  148<o<222 for the two -beam interference method with  metallized samples, 210<0<357 for the two -beam interference method with non -metallized samples, 171« <296  for the wedge method, and 71<a<126 for the multiple - beam method. Hence, the conclusion was drawn that the  multiple -beam method is most accurate, and therefore,  this method was employed in the present experiment.

The etching was carried out in a specially constructed apparatus which provided a close control of the  etchant temperature as well as a means for stirring the  solution at a constant rate. The etching solution was  contained in a 100 milliliter plastic beaker which was  held by clamps connected to a motor. Most of the beaker  was immersed in a tank of water which was kept at constant temperature. Figure 6 shows a schematic of the  apparatus.

The mechanically polished and chemically polished  silicon wafers- 'were= oxidized in, wet oxygen, then etched  together in 12% HF at a temperature of 25 °C and at a stirring rate of 100 rpm. The experimental results  indicated no detectable difference in etching rates for  the two types of surfaces.