Additional Process Info
ADDITIONAL PROCESS INFO
About the "RCA" clean (from the MIT MTL website)
"Contaminants present on the surface of silicon wafers at the start of processing, or accumulated during processing, have to be removed at specific processing steps in order to obtain high performance and high reliability semiconductor devices, and to prevent contamination of process equipment, especially the high temperature oxidation, diffusion, and deposition tubes. In 1970, the RCA Laboratories developed a cleaning procedure for silicon semiconductor device fabrication technology, which has become the industry standard; it uses several reagents containing hydrogen peroxide.
"The RCA cleaning procedure has three major steps used sequentially:
I. Removal of insoluble organic contaminants with a 5:1:1 H2O:H2O2:NH4OH solution (SC-1).
II. Removal of a thin silicon dioxide layer where metallic contaminants may accumulated as a result of (I), using a diluted 50:1 H2O:HF solution.
III. Removal of ionic and heavy metal atomic contaminants using a solution of 6:1:1 H2O:H2O2:HCl (SC-2).
"The RCA cleaning technique does not attack silicon, and only a very thin layer of silicon dioxide is removed (in II) in the process. The procedure was also designed to prevent re-plating of metal contaminants from solution back to the wafer's surface."
At SNF, for ease of use, we use a slightly different formulation of SC-2 (5:1:1). And for cleans which precede film deposition, the order is switched so that the last dip is the 50:1 HF, to minimize any native oxide that might form.
5:1:1 H2O:H2O2:NH4OH, SC-1
This solution is heated to 50 degree C. At this station, this dip serves the same purpose as the first SC-1 step of the RCA standard clean process, in removing any trace organics.
5:1:1 H2O:H2O2:HCl, SC-2
This solution is heated to 50 degrees C. This
HCl clean is a variant of the SC-2 step of the standard RCA clean procedure
(the standard SC-2 uses slightly different proportions). This formulation is
excellent for removing trace metal cations from silicon surfaces.
There are several varieties of HF-based etchants, different in acid strength and in composition. HF-based etchants include the BOE etchants, Pad etch, and ammonium fluoride (NH4F). Before working with any HF-based etchant, read the information and the links provided in the SNF Lab Manual Part II about HF.
HF/Water Mixtures: 50:1 HF and 49% HF
SNF stocks two "straight HF" mixtures (which contain only HF and water): 50:1 HF and 49% HF.
Concentrated hydrofluoric acid is approximately 49% HF and 51% water. 50:1 HF is approximately 2% HF in water, about 25-fold less than 49%. Do not confuse these two (yes, this mistake has been made). The 49% HF bottles should have ‘HF’ handwritten on the cap.
The etch rate of thermal oxide in 50:1 HF is generally nominal (about 50 angstroms/minute.) Because this acid is not buffered, the etch rate may vary with acid lifetime or usage.
Buffered Oxide Etchants: BOE
BOE is the acronym for "Buffered Oxide Etch", which is a mixture of ammonium fluoride, HF, and water. Ammonium fluoride is normally a solid with a low temperature of sublimation, but is very soluble in water (concentrated ammonium fluoride is approximately 40% by weight in water.) In the BOE etchants, the ammonium fluoride acts as a buffer, maintaining the pH of the solution which keeps the etch rate stable/constant over time. High concentrations of ammonium fluoride are used in BOE; in fact, the total fluoride ion content is nearly that of concentrated 49% HF, and so BOE etchants are considered to pose the same toxic hazards as 49% HF.
The temperature of BOE etchants is not controlled in SNF. Depending on the manufacturer, BOE acids may also have an added surfactant to help circumvent surface tension, which can prevent etching in small geometries or in areas with high aspect ratios. The exact formulations of BOE's are generally proprietary, but here is a general summary is below.
20:1 BOE is approximately 20 parts of 40% ammonium fluoride and 1 part of 49% HF. Thus, 20:1 BOE is approximately 38% NH4F, 2.5% HF, and 60% water.
The etch rate of thermal oxide is approximately 300 angstroms/minute.
6:1 BOE is approximately 6 parts of 40% ammonium fluoride and 1 part of 49% HF. Thus, 6:1 BOE is approximately 34% NH4F, 7% HF, and 59% water.
The etch rate of thermal oxide is approximately 900 angstroms/minute.
This solution consists of 90% concentrated sulfuric acid (H2SO4) and 10% hydrogen peroxide (H2O2) and is heated to 120 +/- 10 degrees C. Sulfuric acid is about 95%-98% pure; the hydrogen peroxide in the lab is 30% in water.
This combination is excellent for removing organics. The sulfuric acid converts organic compounds to elemental carbon (which is why the solution may darken temporarily when loaded with photoresist). The peroxide then oxidizes the carbon to carbon dioxide and water (which is why the solution boils and fumes, and eventually clears again.) When the piranha mixture has been around for a while or has been used extensively, the hydrogen peroxide all turns to water, which is a lousy oxidant in this system. So, additional hydrogen peroxide can be added on an as-needed basis to increase the active lifetime of the piranha. However, for the temperatures that we run, additional peroxide can help only so much before the acid becomes diluted and then needs to be changed. For regular usage levels in our lab, a change frequency of about once/week is generally sufficient.Wafers with photoresist or if they have just been scribed must undergo a 20 minute clean in the 90% sulfuric before proceeding to pre-diffusion or pre-deposition cleans.
Photoresist can become very difficult to remove if the wafers have undergone some plasma-hardening process, including plasma etch, ion implant, or extended hardbake. If your wafers have undergone one of these treatments, it is highly recommended that the wafers be processed first in an oxygen plasma (such as the gasonics (clean) or matrix (contaminated)) in order to remove the plasma-hardened resist layer. Usually, any resist which may remain after this treatment is not hardened and can be readily removed in piranha. If your devices are sensitive to ion bombardment, you may opt not to use oxygen plasma, but then resist removal may prove problematic.
Thick photoresist (such as SPR220 or anything thicker than a couple of microns) is very quickly attacked by sulfuric acid, and as the oxidation/gas generation process takes place, the bubbles generated can lift the wafers up out of the cassettes so that they end up in the bottom of the bath. Removing the wafers requires cooling the hot pot, which will take several hours. To avoid this, try to process wafers with thick photoresists (such as SPR220) through an oxygen plasma (gasonics or matrix) prior to running through piranha. Alternatively, you can place an empty, clean cassette gently on top of the wafers in your cassette, to hold them in place. A plasma clean step is preferred; thick photoresist often lifts off in thick blackened sheets, which is not particularly attractive and can only be removed with addition of lots of peroxide.