Systematic Characterization and Fluorescence Threshold Strategies for the Wideband Integrated Bioaerosol Sensor (WIBS) Using Size-Resolved Biological and Interfering Particles

Atmospheric particles of biological origin, also referred to as bioaerosols or primary biological aerosol particles (PBAP), are important to various human health and environmental systems. There has been a recent steep increase in the frequency of published studies utilizing commercial instrumentation based on ultraviolet laser/light-induced fluorescence (UV-LIF), such as the WIBS (wideband integrated bioaerosol sensor) or UV-APS (ultraviolet aerodynamic particle sizer), for bioaerosol detection both outdoors and in the built environment. Significant work over several decades supported the development of the general technologies, but efforts to systematically characterize the operation of new commercial sensors have remained lacking. Specifically, there have been gaps in the understanding of how different classes of biological and non-biological particles can influence the detection ability of LIF instrumentation. Here we present a systematic characterization of the WIBS4A instrument using 69 types of aerosol materials, including a representative list of pollen, fungal spores, and bacteria as well as the most important groups of non-biological materials reported to exhibit interfering fluorescent properties. Broad separation can be seen between the biological and non-biological particles directly using the five WIBS output parameters and by taking advantage of the particle classification analysis introduced by Perring et al. (2015). We highlight the importance that particle size plays on observed fluorescence properties and thus in the Perring-style particle classification. We also discuss several particle analysis strategies, including the commonly used fluorescence threshold defined as the mean instrument background (forced trigger; FT) plus 3 standard deviations (σ) of the measurement. Changing the particle fluorescence threshold was shown to have a significant impact on fluorescence fraction and particle type classification. We conclude that raising the fluorescence threshold from FT + 3σ to FT + 9σ does little to reduce the relative fraction of biological material considered fluorescent but can significantly reduce the interference from mineral dust and other non-biological aerosols. We discuss examples of highly fluorescent interfering particles, such as brown carbon, diesel soot, and cotton fibers, and how these may impact WIBS analysis and data interpretation in various indoor and outdoor environments. The performance of the particle asymmetry factor (AF) reported by the instrument was assessed across particle types as a function of particle size, and comments on the reliability of this parameter are given. A comprehensive online supplement is provided, which includes size distributions broken down by fluorescent particle type for all 69 aerosol materials and comparing threshold strategies. Lastly, the study was designed to propose analysis strategies that may be useful to the broader community of UV-LIF instrumentation users in order to promote deeper discussions about how best to continue improving UV-LIF instrumentation and results. Published by Copernicus Publications on behalf of the European Geosciences Union. 4280 N. J. Savage et al.: Characterization and fluorescence threshold strategies for the WIBS-4A

