Q2IC Technology

Diffusion and ion implantation are the two principal techniques used to introduce dopant impurities into a silicon wafer to form regions of controlled n-type or p-type conductivity, and while both achieve the same broad end goal, they differ fundamentally in mechanism, control, and practical characteristics.

Dry versus Wet Oxidation

Both dry and wet (steam) oxidation grow a layer of silicon dioxide on the silicon surface by reacting oxidizing species with silicon at high temperature, but they use different oxidant chemistries and produce oxides with different growth rates and quality characteristics. Dry oxidation uses pure, dry oxygen gas as the oxidant, following the reaction Si + O2 to SiO2 at typical temperatures of 900-1200 degrees Celsius; because the oxidizing species (molecular oxygen) diffuses relatively slowly through the growing oxide layer, dry oxidation proceeds slowly, but the resulting oxide is dense, has fewer defects, and exhibits a higher breakdown field strength and better interface quality, making dry oxidation the preferred choice for growing thin, high-quality gate oxides in MOS devices where electrical integrity is critical. Wet (steam) oxidation instead uses water vapor as the oxidant, following the reaction Si + 2H2O to SiO2 + 2H2, and because water vapor diffuses through the growing oxide considerably faster than molecular oxygen, wet oxidation proceeds at a substantially higher growth rate for the same time and temperature; however, the resulting oxide tends to be somewhat less dense and of lower electrical quality than a dry oxide, making wet oxidation the preferred choice for growing thicker field oxides used for device isolation, where growth speed and thickness are more important than the ultimate electrical quality of the oxide.

SiO2 Synthesis Process

Thermal oxidation of silicon is most commonly carried out in a horizontal or vertical resistance-heated tube furnace, in which a batch of wafers is loaded onto a quartz boat and inserted into the heated furnace tube, and the desired oxidant gas (dry O2, or O2 bubbled through or mixed with steam for wet oxidation, sometimes with a small percentage of HCl added to getter mobile ionic contamination and improve oxide quality) is flowed through the tube at the process temperature, typically in the range of 900 to 1200 degrees Celsius, for a duration determined by the desired final oxide thickness. The oxidation reaction occurs at the silicon-oxide interface, not at the oxide's outer surface, meaning the oxidizing species must diffuse through the already-grown oxide layer to reach the unreacted silicon beneath, and as the oxide thickens, this diffusion path lengthens, causing the growth rate to progressively slow down; this behavior is quantitatively described by the Deal-Grove model, which predicts oxide thickness growth following the relation x^2 + Ax = B(t + tau), where x is oxide thickness, B is the parabolic rate constant (diffusion-limited, dominant for thick oxides), B/A is the linear rate constant (reaction-limited, dominant for thin oxides), and tau accounts for any initial oxide thickness present before the timed growth interval begins. After the desired oxidation time has elapsed, the oxidant flow is stopped and the furnace tube is typically purged with an inert gas such as nitrogen while wafers are slowly cooled and withdrawn, minimizing thermal shock and stress in the grown oxide layer.

It is further worth noting that both dry and wet oxidation follow the same underlying Deal-Grove growth kinetics framework described above, differing only in the numerical values of the linear and parabolic rate constants B/A and B due to the different diffusivity and solubility of the oxidant species (molecular oxygen for dry oxidation, water vapor for wet oxidation) within the growing oxide layer; because water vapor diffuses considerably faster through SiO2 than molecular oxygen at a given temperature, wet oxidation's parabolic rate constant B is substantially larger than that of dry oxidation, explaining the empirically observed several-fold faster growth rate of wet oxidation compared to dry oxidation under otherwise identical time and temperature conditions, a difference that becomes especially pronounced once the oxide has grown thick enough to enter the diffusion-limited (parabolic) growth regime.

It is also worth noting that beyond the depth-and-dose independence advantage discussed above, ion implantation additionally offers superior batch-to-batch and wafer-to-wafer reproducibility compared to furnace diffusion, since the implanted dose is measured directly and precisely as an integrated beam current over the exposure time (a purely electrical measurement, largely insensitive to the small furnace-to-furnace or run-to-run temperature variations that inevitably affect a thermally driven diffusion process), which is one of the key reasons ion implantation has almost entirely displaced diffusion as the primary method of introducing dopants for source, drain, and well formation in modern high-precision CMOS manufacturing, with furnace diffusion now reserved mainly for drive-in and anneal steps following an initial implantation, rather than as the primary dopant introduction technique itself.