No abstract

Advanced electronic devices currently have a serious problem in their fabrication process due to the presence of organic hydrocarbon contamination on the silicon wafer surface. [1][2][3][4][5][6][7] To develop the technology to achieve a sufficiently clean silicon wafer surface, many kinds of organic species existing on the silicon wafer surface have been studied, such as propionic (propanic) acid ester, 8 trimethyl silanol, 9 hexamethyl disiloxane, 9 cyclosiloxanes (D3-D11), 8-10 trichloroethyl phosphate, 9,11 and phthalates 8,11 [di(2-ethylhexyl)phthalate (DOP) and dibutyl phthalate]. Some of these organic species are considered to be physisorbed 11 from the clean room air 8,12 to the silicon wafer surface. Chemisorption of organic species, such as 1,4-cyclohexadiene, 13 cyclopentene, and cyclohexene, 14 on a silicon (100) surface also has been reported.Recently, several studies 8,11,[15][16][17][18] have reported the existence of a time-dependent change in the concentration of organic species on a silicon wafer surface. Some organic species rapidly show a peak of the surface concentration on the silicon wafer surface; afterward they tend to decrease, indicating gradual replacement by the other organic species. Since organic species seem to compete for the adsorption sites on the silicon wafer surface, this behavior is called the "fruit basket phenomenon." 8 Additionally, their surface concentrations have been reported to depend on the condition of the silicon wafer surface. 8 Although the mechanism of this time-dependent phenomenon has been discussed using the heat of adsorption, the heat of vaporization, 8 the boiling point of the organic contaminant, 18 the polarity of the silicon surface, 18 the sticking probability, and the sticking coefficient, 11,19,20 this phenomenon has not been theoretically expressed in a time-dependent form. Since the analysis and prediction of the fruit basket phenomenon are necessary to advance silicon crystal technology, a theoretical model should be developed.For the first time, this study clarifies the mechanism of the timedependent change in the surface concentrations of organic species, developing the model of multicomponent organic species adsorption-induced contamination (MOSAIC), which is an application of rate theory for organic species contamination on a silicon wafer surface. Using the MOSAIC model, this study demonstrates that the adsorption rate and the desorption rate of organic species on the silicon wafer surface are the dominant mechanisms of the fruit basket phenomenon. Numerical Calculation ModelTo describe the fruit basket phenomenon of various organic species on the silicon wafer surface, the rates of the change in the surface concentrations of the organic species are expressed based on the single-component Langmuir-type expression [21][22][23] [1]where is the surface coverage of the silicon wafer surface with the organic species, c is the gas-phase concentration of the organic species, t is time, and k ad and k de are the rate constants of adsorption and deso...

The time-dependent change in concentrations of organic species on a silicon wafer surface, the fruit basket phenomenon, is studied using a rate theory for a multicomponent system consisting of a large number of organic species. The theoretical model and the rate parameters evaluated in this study are shown to be effective for predicting the trend and the abundance of nine organic species on the silicon wafer surface in a clean room, because the trend in organic species are classified using the desorption rate constant and because their concentrations in a steady state have a clear relationship with the ratio of the adsorption rate to the desorption rate constant.The presence of organic hydrocarbon molecules adsorbed on a silicon wafer surface is widely known to cause a serious problem 1-7 of airborne molecular contamination in the advanced electronic device fabrication process. In order to achieve a sufficiently clean silicon wafer surface, two kinds of studies are needed. The first is for clarification of behavior and the influence 8-12 of the organic species existing on the silicon wafer surface. The second is for measuring and controlling the concentration of the organic species in the clean room air, since these organic species are considered to be mainly adsorbed from the clean room air 8,11,12 onto the silicon wafer surface.Various organic species currently known to exist on the silicon wafer surface 8-11 show a time-dependent change in their concentrations on the silicon wafer surface, which is called the fruit basket phenomenon. 8,11,13-16 Some organic species rapidly show a peak in their surface concentration on the silicon wafer surface; it decreases later, indicating gradual replacement by the other organic species. Therefore, organic species seem to compete for the adsorption sites on the silicon wafer surface. The fruit basket phenomenon has been discussed from various viewpoints, such as the condition of the silicon wafer surface, 8 the heat of adsorption and vaporization, 8 the boiling point of the organic contaminant, 16 the polarity of the silicon surface, 16 the sticking probability, and the sticking coefficient. 11,17,18 However, an appropriate model for the fruit basket phenomenon has not been developed, due to the lack of desorption concepts.In order to describe the decrease in a manner similar to the increase in the concentration of the organic species on the silicon wafer surface, the rate of the desorption and the adsorption should be simultaneously taken into account. Therefore, in our previous study, 19 the mechanism of the fruit basket phenomenon has been discussed using the model of multicomponent organic species adsorption-induced contamination ͑MOSAIC͒, which is an application of rate theory accounting for the adsorption and the desorption of organic species on the silicon wafer surface. The influence of the condition of the silicon wafer surface was also evaluated using the MOSAIC model.Since our previous study 19 discussed the fruit basket phenomenon limited for a small system comp...

