Cleanroom suitable materials
Testing and classification methods

To support industry in manufacturing equipment and components for use in clean areas of manufacturing, there is a need for construction materials which fulfill the required cleanliness specifications. To date no lists of materials are available worldwide which take the specific demands of clean manufacturing into account. As a result, equipment manufacturers and cleanroom fitters are forced to select cleanroom-suitable materials purely from a visual point of view using subjective criteria which have not been scientifically proven. An attempt is now being made to remedy this deficit by introducing standardized tests and assessment methods for evaluating the cleanroom suitability of materials.

1 Introduction

The quality requirements of all products are increasing constantly. As a result, a steadily growing number of industrial fields and associated products manufactured demand clean production methods. All contamination factors relevant to the product need to be controlled in such “clean productions”. However, in many industrial fields, the avoidance of particulate contamination is especially important, e.g. in the semiconductor industry, the food industry and in pharmaceutics. The processes resulting in the emission of particles from production utilities are those of friction and vibration. Frictional processes induced by relative movements between the surfaces of production utilities are the principle cause. To date, there are no scientifically-sound representation models which generate particles from the frictional pairings of technical surfaces. Such a model would enable appropriate materials to be selected for clean manufacturing purposes right from the start of the development phase of components. There is no known research work in existence which clarifies correlations between the technical properties of material pairings and the generation of contamination, especially that of airborne particles from production utilities. In order to be able to group the contamination behavior of basic materials or surface coatings, a procedure for determining the cleanroom suitability of material pairings needs to be developed. 

2 Prioritizing types of contamination 

As well as electrostatic contamination behavior or the outgassing of highly volatile organic substances, the emission of airborne particles in the micron or sub-micron range is also especially relevant. If the human source of particles is ignored, particulate emission makes up approx. 30% to 40% of all the undesired contamination occurring in a cleanroom, thus having the greatest influence on product contamination. In a representative survey where approx. 270 companies were questioned, results showed that particulate contamination was the most important form of contamination as far as research requirements are concerned. Two thirds of all material requirements demand the presence of only a few particles or even the total absence thereof; the surface quality of the materials used is closely related to the emission of particles (see Fig. 1). The release of particles from production utilities is caused either by material vibration or by the relative movements of at least two materials rubbing against one another. The surface systems of production utilities considerably influence the generation of particles due to frictional pairing of the components of these utilities.

3 Aims and objectives

3.1 Development of a test procedure

A standardized test method needed to be developed to clarify the particulate contamination behavior of different material pairings. A contamination-free material test bench for investigating materials for use in cleanrooms was constructed in order to develop this aim from a practical point of view. 
The development of a scientifically-sound test method ensures 
– Reproducibility,
– Comparability,
– Accuracy and
– Correct interpretation of the measurement results.

3.2 Assessment and classification model

In order to assess and classify the measurement data obtained, assessment methods needed to be developed and applied. To some extent here, some of the guidelines and standards from semiconductor and cleanroom technologies already in existence can be used. Analysis algorithms utilized for the measurement data need to show the clear suitability of different material pairings for use in clean production areas.

3.3 Optimization of material pairings
The procedure was divided up into the following points:
– Analysis of the initial state (determination of initial contamination levels),
– Modification of the material pairing (optimization of the tribological system) and
– Verification analysis (repeat determination of contamination levels).
This permitted the systematic further development of the material pairings in order to meet cleanliness requirements. The consideration of airborne particulate contamination permits a statement to be made concerning the suitability of the material pairing for a specific air cleanliness class. This is because internationally-recognized air cleanliness standards only take this form of contamination into account. This fact considerably increases national and international acceptance of the test procedure and its results.

