Arup Bhattacharya1, Mohammad Saleh Nikoopayan Tak2, Shervin Shoai-Naini3, Fred Betz4, Ehsan Mousavi This email address is being protected from spambots. You need JavaScript enabled to view it.2 1 Bert S. Turner Department of Construction Management, College of Engineering, Baton Rouge, LA, USA
2 Nieri Family Department of Construction Development, and Planning, College of Architecture, Arts and Humanities, Clemson, SC, USA
3 Proterra Incorporation, Burlingame, CA, USA
4 Affiliate Engineers Incorporation, Madison, WI, USA
Received:
November 20, 2022
Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.
Revised:
April 2, 2023
Accepted:
May 5, 2023
Download Citation:
||https://doi.org/10.4209/aaqr.220407
Bhattacharya, A., Nikoopayan Tak, M.S., Shoai-Naini, S., Betz, F., Mousavi, E. (2023). A Systematic Literature Review of Cleanroom Ventilation and Air Distribution Systems. Aerosol Air Qual. Res. 23, 220407. https://doi.org/10.4209/aaqr.220407
Cite this article:
Cleanroom ventilation systems are well-established; however, the advantages and limitations of current practices need to be examined and explained further. This study begins by looking over the history of cleanroom ventilation systems that creates the basis for understanding ventilation rate specifications, terminologies employed in ventilation effectiveness, and recognizing scientific studies that correlate ventilation effectiveness with air change rates. This systematic review includes a comprehensive summary that contains a set of historical data and evidence that may be used to specify ventilation requirements in cleanrooms. Scientific articles are classified in terms of laboratory experiments, simulations/numerical analysis, or hybrid. HVAC designers and operators can use published codes and guidelines more efficiently if the terminology is properly understood and the design solutions are easy to implement. The present study aims to provide a deep insight into understanding the role of ventilation on the transport mechanisms of unwanted particles in cleanrooms. Historically, the ventilation rate is typically over-estimated, based on the experience of the designer, to ensure indoor air quality and thermal performance. However, the excess rate significantly impacts the system’s energy consumption. Hence, it is crucial to investigate existing recommendations to ensure if they have a scientific basis or could be proven theoretically before further implementations. Besides, any possible risks and influences associated with the traditional methods must be assessed to guarantee the facility’s performance, sustainability, and energy efficiency.HIGHLIGHTS
ABSTRACT
Keywords:
Cleanroom, Ventilation, Ventilation requirements, Air change rate
Several of the terms in the realm of ventilation systems in cleanrooms may have changed over the course of nearly 70 years of cleanroom history. Early-stage cleanrooms, designed to keep away dust from manufacturing tables, used high efficiency filters (> 99.95%) and positive pressurization (Whyte, 2010). Air distribution patterns, and more specifically unidirectional airflow patterns, were first proposed by Whitfield (1962). The unidirectional air flow system gained immediate attention from cleanroom designers (Whitfield and Whitfield, 1981). This current work does not intend to study how cleanrooms emerged and how they have evolved over time; interested readers can refer to excellent work on the history of cleanrooms (Naughton, 2019). Instead, the aim is to study the scientific articles about cleanroom ventilation systems and analyse how research and evidence in this field have evolved. To that end, all the relevant papers identified in the search phase were sorted based on the year of publication. Early publications (< 1991) mainly focus on developing the basics and theory of the cleanroom ventilation system. At that time, most of these developments were based on theoretical speculations, as well as some experimental validations. From a parallel perspective, it seemed that early papers had a specific focus on industrial and semiconductor cleanrooms, in response to the need of the industry. In the early 2000s, a cleanroom benchmarking study showed that 58% of cleanrooms in the State of California were semiconductor cleanrooms (Tschudi et al., 2001). Moreover, it was found that the early-stage papers mainly focused on evaluating the overall air distribution schemes. Later, researchers began to investigate the specifics of airflow patterns (e.g., velocity and temperature distribution) as well as positive pressurization. In the 1990s, significant effort was invested in finding optimum air velocity, ventilation rates, and positive pressure magnitude. In this period and perhaps with the prevalence of computers, a shift towards numerical studies substantially increased. By early 2000s, there was a shift in the direction of research to study the distribution of contaminants specifically particle containment and removal techniques, as they related to the ventilation system efficiency. Although ventilation efficiency was first introduced in the early 1980s (Tschudi et al., 2001), it took nearly two decades for it to gain significant attention in the field of cleanroom design and operation. More recent publications tend to shift focus towards optimizing system performance. This was an interesting shift, as not only did facilities emphasize the role of air cleanliness, but researchers became more cognizant of high rates of energy consumption by cleanrooms. Hence, attempt was made to optimize the ventilation system’s performance to maintain a clean environment while minimizing energy consumption (Table 1 and Fig. 1). These findings convinced our team to organize this report in the same manner. That is, the report discusses 1) air and ventilation requirements for cleanrooms, followed by 2) particle containment and removal, and finally, 3) optimization of ventilation systems in cleanrooms. This article also discusses the potential of numerical and experimental approaches to advance the understanding of ventilation requirements in cleanrooms. These requirements may significantly impact design, construction, energy efficiency, and maintenance of the ventilation system. We present a systematic approach to reviewing the existing literature on the ventilation system in cleanrooms to optimize the system’s design, operation and maintenance. The research team created a process to develop a bibliography of articles related to ventilation systems for cleanrooms. This process was implemented through weekly meetings and monthly check-ins to ensure that the search process could provide helpful results. The research team began the process by extracting keywords and phrases from ventilation terminology. Specifically, the terms ventilation, ventilation effectiveness, ventilation rate, ventilation requirements, ventilation air, air change rate, and ventilation effectiveness factor (first group) in conjunction with the terms cleanroom, clean room (second group) were investigated. This means in every trial of searching for papers, one keyword was selected from the first category of seven keywords, along with one of the two in the second group; this approach culminated in 14 combinations of searched keywords. These terms were input into academic search engines Engineering Village, Compendex (Elsevier), PubMed, Google Scholar, and Scopus to generate a database of publications relevant to a cleanroom setting. After the first round of review for the papers’ titles and abstracts, in a few cases where the decision to include or exclude the article was not clear based on the title alone, a total of 236 papers were selected to