How clean is clean?

Reducing the possibility of physical contamination from particulate matter is vital for ultra-close tolerance micro-machines. Yet most people forget that it is now a routine requirement in manufacturing the high capacity magnetic hard disc data storage that we take for granted in our desktop and portable PCs – until they fail, that is!

The problem with particulate contaminants is not only their physical interference with sliding or contacting components but the fact they are not chemically inert. This means that micro-chemical reactors can get incorporated into safety critical components. While the particles remain in a stable condition they are of no consequence. However, if the local environment and conditions change, then this is where the trouble starts.

Conducive conditions

Emphasising the pervasive influence of chemically reactive particulate contamination, Fig. 1 shows the result of leaving a single grain of salt on a sterile surgical steel surface exposed to the atmosphere. It is well known that salt solution and oxygen from the air will combine to corrode iron. However, a dry salt crystal (which is not hygroscopic or deliquescent) will have no immediate effect on the steel until a single episode of moisture condensation changes the situation dramatically. Even at normal comfort humidity levels the capillary at the contact point between the salt crystal and the steel encourages moisture condensation. The tiny amount of concentrated saline solution formed then attacks and breaks down the thin surface oxide on the steel, forming iron chloride and oxychloride compounds, and leading to pitting corrosion.

This reaction leaves the sodium ions in sodium hydroxide form, with nowhere to go. The moisture absorbed by this extremely deliquescent compound causes further dissolution of the salt crystal and so promotes a continuing reaction. The image shows the central dark spot of mixed iron oxide/chloride "rust" surrounded by liquid droplets where the sodium hydroxide has spread far from the site of the original salt crystal. The liquid droplets are stable at an ambient relative humidity of only 50%.

Cleanroom filtration can remove particulates down to very low levels and small sizes, but this does not mean that the components assembled or produced will be free from particulate contamination. Although the above example of salt on steel may seem contrived and not relevant to the semi-conductor industry, for example, it is quite common to find minute crystalline deposits on silicon samples prepared in cleanroom conditions. These are not due to airborne particulates evading the filtering systems. They almost invariably arise from drying stain residues where proprietary cleaning and etching solutions have not been rinsed thoroughly. The dissolved material has nowhere else to go as the water evaporates, so even very dilute solutions eventually dry to form solid deposits.

Particles down to microns in size can be seen by microscopy inspection. Monolayer contamination can be just as deleterious but is impossible to see. In such cases surface analytical techniques using ion or electron spectroscopies are then necessary.

In general, these techniques operate in ultra high vacuum spectrometers and so only compounds that have a low vapour pressure, or at least some degree of bonding to the surface, can survive to be analysed. Unfortunately, these are the very species that are difficult to remove and are most usually the culprits in technological problems.

X-ray photoelectron spectroscopy – XPS – involves irradiation of a sample with low energy x-rays to excite core level electrons to be emitted which are element specific. Auger electron spectroscopy is a similar process which uses a focused and rastered electron beam to image the sample and excite electrons. Peaks in the electron energy spectrum allow the elements present to be quantitatively identified and, in most cases, some chemical bonding information can be obtained. Almost all the signal detected originates from within 5-10nm of the outermost surface. This analysis volume would contribute barely 1% of the total x-ray signal obtained by SEM/EDX analysis of the same sample and so would probably be missed in most cases.

Accurate analysis

Time of flight secondary ion mass spectrometry – ToF-SIMS – uses a focused and rastered ion beam to image the sample and to sputter material from the surface monolayers. The incident ion dose is so low that very little damage occurs and intact organic molecular fragments are usually emitted. This technique is remarkably sensitive (to ppb levels) but is not inherently quantitative. Accurate mass analysis of the ion fragments produced can define quite precisely the molecular compounds present on the surface.

The combination of the complementary XPS/AES and ToF-SIMS techniques can therefore provide an extremely detailed analysis of material that could otherwise not be seen on the surfaces of samples. By increasing the incident ion dose rate, at the expense of producing chemical damage in the sample, it is possible to also determine the elemental distribution into the sample by depth profiling. XPS/AES depth profiling or Dynamic SIMS (DSIMS) can be used but DSIMS has a far greater elemental sensitivity and is particularly suited to identifying contaminants present at trace levels.

CSMA has used these and other related techniques for many years to characterise contamination that arises on client's surfaces, despite the components being prepared using stringent cleanroom conditions.

Fig. 2 shows an example where dynamic SIMS depth profiling has been used to discover traces of calcium and magnesium in a critical insulating gate oxide layer in a semi-conductor device. The presence of these ionic species in the barrier oxide causes electrical leakage and leads to failure of the device in service if the contamination exceeds a certain level. Random variation in contamination levels means unpredictable failure of the devices, which of course cannot be tolerated by the end user. Calcium and magnesium are common contaminants of domestic tap water, suggesting that distilled or de-ionised water has not been used in a rinsing stage in the process.

Fig. 3 shows ToF-SIMS spectra obtained from two silicon wafers cleaned using different proprietary cleaning solutions. These no doubt remove particulates but leave behind the telltale ⁵⁵C₃H₃O-, ⁷¹C₃H₃O₂- and ⁸⁵C₄H₅O₂- signatures of acrylate/methacrylate residues which are not removed by either product.

Fig. 4 shows ToF-SIMS spectra obtained from two silicon wafers. Heat treatment is required to drive off some organic residue to improve the adhesion of photoresist. The spectra show ⁴⁴(C₂H₆N⁺) and ⁶²(C₂H₈NO⁺) fragments characteristic of one of the compounds used earlier in the processing.

To summarise, it is clear that using cleanroom technology can visibly reduce interference from particulate contamination in critical manufacturing processes. However, it should be remembered that there are other ways in which particulate and monolayer contamination can be introduced within the cleanroom environment. The origin of this contamination can often be obscure but surface analytical techniques are ideally suited to tracking down the source and helping to alleviate the problems that arise further downstream as a consequence.