Semiconductor Rinsewater Requirements

Specifications for electronic rinse water requirements are growing progressively and becoming more demanding as line widths that characterize silicon chip circuitry become finer and the electrical resistances required of their devices escalate. As time passes improvements in water purification have become more defined. Especially as newer more reliable measurement devices allow for better more reliable measurements and new means to improve water purity become available. High purity water quality specifications are often the limits of what is attainable at a given time. They do not totally reflect the bounds that are present, but rather they are the best levels currently possible and as a result must be tolerated. This is what is meant when using the term "state of the art." As time passes and progress occurs, specifications can be improved upon and they are. As a result there is a constant drive for ideal high purity water. However, in certain contexts its attainment is necessary to certain semiconductor devices and can impose an unnecessary expense and burden.

The American Society for Testing and Materials (ASTM) differentiates the purity of electronics grade water into four different types as shown in Table 1 below.

It should be noted that as the line widths become become thinner the specific numbers and sizes of the particles become smaller. Also, live bacteria which is a special type of particles due to their replicating power are meeting with increasingly restrictive specifications. Along with the concern for particles is a circumscription of the already meager tolerance accorded lower electrical resistivities. So, while an E-IV water may exhibit a resistivity value of of 0.5 microsiemens-cm at 25^oC, the E-I water must not have a electrical resistivity below 17 microsiemens-cm and should have a resistivity value of aboce 18 microsiemens-cm 95% of the time.

Of all the listed cation impurities, electronics grade water defined y ASTM call for decreases of potassium from 5 micrograms per liter to 2 micrograms er liter for the E-IV to E-I types. Sodium goes from 2 micrograms per liter to 0.5 ,micrograms per liter, zinc from 5 micrograms per liter to 0.5 micrograms per liter and lastly nickel from 2 micrograms per liter to 0.1 micrograms per liter.
Analytical lab specifications as it relates to the line widths of semiconductordevices and the particle sizes and numbers appropriate to them can be seen in Table 2 below.

(Organics should be regulated pre and post UV; breakdown by UV may yield ionic species such as Cl ^-, Br^-.)

(Organics should be regulated pre and post UV; breakdown by UV may yield ionic species such as Cl ^-, Br^-.)Line widths of 0.8 microns and (Metal Oxide Subtrate: metal nano material) MOS-1=2 are specified at less than 500 per liter of sizes 0.08 microns and 0.3 microns respectively. Line widths of 0.5 microns and 0.3 microns even though there are no numbers specified their sizes are to be less than 0.05 microns and 0.3 microns respectively. For lines of 0.3 microns, bacteria living and dead are to be a total of zero both at filters and points of use. The 0.5 micron line width allows for fewer than 2 per 100 mL at the points of use, but none at the final filters. For MOS 1=2 widths have slightly higher counts but are less than 10 per 100 mL. The number of living and dead bacteria stipulated for the 0.8 micron line width is less than 3 per 100 mL at the final filter and less than 10 at the point of use. 

Laboratory specifications limit the organic concentrations at both the final filters and point of use to less than 20 ppb for MOS 1=2, dropping to less than 1 ppb for the 0.3 Micrometer width. For silica concentrations in the range MOS 1=2 at 8 micrometers, 0.5 micrometers and 0.3 micrometers are specified to be below 3 ppb, 0.5 ppb, 0.1 ppb and 0.05 ppb respectively.


In regards to water testing, critical testing is not limited to just particles, but to all water testing. This includes tests such as EPI bacteria counting for living and dead bacteria, measuring ions in solutions both metallic and non metallic even down to the ppt levels and measuring as well identifying organics at less than the 10 ppb range. Critical testing should not be limited just final filter and point of use. There are new tests that are beginning to reveal the use of critical testing that is useful for following wafer cleanliness. The required quality standards for pure water is way beyond present testing capabilities. New procedures are being developed. This is occurring because there is more cooperation between vendors, process engineers, facility personnel and the analytical laboratory.

High purity semiconductor-grade water should routinely monitored for bacteria, TOC, resistivity, silica, particulates, trace ions, and residual solids. The system should be inspected weekly at makeup, polishing, and point of use stations. In order to see how stringent the standards have and continue to become over time for electronic rinsewater see Table 3 below that shows comparisons for 1975, 1980, and 1985.

In 1985 the bacterial level that could be achieved was said to be zero per 100 ml. At the time fewer than 6 cfu/100 ml was considered acceptable. In 1975 and 1980 the bacterial levels that were acceptable were 20 cfu/100 ml and fewer than 20 cfu/100 ml respectively. Using Table 4 this information can compared to that of the SEMI (Semiconductor Equipment Manufacturing Institute) recommended bacteria levels for fewer than 6 bacteria per 100 mL in 1985.

It is of interest to note that Table 5 shows the quality of the high purity water produced by three systems at Tohoku University in Japan in 1991. This resulted in new suggestions for an updated version of high-purity water needed for the semiconductor industry as seen in Table 6. These suggestions can be compared with the requirements for high-purity water set forth at the time in Table 7. Notice that the requirements suggested Table 6 are slightly more stringent.

It is evident that when comparing the tables that the trend towards ever lower impurities in water at all levels it evident. As means to measure specific impurities becomes more reliable and as methods of control and removal become more efficient the maximum levels for the presence of these impurities will continue to decrease. No matter if driven by advances in attainability or out of applicational necessity the quest for pure semiconductor rinsewaters will continue as the technology for ultrapure water grows and as technology as a whole grows.