transfer into a spreadsheet, of which 17 were not in English, and 17 were not found. Hence, the actual number of articles to review was reduced to a total of 202 papers. Once the database was complete and organized, the team began to collect the papers that referenced the list of keywords and terms. The research team then divided the publications amongst themselves to review. The first step was to evaluate the methodology used in each article. Tags used to describe the methodology were experimental, numerical, hybrid, or review. If the paper was experimental, a second tag was given based on the conditions of the experiment (i.e., field experiments/evaluations, test chamber). Fig. 2 shows the distribution of the reviewed papers based on their research method. After tagging all the cleanroom papers in Mendeley, the team began to read each paper in the database. A spreadsheet was developed to collect relevant data from each reviewed publication. Any guidance provided by the author(s) about the findings in the study was listed, as well as the recommendations provided based on the author’s findings. A summary of each paper was provided for reference when analysing the data. From the relevant papers gathered, studies were designed to address certain problems that often arose in cleanrooms, including contamination spread, airflow, turbulence, overuse of energy, etc. These studies produced important findings for the future of cleanrooms, which were then used to set new cleanroom regulations and design recommendations. Efforts were invested in characterizing the role of ventilation in cleanrooms. In this section, we aim to explore these studies and how they led to initial ventilation requirements for cleanrooms. Specifically, we investigate (a) temperature and RH, (b) pressurization and ACH, and (c) air distribution requirements in cleanrooms. Studies on thermal comfort investigated variables, including relative humidity and temperature, and how those were dictated by processes such as human comfort (Peyton and Abdou, 1995). Davison (1996) discussed the advantages and disadvantages of four systems that helped to control thermal comfort in photolithography minienvironments: hot-deck/cold-deck system, 100% makeup air system, basic recirculation system, and recirculation farm system. The hot and cold-deck system and the 100% makeup air system controlled settings within –17.6°C temperature and 0.05% relative humidity. The basic recirculation system controlled temperature within –17.4°C, and relative humidity specifications ranged from 33% to 41%. The recirculation system used a combination of 100% makeup air system and a basic recirculation system to allow separate controls for the minienvironments. The system could regulate thermal comfort settings within –17.6°C temperature and 0.5% relative humidity (Davison, 1996). Standards for thermal comfort have become increasingly stricter, as discussed by Naughton (1992). Due to the increasingly strict standards, he introduced a more complex commissioning process in microelectronics cleanrooms (Naughton, 1992). In a study evaluating adiabatic humidification in HVAC systems, Jo et al. (2017) found that pressurized water atomizers save energy in the thermal comfort of semiconductor cleanrooms. Almost 23% of energy savings occurred when a pressurized water atomizer was installed in the return ducts, as opposed to a steam humidifier attached to a make-up air unit (Jo et al., 2017). In 2017, Rotheudt et al. (2017) studied near-wall flow separation in vertical unidirectional flow cleanrooms numerically, considering typical inlet velocity from 0.25 to 0.45 m s–1 with 200 ACH. It was found that due to temperature differences between the air flow and ambient environment, the boundary layer separated from the wall, creating a recirculating zone prone to contaminant stagnation. The simulation results showed that wall insulation decreased the thickness of the flow separation area (Rotheudt et al., 2017). Al Beltagy et al. (2018) conducted a study assessing the use of radiant cooling systems in cleanroom applications. It was found that the system offers uniform temperature distribution. Using Predicted Mean Vote and Predicted Percentage of Dissatisfied, the study showed that thermal comfort could be achieved through radiant cooling systems (Al Beltagy et al., 2018). Another study on radiant panel systems showed that radiant panels were inefficient in uniformly cooling the room when combined with displacement ventilation systems, where cool air is supplied from the floor, creating separate flow zones (Choi et al., 2019; Davidson, 1989). For the unidirectional airflow systems, however, the air supplied above the working table should be at a higher temperature to provide uniformity (Alhamid et al., 2018). In a study on the effect of an air warming blower on ventilation air in an operating room, it was found that the warm air caused thermal plumes and 60% turbulence intensity when the blower was on. This causes particles to actually rise toward the operating table, whereas with the blower off, no particles rose towards the table (He et al., 2018). Studies on thermal comfort indicated that thermal uniformity leads to more comfortable cleanroom environments. Positive pressure differentials dictate important details for clean room ventilation design. Safety precautions, equipment design, and HVAC system calibration can be affected by inaccurate positive pressure differentials. Most importantly, positive pressurization allows external sources of contaminants to be controlled (Bharath and Reddy, 2013). To be preemptive to the risks of gas leaks in a cleanroom, Campbell (1996) suggested maintaining pipe pressure above or at –24.9 Pa. In 1998, Karimipanah (1998) used pressure as a proxy to predict flow parameters, such as turbulence, deflection, and impingement. Yamaguchi et al. (2010) recommended using a simulation to calculate pressure change in a room caused by air conditioning equipment. The new method uses an electrical circuit network instead of a pressure propagation mechanism, considering air volume, lifted pressure, and friction loss as electrical properties. Implicating this simulation will allow for more accurate settings in pressure standards as opposed to traditional analytic methods. The suggestions made in this study give options to older cleanrooms for remodeling, as well as aid new cleanrooms in calibrating and use equipment efficiently (Yamaguchi et al., 2010). In 2009, Chien et al. (2009) showed that greater pressure inside the room allowed smoke to ventilate properly, meaning that smoke can flow out of an isolated room only if the pressure inside the room is greater than the pressure outside the room. Fan filter units and raised grating coverage ratio affected the pressure distribution in the room and should be investigated to provide ventilation (Chien et al., 2009). Investigations into cleanroom energy efficiency showed that inaccurate pressure differentials could become a major factor in energy overuse. Mathew et al. (2010) showed that low-pressure drop HEPA filters could significantly reduce the energy used by recirculation units in cleanrooms. Lakshmana Prabu et al. (2017) discussed “good manufacturing practices” when dealing with pressure differentials in pharmaceutical cleanrooms. Manometers are used during commissioning to measure pressure differentials and ensure that they are accurate, which is normally between 0.67 and 2.7 Pa depending on the cleanliness class (Lakshmana Prabu et al., 2017). Investigations into differential pressure setpoints for chilled water pumps in HVAC systems show that the method for quantifying the setpoint is