i.e. users often must write data analysis code themselves and processing large data sets can push 148 the limits of standard laboratory computers. Discriminating based on fluorescence intensity also 149 requires more detailed investigations into the strategy by which fluorescent thresholds can be 150 applied to define whether a particle is considered fluorescent. Additionally, relatively little 151 attention has been given to the optical properties of non-biological particles interrogated by the 152 WIBS and to optimize how best to systematically discriminate between biological aerosol of 153 interest and materials interfering with those measurements. 154 Here we present a comprehensive and systematic laboratory study of WIBS data in order to 155 aid the operation and data interpretation of commercially available UV-LIF instrumentation. This 156 work presents 69 types of aerosol materials, including key biological and non-biological the electrical pulse detected by a photomultiplier tube (PMT) located at 90 degrees from the laser 172 beam. Particles whose measured cw laser-scattering intensity (particle size) exceed user-173 determined trigger thresholds will trigger two xenon flash lamps (Xe1 and Xe2) to fire in 174 sequence, approximately 10 microseconds apart. The two pulses are optically filtered to emit at 175 280 nm and 370 nm, respectively. Fluorescence emitted by a given particle after each excitation 176 pulse is detected simultaneously using two PMT detectors. The first PMT is optically filtered to 177 detect the total intensity of fluorescence in the range 310-400 nm and the second PMT in the 178 range 420-650 nm. So for every particle that triggers xenon lamp flashes, Xe1 produces a signal 179 in the FL1 (310-400 nm) and FL2 (420-650 nm) channels, whereas the Xe2 produces only a 180 signal in the FL3 (420-650 nm) channel because elastic scatter from the Xe2 flash saturates the 181 first PMT. The WIBS-4A has two user defined trigger thresholds, T1 and T2 that define which 182 data will be recorded. Particles producing a scattering pulse from the cw laser that is below the 183 T1 threshold will not be recorded. This enables the user to reduce data collection during 184 experiments with high concentrations of small particles. Particles whose scattering pulse exceeds 185 the T2 threshold will trigger xenon flash lamp pulses for interrogation of fluorescence. Note that 186 the triggering thresholds mentioned here are fundamentally different from the analysis thresholds 187 that will be discussed in detail later.  (1) 193 This parameter relates to a rough estimate of the sphericity of an individual particle by 194 measuring the difference of light intensity scattered into each of the four quadrants. A perfectly 195 spherical particle would theoretically exhibit an AF value of 0, whereas larger AF values greater 196 than 0 and less than 100, indicate rod-like particles (Kaye et al., 1991;Gabey et al., 2010;Kaye et 197 al., 2005). It is important to note that this parameter is not rigorously a shape factor like used in with fungal spores during their ambient study and because they observed that these particles 255 scaled more tightly with observed ice nucleating particles. The authors classified a particle in the 256 FP3 category if the fluorescence intensity in FL1 > 1900 arbitrary units (a.u) and between 0-500 257 a.u for each FL2 and FL3. All materials utilized, including the vendors and sources from where they were acquired, 262 have been listed in supplemental Table S1, organized into broad particle type groups: biological 263 material (fungal spores, pollen, bacteria, and biofluorophores) and non-biological material (dust, 264 humic-like substances or HULIS, polycyclic aromatic hydrocarbons or PAHs, combustion soot 265 and smoke, and miscellaneous non-biological materials). Combustion soot and smoke are 266 grouped into one set of particles analyzed and are hereafter referred to as "soot" samples.    with dimensions 20.5 L x 10.25 H x 12.5 W in (supplemental Fig. S1). Soft rubber beading seals 293 the top panel to the walls, allowing isolation of air and particles within the chamber. Two tubes 294 are connected to the lid. The first delivers pressurized and particle-free air through a bulkhead  For each experiment, an agar plate with a mature fungal colony was sealed inside the 301 chamber. The air delivery nozzle was positioned so that a blade of air was allowed to approach 302 the top of the spore colony at a shallow angle in order to eject spores into an approximately 303 horizontal trajectory. The sample collection tube was positioned immediately past the fungal 304 plate to aspirate aerosolized fungal particles. Filtered room air was delivered by a pump through 305 the aerosolizing flow at approximately 9 -15 L/min, varied within each experiment to optimize 306 measured spore concentration. Sample flow was 0.3 L/min into the WIBS and excess input flow 307 was balanced by outlet through a particle filter connected through a bulkhead on the top plate.

308
Two additional rubber septa in the top plate allow the user to manipulate two narrow metal 309 rods to move the agar plate once spores were depleted from a given region of the colony. After 310 each spore experiment, the chamber and tubing was evacuated by pumping for 15 minutes, and 311 all interior surfaces were cleaned with isopropanol to avoid contamination between samples.  The setup was modified (method P2) for a small subset of samples whose solid powder was 336 sufficiently fine to produce high number concentrations of submicron aerosol particles that could 337 risk coating the internal flow path and damaging optical components of the instrument. In this 338 case, the same small vial with powder and stir bar was placed in a larger reservoir (~0.5 L), but 339 without vial lid. The lid of the larger reservoir was connected to filtered air input and an output 340 connection to the instrument. The additional container volume allowed for greater dilution of 341 aerosol before sampling into the instrument.    Once the flame from the combusting sample was naturally extinguished, the smoldering sample 361 was waved at a height ~5 cm above the WIBS inlet for 3-5 minutes during sampling.   The WIBS is routinely used as an optical particle counter applied to the detection and 372 characterization of fluorescent biological aerosol particles (FBAP). Each interrogated particle 373 provides five discreet pieces of information: fluorescence emission intensity in each of the 3 374 detection channels (FL1, FL2, and FL3), particle size, and particle asymmetry. Thus, a thorough 375 summary of data from aerosolized particles would require the ability to show statistical 376 distributions in five dimensions. As a simple, first-order representation of the most basic 377 summary of the 69 particle types analyzed, Figure 2 and Table 1 show median values for each of 378 the five data parameters plotted in three plot styles (columns of panels in Fig. 2).

