Tube diffusion

Tube or batch diffusion furnaces are a common industrial tool used for doping silicon wafers. Following a cleaning step, silicon wafers are loaded vertically onto a quartz carrier boat with equidistant spacing to allow for gas flow between wafers. By stacking wafers vertically in a boat, a lower limit is placed on wafer thicknesses below which the yield will decrease. There are three different forms of tube diffusion furnaces; solid source diffusion, gas source diffusion and the more common liquid source diffusion.

Solid Source Diffusion

In solid source diffusion, the boat carrying the silicon wafers is loaded into the diffusion tube alongside the solid source (e.g. SiP) comprising of a phosphorus and silicon oxide, in the instance of n-type diffusion [1]. The source can either be loaded in the boat with the wafers, or else in a separate platinum carrier. During the heating process, the phosphorus source evaporates producing P vapour. This vapour is transported via carrier gas and deposited uniformly on the surface of the heated silicon. Further heating allows the diffusion of the phosphorus atoms into the surface.

Gaseous Source Diffusion

Gaseous source diffusion is a process where a gaseous dopant source—such as phosphorus oxychloride, phosphorus pentoxide or phosphorus tribromide—is directly introduced with other carrier gases into the diffusion tube.

Boron diffusion using boron trichloride (BCl₃) is currently the standard method used in the photovoltaic industry to create p-type silicon regions. In this process, BCl₃ is introduced into a high-temperature furnace (800–1100°C) along with oxygen, where it reacts to form boron oxide (B₂O₃) on the silicon surface through the reaction

4BCl₃+3O₂→2B₂O₃+6Cl₂4​

The boron oxide acts as a diffusion source, allowing boron atoms to penetrate the silicon lattice. This creates controlled p-type doping, essential for forming emitters in solar cells. While the method provides precise doping and uniformity, the highly corrosive nature of BCl₃ and its by-products, such as Cl₂​, necessitates robust safety measures, including gas scrubbing and proper ventilation. Despite these challenges, BCl₃ diffusion is a very efficient technique for high-throughput manufacturing.

Liquid Source Diffusion

Liquid source diffusion is the most common form of diffusion process used in the industry. Commonly known as POCl₃ diffusion, the dopant source consists of a colourless liquid called phosphoryl chloride (or more commonly called phosphorus oxychloride). N₂ carrier gas is bubbled through the liquid POCl₃ and into the diffusion tube. At high temperatures, the liquid evaporates and reacts with O₂ gas to produce phosphorus pentoxide (P₂O₅) in the following reaction,

4POCl_3(g) + 3O_2(g)  → 2P₂O₅ + 6Cl_2(g)                   (1)

where g indicates that the molecules are in the gas phase. During the pre-deposition phase of the diffusion, phosphorus pentoxide is deposited on the surface of the silicon wafers in what is known as the infinite source. During drive-in, phosphorus pentoxide is reduced by the surface Si to form phosphorus which can be driven into the wafer. This reaction is outlined below.

P₂O₅ + 5Si_(s) → 5SiO_2(s) + 4P_(s)                                 (2)

where s indicates that the atoms or molecules are in the solid phase.

A video demonstrating the tube diffusion process at UNSW Sydney is shown below.

Post-diffusion cleaning

Following a phosphorous/boron diffusion, typically, a phosphorous/boron doped silicon oxide layer, known as phosphosilicate glass (PSG) or boron silicate glass (BSG), remains on the surface of the wafer. This layer should be removed prior to surface passivation, and is typically removed in a short wet-chemical etch consisting of an immersion in dilute hydrofluoric (HF) acid for 1-2 minutes, or until a hydrophobic surface is observed. The wafers will then undergo a rinse in de-ionized water to remove any hazardous HF acid.

Following this process, the samples may undergo further cleaning in an RCA cleaning procedure prior to passivation, although the wafers should still be considered ‘clean’ immediately after removal from the furnace.

Emitter etch back

The diffusion process typically takes place on both the front and rear of the wafers. The emitter contains a very high concentration of dopant atoms, which are not necessarily all electronically active as, e.g. the concentration may exceed the solid solubility limit. Therefore, it may be required to etch a few nanometers of an emitter layer that contains the inactive dopants. The etch-back process should ensure that emitter is not removed entirely. This process can be incorporated into PSG/BSG cleaning or edge junction isolation equipment. Removing the electrically inactive dopants is critical. This can reduce the dead layer, which can improve the J_sc and V_oc of the solar cell. Furthermore, the quality of the surface passivation by e.g., SiN_x:H can be improved as well.

Inactive dopants can be determined using secondary-ion mass spectrometry (SIMS) and Electrochemical capacitance-voltage (ECV) measurements where SIMS measure the total dopant concentration and ECV measures only the electrically active dopants. Comparison of the two measurements will represent the inactive dopants that are present at the surface.

[1] – Y. Murata, C. Mcmurtry, Solid diffusion sources for phosphorus doping, US3852086 A, 1974.