inadequate, meaning that the previous methods were insufficient in achieving the desired energy and economic savings since they may not have taken into account important factors such as pipeline topology, material, chilled water flow demand, heat loading, and equipment aging. Therefore, an equation (please refer to Table S1 for pump demand capacity and differential pressure setpoint equations) is created to accurately calculate the setpoint for energy efficiency and cost savings (Su and Yu, 2013). Findings on pressure differentials allow pressure standards to be reevaluated for more efficient use in cleanrooms. Airflow can be controlled through air velocity and air changes per hour (ACH). Bugaj and Przydrozny (1986) suggested that for a supply grille width 1.9 m and height 0.8 m, lower edge at 1.5 m above floor and exhaust through floor level ports, supply air velocity and ACH should be equal to 0.45 m s–1 and 18.9 ACH, respectively. For supply grille width 1.9 m and height 1.1 m, lower edge at 1.2 m above the floor level and exhaust same as before, satisfactory supply velocity should be 0.35 m s–1 with the corresponding ACH of 21.5 (Bugaj and Przydrozny, 1986). Supply plenums in cleanrooms contain sound absorption plates and ultra-low penetration air filters that affect the uniformity of airflow. Increasing the size of the chamber and the size of the opening beyond the sound absorption plate is recommended to decrease nonuniformity in airflow. Also, a decrease in air supply velocity could decrease nonuniformity. In Sadjadi and Liu’s (1991) experimental tests, decreasing average velocity from 0.46 m s–1 to 0.41m s–1 offered a 3% decrease in air velocity nonuniformity. A study on optimizing cleanroom air velocity found that 0.30–0.33 m s–1 can control particle contamination and airflow. This study tested many different configurations and recommended that airflow velocities be tested for all configurations because of varying contamination source locations (Maxwell et al., 1994). Kalliomäki et al. (2016) investigated the effect of different doors on cleanroom airflow. It is recommended to use single sliding doors to produce the least effect on air velocity distribution and ventilation rates (Kalliomäki et al., 2016). Campbell (1996) recommended 240 to 300 ACH to mitigate the risk of gas leaks in semiconductor cleanrooms. Nam (2000) also tested airflow in relation to safety measures in cleanrooms. According to the study, the 0.91 m s–1 airflow rate can pose difficulties for smoke detection compared to the other airflow rates tested (0.1 m s–1 and 0.46 m s–1). This is due to the reduction of the CO2 cloud radius, which dilutes the concentration of smoke or fire particles in the air and can compromise the sensitivity of smoke detectors, potentially resulting in delayed detection. To optimize area coverage, smoke detectors should be positioned near the source of airflow or fire, while the effect of airflow rate on smoke detection effectiveness could vary depending on factors such as fire source size and location, detector sensitivity, and detector placement relative to the airflow pattern (Huo et al., 2009; Nam, 2000). Re-entrainment of particles or contaminants can be caused by excessive air change rates. Wang et al. (2007) suggest lowering the air changes per hour to create uniform airflow. One study presented airflow calculations to determine cleanliness levels, showing that lowering airflow could be beneficial (Noh et al., 2008). For instance, to achieve ISO 6 clean class, Wang et al. (2015) stated that the air change rate should be 50 ACH, while Lin et al. (2010b) recommended a 70–139 ACH with ventilation systems of wall-return and fan dry coil unit-return for non-unidirectional airflow cleanrooms. Using numerical simulations, Rui et al. (2008) found that 0.25 m s–1 air supply velocity controls the spread of particles. Khoo et al. (2012) conducted experiments in a full-scale, raised floor, non-unidirectional flow industrial cleanroom with mini environments (an enclosed space with a fan and HEPA filter to separate the sensitive region from the operator and the rest of the room) to examine the effects of ACH and free area ratio of the raised floor on particle concentration. Studying four ACH levels and four free area ratios of raised-floor, it was concluded that increasing both decreases particle concentration. Furthermore, it was determined that the influence of the raised floor's free area ratio on particle concentration is more significant at low ACH cases than it is at high ACH scenarios (Khoo et al., 2012). Yang et al. (2015) numerically simulated a unidirectional protection isolation room to determine recommended supply air flow rate to achieve ISO-5 cleanliness with the positions of a manikin standing, sitting or lying. With a standing and sitting body, it was found that a minimum of 360 ACH could be effective in controlling particles and achieving ISO-5-level cleanliness. For a lying body, to maintain cleanliness, the ACH had to be equal to or greater than 288 ACH to achieve ISO 5 level cleanliness. Hence, Yang et al. (2015) suggest supplying air at 360 ACH during the day and 288 ACH during the night. Khankari (2017) showed using CFD simulations that minienvironments can provide the required cleanliness in the product manufacturing region for semiconductor manufacturing rooms with reduced ACHs of 37.5 h–1 and 150 h–1 for the supply airflow rates for the room and for the minienvironment, respectively, provided lower cleanliness in other areas of the cleanroom is acceptable. In 2019, Loomans et al. (2019) suggested that ACH be lowered using demand-controlled filtration or finetuning. They set a minimum of 6 ACH to combat any buoyancy forces or temperature differences (Loomans et al., 2019). Findings from these papers allow recommendations for air velocity and ACH to combat nonuniformity in airflow. In a later study in 2020, Loomans et al. (2020) investigated the effects of lowering ACH and pressure difference within a demand-controlled filtration (DCF) on particle concentration levels in cleanrooms during non-operating hours. They demonstrated that as long as particle concentration conditions outside the cleanroom are consistent with normal conditions, meaning the typical environmental conditions outside the cleanroom that would not pose a significant risk of contaminating the cleanroom, having the ACH and Pressure difference at zero level (turning-off the fan), the quality of the cleanroom would not be compromised. Moreover, they discussed that an ACH of 4 h–1 and a pressure difference of 7.5 Pa could further secure the air quality condition of the ISO class 4 cleanroom. This strategy can save the energy consumed by fan systems up to 30 to 70 percent (Loomans et al., 2020). In an experimental study by Behrens et al. (2021), the appropriate ACH in a pharmaceutical cleanroom Class 100,000 (ISO 8) was investigated to meet the required limit of particle concentration and the required clean-up time of 20 minutes. Testing four different rates of 10, 12, 15 and 20 ACH and four different kinds of operator garments, it was shown that 10 ACH could meet the needs of this cleanroom class. It is worth noting that a 50% reduction in ACH can lead to 25–30% less energy consumption by the HVAC unit (Behrens et al., 2021). Another recent study investigated the effect of lowering supply airflow of electronic cleanrooms on the dispersion of particles and conducted experiments for four setups of generation at two different locations with two different heights, two different fan filter unit (FFU) area ratios of 50% and 25%, and three different FFU speed ratios of 80%, 50%, and 40%. A fan filter unit