Natural sources and experimental generation of bioaerosols: Challenges and perspectives

Abstract

Experimental aerosol generation methods aim to represent natural processes; however, the complexity is not always captured and unforeseen variability may be introduced into the data. The current practices for natural and experimental aerosol generation techniques are reviewed here. Recommendations for best practice are presented, and include characterization of starting material and spray fluid, rational selection of appropriate aerosol generators, and physical and biological characterization of the output aerosol. Reporting of bioaerosol research should capture sufficient detail to aid data interpretation, reduce variation, and facilitate comparison between research laboratories. Finally, future directions and challenges in bioaerosol generation are discussed.

Copyright © 2020 American Association for Aerosol Research

1. Introduction

Bioaerosols are airborne entities that either contain microorganisms or biological materials derived from living organisms, mixed with solids or fluids (Depres et al. 2012). When referring to bioaerosols, it is important to distinguish between two primary types: dry (particles) and liquid bioaerosols (droplets). The latter can remain in liquid phase, or evolve into dry particles depending on their local environment (Lighthart 1994). The nature of the dried bioaerosols resulting from the evaporation of droplets is distinct from that of bioaerosols originally formed as dry particles (Cox 1970 1987). Understanding the nature and impact of bioaerosols requires recognition of interactions between several aspects of the aerosol system, such as (i) source of aerosolized material, (ii) aerosol generation method, (iii) atmospheric transport/processing of bioaerosol particles, (iv) aerosol sampling/deposition, and (v) down-stream quantification techniques. This review focuses on the initial two aspects: sources and mechanisms of generation of bioaerosols. The combination of source and mechanism defines the process of aerosol generation, and the aim in experimental aerosol generation studies should be to mimic the natural process. For example, liquid aerosol generation defines the initial droplet which is the micro-environment where biological components reside (Haddrell and Thomas 2017).

In this review, natural sources and mechanisms of bioaerosol generation in the environment are described with respect to experimental techniques. Recommendations are made regarding experimental design, future direction, and reporting of experimental methods with the aim to aid interpretation of data, and reproducibility between laboratories undertaking bioaerosol research. To fully understand and standardize reporting of bioaerosol data, the entire aerosol system from generation of droplets to sampling and quantification of the biological components requires consideration. In view, readers are directed to reviews in this special issue covering Bioaerosol Research: Methods, Challenges, and Perspectives.

2. Natural sources and generation of bioaerosol

Bioaerosols ejected into the air can be transported thousands of kilometers before deposition and are even found in cloud droplets (Morris et al. 2008). They are thus an integral part of dispersal of biological material across the globe. Natural sources of bioaerosols are diverse, and emissions of bioaerosols into the atmosphere can be segregated into two major categories based on their nature upon formation and ejection from their source: liquid (droplets) or dry (particles). They are formed via a number of mechanisms falling in two groups: natural or anthropogenic systems, including the built environment (Figure 1a). The built environment in which humans live includes buildings, parks, transportation systems, utilities, and associated infrastructure such as ventilation systems (Prussin and Marr 2015). In most cases, an individual microorganism is not launched by itself but may be accompanied by liquid, solutes, and/or abiotic solid material when dispersing short or long distances before eventually depositing. Following deposition, on surfaces, bioaerosols can also be resuspended by airflow, including from wind outdoors, or ventilation such as walking or opening doors indoors. (Bhangar et al. 2016; Inizan 2018). Furthermore, peaks in aerosol concentrations, particle size distributions, and variations in biodiversity are often observed in natural situations where they evolve and vary as a function of time and seasonality (Roses-Codinachs et al. 1992; Burrows et al. 2009; Celenk et al. 2009; Simon and Duquenne 2013; Lofgren et al. 2007; Caliz et al. 2018).

Figure 1. Natural and experimental mechanisms of bioaerosol generation.

2.1. Dry aerosols: Particles

Dry natural aerosols are produced by mechanical processes that can lift biological material from the source surface into the atmosphere by direct suspension or by saltation, as in dust storms or wind in cities (Kellogg and Griffin 2006; Griffin and Kellogg 2004; Griffin et al. 2007; Polymenakou et al. 2008; Soleimani et al. 2016); although the detailed mechanisms behind these processes remain poorly understood, a rich interplay between fluid and mechanical properties of complex matter, fracture, and adhesion are involved (Depres et al. 2012). Dry bioaerosols can be generated from wind-induced mechanical motion, from resuspension (Langre 2008; Inizan 2018), and from wet/liquid mechanically induced processes. An example of the latter is the impact of raindrops on surfaces that can generate convective flows, vortices, or mechanical explosive ejections promoting efficient dispersal of pollen grains and spores (fungal, ferns) (Kim et al. 2019; Roper and Seminara 2019; Trail, Gaffoor, and Vogel 2005; Niklas 1985; MacCartney 1994; Roper et al. 2010).

Dry bioaerosols may also be generated by organisms, including humans, livestock, wildlife, and microorganisms. Humans shed on the order of 10⁸ cells and commensal microorganisms per day, and have been shown to be a major contributor to bioaerosols by release of skin and hair fragments during movement (Milstone 2004; Hospodsky et al. 2012; Adams et al. 2015; Qian et al. 2012; Colbeck and Whitby 2019). Municipal waste handling contributes to substantial amounts of bioaerosols, including bacteria, fungal spores, and endotoxins (Pillai and Ricke 2002; Folmsbee and Strevett 1999). During mechanical agitation in a green compost facility, a 3-log increase in airborne Aspergillus fumigatus concentrations was reported (Taha et al. 2006). Elevated bioaerosol concentrations have also been detected in the air around landfills (Huang et al. 2002; Lis et al. 2004).

2.2. Liquid aerosols: Droplets

Liquid bioaerosols, emitted in the form of droplets, are ubiquitous in the environment as more than 70% of the earth is covered by water and bacteria concentrations in the ocean are >10⁸ cells per liter (Azam et al. 1983). It is therefore likely that the majority of bioaerosols in the atmosphere originate from natural bodies of water, including both saltwater and freshwater systems. Droplet bioaerosols are produced from a range of mechanisms involving fluid fragmentation (Bourouiba and Bush 2013; Eggers and Villermaux 2008). Woodcock and Gifford (1949) were pioneers in showing that the aerosols are generated from sea-spray (Blanchard and Syzdek 1982; Blanchard 1983). The largest aerosols produced from natural bodies of water are typically spume drops with diameters up to a few millimeters (Andreas 1998; Veron 2015).

