This article summarizes and analyzes current issues of concern, such as determining cleanroom airflow velocity, the application of FFU systems, and the control of suspended molecular contamination.

I. Introduction

Cleanroom airflow velocity and air change rate have long been a key concern in cleanroom design. With the increasing effectiveness of cleanroom contamination source control and improvements in final-stage filter efficiency, there has been considerable discussion as to whether the recommended or reference values in relevant standards and guidelines are conservative. Concerns regarding noise and damage repair in FFU applications have been addressed in practice. With the continuous improvement of FFUs, the use of FFU return air systems has become a hot topic. Control of suspended molecular contamination (AMC) has become increasingly prominent and garnered attention in the microelectronics and IC industries. The following summarizes and analyzes these issues.

2. Airflow velocity

2.1 Application of relevant recommended or reference values

The determination of airflow velocity at a certain cleanliness level in a cleanroom varies with the specific conditions such as the use of the cleanroom. It is affected not only by the amount of dust generated indoors and the efficiency of the filters, but also by other factors. For industrial cleanrooms, the factors affecting cleanliness and the selection of airflow velocity are mainly:

(1) Indoor pollution sources: building components, the number of personnel and operating activities, process equipment, process materials and process processing itself are all sources of dust particle release, which vary greatly according to specific conditions;

(2) Indoor airflow pattern and distribution: unidirectional flow requires uniform and equal streamlines, but will be disturbed by the layout and position changes of process equipment and personnel activities, forming local vortices; while non-unidirectional flow requires full mixing to avoid dead corners and temperature stratification;

(3) Control requirements for self-cleaning time (recovery time): accidental release or introduction of pollutants in the cleanroom or interruption of air flow or normal operation Intermittent convection during operation or the movement of people and equipment will cause the cleanliness to deteriorate. The self-cleaning time to return to the original cleanliness is determined by the airflow velocity. The control requirements for the self-cleaning time depend on the tolerance of the impact on the quality and yield of product production within this time frame (under deteriorated cleanliness). (4) Efficiency of the final filter: Under a certain amount of indoor dust, a higher efficiency filter can be used to reduce the airflow velocity. To save energy, it is necessary to consider using a higher efficiency filter and reduce the airflow velocity, or using a lower efficiency filter and a higher airflow velocity to minimize the product of flow and resistance. (5) Economic considerations: Excessive airflow velocity will increase investment and operating costs. The appropriate airflow velocity is a reasonable combination of the above factors. Excessive airflow velocity is often unnecessary and may not be effective. (6) For clean rooms with low cleanliness requirements, the number of air changes is sometimes determined by the indoor heat removal requirements. The above factors are difficult to quantify and can only be analyzed, compared and estimated. Therefore, in engineering applications, cleanroom airflow velocity is often determined by referring to recommended or reference values in relevant standards and guidelines, and then by comprehensively considering the aforementioned influencing factors based on specific circumstances.

Airflow velocity is used in unidirectional cleanrooms; air changes are more appropriate for non-unidirectional cleanrooms, as airflow velocity is difficult to accurately measure. Alternatively, the final filter fill rate can be used to reflect this, which can be applied to cleanrooms with various airflow patterns. Generally, 100% fill rate corresponds to a flow rate of 0.5 m/s (100 fpm), and 25% corresponds to 0.125 m/s (25 fpm). Current recommended or reference values in relevant standards and guidelines are shown in Table 1.