is a self-contained air filtration system consisting of a fan, a filter, and a housing, which is designed to be mounted in a ceiling, floor or a cleanroom wall aiming to create positive pressure and help to maintain the required level of cleanliness. Each device was positioned within the system’s ceiling or floor grid and could work independently of the others. Experiments were conducted in a mini-environment with 16 FFUs in the ceiling of an experimental chamber. For FFU area ratios of 50% and 25%, only half and a quarter of all FFUs were operating in the selected environment. The FFU speed ratio refers to the rate at which each FFU supplies air to the cleanroom with respect to the maximum speed of FFU; because the FFUs work independently, a decrease or increase in one’s speed ratio would not affect the others. It was concluded that FFU area ratios of 50% and 25% could increase particle dispersion to the sides when the air supply volume was reduced. It was shown that at an area ratio of 50%, most subzones were well protected, no matter where the source was, and they showed the rate of air distribution to FFUs could be cut by more than 40% (Shao et al., 2021). Various ACH magnitudes in time were summarized in Fig. 3. Providing a controlled air distribution pattern is one of the most important goals for cleanroom ventilation, which can be controlled by room design, including the type of ventilation system and where the supply and exhaust are located (Thatiparti et al., 2017). The most popular air distribution systems used in cleanrooms are unidirectional flow, to prevent particle settling, and ceiling diffuser systems, to sweep particles on surfaces (Sandle, 2011; Sooter, 1964). A study on energy saving in vertical unidirectional flow systems led to design recommendations, which stated that return outlets are most efficient when placed on the long walls of a cleanroom design (Zhou et al., 2003). One study showed that a unidirectional downward flow ventilation design is recommended for removing most of particles (Wan and Chao, 2007). Suwa et al. (2011) conducted a study on airflow in Front-Opening Unified Pod Systems (FOUPs). These systems are used in very-large-scale integration production to hold the silicon wafers in a controlled environment. FOUPs are often used in cleanrooms as part of the manufacturing process for semiconductor devices and are typically opened and closed inside the cleanroom. To obtain optimal airflow in these systems with reduced supply air filters, a spread-type air supply was recommended. By dispersing the supply air over a wide area using a diffuser or other equipment, spread-type air supply helps to reduce the presence of stagnant or dead air zones where contaminants can accumulate (Suwa et al., 2011). Eslami et al. (2016) tested many different design configurations of supply and exhaust inlets. It was suggested that the supply inlets should be located on the ceiling of the cleanroom along the short wall and oriented in a vertical direction, perpendicular to the ceiling and the exhaust inlets should be placed symmetrically along the long walls (Eslami et al., 2016). Three different arrangements of exhaust grills were tested in a simulation study by Metwally et al. (2018) in a pharmaceutical cleanroom with 12 air supply ceiling diffusers to identify the best arrangement that provides unidirectional flow to avoid particle settlement or extra dispersion which can occur if the airflow patterns are not optimized, leading to turbulence or recirculation of air and particles in the room. Three different exhaust arrangements were studied a) two exhaust grills on the same wall, b) two exhaust grills on opposite walls, and c) four exhaust grills in pairs of two on opposite walls. The case of two exhaust grills on opposite walls provided the best unidirectional airflow, and no vortexes were generated in both vertical and horizontal sections (Metwally et al., 2018). Zhao et al. (2020) conducted a simulation study to investigate the airflow distribution, air velocity, thermal comfort, and cleanliness recovery time in ISO Class 7 cleanroom. The results demonstrated that thermal discomfort due to temperature gradient did not occur, even at the highest tested ACH (30), and that the air distribution was generally uniform. It was discovered that the flow pattern is mostly governed by the value of ACH, demonstrating that ACH is a crucial element in the indoor airflow field and significantly impacts the deflection and dispersion of thermal plumes (Zhao et al., 2020). Ventilation arrangement can cause flow stagnation and nonuniform velocity distribution (Vutla et al., 2019). The design and layout of the cleanroom, the location and number of air supply and exhaust vents, the arrangement of equipment and fixtures, and the presence of personnel or other sources of airflow disruption all can impact the flow of air in a cleanroom. If the flow of air is not optimal, particles could travel around the room. Often, cleanroom ventilation is measured through ventilation efficiency. Kato et al. (1992) investigated locally balanced supply and exhaust airflow rate systems. By measuring the three measurements of ventilation efficiency (spatial average concentration, mean radius of diffusion, and concentration across the room), they found that balancing the supply and exhaust airflow rates allowed airflow to be uniform (Kato et al., 1992). Ventilation efficiency can be calculated through different variables. In one study by Quarini et al. (1997), two “Figures of Merit” are calculated to determine the efficiency of ventilation in a cleanroom (Table S1). They compared the volume usage and the average age of air in the ideal situation and the new design (Quarini et al., 1997). Rouaud and Havet (2005) derived an equation recommended to find the correct ACH for the desired cleanliness level. In 2014, Whyte et al. (2014) investigated ISO 14644-3, which helped define and calculate Air Change Effectiveness (ACE) to measure particle decay rate. The ACE index compares the ACH or recovery rate or decay rate at a critical location with the overall ACH of the cleanroom; the Performance Effectiveness index (PE), which compares the concentration of particles at a critical location with the average concentration of particles at the cleanroom, and Contamination Removal Effectiveness index (CRE) compares the concentration of particles at the exhaust grill with the concentration of particles at a critical point to measure the effectiveness of the ventilation system in a cleanroom (Whyte et al., 2014). It was discussed that ACE and PE were more accurate ways of measuring the effectiveness of removing contaminants as opposed to CRE. When the production occurs between the point of particle measurements and the exhaust, the CRE can show excellent ventilation system performance (Whyte et al., 2018). Gavrilin et al. (2018) proposed equations for the efficiency of the system and the air removal efficiency in microelectronic cleanrooms. Loomans et al. (2019) investigated demand-controlled filtration to maximize the efficiency of ACH in a pharmaceutical cleanroom. Demand-controlled filtration is an air filtration system that changes its filtration rate according to the level of pollutants in the air, as opposed to using a fixed filtration rate. This results in better filtration effectiveness and lower energy usage (Loomans et al., 2019). Another technology that can improve ventilation efficiency for energy and economic interests is the adaptive full-range decoupled ventilation strategy (ADV). This system incorporated three ventilation strategies in one: the dedicated outdoor air ventilation strategy, the partially