379
For the sake of WIBS analysis, each pollen type was broken into two size categories, because 380 it was observed that most pollen species exhibited two distinct size modes. The largest size mode 381 peaked above 10 µm in all cases and often saturated the sizing detector (see also fraction of 382 particles that saturated particle detector for each fluorescence channel in Table 2). This was 383 interpreted to be intact pollen. A broad mode also usually appeared at smaller particle diameters 384 for some pollen species, suggesting that pollen grains had ruptured during dry storage or through 385 the mechanical agitation process. This hypothesis was supported by optical microscopy through 386 which a mixture of intact pollen grains and ruptured fragments were observed (Fig. S2). For the 387 purposes of this investigation, the two modes were separated at the minimum point between 388 modes in order to observe optical properties of the intact pollen and pollen fragments separately.

389
The list number for each pollen (Tables 2, S1) is consistent for the intact and fragmented species, 390 though not all pollen exhibited obvious pollen fragments.

391
The WIBS was developed primarily to discriminate biological from non-biological particles,   whereas the median FL1 intensity is 543 a.u., at which point there is no specific peak. In this 482 way, the median value only broadly represents the data by weighting both the broad distribution 483 and saturating peak. To complement the median values, however, Table 1 also shows the fraction 484 of particles that were observed to saturate the fluorescence detector in each channel.

485
The representation of median values for each of the five parameters (Fig. 2) shows broad 486 separation between particle classes, but discriminating more finely between particle types with 487 similar properties by this analysis method can be practically challenging. Rather than 488 investigating the intensity of fluorescence emission in each channel, however, a common method 489 of analyzing field data is to apply binary categorization for each particle in each fluorescence and Dust 4) are relatively similar and show ~75% fluorescent particles <4 µm, with particle type 512 divided nearly equally across the A, B, and AB particle types (Fig. S4I). The two others (Dust 2 513 and Dust 6) show very few similarities between one another, where Dust 2 shows size-dependent 514 fluorescence and Dust 6 shows particle type A and B at all particle sizes (Fig. S4I). As seen by 515 the median fluorescence intensity representation (Fig. 2, Table 1), however, the relative intensity 516 in each channel for all dusts is either below or only marginally above the fluorescence threshold.

517
Thus, the threshold value becomes critically important and can dramatically impact the 518 classification process, as will be discussed in a following section. Similarly, HULIS 5 ( Fig. S4K) 519 is the one HULIS type that shows an anomalously high fraction of fluorescence, and is 520 represented by B, C, BC particle types, but at intensity only marginally above the threshold value 521 and at 0% detector saturation in each channel. analyzed, four showed >69% of particles to be fluorescent at sizes >4 µm, most of which are 528 dominated by B particle types. Two samples of combustion soot are notably more highly 529 fluorescent, both in fraction and intensity. Soot 3 (fullerene soot) and Soot 4 (diesel soot) show 530 FL1 intensity of 318 a.u. and 751 a.u., respectively, and are almost completely represented as A 531 particle type. The fullerene soot is not likely a good representative of most atmospherically 532 relevant soot types, however diesel soot is ubiquitous in anthropogenically-influenced areas 533 around the world. The fact that it exhibits high median fluorescence intensity implies that 534 increasing the baseline threshold slightly will not appreciably reduce the fraction of particles 535 categorized as fluorescent, and these particles will thus be counted as fluorescent in many 536 instances. The one type of wood smoke analyzed (Soot 6) shows ca. 70% fluorescent at >4 µm, 537 mostly in the B category, with moderate to low FL2 signal, and also presents similarly as 538 cigarette smoke. Additionally, the two smoke samples in this study (Soot 5, cigarette smoke and 539 Soot 6, wood smoke) share similar fluorescent particle type features with two of the brown AB, and ABC as particle size gets larger. Pollen 9 (Phleum pretense) has a physical diameter of 550 ~35 µm, so the mode seen in Figure 3a may be a result of fragmented pollen and due to the upper 551 particle size limit of WIBS detection, intact pollen cannot be detected (Pöhlker et al., 2013).