One of the main aerosolization mechanisms from waters are bursting bubbles. Air bubbles are ubiquitous and populate the surfaces of pools, rain puddles, and wastewater treatment plants, in addition to the oceans and the fresh water bodies. They are produced during wave breaking or during rainfall impacts causing splashes that generate bubbles in addition to droplets, with an estimated 10¹⁹ bubbles created every second in Earth’s oceans and seas (Poulain and Bourouiba 2019). Bubble produce two groups of droplets: jet drops (up to 100 µm) and film drops (up to 10 µm), with jet drops being thought to have narrower size distributions and be overall larger than film drops (Veron 2015). Small bubbles below the capillary length produce jet drops via cavity collapse, while bubbles above the capillary length produce film drops from disintegration and destabilization of their cap (Walls, Bird, and Bourouiba 2014; O'Dowd and De Leeuw 2007; Resch, Darrozes, and Afeti 1986; Löndahl 2014). The produced droplets can carry biological material, salts, and other organic materials (Fitzgerald 1991; O'Dowd et al. 2004), and their load was shown to be directly affected by plankton blooms (Fitzgerald 1991). More recently, it became clear that the natural processes of generation of such droplets involve a subtle interplay between the fluid phase and the biological organisms: the latter can manipulate the underlying interfacial physics to enhance their own dispersal, via bubble bursting by producing secretions that can ultimately change by orders of magnitude, the number, sizes, speed of ejections, and compositions of the droplets launched (Poulain and Bourouiba 2018). Similarly, changes in salinity, volatile compounds, surfactants, and temperature can dramatically modify the production of droplets from bubbles, radically changing the lifetime of bubbles, which in turn controls the number and sizes of droplets (Poulain, Villermaux, and Bourouiba 2018; Poulain and Bourouiba 2019). These processes highlight the importance of the interplay between the organisms and the fluid phase in shaping natural or engineered emission of bioaerosols.

Similarly to gas bubbles that burst, impacts by raindrops can also generate droplet bioaerosols of a wide range of sizes (Fitt, McCartney, and Walklate 1989; Madden 1997; Joung, Ge, and Buie 2017). Splashes generate secondary droplets upon raindrops impacting on soil, vegetation, or animal manure (Aylor 1990; Hau and de Vallavieille-Pope 2006; Dungan 2010; Millner 2009; Gilet and Bourouiba 2014). These splash impactions contribute to disease transmission in crop systems, in addition to dispersal of organisms broadly (Madden 1997; Gilet and Bourouiba 2015; Wang and Bourouiba 2018a; Lejeune, Gilet, and Bourouiba 2018).

Wastewater treatment plants, especially aeration basins, have been shown to release bacteria and viruses into the atmosphere (Wang et al. 2018b; Karra and Katsivela 2007; Sánchez-Monedero et al. 2008). Furthermore, these microorganisms may carry antibiotic resistance genes, facilitating spread of resistance in the environment (Li et al. 2016). Another source are cooling towers that are notorious for being the source of bacteria during many Legionnaires’ Disease outbreaks (Fields, Benson, and Besser 2002; Hamilton et al. 2018; Fitzhenry et al. 2017; Gallagher 2017). Liquid bioaerosols are also released during human hygiene practices such as showering, toilet flushing, and operating taps often due to biofilm growth on appliances and microbes present in water (Johnson et al. 2013a; Barker and Jones 2005; Zhou et al. 2007; Verani, Bigazzi, and Carducci 2014; Feazel et al. 2009; Thomson et al. 2013a; Traverso et al. 2013), resulting in a wide range of droplet sizes (Johnson et al. 2013b; Bollin et al. 1985). Finally, humans and animals produce liquid bioaerosols when breathing, coughing, sneezing, or vomiting (Fabian et al. 2008; Heo et al. 2017; Johnson and Morawska 2009; Xie et al. 2009; Bourouiba, Dehandschoewercker, and Bush 2014; Scharfman et al. 2016; Bourouiba 2016; Alsved et al. 2019). Exhalations are emitted in the form of a discrete high-momentum turbulent cloud, a ‘puff cloud’, that is multiphase, comprised of gaseous phase: the turbulent air exhaled; the liquid phase: the forcibly ejected respiratory tract liquid fragmenting into droplets; and eventually, the solid phase: the droplet residues remaining suspended in the turbulent cloud (Bourouiba, Dehandschoewercker, and Bush 2014; Scharfman et al. 2016; Bourouiba 2016, 2018) Here, similar to wind-dispersal, it is not primarily the droplet characteristics that determine their range of dispersal, but the properties of the turbulent puff cloud laden with the droplets that governs the range of dispersal of the bioaerosols (Bourouiba, Dehandschoewercker, and Bush 2014; Bourouiba 2016, 2018).

3. Experimental generation of bioaerosol

Experimental aerosol generation aims to replicate aspects of natural bioaerosol generation in a controlled environment. These experiments enable greater understanding of airborne phenomena such as viability or transformation of microorganisms and biological material in the atmosphere. Biodiversity and peaks in concentration are often not considered or replicated in laboratory experiments, yet are important to the application of research to real bioaerosols, for example, biocollector efficiencies, risk assessment of microbiological or toxicological exposure and effectiveness of protective technologies against bioaerosols (Simon and Duquenne 2013; Degois et al. 2019). Reproducible particle concentrations in laboratory experiments are typically higher than those that occur naturally to facilitate detection and statistical analysis. Experimental generation of bioaerosol can be considered as a number of steps, each potentially critical to variability and interpretation of data: preparation and storage of material pre-aerosolization, followed by aerosol generation and characterization. It is critical to account for the key interplay between the coupled organisms and the fluid phase in determining selection of the droplet size distributions, numbers, and compositions produced by a given fluid fragmentation process. The underlying physics of the fluid fragmentation, either impact-driven or shear-driven, steady or unsteady, involving Newtonian or non-Newtonian fluids, involves a particular relationship between the properties of the mixture, fluid, and organisms, and the resulting droplets generated to form the final spray.

3.1. Preparation and storage of material

Preparation, storage, and characterization of material before aerosolization is necessary for understanding the outcome of a study, which may vary depending upon the choice of aerosol generation device. For example, pollen and fungal spores represent dried products and therefore should be aerosolized as such, while bacteria and viruses are commonly found naturally in liquid suspensions and are hence aerosolized from liquids. Culture conditions of microorganisms can influence phenotype and survival within experimental systems. Hence, aspects such as media type, cell line, temperature, humidity, pH, aeration, incubation time, and final concentration must be considered as important variables between laboratory studies (Cox, Bondurant, and Hatch 1971; Cox 1987; Handley and Webster 1995; Hogan et al. 2005; Faith et al. 2012). Storage conditions (e.g., media, temperature, humidity, and time) may all influence the quality of the material to be aerosolized and introduce variability. Standard conditions of laboratory preparation may be very different from the nutrient-sparse, stressful conditions of natural environments and this may impact microbial physiology in the spray suspension and in the aerosol phase. Aggregation of particles during storage is a problem for dried material due to the presence of moisture, and various inter-particle attractive forces and can influence efficiency of dispersal (Calvert, Ghadiri, and Tweedie 2009; Masuda 2009).

Spray fluid characteristics such as solute/solvent composition, fluid density, viscosity, and surface tension alter the initial droplet size, hygroscopic growth, and evaporation kinetics (Gilet and Bourouiba 2015; Wang and Bourouiba 2018b; Eggers and Villermaux 2008; Johnson, Pearce, and Esmen 1999; Yang and Marr 2012; Haddrell and Thomas 2017; Vejerano and Marr 2018; Kooij et al. 2018). These characteristics, in turn, influence aerosol transport and atmospheric processing before sampling. For example, differences occur in survival of microorganisms when comparing experimental spray fluids to those replicating natural sources (Barlow and Donaldson 1973; Donaldson 1972; Trouwborst and Kuyper 1974; Ijaz et al. 1985; Lever, Williams, and Bennett 2000; Zuo et al. 2014). The starting concentration and aggregation of biological material within spray fluid impacts particle loading and size. This in turn affects physical deposition rates, aerosol survival, and infectivity (Hogan et al. 2005; Eninger et al. 2009).