decoupled control strategy, and the adaptive economizer control. Combining each system’s advantage, this strategy could enforce the strict humidity guidelines without using as much energy as outdoor air systems (Zhuang et al., 2019a). The ADV system can reduce 18.2% and 13.6% of the mean total cost compared to the dedicated outdoor air strategy and the partially decoupled control strategy, respectively (Zhuang et al., 2019b). These findings on ventilation efficiency can be used as guides for creating more efficient cleanrooms in terms of maintaining cleanliness and preventing contamination while minimizing energy consumption. Recommendations about additional ventilation approaches were also made, including equipment and natural ventilation (Joyce and Iliria, 1998; Plesu et al., 2018). In 1974, Clark et al. (1974) introduced a way to create airflow in a cleanroom naturally by lowering the wall temperature compared to the ambient temperature. It was found that when the walls were 11°C cooler than the ambient temperature, air movement equivalent to 30 ACH was created (Clark et al., 1974). Semiconductor cleanrooms are difficult to ventilate because heat exhaust interferes with airflow. A Fan Dry Coil Unit system, a type of air-handling unit that uses a fan to circulate air over a dry coil for cooling or heating without water or refrigerant, was proposed to combat the heat exhaust and to achieve higher particle removal efficiency (Lin et al., 2010a). Additional technology can be applied to help control pollutants. Kanaan et al. (2014) investigated upper room ultraviolet germicidal irradiation (UVGI) to improve return air quality in cleanrooms. They found that this technology can help reduce the concentration of airborne bacteria without increasing energy usage (Kanaan et al., 2014). When using UVGI, The UV disinfection rate attained in the top zone reduced from 88% to 78% when pathogen-carrying particle size grew from 2.5 mm to 20 mm. This implied that bigger particles are more likely to be removed by deposition. Therefore, UVGI is still effective in reducing bacteria in the upper zone, but its effectiveness decreases when larger particles are present. Using upper-room UVGI was recommended as it allowed for a higher fraction of return air, leading to more energy savings (Kanaan et al., 2015). In 2015, Idris et al. (2015) investigated the effect of window placement on natural ventilation efficiency. Large eddy simulations showed that the highest airflow rate was created by two windows side-by-side (Idris et al., 2015). Ciuzas et al. (2016) investigated indoor air quality and suggested combining air-cleaning equipment and ventilation systems to achieve proper indoor air quality. Window and door designs can provide natural ventilation in cleanrooms. In some instances where window openings are permitted to air out toxins, it was found to be more efficient to increase the number of openings rather than the size of the openings (Liang and Qin, 2017). These additional resources offer approaches other than ventilation systems to achieve cleaner air. The main aim of ventilation in cleanrooms is to minimize the introduction, generation, and retention of contaminants. Therefore, various indices have been defined to capture the role of ventilation in removing contaminants from the cleanroom (Zhou et al., 2017). These indices were used in the studies to assess the contamination concentration qualitatively and quantitatively in cleanrooms and provided recommended practices that can be able to maintain the desired cleanliness as shown in Fig. 4 and Table 2. The idea of ventilation-driven pollutant removal is that the airflow should carry the contaminants toward the exit, where the contaminants were assumed to be transported by convective transport, which is the transport of passive particles through air streamlines. Passive particles can refer to pollutants, such as smoke or dust, that are carried by the airflow and are not actively emitted by a source within the space (Ljungqvist and Reinmuller, 1996). With advancement in computational powers, simulation studies came into the picture and provided the option to collect data using numerical simulation. First, 2-dimensional studies of airflow patterns were conducted, and eventually, 2-dimensional simulation studies proved insufficient to capture flow patterns (Suesada et al., 1990). Kuehn et al. (1992) recommended 3-dimensional models to understand airflow and contaminant concentration relations. The accuracy of cleanroom airflow simulation was higher using RNG k-ε than standard k-ε, especially when turbulence was a factor (Rouaud and Havet, 2002). The airflow in cleanrooms is broadly divided into two categories—unidirectional, where air flows from the supply to the exhaust in streamlines, and non-unidirectional, where a non-perfect mixing of airflow exists. One study showed that contaminant dispersion from the source was uniform for unidirectional flow cleanrooms. In the presence of isotropic turbulent vortex, the concentration was accumulated non-linearly (Ljungqvist, 1979). In 1979, Kundsin et al. (1979) compared the concentration of microorganisms in unidirectional Air Flow Room with a ventilation rate of 240–360 ACH, and an ICU room with minimal airflow 34–43 ACH, with a single patient as the occupant. It was found that LAFR had the least fallout of viable particles resulting from the filtration used (Kundsin et al., 1979). Different methods of particle deposition in a cleanroom were investigated by Whyte et al. (2015) through an experimental study by measuring the PDR (particle deposition rate) via witness plates, which are smooth and clean surfaces used to collect particles and monitor contamination levels in a cleanroom environment, in three different ventilation schemes: unventilated cleanroom, turbulent airflow with 13 ACH, and a unidirectional airflow. The results showed that about 82% of the total deposition of particles is due to gravitational settlement, and about half of the rest is because of turbulent flows, while the rest can be attributed to electrostatic deposition (Whyte et al., 2015). In their simulation study, Khalil (2011) used local air age to demonstrate ventilation effectiveness in terms of air distribution and scavenging passive contaminants and showed that local air age depends upon the supply characteristics and locations of supply vents. Eslami et al. (2016) simulated a full-scale contaminated cleanroom to observe airflow patterns and particle dispersion with different supply and exhaust settings through both Eulerian and Lagrangian methods. It was demonstrated that vertical and horizontal configurations (unidirectional) push the contaminants towards the exhaust grille, and hence more particles could escape, whereas the non-unidirectional configuration aided in the mixing of the particles and lower ventilation effectiveness was achieved (Eslami et al., 2016). Kang et al. (2017) observed the optimization of unidirectional clean air conditioning system using three-stage fans, i.e., fans with three different speed settings, increased the indoor airflow and in turn, the ventilation efficiency in removing pollutants. In a first-of-its-kind study using numerical simulation, Kato et al. (1992) investigated unidirectional flow cleanrooms where supply and exhaust flow rates were balanced locally (Kato and Murakami, 1988). In this study, the ventilation effectiveness was analyzed using the scale of ventilation efficiency (SVE 1-3) (Table S1). Results showed that locally balanced airflow system was more efficient in removing contamination while allowing less extensive diffusion. It was also noted that minor changes in the