552
Pollen 8 (Fig. 3d) shows a mode peaking at ~10 µm in diameter and comprised of a mixture of 553 B, AB, BC, and ABC particles as well as a larger particle mode comprised of ABC particles. The 554 large particle mode appears almost monodisperse, but this is due to the WIBS ability to sample 555 only the tail of the distribution due to the upper size limit of particle collection (~20 µm as  An extension of observation from the many particle classes analyzed is that particle type (A, 569 AB, ABC, etc.) varies strongly as a function of particle size. This is not surprising, given that it 570 has been frequently observed and reported that particle size significantly impacts fluorescence  The general trend of fluorescence dependence on size is less pronounced for FL1 than for 583 FL2 and FL3. This can be seen by the fact that the scatter of points along the FL1 axis in Figure   584 2b is not clearly size-dependent and is strongly influenced by particle type (i.e. composition 585 dependent). In Figure 2c, however, the median points cluster near the vertical (size) axis and 586 both FL2 and FL3 values increase as particle size increases. It is important to note, however, that 587 the method chosen for particle generation in the laboratory strongly impacts the size distribution shown here is not a key message, the relative fluorescence at a given size can be informative. As discussed, each individual particle shows increased probability of exhibiting fluorescence 594 emission above a given fluorescence threshold as size increases. Using Pollen 9 (Phleum 595 pratense, Fig. 3a) as an example, most particles <3 µm show fluorescence in only the FL1 596 channel and are thus classified as A-type particles. For the same pollen, however, particles ca. 2-597 6 µm in diameter are more likely to be recorded as AB-type particles, indicating that they have 598 retained sufficient FL1 intensity, but have exceeded the FL2 threshold to add B-type 599 fluorescence character. Particles larger still (>4 µm) are increasingly likely to exhibit ABC 600 character, meaning that the emission intensity in the FL3 channel has increased to cross the 601 fluorescence threshold. Thus, for a given particle type and a constant threshold as a function of 602 particle size, the relative breakdown of fluorescence type changes significantly as particle size 603 increases. The same general trend can be seen in many other particle types, for example Pollen 8 604 (Alnus glutinosa, Fig. 3d), Fungi 1 (Aspergillus brasiliensis, Fig. 3b), and to a lesser degree 605 HULIS 3 (Suwannee fulvic acid, Fig. 3j) and Brown Carbon 2 (Fig. 3i). The "pathway" of 606 change, for Pollen 9, starts as A-type at small particle size and adds B and eventually ABC 607 (AABABC), whereas Pollen 8 starts primarily with B-type at small particle size and 608 separately adds either B or C en route to ABC (BAB or BCABC). In this way, not only is 609 the breakdown of fluorescence type useful in discriminating particle distributions, but the 610 pathway of fluorescence change with particle size can also be instructive.

611
To further highlight the relationship between particle size and fluorescence, four kinds of 612 particles (Dust 2, HULIS 5, Fungi 4, and Pollen 9) were each binned into 4 different size ranges, 613 and the relative number fraction was plotted versus fluorescence intensity signal for each channel 614 (Fig. 4). In each case, the fluorescence intensity distribution shifts to the right (increases) as the 615 particle size bin increases. This trend is strongest in the FL2 and FL3 (middle and right columns 616 of Fig. 4) for most particle types, as discussed above.

617
The fact that particle fluorescence type can change so dramatically with increasing particle 618 size becomes critically important when the Perring-style particle type classification is utilized for

Fluorescence threshold defines particle type 629
Particle type analysis is not only critically affected by size, but also by the threshold 630 definition chosen. Figure 5 represents the same matrix of particle types as in Figure 3  Biofluorophore 11 (tryptophan) follows the pathway ABCABC.

647
By extension, the choice of threshold bears heavily on how a given particle breakdown 648 appears and thus how a given instrument may be used to discriminate between biological and 649 non-biological particles. A commonly made assumption is that particles exhibiting fluorescence 650 by the WIBS (or UV-APS) can be used as a lower limit proxy to the concentration of biological 651 particles, though it is known that interfering particle types confound this simple assumption  (Fig. 6d), by increasing the threshold from 3σ (red traces) to 6σ (orange traces), 661 the fraction of dust particles fluorescent in FL1 decreases from approximately 50% to 10%.