4 Development of the method

Nearly all movement sequences in manufacturing equipment lead to frictional processes between two materials; friction is the most common cause of particle generation. In order to make this scientifically tangible, the emission of particles needs to be measured and assessed. 
The first step involved is to identify the main source(s) of particulate emission from real parts/components. In order to reduce to a maximum the effect of external influences on the measurement results obtained from the main sources of emission detected, as many sources of disturbance need to be eliminated as possible. This is best achieved by “extracting/separating” the frictional pairing site from the real component. To do this, the frictional pairing site must be “recreated” on a test bench under standardized, reproducible laboratory conditions (Figure 2). The method for assessing the cleanliness level of the material pairings investigated must be capable of supplying the two following results:
• The air cleanliness class (in accordance with international regulations for assessing air cleanliness) in which the “separated” material pairings may be implemented and 
• To what extent the test results obtained under laboratory conditions may be applied to real parts/components. In accordance with established procedures laid down in VDI 2083 Part 8 “Cleanroom suitability of operating utilities“, it is possible to investigate real parts and components with regard to their suitability for use in cleanrooms. This procedure has been developed for assessing complex parts and cannot be applied to the investigation of individual material pairings. 

Transferal of the results to real components

Once frictional partners have been identified and transferred to a tribological system on a laboratory test bench, typical work loads are then simulated to which the real components are actually subjected. Classification measurements are then performed and analyzed in accordance with VDI 2083 Part 8 (real component) and also in accordance with the classification model for material pairings using sets of stress parameters representing three typical work loads. To determine the transferability of the different classification results obtained, a model needed to be developed. However, this could only be developed in the second step once the classification model for assessing the cleanroom suitability of material pairings had been proved on a test bench. 
All the measurement values and statements obtained can be verified using appropriate statistical methods. 

5 Design of a material test bench 

5.1 Frictional processes

Three main test methods can be used to achieve friction between two bodies (see Fig. 3 and Fig. 4). With the ball-on-disk test, a ball is pressed against the face of a disk. The area of contact is punctiform and the ball is fixed. With the disk-on-disk test, a rotatable disk run on bearings is advanced laterally towards the driven disk and then rolls against it. The area of contact is in the form of a line and both surfaces are curved at the site of contact. With the roller-on-disk test, a roller (e.g. stainless steel or PA6 roller) is pressed onto the face of a disk which has been treated with the coating under investigation. This method is used for testing coverings and coatings over which rollers actually move. For example, the roller-on-disk test is implemented to test a cleanroom floor covering where carts on rollers are used. 

5.2 Principle of the test

As the ball-on-disk test is often implemented in literature and in practice, comprehensive data records are available for correlation with particle generation. Due to a further advantage, i.e. the presence of a punctiform contact area between the materials, it is possible to essentially standardize this method. As a result, the ball-on-disk test can be utilized for the considerations described below. With this test, a test disk rotates with a frequency f beneath a ball with a diameter d. The latter is pressed onto the disk and a normal force FN is applied and a radius r used. At the site of contact between the ball and the disk, material abrasion occurs which is accompanied by the emission of particles. Abrasion marks are formed. The recordable load values on real components (area of applied force, distance traveled, multiple loads on partial distances, etc.) can be correlated with the characteristics of the abrasion marks and the contact site both for analysis purposes and for fine differentiation. 

5.3 Identification of relevant parameters 

With the ball-on-disk test, the test bodies concerned (ball and disk) represent a system of abrasion. This can be described using a set of parameters. For the tests carried out here, the relevant parameters of the abrasion system were identified in accordance with DIN 50320 (old) as shown in Figure 5. At present no data in literature exists which is concerned with the particulate emission from frictional pairings or which links tribological values with the particulate emissions observed. In DIN 50324 (old), a cooperative test is described where stainless steel pairings were investigated using the ball-on-disk method. The input values/parameters were used as a basis for the preliminary tests.
In the preliminary tests, the dependence of the parameters tested (particle emission, tribological values) on the normal force was investigated and the chronological development was also recorded over an extended period of time.

5.4 Implementation of the test bench 

Figure 6 shows a test bench designed using the ball-on-disk test principle. The test bench itself also fulfills cleanroom and cleanliness suitability requirements.