Clearly, characterizing the material to be aerosolized in terms of quality such as viability of the organism, stability of chemicals and phase, and presence of aggregations is important for reproducibility. It is recommended that sufficient material be prepared and if necessary aliquoted for the duration of a study to aid consistency. How the biological material used is prepared and stored should be documented and standardized within studies and where appropriate between laboratories. Rationales for selection of methods (including standardization) should be supported by experiments exploring variability within preparation and storage parameters. Experimentally, the physicochemical properties of spray fluid should represent the natural source in experiments aiming to model ambient phenomena. These principles should be considered for all bioaerosols with full details provided in methodology.

3.2. Laboratory liquid bioaerosol generation

3.2.1. Fluid fragmentation into droplets

Selection of an aerosol generation technique involves consideration of several aspects related to both the aerosol and the biological material being investigated. The aerosolization principle being simulated, size and polydispersity of aerosol particles, and required particle concentration are important considerations. Biologically, the quantity of material available, size, and sensitivity of the biological entities, and the stresses to which the bioaerosol is exposed during generation are important (Figure 1b). These aspects are usually dictated by the natural phenomenon under research. Numerous devices for generation of bioaerosols are available (Table 1). However, many are adapted from aerosolization techniques for non-biological material and thus not optimized for bioaerosol studies.

Table 1. Common methods for generation of experimental bioaerosols.

Aerosolization mechanism^a	Example device (s)	Properties of material	Experimental considerations	Particle size^b	Airflow (L/min)	Fluid flow (mL/min)	Material recirculation	References
Liquid atomization
Twin-fluid atomization (pneumatic recirculation)	Collison nebulizer	Liquid suspension	Fluid forces and recirculation may damage some biological material	0.032–1.3 µm (GSD 1.8)^c	2–7 (per jet)	0.02–0.08	Yes	May (1973); Liu and Lee (1975); Reponen et al. (1997); Stone and Johnson (2002); Wang et al. (2004); Yao and Mainelis (2006); Grinshpun et al. (2007); Rule et al. (2009); Turgeon et al. (2014); Zhen et al. (2014); Ibrahim et al. (2015); Bowling et al. (2019)
Twin-fluid atomization (bubble bursting)	Sparging liquid aerosol generators (SLAG)	Liquid suspension	Fluid forces may damage some biological material	<5 µm	2–30	2	Generally no^d	Reponen et al. (1997); Mainelis et al. (2005); Rule et al. (2009); Simon et al. (2011, 2013); Zhen et al. (2014)
Twin-fluid atomization	Centered flow; Single pass aerosolizer	Liquid suspension	Fluid forces may damage some biological material	<2 µm	0.3–1.5	0.2	Generally no^d	Zhen et al. (2014)
Twin-fluid atomization/Jet nebulization	Commercial nebulizers (e.g., Sidestream nebulizer, Pari LC Sprint)	Liquid suspension	May recirculate material; venturi principle	<3.5 µm	<8	8	Yes/No	Najlah et al. (2014); Astudillo et al. (2018); Ghazanfari et al. (2007) Pritchard et al. (2018)
Twin-fluid atomization (flow-focusing)	Flow focusing monodisperse aerosol generator (FMAG)	Liquid suspension	Favourable monodispersity; particle size depends on pressure and flow rate	9–100 µm (initial wet distribution)	300–2000^e	0.035–0.5	No	Ganan-Calvo and Gordillo (2001); Martin-Banderas et al. (2005); Thomas et al. (2008b, 2009); Duan et al. (2016)
Centrifugal nebulization	Spinning top aerosol generators (STAG)	Liquid suspension	Generate larger particle sizes; size range is controlled by modifications such as impaction of material before exit from device	0.95–6.7 µm	N/A	1	No	May (1949, 1966); Ellison (1967); Young, Larson, and Dominik (1974); Mitchell (1984); Eisner and Martonen (1988); Melton, Burnell, and Harrison (1989); Biddiscombe et al. (2006); Bohannon et al. (2015)
Ultrasonic nebulization^f (acoustic waves; vibrating mesh or orifice generation)	Commercial nebulizers (e.g., Sonotek™, DeVilbiss Pulmosonic, Omron microair, Rhône-Poulenc Fisoneb)	Liquid suspension	Conversion of high frequency sound waves into mechanical energy that disrupt liquids into droplets.	1–15 µm (initial wet droplet size 10–60 µm)	<100 where applicable	0.001–2.2	No	Berglund and Liu (1973); Wiedmann and Ravichandran (2001); Katial et al. (2000); Najlah et al. (2014); Ratnesar-Shumate et al. (2011); Santarpia et al. (2012); Dybwad (2010); Dybwad and Skogan (2017); Dabisch et al. (2017)
Vibrating mesh nebulization (active or passive)	Commercial nebulizers (e.g., Omron, Pari, Aeroneb™)	Liquid suspension	May be active or passive. Piezo-element vibrates a mesh substrate contacted by liquid. Movement pushes liquid through mesh into airstream.	<6 µm	<4	N/A	No	Turgeon et al. (2014); Najlah et al. (2014); Allegra et al. (2016); Gowda, Cuccia, and, Smaldone (2017); Ghzanfari et al. (2007); Pritchard et al. (2018); Astudillo et al. (2018); Bowling et al. (2019)
Electrospray ionization	Model 3480, TSI Inc.	Liquid suspension	Very small droplets generated; Reduced virus and protein agglomeration	0.002–2 µm (initial wet droplet size 0.13–5 µm)	<2	0.008–0.03	No	Mulholland, Chen, and Pui (2001); Kim et al. (2008); Eninger et al. (2009); Guha et al. (2012); Almeria and Gomez (2014); Rutkowski et al. (2018); Ganan-Calvo et al. (2018)
Pulsed droplet ejection (piezoelectric or pneumatic)	Ink jet droplet-on-demand dispensing technology	Liquid suspension	Very reproducible monodispersity; Large initial droplets controlled by orifice size	10–400 µm (initial wet droplet size)	N/A	0.0001–0.005	No	Ulmke, Wriedt, and Bauckhage (2001); Amirzadeh and Chandra (2010); Minov et al. (2015); Harris, Liu, and Bush (2015); Vaughn et al. (2016); Ionkin and Harris (2018); Otero-Fernandez et al. (2019)
Solid powder
Air flow dispersal (i.e., ejector or venture)	Powder disperser/Agar tube disperser	Powdered/dried material or fungal spores on agar surface	Airflow on powder/surface	Depends on material size and quality	<5–>15	N/A	No	Cox et al. (1970); Reponen et al. (1996, 1997); Lee et al. (2008); Tang et al. (2008); Jung, Lee, and Kim (2009); Tsai et al. (2012); Tiwari, Fields, and Marr (2013); Pokharel et al. (2019)
Air flow dispersal	Swirling flow disperser	Dried materials or fungal spores on agar surface	Stable aerosol concentrations of smaller spores	As above	15	N/A	No	Reponen et al. (1997); Afanou et al. (2014)
Scraping/mixer	Rotating brush generator (RBG)	Powdered/dried material scraped by blade or impeller	Packing density can vary output	As above	<15	N/A	No	Calvert, Ghadiri, and Tweedie (2009); Masuda (2009); Verreault et al. (2012); Tsai et al. (2012); Wang et al. (2014)
Fluidized bed	Vilnius aerosol generator (VAG); PITT-3	Powdered/dried material forced through bed of heavy particles fluidized by air	Range of aerosol concentration and particle sizes	0.5–40 µm	<10	N/A	No	Marple, Liu, and Rubow (1978); Prenni et al. (2000); Calvert, Ghadiri, and Tweedie (2009); Masuda (2009); Tsai et al. (2012); Harnish et al. (2014)