Note: 1. ISO 14644-4 explicitly uses airflow velocity/air changes as a reference. The values listed in the table are only applicable to microelectronics and IC factories. For pharmaceutical factories, only ISO Class 5 airflow velocity > 0.2 m/s is listed; no reference values are provided for ISO Classes 6 through 8. 2. (M) refers to mixed flow, N refers to non-unidirectional flow; * refers to a clean area where effective isolation measures have been taken for the pollution source. The recommended or reference values for air flow velocity and ventilation frequency in Table 1 should be considered to be a reflection of experience. For example, the values proposed by ISO/DIS14664-4 are clearly applicable to that type of clean room; the recommended values of IEST are also considered by some authoritative organizations to be only applicable to semiconductor factories. Due to the large changes in specific conditions, some empirical values may no longer be suitable for current indoor dust source control measures and filter efficiency improvements. 2.2 Discussion on the recommended or reference values In recent years, many people have believed through experiments that these recommended or reference values are too conservative. Their arguments can be summarized as follows: (1) The lateral diffusion of air flow in a clean room is only possible at very low flow rates. Under reasonable air flow organization, a flow rate of 0.05 to 0.1 m/s is sufficient to carry away pollutants. At this flow rate, the diffusion performance of submicron particles is much lower than the convection performance; and air flow velocities greater than 0.36 m/s are prone to millions of vortices, causing pollutants to be re-involved. Therefore, the ideal self-cleaning time Tr of a clean room is equal to volume/flow rate. After reaching a certain value, due to the re-involvement of pollutants, the actual Tr will no longer decrease significantly even if the airflow velocity is increased. (2) The effect of the efficiency of the final filter on cleanliness is worth noting. Some airflow velocity/air change frequency recommendations or reference values often do not take into account the factors that improve the efficiency of the final filter. Currently, the efficiency of HEPA/ULPA can be selected from 99.67%, 99.99%, 99.999%, 99.9995% to more than 8 9s. In addition to the above-mentioned effects of its efficiency on airflow velocity, the following aspects are also worth noting. Under non-unidirectional flow conditions, the formula for the stable dust concentration in a clean room based on the equilibrium release principle can be obtained: (a) When the indoor dust generation volume is high, the change in the efficiency of the final filter has little effect on the cleanliness. Therefore, in this case, excessively high filtration efficiency is unnecessary. (b) When the indoor dust generation volume is low, the change in the transmission rate of the final filter efficiency at a low airflow velocity increases the impact on the cleanliness. The above situation can be seen in Figures 1a-1c.

Relevant data for the plot:

Dust concentration before fresh air enters the final filter: 1.75×10⁶ particles/m³

Indoor emission volume: G1 = 350 particles/m³/min

G2 = 3500 particles/m³/min

G3 = 35,000 particles/m³/min

G4 = 350,000 particles/m³/min

Ratio of fresh air volume to total air volume: 0.03

Currently, some IC fabs utilize an ISO Class 5 (0.3μm) cleanroom with an FFU system equipped with ULPA (99.9995%, 0.12μm). The outlet air velocity is 0.38m/s, and the occupancy rate is 25%. This results in an average indoor air velocity of 0.095m/s, which is within the lower limit of all recommended or reference values. The processing in these cleanrooms is performed within the micro-environment, and with relatively few people inside, it can be assumed that the emission volume is low. In this case, a lower air velocity may be desirable. According to reports, the IEST lower limit for cleanroom airflow velocity recommendations has been lowered. For example:

ISO Class 5 or below: Airflow velocity 0.2-0.5 m/s;

ISO Class 6 or 5 (non-unidirectional airflow): Air changes > 200/h;

ISO Class 7: Air changes 20-200/h;

ISO Class 8: Air changes 2-20/h;

III. Application of FFU Systems

3.1 Current Status of FFUs

FFUs have proven themselves to be reliable in terms of service life and maintenance. The current improvements are mainly: (1) taking measures to equalize the current and reduce the noise, so that the noise can be within 50db; (2) the motor uses DC/EC (electronic rectifier motor), which saves nearly 50% of the energy compared with the original AC motor. Because the small capacity (power <1/2HP) AC motors used in small fans are generally capacitor split phase or hidden pole type, their efficiency is only about 40%, while the efficiency of DC/EC motors can reach 75-80%; in terms of speed control, each unit can be controlled by filter pressure reduction to save energy, but the current investment recovery period is still long and it has not been widely adopted. Generally, group control or full group control is commonly used. (3) However, the static pressure at the outlet of the FF flap cannot be too large. Generally, the outlet wind speed is 0.38m/s, at which time the static pressure is generally within 250Pa. 3.2 Advantages of FFU return air system compared with other methods

3.2.1 General evaluation

Advantages:

(1) High flexibility and easy to modify;

(2) Less space occupied in the building;

(3) The air pressure in the clean room is greater than that in the return air static pressure room, eliminating the possibility of the static pressure room contaminating the clean room.

Disadvantages:

(1) The total resistance of the return air duct (including the porous floor, grille and air duct), the resistance of the dry surface cooler and the resistance of the final filter (at the initial resistance) should be controlled at about 165Pa in total to ensure that the maximum resistance during operation is within 250Pa. Therefore, the heat transfer area of the dry surface cooler should be larger, the size of the return air duct should also be larger, and the resistance of the porous floor and grille should be small. The general practice is to control the resistance of the dry surface cooler to about 50Pa and the resistance of the return air duct to within 15Pa. Otherwise, a pressurized fan system needs to be added, which reduces the comprehensive advantages of the FFU system.

(2) After adopting DC/EC motors, the energy consumption per unit air volume may be lower than that of the current centralized system with general large centrifugal fans. However, studies have shown that the energy consumption is still higher than that of the return air system with improved large axial fans. Therefore, it is necessary to pay attention to the factors of improving the efficiency of large axial fans and reducing the resistance of their systems.