locations of air inlet/outlet did not affect the ventilation. Whereas a large imbalance between supply and exhaust due to excessive air supply or excessive air exhaust per flow unit might decrease the ventilation efficiency (Kato et al., 1992). Saidi et al. (2011) modeled a full-scale cleanroom numerically to measure the effects of source movement on contaminant spread and introduced performance measurement indices called ‘final efficiency’ and spreading radius. Final efficiency is defined in Table S1. The spreading radius (SR) was calculated from the first moment of the contaminant distribution function as described in Saidi et al. (2011). It was shown that the contaminant source motion and its path impact the contaminant distribution. They recommended shifting the source motion path to the dominant direction of the ventilation airflow for improved contaminant distribution and final efficiency. In terms of configurations, the best overall performance was achieved with the lateral inlet/outlet configuration (Saidi et al., 2011). Chen et al. (2022) conducted a simulation study for an ISO class 8 cleanroom to investigate the appropriate ACH for meeting a self-cleaning time of 20 minutes as required by current good manufacturing practices and analyzed the field distribution of particles for four sets of inlet and outlet setups. The CFD results showed that ACHs below 15 correlate to higher than 20 minutes of self-cleaning time, and increasing ACH can drastically reduce self-cleaning time. Furthermore, raising ACH above 25 has minimal effect on lowering the air cleaning time. It was also concluded that the top supply down return airflow arrangement prevents any specific concentrated area of suspended particles (Chen et al., 2022). Zhou et al. (2020) established a mathematical model to determine particle concentrations and minimum ACH in cleanrooms with non-unidirectional airflows addressing energy efficiency according to the cleanroom air cleanliness class. The mathematical model can take the effects of indoor particle emission rate, return air particle concentration, and ACH, which are primary factors influencing particle concentration (Table S1). It was shown that by increasing ACH from 40 to 120 per hour, the particle concentrations decreased by 65%. Similarly, it was shown that by increasing the ratio between return particle concentrations to indoor particles from 0.7 to 1.3, the concentration decreased by 46% (Zhou et al., 2020). It was established that human occupants were the major source of contamination inside cleanroom and the contamination source strength inside the cleanroom was linear with occupancy (Strauss et al., 2011). In discussions about factors affecting airflow in cleanrooms, Zhao et al. (2004) demonstrated through the simulation study that contaminant concentration varied based on occupant level and furniture inside a cleanroom because obstacles significantly influenced airflow. They provided the particle released rate of 5 × 104 ± 1 × 104 particles min–1 person–1, and recommended considering the generation rate during design (Strauss et al., 2011). In 2019, Ogunsola et al. (2019) ran experiments in Pharmaceutical and Biomedical cleanrooms and found that the generation of particles is process-specific, and biotechnological cleanrooms generate smaller size particles than pharmaceutical cleanrooms). Shao et al. (2022) investigated the change in particle dispersion due to changing the location of the particle generation source. They released particles once at a point in the middle of the cleanroom at the hypothetical upper zone and other times at the top position of the upper zone. Due to non-unidirectional airflow by the FFUs, the patterns of concentration of particles at different points were shown to vary greatly by airflow speed and source location. Switching off the FFUs in the zone without the generation source is more effective than lowering the speed of all FFUs (Shao et al., 2022). The type of inlet grills (with or without diffusers), the temperature difference between the air supply and room temperature, airflow velocity as well as the position of contamination source were studied in a cleanroom by Whyte et al. (2010). It was concluded that the type of air inlet had the most significant influence on the concentration of particles. Although particle concentrations were higher below the air inlet with a 4-way diffuser than with no diffuser inlet, it provided better air mixing within the room (Whyte et al., 2010). Bhattacharya et al. (2020) conducted a series of experiments to understand the effects of door opening, human movement, and pressure differential in containing cross-contamination in a cleanroom. Higher pressure differentials were more effective in preventing particles from entering the cleanroom from the outside. The traffic entering the cleanroom was found to carry more contaminants inside the cleanroom than through the movements out (Bhattacharya et al., 2020). Subsequently, Nikoopayan Tak et al. (2023) conducted an experimental investigation in a cleanroom to determine the impact of pedestrian traffic, airflow rate, and filtration on ventilation performance when the contamination source was within the room. The experiments were conducted in two distinct airflow conditions and with three different occupant movement types. Results showed that increasing the filter efficiency could create flow resistance and hence reduced ventilation effectiveness by almost 50% (Nikoopayan Tak et al., 2023). Yang et al. (2021) showed that the increase in the number of people in the cleanroom elevated the concentration of particles to almost about 2000 particles m–3 with each extra person. They demonstrated that the air supply of the cleanroom could be lowered while preserving cleanliness both in operation and at rest. In the at-rest condition of the cleanroom, the air supply could be cut in half, resulting in 41.7 percent less power supply in this mode (Yang et al., 2021). A comparative study between the displacement and LAF ventilation systems showed that displacement systems yielded a significant correlation between air and surface contamination, which was not apparent for the LAF systems (Friberg and Friberg, 2005). Lin et al. (2010a) studied the ventilation effectiveness of a proposed new ventilation scheme for semiconductor cleanrooms where the cooling load was high. Changing the conventional mixing system to fan dry coil units (FDCUs) placed above the process tools resulted in removing 20–70% more sub-micron particles (Lin et al., 2010a). A similar study by Hu et al. (2014) demonstrated that employing FDCU system resulted in higher ACH and increased particle removal efficiency as FDCU, which removed 50% more 0.1-micron particles than the mixing systems (Hu et al., 2014). For chilled ceiling displacement ventilation type cleanrooms, Kanaan et al. (2014), developed a CFD model to predict airborne bacteria dispersion and suggested utilization of upper-room ultraviolet germicidal irradiation, resulting in improved energy efficiency and acceptable air quality. In other attempts to reduce contaminants from cleanrooms, Sadrizadeh and Holmberg (2015) recommended the use of mobile unidirectional air flow units to reduce microbial concentration when sampling shows unacceptable concentration. The study discovered that the extra mobile ultra-clean unidirectional airflow unit lowers the counts of airborne bacteria and surface contamination to an acceptable level for infection-prone procedures (Sadrizadeh and Holmberg, 2015). Ciuzas et al. (2016) tested the ventilation efficiency as measured by the capability to remove aerosol and VOC particles, for air cleaners combined