Notes: Values in table are for standard settings of aerosol generator. For many generators, there is a possibility to vary output by changing operational parameters including flow rate (gas or liquid), disc rotation speed, orifice size, and vibrational frequency.

^a Details of mechanisms can be found within references.

^b Droplet sizes are derived from references. It is often not denoted whether these represent initial or equilibrium sizes. It is anticipated that they represent equilibrium particle size unless stated.

^c Larger dried particles such as fungal spores and beads can be generated up to approximately 5 µm.

^d Some generators operate in both recirculating and non-recirculating modes.

^e Pressure in millibar. N/A, not applicable.

^f Vibrating orifice aerosol generator (VOAG) is no longer commercially available; however, the mechanism is well documented as a means of generating precise aerosol concentration standards in laboratory studies.

Many natural sources of bioaerosol arise from wet environments, as described in the previous section, and these are replicated in the laboratory by fragmentation of liquids. Fragmentation is the breakup of bulk fluid into droplets which occurs when forces imposed on the system overcome surface tension forces that tend to minimize creation of new surface area. The Weber number (We) is the non-dimensional number that quantifies competition between kinetic energy and surface energy, defined as We = (ρv²L)/σ, linking fluid density (ρ), speed (v), length-scale (L), and surface tension (σ). When We is high, creation of new surface in the form of fragmentation of a bulk fluid into droplets is possible (Lefebvre and McDonnell 2017; Bourouiba and Bush 2013). Fragmentation is induced by (i) impacts, transforming a bulk fluid into a sheet, then ligaments, and then droplets via a series of surface-tension dominated interfacial instabilities and processes (Wang and Bourouiba 2017, 2018b; Eggers and Villermaux 2008); (ii) shearing, from an airflow or one fluid moving faster over the interface of another, leading to classical hydrodynamic instabilities (i.e., Kelvin-Helmholtz), resulting in ligament, and then droplet formation (Eggers and Villermaux 2008); (iii) or bubble bursting, leading to the creation of secondary droplets, for example, from film rupture and destabilization into ligaments, and then droplets (Walls, Bird, and Bourouiba 2014; Poulain and Bourouiba 2018).

In these processes, fragmentation is influenced by both the fluid properties (Newtonian vs. non-Newtonian rheology) and the regimes in which fluid destabilization occurs, such as inertial versus viscous regimes, or steady versus unsteady processes (Wang et al. 2018a; Wang and Bourouiba 2018b). A frequently used bioaerosol generator is the twin-fluid Collison nebulizer that generates droplets by physical shearing and impaction onto a vessel wall. The resultant droplets that exit the orifice of the Collison nebulizer eventually become fine dried particles of average sizes generally less than 2 µm (May 1973). The Collison nebulizer has been used to generate larger dried particles including polystyrene latex beads and fungal spores up to around 3 µm diameter (Wang et al. 2004; Yao and Mainelis 2006; Grinshpun et al. 2007); however, particles greater than 5 µm presented problems (Kanaani et al. 2008). Benefits of the Collison nebulizer are user-friendliness, the relatively small volume of material needed, high reproducibility, high particle output (especially for the multiple-jet devices), and its widespread application, which facilitate comparison among studies (May 1973; Liu and Lee 1975; Reponen et al. 1997; Ibrahim et al. 2015). Several other devices use a similar principle of twin-fluid fragmentation (Table 1); however, a main disadvantage of most of these devices is the repeated recirculation of the liquid. Biological material may be damaged and/or lose viability due to repetitive exposure to shear forces during atomization and impaction against the reservoir wall resulting in gradual degradation of the starting material in a time- and pressure-dependent manner (Stone and Johnson 2002; Zhen et al. 2013, 2014; Turgeon et al. 2014; Ibrahim et al. 2015; Haddrell and Thomas 2017; Astudillo et al. 2018; Otero-Fernandez et al. 2019). Recirculation has been found to be most detrimental to microbes with a cell membrane, such as bacteria. Conversely, non-enveloped viruses and bacterial or fungal spores with more rigid exteriors may be less damaged by mechanical forces (Zhen et al. 2014; Turgeon et al. 2014). Even aerosols of lipid-based liposomes and microtubules demonstrate size reduction due to breakage (Niven, Carvajal, and Schreier 1992; Johnson et al. 1999). However, a recent report demonstrates the validity of comparing generation effects across viruses and bacteria. A vibrating mesh aerosol generator when compared with a 3-jet Collison nebulizer demonstrated increased viability for influenza virus, rift valley virus, and encephalitic alphaviruses. This effect was not observed for a vegetative bacterium, Francisella tularensis, aerosolized similarly (Bowling et al. 2019). It is recommended that the effect of aerosol generation be assessed for each bioaerosol, and a range of generation mechanisms assessed to select the least detrimental method (Zhen et al. 2014; Turgeon et al. 2014; Astudillo et al. 2018; Bowling et al. 2019). It is also recommended that the reported methodology should include details of spray system, including nozzle type, characteristic dimensions, and materials of the system (including nozzle/impaction surface), and operating parameters, including fluid composition and associated properties such as liquid and gas flow rates, pressures, timescales, and liquid volumes used. Furthermore, it is recommended that the studies are conducted to better understand the details required to improve repeatability in bioaerosol studies; for example, understanding the effects of fluid composition and associated properties such as static surface tension, density, rheology, and dynamical parameters such as the Weber number.

3.2.2. Replication and validity

Replicating natural bioaerosol formation mechanisms is not simple, but does have the advantage of ruling out a potential source of variation when extrapolating laboratory generated data to natural processes. Pollen and spores often have hydrophobic surfaces, and therefore predominantly reside at liquid–air interfaces (Reponen et al. 1996), where they can be aerosolized by bubble bursting (Simon et al. 2013). Various set-ups have been constructed to replicate natural bubble bursting processes such as ‘bubble tanks’ and film drop generation (Fuentes et al. 2010; Perrott et al. 2017; Alsved et al. 2018; Joung, Ge, and Buie 2017; Poulain and Bourouiba 2018). Additionally, wind tunnels have been used to replicate wind-blown generation of bioaerosol (Taha et al. 2005). A difficulty is that these systems generally produce very low aerosol concentrations.