662
Increasing the fluorescence threshold even higher to 9σ, reduces the fraction of fluorescence to 663 approximately 1%, thus eliminating nearly all interfering particles of Dust 3. In contrast, for 664 biological particles such as Pollen 9 (Fig. 6b), increasing the threshold from 3σ to 9σ does very 665 little to impact the relative breakdown of fluorescence category or the fraction of particles 666 considered fluorescent in at least one channel. Changing threshold from 3σ to 9σ decreases the 667 FL1 fraction minimally (98.3% to 97.9%), and for FL2 and FL3 the fluorescence fraction 668 decreases from 90% to 50% and from 60% to 42%, respectively. Figure 6 also underscores how 669 increasing particle size affects fluorescence fraction, as several particle types (e.g. Pollen 9 and 670 HULIS 5) show sigmoidal curves that proceed toward the right (lower fraction at a given size) as 671 the threshold applied increases and thus removes more weakly fluorescent particles.

672
To better understand how the different thresholding strategies qualitatively change the 673 distribution of particle fluorescence type, Figure 7 shows stacked fluorescence type distributions 674 for each of the four thresholds analyzed. Looking first at Dust 3 (Fig. 7d), the standard threshold 675 definition of 3σ shows approximately 80% of particles to be fluorescent in at least one channel, 676 resulting in a distribution of predominantly A, B, and AB-type particles. As the threshold is 677 increased, however, the total percentage of fluorescent particles decreases dramatically to 1% at 678 9σ and the particle type of the few remaining particles shifts to A-type particles. A similar trend 679 of fluorescent fraction can also be seen for Soot 6 (wood smoke) and Brown Carbon 2, where

748
By analyzing the five WIBS data parameter outputs for each interrogated particle, we have 749 summarized trends within each class of particles and demonstrated the ability of the instrument 750 to broadly differentiate populations of particles. The trend of particle fluorescence intensity and 751 changing particle fluorescence type as a function of particle size was shown in detail. This is 752 critically important for WIBS and other UV-LIF instrumentation users to keep in mind when 753 analyzing populations of unknown, ambient particles. In particular, we show that the pathway of 754 fluorescence particle type change (e.g. A  AB  ABC or B  BC  ABC) with increasing 755 particle size can be one characteristic feature of unique populations of particles. When 756 comparing the fluorescence break-down of individual aerosol material types, care should be 757 taken to limit comparison within a narrow range of particle sizes in order to reduce complexity 758 due to differing composition or fluorescence intensity effects.

759
The fluorescence threshold applied toward binary categorization of fluorescence or non-760 fluorescent in each channel is absolutely critical to the conceptual strategy that a given user 761 applies to ambient particle analysis. A standard WIBS threshold definition of instrument 762 background (FT baseline) + 3σ is commonly applied to discriminate between particles with or 763 without fluorescence. As has been shown previously, however, any single threshold confounds while the fraction of interfering material is likely to be reduced almost to zero for most particle 782 types. Several materials exhibiting outlier behavior (e.g. HULIS 5, diesel soot) could present as 783 false positive counts using almost any characterization scheme. It is important to note that 784 HULIS 5 was one of a large number of analyzed particle types and in the minority of HULIS 785 types, however, and it is unclear how likely these highly fluorescent materials are to occur in any 786 given ambient air mass. More studies may be required to sample dusts, HULIS types, soot and 787 smoke, brown organic carbon materials, and various coatings in different real-world settings to 788 better understand how specific aerosol types may contribute to UV-LIF interpretation at a given 789 study location. We also included a comprehensive supplemental document including size 790 distributions for all 69 aerosol materials, stacked by fluorescent particle type and comparing the 791 FT + 3σ and FT + 9σ threshold strategies. These figures are included as a qualitative reference 792 for other instrument users when comparing against laboratory-generated particles or for use in 793 ambient particle interpretation.

794
It should be noted, however, that the presented assessment is not intended to be exhaustive, The study presented here is meant broadly to achieve two aims. The first aim is to present a 811 summary of fluorescent properties of the most important particle types expected in a given 812 sample and to suggest thresholding strategies (i.e. FT + 9σ) that may be widely useful for 813 improving analysis quality. The second aim is to suggest key analysis and plotting strategies that 814 other UV-LIF, especially WIBS, instrumentation users can utilize to interrogate particles using 815 their own instruments. By proposing several analysis strategies we aim to introduce concepts to 816 the broader atmospheric community in order to promote deeper discussions about how best to 817 continue improving UV-LIF instrumentation and analyses.     Table 1