with ventilation systems. The air cleaners could remove 97% of aerosol particles (21–45% of VOC) from 0.1 to 1.2 micron within 30 minutes when the generation source is inside the room (Ciuzas et al., 2016). Emphasis on reducing fossil fuel energy consumption and greenhouse gas emissions creates a demand for an advanced cleanroom ventilation system. Several techniques have been developed to mathematically describe the thermal and dynamic behaviors of complicated ventilation systems in various operating conditions. Engineers have been able to develop complicated mathematical models and computational simulations to estimate the airflow field, air velocity, and temperature distributions with any control volume. These models have been widely used as a reliable tool for optimizing the system’s performance and sizing. Similarly, the advent of electromechanical actuators allows the integration of linear and non-linear control algorithms in ventilation systems. The classical and modern control strategies have been implemented to minimize actuators’ operating time subject to different ambient conditions. This section aims to review the steps taken in the last decades to optimize ventilation systems subject to cleanroom operating conditions and limitations. Different air distribution systems have been investigated for cleanrooms (Melikov et al., 2011). Simulations based on the actual physics of the ventilation system continue expanding to replace expensive real-world testing. For example, in one of the earlier CFD studies, Quarini et al. (1997) could identify the stagnation areas where the particles may stay, estimate the airflow field and calculate the temperature distribution in cleanrooms (Quarini, 1996). The optimum particle counting sensor location in a rectangular duct (Zhou et al., 2003) and within the room (Mousavi et al., 2020) was numerically modeled. Similarly, computational modeling was used to calculate the optimum locations for inlets and outlets such that suitable air velocity magnitude and contamination level are achieved in an enclosure. The computational results were then validated using experimental testing. Arabian and Brehob (2010) modeled the dispersion of particles using computational simulation to find out the relationship between air cleanness and surgical site infection. Also, the detection of aerosol particles may be monitored to maintain the cleanliness of cleanrooms. An incorrect sensor placement in the ducts may result in faulty results leading to inefficient particle removal. The model, validated by experiments, showed that this approach could significantly reduce the computing time while providing accurate results. Computational modeling could enable the optimization of cleanroom ventilation systems at early design stages (Wrigglesworth, 2002). The evaluation of contaminant distribution in cleanrooms was estimated through combined experimental and numerical work to find the optimum air velocity to achieve the desired contamination level (Maxwell et al., 1994). This study evaluated the contaminant level as a function of supply air velocity, which varies between 0.2 m s–1 and 0.6 m s–1. The concentration of synthetic particles was then experimentally measured using 8 laser-based particle counters. Also, Quarini (1996) explored the optimal areas of recirculation in a semiconductor manufacturing cleanroom. Researchers have also investigated the optimal use of local air-cleaning equipment to reduce contamination. One study showed the tradeoff in the air jet velocity of air cleaners; low velocities were not strong enough to defy the body thermal plume, and too high velocities compromised thermal comfort (Gao and Zhang, 2010). Perhaps to least sophisticated control tool was using a system of dampers to reduce the ventilation systems’ energy consumption (Tschudi et al., 2002). Kim et al. (2005) designed a computer-controlled damper system that features a micro gas sensor with an auto ventilation controller to keep the level of contaminations and toxic gases below the prescribed value. This system traced small variations in toxic gases while automatically removing excess concentrations by allowing more clean air into the room. Koidl (2006) offered a design for high efficiency cleanrooms featuring optimized vacuum pumps with proper air change rates and demand-controlled filtration. The energy efficiency of cleanrooms with five different HVAC configurations has been explored by Hu and Tsao (2007). It was shown that allowing air recirculation could reduce energy consumption by up to 50%. However, particle concentrations in the return air and outdoor air could be filtered independently before the mixing. Substituting the HEPA filters with fine filters for the return air, the system’s energy consumption could be reduced by 45%; It was possible to save energy (33–50%) by reducing the return airflow as the cleanroom could still be maintained under the required concentrations for class 5 ISO (Ma et al., 2020). Zhuang et al. (2021) suggest that coordinating outdoor and supply air ventilation systems is critical for energy-efficient cleanroom operation as the supply air volume could be adjusted based on indoor particle concentration and sensible cooling demand, while the exterior air volume could be handled by an adaptive full-range decoupled ventilation (ADV) system. They demonstrate that a coordinated demand-control ventilation technique can save up to 89.6 percent of reheating and 63.3 percent of total energy (Zhuang et al., 2021). Ma et al. (2021) investigated the potential approaches for reducing energy consumption by doing experiments in four semiconductor cleanrooms. They showed that the cooling, heating, and humidification coils account for 31 to 41.9 percent of the overall pressure drop in a clean makeup air unit (MAU) system and suggested that the outdoor air-treatment process in the MAU system could be modified based on seasonal demand repositioning coils and eliminating the cold-heat offset. This technique could lower airflow resistance by 31.0–42%, resulting in 10–17% energy saving (Ma et al., 2021). Zajic et al. (2010) developed a semi-physical simulation model to reduce the cooling system energy consumption through the regulation of cooling coil unit operation with the integration of theories of classical control. The results showed that the design ventilation system could reduce the concentration of sub-micron-sized particles with lower power usage when compared to traditional ceiling supply systems. A predictive-occupancy system offers demand-controlled ventilation for enhanced efficiency while maintaining the contamination within the prescribed range. This experimental study measured the airborne particle concentration levels using a hand-held particle counter for real-time measurement of particle concentrations and found that in a cleanroom setting, occupants are the primary cause of variable contamination (Strauss et al., 2011). The influence of room ACH and area ratio of raised-floor on particle concentrations have been subject to an experimental study conducted by Khoo et al. (2012). Su and Yu (2013) created a numerical model to find out the optimized differential pressure setpoint which corresponds to the desired temperature and humidity in the cleanroom. The operation of the water pump may be optimized to reach the desired temperature and humidity in a cleanroom. The differential pressure setpoint of the water pump is typically determined through trial and error. However, the pressure is often set excessively to ensure a pass of commissioning tests. The proposed controller loop requires a differential pressure sensor and a variable frequency drive to