Replication of the bubble bursting mechanism has been achieved in the laboratory with a scaled device (Reponen et al. 1997) with further enhancement preventing recirculation of biological material (Mainelis et al. 2005) extending the range of experimentation possible in the field (Rule et al. 2009; Simon et al. 2011, 2013; Alsved et al. 2018). Bioaerosol generation using bubbling principles that mimic natural formation is used to study preferential aerosolization mechanisms and to compare the differential behavior of bacterial or viral strains (Perrott et al. 2017; Gauthier-Levesque et al. 2016).

Reducing stress or damage to microbes during the generation process would aid reproducibility in bioaerosol studies and can be achieved in a number of ways. Assessment of operational parameters such as spray fluid (i.e., viscosity or solute composition), air flow velocity/pressure, or spray/drying time, on damage to biological materials will enable selection of appropriate conditions for a particular bioaerosol generation system (Zhen et al. 2014). Additives, such as artificial mucus or allantoic fluid from embryonated eggs (Turgeon et al. 2014) may be included in the nebulization fluid to reduce aerosolization stress and reproduce natural environments. Many spray devices that operate by alternative fragmentation mechanisms in a single-pass mode are now available, where the biological material only passes through the nozzle once before aerosolization with resultant low rates of damage to the material (Table 1). However, even gentle processes may be detrimental and each device requires an assessment against a particular biological material. A potential disadvantage of single-pass techniques is the requirement for larger volumes of valuable starting material, where only a small fraction is aerosolized. Many devices offer advantages over the Collison nebulizer in terms of expanding the initial particle size distributions closer to those represented by natural sources. For example, ultrasonic nebulization and centrifugal atomization techniques have been used to generate reproducible aerosol concentrations and distributions greater than 5 µm (Dybwad and Skogan 2017; Bohannon et al. 2015). Piezoelectric droplet-on-demand generators offer excellent monodispersity in a highly controlled reproducible manner (Ulmke, Wriedt, and Bauckhage 2001; Vaughn, Tracey, and Trevitt 2016; Otero-Fernandez et al. 2019), that provide promise in covering the range of droplets generated from exhalation events (Xie et al. 2009).

Some studies investigate natural processes within laboratory settings, for example, measuring the size and concentration of exhaled bioaerosols generated from human respiratory activities (Fennelly et al. 2004; Wainwright et al. 2009; Xie et al. 2009; Yan et al. 2018; Scharfman et al. 2016; Bourouiba, Dehandschoewercker, and Bush 2014). Such studies provide a better physicochemical representation of the source and regime of dispersal of respiratory bioaerosols; however, there may be aspects worthy of further research such as inter- and intra-personal variability (healthy versus infected, changes over disease course) and interventions that can modify bioaerosol generation (Lindsley et al. 2012; Bischoff et al. 2013; Lofgren et al. 2007; Asadi et al. 2019).

3.3. Laboratory dry bioaerosol generation and validity

In comparison with bioaerosols from liquid, bioaerosols from dried material are much less studied in laboratory settings. However, this mode of generation is important as bioaerosols of pollen grains and fungal/bacterial spores are naturally generated from dry environments due to disturbances by airflows, and frequently exist as aggregates (Niklas 1985; Lacey 1991). Experimental techniques for dry bioaerosol generation often use an airflow pointed toward either powdered material or a sporulating fungal agar culture (Cox et al. 1970; Reponen et al. 1997; Lee et al. 2008), or alternatively using a scraping/brushing mechanism to detach material into an air flow (Wang et al. 2014). Pollen dispersal has been reproduced in laboratory studies by taking grass or catkins from the environment into a controlled chamber and replicating the moisture-drying cycle causing release of pollen (Taylor et al. 2002, 2004). There is great potential to translate techniques developed for aerosolization of non-biological particulates such as metal oxides, nanoparticles, and therapeutics (Tang et al. 2008; Calvert, Ghadiri, and Tweedie 2009; Masuda 2009; Tsai et al. 2012; Tiwari, Fields, and Marr 2013). The primary experimental dispersion mechanisms for dried powders include individual or combinations of the following processes: (1) entrainment into accelerating or decelerating air flow i.e., eductor or venturi, (2) break-up of particles through impaction onto a target that may be stationary or moving i.e., fluidized bed, and (3) mechanical disruption of powdered material into air flow i.e., scraping (Table 1; Calvert, Ghadiri, and Tweedie 2009). In addition, the dried material is required to be delivered to the dispersal mechanism, for example, by a vibrating tray or hopper, before being aerosolized (Calvert, Ghadiri, and Tweedie 2009; Masuda 2009; Pokharel et al. 2019).

In the above processes, there are technical challenges that can affect data interpretation. Consideration of production and storage conditions for dried material is important as humid atmospheres may cause material to clump and compact (Inizan 2018). Thus, storage can affect the flowability of the material within feeding mechanisms and break-up during aerosolization (Gόrny et al. 2002; Masuda 2009). Irrespective of storage conditions, particles <10 µm have high attractive forces between particles and to surfaces that must be overcome for efficient dispersal (Beaudoin et al. 2015), and these effects are highly sensitive to humidity (Inizan 2018). Preparation method can affect the size of the dried particles, for example, milling parameters influenced the final size of Bacillus thuringiensis particles used as a biopesticide (Kim and Je 2012). Methods that use a feeding mechanism should have continuous stable flow of the dried material in quantities that facilitate delivery of a constant concentration of particles to the dispersal mechanism (Masuda 2009; Pokharel et al. 2019). Problems with these factors can provide difficulties in the maintenance of stable particle concentrations in the aerosol, particularly at the beginning of aerosolization, where concentration peaks may occur that subside and stabilize over time (Tang et al. 2008; Calvert, Ghadiri, and Tweedie 2009). Generally introduction into a moving air stream with acceleration through small exits will deagglomerate loosely attached clumps as occurs in devices using the eductor and venturi principles (Masuda 2009; Calvert, Ghadiri, and Tweedie 2009; Tiwari, Fields, and Marr 2013). Issues with agglomeration and generation of particles of the correct size can be reduced by incorporation of systems downstream that remove larger particles such as impactors or cyclones before measurement (Pokharel et al. 2019). The development of the swirling flow disperser overcame this technical challenge for some bioaerosols, delivering stable particle concentrations for up to an hour facilitating reproducibility (Reponen et al. 1997).

3.4. Aerosol characterization

3.4.1. Liquid bioaerosols characteristics at the source

The size distribution of aerosolized biological material is vital for its transport, transformation in the air, and deposition on surfaces or in a sampler. It is not always controlled by the physical size of the organism. An example is that some biological components of aerosol droplets are so small (i.e., viruses and proteins) that the resultant droplet size is initially controlled by the properties of the spray fluid and method of aerosol generation. Thereafter, the resulting droplet size distributions will evolve in a manner specific to the particular droplet fragmentation process. Thus, subtle design differences and variation in operational parameters (e.g., pressure and viscosity) lead to variation in physical aerosol characteristics both at the source and downstream (Gussman 1984; Hogan et al. 2005; Bourouiba and Bush 2013; Poulain and Bourouiba 2018; Wang and Bourouiba 2018b; Wang et al. 2018a). Physical characteristics of the aerosol should be described, including the full droplet size probability density function, the concentration of droplets generated, and for consistency with current literature, the resulting moments of the distribution, such as the mass median aerodynamic diameter (MMAD) or the geometric standard deviation (GSD). Many particle sizing technologies exist and an appraisal is outside scope of this review, other than to strongly recommend that capturing and reporting such information becomes standard across bioaerosol research communities.