The etch rate, chemical reactions and etched surface of -silicon carbide are studied in detail using chlorine trifluoride gas. The etch rate is greater than 10 mm min À1 at 723 K with a flow rate of 0.1 ' min À1 at atmospheric pressure in a horizontal reactor. The maximum etch rate at a substrate temperature of 773 K is 40 mm min À1 with a flow rate of 0.25 ' min À1 . The steplike pattern that initially exists on the -silicon carbide surface tends to be smoothed; the root-mean-square surface roughness decreases from its initial value of 5 mm to 1 mm within 15 min; this minimum value is maintained for more than 15 min. Therefore, chlorine trifluoride gas is considered to have a large etch rate for -silicon carbide associated with making a rough surface smooth.

The silicon etching rate by chlorine trifluoride gas is systematically studied using a reactor having a very small cross section above the silicon substrate and achieving a very high efficiency of etchant gas consumption and very large etching rate, larger than 20 m min Ϫ1 . The silicon etching rate is shown to be proportional to the flow rate of the chlorine trifluoride gas. However, this rate is, for the first time, found to be independent of the initial silicon substrate temperature. This study shows that the silicon substrate is automatically heated to the temperature determined by the balance of the reaction heat and the heat transport in the reactor. Since this temperature increment processes an extremely large-surface chemical reaction rate, the etching rate is governed by the transport rate of the chlorine trifluoride gas. This study concludes that a high efficiency silicon etching by chlorine trifluoride gas is possible without any supplemental heating.Chlorine trifluoride (ClF 3 ) gas has a very high reactivity for various materials. 1-5 This gas is especially suitable for plasmaless etching 2,6-10 near room temperature at atmospheric and reduced pressures. In silicon crystal technology, chlorine trifluoride gas has been known to be used for the in situ cleaning 2,11,12 of a chemical vapor deposition ͑CVD͒ reactor in order to remove any polysilicon film deposited on the susceptor and on the inner wall of the chamber.For the further development of new industrial etching and cleaning processes using chlorine trifluoride gas, its chemical reaction should be systematically studied. For this purpose, our previous study 13 reported the chemical reaction between a silicon surface and chlorine trifluoride gas in ambient nitrogen at atmospheric pressure. It also reported that chlorine trifluoride gas has been shown to work as a source of active fluorine atoms to form inorganic fluorides, for example, silicon tetrafluoride for silicon etching. However, this previous study was performed only to evaluate the overall chemical reaction and the produced gas species. The other fundamental properties of the chemical reaction by chlorine trifluoride gas, such as the etching rate and the rate-determining parameters, have, unfortunately, not been studied.Therefore, in this study using the chlorine trifluoride gas in the CVD reactor designed for achieving the industrially applicable highperformance process, the silicon etching rate and its ratedetermining parameters are experimentally evaluated. ExperimentalIn order to etch silicon by chlorine trifluoride gas, the horizontal cold-wall CVD reactor shown in Fig. 1 was used. This reactor consists of a gas supply system, a quartz chamber, and infrared lamps. A 30 ϫ 50 mm silicon substrate is horizontally held on the bottom wall of the quartz chamber. The silicon substrate is cut from the n-type ͑100͒ 200 mm diam semiconductor silicon wafer, which was grown using the Chzochralski method.The silicon substrate is heated by infrared rays from the infrared lamps through the transpar...

This study theoretically and experimentally evaluates the role of the coexisting organic compounds on the time-dependent airborne organic contamination on a silicon wafer surface in the cleanroom air. The maximum contamination and the ratio of the desorption to the adsorption are independently described from each other, using simple equations consisting only of the surface concentrations of the organic compounds on the silicon wafer surface. These parameters are consistent with the experiment using the silicon plate sampling method. Additionally, the suppression of the increase in the surface concentration of bis͑2-ethylhexyl͒phthalate theoretically predicted due to the coexisting organic compounds is experimentally observed. The time-dependent behavior of the airborne organic contamination is concluded to occur following the simple rate theory.