5.5 Integrating particle tests with the determination of tribological values

Parallel to the particle emission tests, tribological values were also determined. Particle emission was used as a basis for comparing the results. From the measurement values obtained from the various particle size channels (>0.2 / >0.3 / >0.5 µm and >5.0 µm), it was possible to determine “particle volume” values. Tribological values were related individually and in combined form (as derived values such as the volume of abrasion) to this value. In this way the correlation between tribological values and particle emission values could be assessed.

6 Classification model for material pairings

A major objective is to observe and characterize the development of particle emission during a tribological load. The optical particle counters implemented supply particle measurement values differentially, i.e. in the form of counting events per measurement period or per measurement volume. In the preliminary tests, the graphical representation of particle measurement values showed a highly inconstant pattern of development in relation to the number of rotations. Such measurement values were therefore not suitable as a basis for characterization (see Fig. 7). During the first rotation of the test ball on test disk, the ball travels over an intact surface with the result that only a small amount of particle emission takes place. During the next rotation the ball travels over a surface which has been subjected to a load, thus resulting in an increase in particle emission. This effect continues during each of the subsequent rotations and explains the inconstant pattern of emission values observed over time. This fact means that the pre-requisites for analysis using student-t and Poisson’s statistics as required in VDI 2083 Part 8 could no longer be fulfilled. In general, it could be observed that measurement values tend to increase as the number of rotations also increases (see Fig. 7). Provided all the particles generated are added together up to a discreet number of rotations, this equates to an integral representation of the particle measurement values. Using this representation, the resulting curves show a steady increase (see Fig. 8). With the aid of the integral representation, graphs can be created which form a basis for characterizing particle emission. To clarify the increase in particle emission observed over time, it is assumed that particles are already generated as abrasive matter during the first rotation of the material pairing. During the next rotation, together with the previously generated particles, the material pairing generates more particles faster than before resulting in an avalanche effect. Using mathematics based on an exponential concept, an attempt is made to represent this rapid increase in particle emission. 

After having approximated the cumulated particle emission graphs using non-linear regression, the curves can be further considered from a mathematical point of view. 
The parameters a and b, or the pair of variates (a/b) of the regression curves represent a scale for the emission rate and thus for the cleanroom suitability of the material pairing under consideration. As the number of particles generated is directly related to the number of rotations and only indirectly related to the time elapsed, it is the number of rotations rather than the measuring time which is used as a new control variable. The limiting values for the maximum particle counts permitted for the individual air cleanliness classes in accordance with DIN EN ISO 14644-1 are then converted in relation to the number of rotations. This results in a graph of air cleanliness classes which is adapted to the problem (Figure 9). If the sum of particle values is represented using regression curves, these then always intersect with the limiting lines of air cleanliness classes. In order to reduce the pair of variates (a/b) from the regression curves to a single parameter, and also to obtain a sound statement concerning the cleanroom suitability of the material pairing in question, the number of rotations needs to be fixed. In the example shown, with a reference number of rotations of N = 500, the value from the tested material pairing lies between the limiting lines of the air cleanliness classes ISO Class 5 and ISO Class 6. This means that this material pairing can be safely used in cleanrooms fulfilling ISO Class 6 specifications. 

7 Advantages of the classification method

The advantages of the classification model for assessing the cleanroom suitability of any chosen material pairing are summarized below:
(du hattest vergessen sie einzufügen!!!) 
- Pre-assessment of the cleanroom suitability of a material pairing using tribological preliminary tests
- Correlation with tribological data bases
- Material tests without needing to make expensive molds for the actual production equipment
- Process-independent assessment of the cleanroom suitability of material pairings for use in discreet air cleanliness classes
- No further requirement for empirics when selecting materials.

8 Outlook

In further research, work will especially be carried out on the development of the interactive relationship between classification results on real component as they have been classified in accordance with VDI 2083 Part and on “separated” material pairings on laboratory test benches. 

From: 
wt Werkstatttechnik, Edition 03/2005 
http://www.technikwissen.de/wt/aktuell/ausgabedetail.asp?id=16534&heft=03/2005
Article as pdf-download