3.4.2. Liquid bioaerosols characteristics: Monitoring evolution from source

Liquid aerosolization initially generates wet droplets that evaporate in a time-dependent manner as a function of relative humidity, volatility and composition of the liquid, and temperature of the surrounding environment. At relative humidities below 30%, equilibrium size (dried particle) is reached within a few seconds for micrometer-sized droplets. This process is important because microorganisms often lose viability rapidly within the first minutes of aerosolization (Hayakawa and Poon 1965; Cox 1987), presumably as the initial large droplets evaporate to reach equilibrium size (Haddrell and Thomas 2017). Particle size distribution information is generally collected at a distance from the orifice of the aerosol generator and would be assumed to represent equilibrium particle sizes. However, without monitoring how the distribution changes over distance within the experimental apparatus, this may not be a safe assumption for the reader without explicit statement within an article. Factors that impact evaporation and equilibrium particle size such as humidity, temperature, and how they are controlled within the experimental system should be reported in methodology. Furthermore, it is recommended that consideration be given to how evaporation stresses can impact an experiment with bioaerosol. Monitoring the aerosolized particles for the duration of an experiment can provide information on whether the experimental system and the spray device is performing as expected and aid interpretation of data; for example, nebulizer fluid may evaporate over time causing concentration of solution and a change in particle size distribution (Chen and John 2001).

When monitoring bioaerosol particle size, concentration, and biodiversity in natural environments, temporal and seasonal variations are observed (Burrows et al. 2009; Caliz et al. 2018). Replication of such natural factors in laboratory studies has been rare. However, reproducible and variable peaks in concentration of aerosol particle concentration, with consistent particle median diameter and GSD, have been achieved by varying airflow rate through a liquid bubbling aerosol generator using Escherichia coli and Penicillium brevicompactum spores. Detailed analysis of repeatability and reproducibility was not undertaken, although limited repetitions demonstrated comparable peak concentration intensities (Simon and Duquenne 2013).

3.4.3. Liquid bioaerosol characteristics: Loading and survival

Most devices applicable for bioaerosol generation produce polydisperse size distributions. Also natural bioaerosols are generally polydisperse and understanding where the biological material predominantly resides within the particle size distribution is important for modeling efforts to understand the effects of bioaerosols. Representative experimental setups should replicate the particle size distribution generated by the natural source under investigation. Microbial viability has been found to increase with particle size (Cox 1987; Handley and Webster 1995; Lighthart and Shaffer 1997), potentially due to less surface exposed to the atmosphere (Jones and Harrison 2004), thus generating particles that are different than what is found in nature can introduce bias. Nonetheless, aggregations of cells or cells attached to non-biological particles are found in the atmosphere, where it has been shown that shielding from UV light and desiccation resulted in higher preserved viability (Clauß 2015; DasSarma and DasSarma 2018).

The concentration of biological material in the aerosol divided by the concentration in the original spray fluid derives a term called the spray factor (Bowling et al. 2019). The spray factor thus measures the effect of aerosol generation process on the concentration of biological material in the aerosol. Faith et al. (2012) used spray factor to demonstrate that media composition and relative humidity influences the aerosol stability but not the infectivity of Francisella tularensis. However, spray factor can be a source of variability between laboratories as it is dependent on the sampling method. For example, sampler type, sampling fluid, and distance of sampler from aerosol generator (i.e., time in aerosol phase) can affect recovery efficiency in a species-dependent manner (Marthi et al. 1991; Dabisch et al. 2012a; Dybwad et al. 2014). Care must be taken in interpretation of spray factor values between systems as the derivation is a reflection of both the experimental system (i.e., physical losses) and the susceptibility of biological material (degradation from aerosol generation, sampling and enumeration).

Understanding how differences in experimental design influence spray factor will enable selection of appropriate conditions for a particular biological material and aid inter-laboratory comparison. It would be prudent to characterize the system with a physical tracer to determine losses within the experimental system due to aerosol deposition alongside biological assessment. Examples of physical tracers include chemicals (i.e., fluorescein or uranine), bacterial spores, radiolabeled biological material or liquid, and fluorescent microspheres (Miller et al. 1961; Cox et al. 1970; Zhao et al. 2011; Dabisch et al. 2012b; Bowling et al. 2019). The particle size distribution formed by the physical tracer should be closely matched to the bioaerosol under examination (Bowling et al. 2019). Studies reporting spray factor should include sufficient details of the experimental system including operational parameters of both aerosol generation and sampling techniques to enable reproduction elsewhere. Variation in air and liquid flows, liquid volumes in aerosol generator and sampler, and sampling fluid composition may make inter-laboratory comparison difficult and should be reported in methodology. Comparisons between microbial species should be balanced by the knowledge that the experimental system may influence observed viability measurements (Terzieva et al. 1996; Rule et al. 2009). When it comes to culturability, diluent selection used for enumeration can affect viability (Won and Ross 1966). It would be recommended that the assessment of culturability be conducted across a range of nonselective and supplemented media and diluents while enumerating to ensure the microbes are not adversely influenced.

3.4.4. Dried bioaerosols: Aerosol characterization from source to study

Similar issues described for liquid generated bioaerosols exist for laboratory generated bioaerosols from materials such as dried bacterial or fungal spores, pollen, and dust-associated endotoxins (Thorne 1994; Kim and Je 2012; Afanou et al. 2014; Vimala Devi, Duraimurugan, and Chandrika 2019). Natural bioaerosols cover the whole size range of aerosols, from a few nanometers up to 100 µm, with many of the largest sizes due to the attachment to fragments of biotic or abiotic material (Löndahl 2014; Clauß 2015). Particle concentration and size of the composite dried particles is of utmost importance and should be characterized, preferably at the point where samples are taken for subsequent analysis. As previously mentioned, the quality and method of preparation and storage, as well as choice of aerosol generation will facilitate ease of disaggregation of clumped biomaterial and contribute to reproducibility of the aerosol (Calvert, Ghadiri, and Tweedie 2009; Masuda 2009). Certainly, it is recommended that the preparation and storage methods for dried biological material be standardized to aid flowability and reproducibility of the aerosolized material.

Few studies have been conducted comparing survival of liquid compared with dry aerosolized biological material. However, comparative studies of wet and dried generated bioaerosols demonstrate differences in aerosol survival which infer the preparation, aerosolization, and rehydration during sampling of the biological material which can influence downstream viability (Cox 1970, 1971, 1987; Thorne 1994). These studies demonstrate that the liquid preparations cannot be used as faithful representation of how dried preparations would behave biologically.