A method, called the silicon plate method, has been developed using a small sampling device with a clean simple process, in order to directly evaluate organic contamination on a silicon wafer surface that came from the cleanroom air. Using this method, the concentration of bis͑2-ethylhexyl͒phthalate on the silicon wafer surface is experimentally shown, for the first time, to reach a steady state which has a relationship with its concentration in the cleanroom air. The experimental results are consistent with those theoretically predicted using the model of multicomponent organic species adsorption-induced contamination; therefore, the silicon plate method is concluded to be effective for evaluating the time-dependent behavior of organic species on the silicon wafer surface.The presence of organic hydrocarbon molecules adsorbed on a silicon wafer surface is widely known to cause a serious problem 1-7 of airborne molecular contamination in advanced electronic device fabrication processes. Various organic species [8][9][10][11][12] have been reported to show a time-dependent change in their concentrations on the silicon wafer surface, this is called the fruit basket phenomenon. 11-17 Some organic species show a sharp peak in their surface concentration on the silicon wafer surface which afterward, tends to decrease, indicating gradual replacement by other organic species. Therefore, organic species seem to compete for the adsorption sites on the silicon wafer surface. In order to experimentally evaluate such timedependent organic contamination on the silicon wafer surface, an appropriate and practical sampling and measurement method is required.The organic contamination on the silicon surface is ordinarily measured using wafer thermal desorption gas chromatograph mass spectrometry ͑WTD-GC-MS͒ 18,19 comprising the following steps: (i) thermal desorption of the organic species from the contaminated silicon wafer surface, (ii) transport of the organic species via the gas phase using a carrier gas stream, (iii) collection of the organic species by an adsorbent solid trap ͑Tenax͒, (iv) thermal desorption of the organic species from the adsorbent solid trap, (v) transport of the organic species to the trap at the entrance of the gas chromatograph mass spectrometer, and (vi) transport of the organic species from the trap to the gas chromatograph-mass spectrometer by heating the trap.However, because the fragments caused by the thermal decomposition of the organic molecules have been detected, 18 the high temperature steps in the WTD-GC-MS method are believed to decrease the yield of the organic species and to make the quantitative analysis complicated. Additionally, and unfortunately, in the complicated and large equipment for the WTD-GC-MS method having the cold wall of the infrared furnace and a very long tube connecting the furnace with the adsorbent solid trap, the organic species thermally desorbed from the silicon wafer can be trapped. Therefore, this ordinary method is considered to have some problems in accuracy.Fo...

Rapid thermal processing (RTP) using radiative heat transfer has advanced after the early studies 1,2 and is currently a very popular technology. It is widely used for many applications in semiconductor manufacturing processes including chemical vapor deposition (CVD) on silicon substrates. Although a new modification of RTP has been developed 3 and RTP has been extended for various materials including gallium arsenide, 4 major problems in the current RTP system or process still exist, such as thermal budget, 5 temperature reproducibility and uniformity 6 which lead to nonuniform film thickness, slip lines, and warpage of the silicon substrate. [7][8][9][10] The reproducibility and the uniformity of the substrate temperature are affected by parameters 6 such as (i) sensing and control of substrate temperature, (ii) process chamber (substrate properties, film properties, dimensions, shape, wall material properties, gas flow dynamics, etc.), and (iii) heat source (size, shape, location, type, reflector set). For (i), temperature measurement systems have been developed by many researchers. 11-18 For (ii) and (iii), transport phenomena including gas flow dynamics and film formation rate in the RTP system have been intensively studied 8,19-25 using numerical calculations, in which the substrate temperature and its uniformity were assumed as the boundary conditions.For further improvement of these numerical calculation models and for obtaining a flat temperature profile by optimizing the radiation of heat from the heat source, a model based on the physics of the RTP system is required. 26 In particular, the numerical model for evaluating the infrared radiation heat should be applicable to the complicated three-dimensional geometry of a RTP system. For this purpose, many researchers have studied models for evaluating the temperatures of the silicon substrates in the furnaces for oxidation and diffusion by accounting for the diffusive reflectors, infrared radiation heat, and the reflections between the walls of the furnace and the substrates. 7,9,12,[27][28][29][30][31][32] However, very few models which take into account the three-dimensional configuration of the RTP system using reflectors, tungsten/halogen filament lamps, and a silicon substrate have been discussed, 26 because of the difficulty in accounting for the view factor under the complicated reflections of the rays in the RTP system. 7 Although an image source of the lamp has been used to account for the rays reflected at the surface of the specular reflector near the real source of the lamp, the image construction is not practical in three-dimensional systems. 7 Additionally, the effect of the second or later reflections at the polished surface of the silicon substrate has not been evaluated, since the rays after the reflections at the surface of the silicon substrate have been unfortunately assumed to show only a negligible effect because of the small reflectivity. 32 Therefore, a model which is able to account for these factors should be developed.For evalu...