continuously measure the operating pressure and adjust the pump speed accordingly for maintaining the desired conditions accurately with minimum power. Maintaining the pressure gradient between the inside and the ambient environment is an effective method to prevent external pollution and cross contamination from entering the secured room. An experimental study is performed to control and maintain this pressure gradient through a classical control method which ensures a steady and maintains continuous pressure gradient. The results show that the cleanliness level may be kept by preventing outdoor air from entering the controlled volume while the energy efficiency is enhanced due to the reduced air leakage (Wang et al., 2015). A recent work led by Liu et al. (2021) presented a clean-air conditioning pressure gradient control system of cleanroom for multizone conditioning that can save up to 39.8 percent of power usage due to the reduced fan operation frequency by transitioning from working mode to non-working mode. The improved system includes a makeup air unit (MAU), which works in tandem with the supply and exhaust fans to enable integrated frequency conversion (Liu et al., 2021). In some cases, occupants could be the main contributor to increasing the number of particles in cleanrooms. Loomans et al. (2019) proposed three control strategies, including fine-tuning, demand-controlled filtration, and optimizing airflow patterns to minimize the power consumption while maintaining the minimum required cleanliness. Their experimental and numerical findings showed that there are multiple ways to save energy costs in cleanrooms; however, this study requires further research to investigate if there is a tradeoff between reducing ACR and other cleanroom requirements such as pressure. A systematic literature review was conducted to reevaluate the evolution of research and evidence related to cleanroom ventilation where 186 papers were reviewed. The focal points of this review study were the ventilation requirements, contamination control, and optimizing the ventilation system in cleanrooms. The findings summarized here discussing these aspects of cleanroom ventilation offer insights into future research directions and recognized room for improving the development process of ventilation standards. As different standards strictly dictate the cleanroom ventilation design, it is pertinent and substantiated by research that these standards should constantly evolve to move from best practices to achieve efficient cleanliness. The studies on thermal comfort in cleanroom environments discussed the potential of different ventilation system types. Maintaining thermal uniformity was a general theme for a more comfortable environment inside the cleanrooms. Even though, generally speaking, higher pressure differential leads to better containment of airborne contaminants in cleanrooms, studies demonstrated that evaluating proper pressure differential was crucial not to end up with overdesign and higher energy consumption. Unidirectional flow cleanrooms were found to be most efficient in contaminant removal, and several recommendations were identified to maintain flow uniformity and desired containment whenever possible. Some of those suggestions include maintaining low supply velocity, balancing inlet and exhaust flow rates on a case-by-case basis, optimizing the contaminant source location, and maintaining a sufficient pressure differential for directional airflow were some of the most significant ones. There were a number of alternative methods suggested by the authors to ensure effective contamination control in non-unidirectional cleanrooms as well. Creating microenvironments inside cleanrooms seemed to have attracted significant attention to achieving desired sterility and energy efficiency, instead of altering the flow properties of the entire cleanroom, especially for large-scale manufacturing cleanrooms. A significant proportion of the reviewed manuscripts discussed air supply rates both in the context of maintaining flow uniformity and by extension, for contamination control. Depending on the use of the cleanroom, a wide range of airflow rates have been studied and recommended, but no specific value was pinpointed. This is interesting as this finding demonstrates that the ACH may not be the sole deciding factor for cleanroom ventilation. Instead, a multi-factor examination of the system is a holistic approach for design. Another recurring theme emerged from this review study was that the location of the contaminant source needed to be optimized in line with the principal flow direction for effective contamination control. Additionally, the locations of the inlet and exhaust grilles must be carefully examined when analyzing air distribution in cleanrooms. For example, having the exhaust close to the identified contamination source provides better ventilation performance for contamination control. Proper sizing of filtration systems and ducting was also crucial for satisfactory containment performance (Vlasenko et al., 1998). Occupants have been identified as the most prominent source of contamination, and thus, occupant-centric ventilation design seemed to be an interesting development in the overall design process. It is worth noting that the research on cleanroom ventilation did not stop at discussing ventilation performance in terms of contamination control and air distribution patterns. Significant cognition was also spent on evaluating how the overall system design can be optimized, including the aspect of energy consumption. The optimization of cleanrooms first began in the 1980s with numerical modeling. Numerical simulations provided an accessible alternative to holistically study the airflow properties in cleanrooms, providing a way of obtaining clarity with more data and visualization capabilities. Experimental methods were still prevalent, especially with digital sensors, including fast response thermocouples, humidity sensors, pressure transducers, etc., and were widely employed by many researchers, engineers, and architects. Experimental methods were initially the principal mode of investigating the system’s performance and were then used as a supplementary method to validate numerical and computational results. Since the 2000s, CFD has been used for modeling the ventilation system in cleanrooms to estimate the temperature distribution, airflow pattern, and velocity. The integration of control theories and advanced instruments, coupled with a better understanding of ventilation parameters through computational exercises and visualization, must be used to design and operate future cleanrooms. The authors would like to thank the American Society of Heating, Refrigeration, and Air-conditioning Engineers (ASHRAE) for their initial support and encouragement of this study.1 INTRODUCTION
Fig. 1. Historical evolution in cleanroom ventilation and air distribution systems.
2 METHOD
Fig. 2. Distribution of reviewed articles based on research method.
3 RESULTS
3.1 Ventilation Requirements for Cleanrooms
3.1.1 Thermal comfort
3.1.2 Pressurization and ACHFig. 3. Evolution of ACH in time. Only ACHs retrieved from original works are shown here, please refer to Table S2 for data used to produce this figure. ACH’s greater than 80 were not shown in this diagram for ease of illustration.
3.1.3 Air distribution Systems
3.1.4 Additional approaches to increase ventilation efficiency
3.2 Contamination in CleanroomsFig. 4. Best practices to control contaminants in cleanrooms.
3.3 Optimization and Controls of Cleanroom Ventilation
3.3.1 Optimization
3.3.2 Controls
4 CONCLUSIONS
ACKNOWLEDGMENTS
REFERENCES