3.5. Surrogates and biosafety

Aerosol research with pathogens requires appropriate biocontainment according to the biosafety classification of each microorganism. It is important to evaluate exposure risks of the laboratory personnel. It is recommended that the aerosol chambers are kept under negative pressure, and with HEPA filter exhausts, in case of system failure. Leakage tests should preferably be performed under high pressure to assess worst case leakage rates (Perrott et al. 2017; Verreault et al. 2014). When pathogens at Biosafety Level-2 and higher are nebulized in high concentrations, additional safety precautions are required to prevent exposure (Bohannon et al. 2016; Perrott et al. 2017). Whenever possible, the use of standardized and validated nonpathogenic surrogates is recommended to facilitate aerosol studies. Examples include bacteriophages or nonpathogenic bacteria as surrogates for human or animal pathogens (Turgeon et al. 2016; Bishop and Stapleton 2016).

Another example of surrogate requirement is the representation of biodiversity of natural air samples in the laboratory for better assessment of biological detection systems or understanding occupational risk (Ratnesar-Shumate et al. 2011). The variation and concentrations of individual microbial or pollen taxa from different ecosystems is important in this context. Known concentrations of bacteria, viruses, or fungal spores can be mixed and aerosolized, generating a complex bioaerosol background (Ratnesar-Shumate et al. 2011; Degois et al. 2019). Berchebru et al. (2014) extended the concept demonstrating good reproducibility of production, storage, and reconstitution of a standardized mixture of 10 bacterial species dried under vacuum and stored for up to a year. However, replicating the biodiversity and variability of natural bioaerosols represents a major challenge and would be worth increased research to enable laboratory studies more representative of specific ecosystems.

4. Recommendations for experimental design and reporting of methodology

Understanding and rationale behind selection of an experimental system is critical for interpretation of the final data and future reproducibility. Aerosolization of biological material has a number of steps where recommendations for good research practice and increased reporting of methodologies would enhance reproducibility within bioaerosol research communities.

1. Preparation and storage of material. The quality and reproducibility of the starting material influence variability of data and should therefore be standardized within a laboratory.

   • Species (and strain) specific effects should be considered during preparation/storage of biological material and aerosol generation. For example, storage conditions for E. coli have been demonstrated to affect viability dependent on additives (Clement 1961), and growth conditions (i.e., preparation method and growth phase) can influence aerosol survival (Hood 1961; Dark and Callow 1973; McDermid and Lever 1996; Faith et al. 2012).

   • Preparation and storage methods that are most appropriate for a particular biological material must be based on an experimental rationale that should be reported in methodology.

   • Preparation and storage methodologies should contain sufficient detail to enable direct comparison between experiments and across laboratories.

   • Spray fluid composition should be characterized and reported in the methodology and matched as closely as possible to the complexity of natural phenomenology including chemical composition and concentration of biological material.

   • Aerosol generation. The complexity of natural processes makes replication of all the aspects in aerosol generation difficult. Ensuring the quality of the biological material exiting the aerosolization device and the reproducibility and accuracy of the measurement of the bioaerosol size distributions are perhaps the most critical parts to accurately compare with the natural processes.

   • Design and operational parameters of the bioaerosol generator that influence the size distribution and quality of the aerosolized material should be reported in the methodology including nozzle material and orifice size, spray pressure, spray time, volume, and fluid composition.

   • Furthermore, for natural aerosol generation processes, understanding the impact of dynamic parameters, such as the Weber number, and properties, such as rheology and surface tension, will enable better rationale for the selection of aerosol generators used during laboratory experiments designed to mimic natural outcomes.

   • Controls that allow physical effects of aerosol generation to be separated from biological effects should be included in experimental design. The non-biological tracer may be included in the same spray preparation as the biological material; however, demonstration that the biological material is not adversely affected should be undertaken (Zhao et al. 2011).

2. Aerosol characterization. Physicochemical and biological attributes of the aerosol should be captured in the methodology to enable comparison between laboratories and support data interpretation and utilization within computational models.

   • Particle concentration and entire size distribution and resulting moments such as MMAD or GSD at equilibrium should be recorded and reported, along with the temperature and humidity. Preferably the measurements should be continuous and for the duration of the experiment.

   • Accuracy of sensors is critical for providing correct measurements such as particle size, relative humidity, and temperature. It is recommended that all the measurement equipments are appropriately and periodically checked and calibrated against appropriate standards.

Consideration should be afforded to understanding the quality of the aerosolized biomaterial using additional techniques and assays to explore viability, injury, sub-lethal damage, or death that may affect subsequent long-term characterization studies.

5. Challenges and future directions

The future prospects for bioaerosol research are bright, and there will be great benefits in diverse communities interacting to solve the complex research challenges. Aerosol generation is a key component that extrapolates between the natural environment and generation of experimental data.

1. Principles and mechanisms of natural bioaerosol generation. A greater understanding of the underlying principles and mechanisms that govern natural dispersal of bioaerosol will support rational design of aerosol generators. This could result in laboratory aerosol generators that replicate the entire process or key interfaces between droplet generation mechanism, fluid composition, and local environment (i.e., airflow and humidity) that epitomize the emitted natural bioaerosol (Bourouiba, Dehandschoewercker, and Bush 2014; Bourouiba 2016; Wang and Bourouiba 2018a, 2018b; Wang et al. 2018a; Poulain and Bourouiba 2018, 2019; Poulain, Villermaux, and Bourouiba 2018; Jung et al. 2016; Gilet and Bourouiba 2015; Traverso et al. 2013). Fluid properties and their influence on aerosol generation are less defined and should be the focus of research effort to understand their impact on both natural and experimental bioaerosol generation processes. Examples of fluid properties include rheology, surface tension, density, and associated dynamic parameters that govern the fragmentation of fluid into droplets, such as the Weber number.

2. Rational design of liquid bioaerosol generators. Empirical, theoretical, and modeling tools can be used to mechanistically understand aerosol generation, aerosol output, and limit stress on organisms during generation. Understanding the interfacial physics leading to fluid fragmentation in a range of configurations, such as unsteady processes (Wang and Bourouiba 2018b) or bubble bursting (Poulain, Villermaux, and Bourouiba 2018), is critical to building the foundation required to quantitatively capture, and thus predict, the complex fluid processes governing spray generation. Such understanding can be integrated into high fidelity numerical fluid dynamic simulations of interfacial processes (Popinet 2018) and to accurately assess and optimize the environment bioaerosols experience during atomization (Fife et al. 2005; Ruzycki, Javaheri, and Finlay 2013). Particular attention should be paid to the areas of high velocity and complex flows that may introduce high shear forces onto biological material (Fife et al. 2005). Theoretical and computational modeling and experimentation have to be used iteratively to develop an optimized system for a given application or natural aerosolization process. A range of physical mechanisms should be considered including the interfacial physics and instabilities, shear forces applied to particle or organisms, inertial and turbulent deposition, gravitational settling, and electrostatic effects.

   Recognizing no single aerosol generator will serve all requirements, the generation technique should be designed to meet experimental needs for reproducibility such as requisite output, droplet dispersity, generation of low background of non-biological particulates, and reproducible surface coating of spray fluid components. Ultimately, the process should be as reflective of the natural process as possible.