(3) Since the energy consumption per unit air volume of general FFU systems is relatively high, the cooling load of the clean room also increases accordingly.

3.2.2 Evaluation in specific situations

(1) When FFU is used to transform old buildings into clean rooms, its comprehensive economic performance is generally desirable.

(2) For clean rooms with strict cleanliness requirements, when the final filter coverage rate is 100%, it is currently not economical to use FFU for large systems; it is meaningful to make specific comparisons for small systems. (3) For cleanrooms with less stringent cleanliness requirements, the overall economic benefits of the large system are often similar when the final filter coverage rate is ≤40%. However, for IC factories, the flexibility of the FFU system is important. Therefore, it is now common for IC factories to use FFU systems when the filter coverage rate is ≤40%.

IV. Airborne Molecular Contamination (AMC)

4.1 Classification and Control Requirements of AMC

AMC was first proposed by the Japanese 20 years ago as a concern for IC factories. In recent years, the diameter of IC production wafers has reached φ300mm, and the process size (line width) has been less than 0.15μm. In certain processing steps and in the wafer transfer and storage environment between processes, AMC has become a problem that seriously affects the yield rate. It has been clearly recognized. Therefore, the control of AMC has shifted from being a topic of discussion to a necessity.

In IC production, AMC is categorized into four categories: A, B, C, and D:

A—Acidic substances, such as HCl;

B—Alkaline substances, such as NH3;

C—Substances with boiling points above room temperature that can condense on smooth surfaces, primarily hydrocarbons. Water vapor in certain processing environments also needs to be considered;

D—Doping substances, such as arsenic, boron, and phosphorus, that can adsorb or react with wafer surfaces.

AMC poses a much wider range of potential contamination risks to current IC production than particle contamination. Particle contamination control simply involves determining particle size and count, but AMC control is not only affected by shrinking chip line widths but also by process technology, process equipment, process materials, and wafer transport systems. Furthermore, trace amounts of process materials (chemicals, specialty gases, etc.) used in one process can often be contaminants in the next process. With wafer processing now comprising over 300 independent steps, determining AMC control indicators is even more complex. Therefore, the control of AMC in IC production will have different requirements for different products, different processes, different procedures and different process materials. The general requirement for various pollutants is to control them between sub-pptm and 1000pptm.

4.2 Implementation of AMC Control

For IC production with a line width of 0.25μm, activated carbon filters are usually installed in the fresh air treatment. For key processes and the transfer and storage of wafers between processes, some manufacturers have adopted AMC control, while others have not. This is mainly due to the measurement of economic effects. There are few reports on specific control requirements and measures, probably due to confidentiality reasons. However, one thing is certain: control can only be carried out in a local environment.

In order to meet the processing requirements of φ300mm wafers and line widths of less than 0.15mm, in recent years, the focus of AMC control has been on the following three aspects:

(1) Establishment of precise measurement technology and standard test methods. Because this is the basis for mastering AMC control, it must be done first; (2) According to future IC production requirements, the equipment in the production line will use micro-environment isolation, and the wafers between each device will be transported using a front-opening standard wafer box (FOUPs) system to isolate the wafers. Therefore, the materials used in the equipment, FOUPs system and micro-environment are required to not release and adsorb related suspended molecular pollutants, and the removal measures for these pollutants have been developed and continuously improved; (3) Filters for controlling AMC. In recent years, especially in the past 2 to 3 years, there has been little progress in the development and introduction of filters for controlling AMC; A. HEPA/ULPA that does not release AMC substances; a. Low-boron ultrafine glass fiber filters, which are now widely used in IC factories in Asia and Europe; b. Porous polytetrafluoroethylene (ePTFE) filters, which are thin film structures and are about ten times more expensive than a. Currently, they are not widely used, and the next generation is being developed. B. Chemical Filters

Currently available chemical filters mainly include:

a. Activated carbon filters, most of which are granular and available in disc and honeycomb types. Activated carbon fiber filters are also available, offering fast adsorption rates but at a relatively high price. Filters with bonded granules and fibers are also available.

b. Non-woven synthetic fabrics are impregnated with various functional granules (such as activated carbon and activated aluminum, but primarily activated carbon) to adsorb AMC.

To date, it has been reported that, in addition to two pilot production lines, four φ300mm wafer processing lines (one in Germany, one in the United States, and two in Taiwan) have commenced operation. While the details of AMC control are unclear, the cleanroom environment is ISO Class 5-6, making cleanroom design relatively simple. It is clear that in the future, the focus of pollution control in the IC production environment will inevitably shift to the research and development and manufacturing of process equipment and wafer transport and storage systems.