3. Dried aerosol generation. Laboratory studies with dried bioaerosols are less prevalent than with liquid bioaerosols. Technical difficulties in preparing and storing dried biological material in ways that retain viability and support efficient aerosolization of the material represent challenges. However, parallel fields that regularly study aerosolized dry powdered material offer a rich resource of new technologies for research with dried bioaerosols. Examples include pharmaceutical drug delivery and risk assessment of occupational exposure to aerosolized nanoparticles (Calvert, Ghadiri, and Tweedie 2009; Masuda 2009; Tsai et al. 2012).

4. Physiological and molecular characterization of biological material during aerosol generation. A range of aerosol generation techniques exist (Table 1) and understanding the effect of each aerosol generation mechanism on the particular biological material investigated would enable rational selection of an appropriate device that minimizes damage. Damage could occur at the sub-lethal level and, therefore, alongside traditional culture-based methods, orthogonal techniques for investigating bioaerosols should be used. Microbiology and virology analysis methods offer a variety of assays that target specific cellular functions by dye inclusion/exclusion or PCR methods for genetic analyses (Stone and Johnson 2002; Zhen et al. 2013, 2014; Alsved et al. 2018; Turgeon et al. 2014; Allegra et al. 2016). Molecular tools are progressing rapidly and are being implemented to interrogate bioaerosol diversity and complexity (Brodie et al. 2007), as well as survival and activity during aerosol generation and transport (Ng et al. 2018; Šantl-Temkiv et al. 2018). Molecular techniques could be useful investigational tools for deconvolution of the relative importance of interacting stresses during aerosol generation, transport, and sampling that occur within experimental procedures and that may differ between natural processes. Furthermore, researchers should continually evaluate advances in assay development for probing prokaryotic and eukaryotic physiology (Cao-Huang et al. 2008) to better understand the impact of aerosol generation and facilitate selection of the most appropriate aerosol generation mechanism for a particular bioaerosol.

5. Polydispersity of aerosols. Natural and experimental bioaerosols are generally polydisperse; however, this can present problems in understanding particle size dependent phenomena. Monodisperse generators such as the droplet-on-demand technologies enable refined exploration of aerosols by reducing variability within individual droplets. Next generation dispensers that extend the range of applications for bioaerosols would be beneficial with properties such as operability with highly viscous, complex biological fluids and capacity to generate smaller droplet sizes.

6. Surrogate biomaterial and spray fluid. Surrogates can be used to understand characteristics of hazardous aerosolized material that is needed to develop risk management strategies (Sinclair et al. 2012; Berchebru et al. 2014; Turgeon et al. 2014; Bishop and Stapleton 2016). Selection of an appropriate surrogate should be tailored to the properties investigated such as aerodynamic properties (Phillpotts et al. 2010; Turgeon et al. 2016; Dybwad and Skogan 2017). However, care must be taken in extrapolating data between surrogate and infectious material to prevent under- or over-estimation of risk (Sinclair et al. 2012). Ideally, parallel studies should be performed that evidence extrapolation between the surrogate and the pathogenic material. Numerous studies have developed respiratory secretion liquids that simulate the chemical composition of natural secretions (Bose et al. 2016; Pytko-Polonczyk et al. 2017), and these may be used as spray fluid surrogates in aerosol studies. However, research is required to better understand and reflect on effects such as the health status and subject variability which can influence the respiratory secretion composition (i.e., mucin type and content). Furthermore, research to better identify and standardize surrogates would benefit communities involved in risk management. Indeed, studies using human or cell-line–derived respiratory secretions have demonstrated increased survival of influenza virus deposited as droplets onto banknotes (Thomas et al. 2008a) or as aerosol droplets (Kormuth et al. 2018, 2019).

7. Understanding and representing variation and diversity in bioaerosols. Natural bioaerosols are complex mixtures of many microbial species in addition to other biological components (i.e., allergens, endotoxins, and glucans) and abiotic material (i.e., dust). Understanding the health effects of airborne bioaerosols such as allergens would be supported by correlative laboratory studies (Douwes et al. 2003). Microbiome analysis has indicated the large biodiversity present in aerosols (Caliz et al. 2018; Li et al. 2019; Mescioglu et al. 2019); however, few studies have attempted to replicate this complexity in the laboratory (Ratnesar-Shumate et al. 2011; Degois et al. 2019). Including representative dried abiotic material, for example, dust from livestock or composting facilities, is advantageous in generating a more realistic bioaerosol. Some studies have replicated this complexity by collecting samples from the natural environment including the microbiological flora to use in laboratory aerosol studies. For example, compostable waste, water from lakes and seas, and soils for replicating splashes from rain drops have been aerosolized in laboratory settings (Heldal et al. 2001; Aller et al. 2005; Joung and Buie 2015; May et al. 2018; Joung, Ge, and Buie 2017).

8. Droplet microfluidic platforms. Microfluidics though underutilized within the aerosol community, offer an attractive platform for the development of novel droplet generators and understanding the complexity of natural droplet formation (Cole et al. 2017; Metcalf, Narayan, and Dutcher 2018). Favorably, monodispersity can be achieved with microfluidic nebulizers due to the confined geometry of the microfluidic channels (Anna 2016).

9. Diversity of aerosol research. A major challenge is the diversity of aerosol research and the translation of advances in related areas that could benefit bioaerosol research. Opportunities created through conferences, symposia, and meetings that bring together different communities will facilitate awareness, interaction, and integration between the fields. Pertinent examples include advances in physics of complex fluid fragmentation, pharmaceutical delivery methods, microfluidics, and molecular analysis techniques. Structured training of aerosol scientists to incorporate interfaces to other disciplines would facilitate multidisciplinary knowledge and skills.

6. Conclusions

Bioaerosol research is challenging due to the number of points within the experimental system where variability can be introduced and due to the dynamic nature of aerosol formation and evolution over time and distance. Despite this, recommendations are proposed for best research practice and rigor in methodological reporting of bioaerosol research. This aims to minimize variation in bioaerosol data, aid data interpretation by experimenters and reviewers, and prompt greater representation of the natural aerosol generation process that is researched. Examples include consideration and characterization of the starting material, a robust rationale for selection of an aerosol generator, consideration of the impact of the aerosol generation mechanism has on the biological material and appropriate characterization and reporting of the output aerosol.

Developments in related research fields will undoubtedly offer a rich scope for designing aerosol generators that are specific to a research challenge and to improve our understanding of the impacts natural and experimental generation techniques have on biological material. Efforts should be placed on maintaining and enhancing the multidisciplinary skills required in future bioaerosol research.

Acknowledgments

The authors acknowledge organizers of this special issue ‘Bioaerosol Research: Methods, Challenges and Perspectives’, including Shanna Ratnesar-Shumate and Alex Huffman, based on requests from the Bioaerosol Working Group and after discussion during the Bioaerosol Standardization Workshop at the International Aerosol Conference in St. Louis, Missouri, in September 2018.

Additional information

Funding

This work was supported by the Ministry of Defense, UK (to R.J.T. and S.T.P.); the MIT-Lincoln Laboratory, Ferry Fund, and the Richard and Susan Smith Family Foundation (to L.B.); and the Swedish Research Council FORMAS and AFA insurance (to M.A. and J.L.).