Simultaneous inboard and outboard, inflight measurements of ultrafine particle concentrations

Abstract Inboard and outboard particle number concentrations are reported from 12 flights from the FAAM BAe-146. The data show that the concentrations pre take off and post landing are dominated by external sources. The concentrations in the cabin are generally lower than outside and pre-take off peak after the doors have closed in 75% of cases. In flight cabin events were associated with either ingress of external particles, or particles generated by the airframe. The airframe events were sporadic in nature, and often associated with changing engine conditions. In over 50% of cases, in cabin particle events were observed on approach, with no increase in outboard concentrations. Supporting aerosol composition data showed that non-exhaust, non-airport sources also influence the cabin air quality, including the presence of lubrication oil. The average number concentration exposure was between 6800 and 10,800 cm−3 per flight and, during a fume event, the average organic mass concentration exposure was 5.85 µg m−3 for the flight. The mass concentration exposure and associated average concentration are below UK workplace exposure limits for non-hazardous dusts and PM10 24-h air quality standards. Copyright © 2021 American Association for Aerosol Research


Introduction
Since 2017, the number of passengers traveling on scheduled flights per annum has exceeded 4 billion (4 Â 10 9 ; International Civil Aviation Organisation [ICAO]; International Air Transport Association Report [IATA]). These passengers are exposed to a relatively poorly characterized environment within aircraft, which sit at a unique indoor-outdoor interface when stationary on the ground. In addition, unlike domestic and occupational environments, there are currently no particle exposure limits for the cabin environment.
Early work on cabin air quality (CAQ) found that the predominant source of particles was from passengers in the years before the smoking ban on aircraft (for example, Dechow, Sohn, and Steinhanses 1997). More recent work on CAQ has been driven either by a need to understand the effects of particulates on an aircraft's Environmental Control System (ECS; Cao et al. 2017Cao et al. , 2018 or concerns about aircrew health during so-called fume events (Winder, Fonteyn, and Balouet 2002). During flight, to compensate for the reduced pressure, compressed outside air is fed via the bleed air system to the cabin (see Cao et al. 2017 and reference therein), and concerns have been raised as to whether engine exhaust products, hydraulic fluids, and/or lubrication oil (present either as oil or a decomposed by-product) can contaminate the cabin air via the bleed air supply, causing a fume event (Howard et al. 2018).
were higher than the supply air to the cabin (a combination of the filtered, recirculated cabin air, and the bleed air), concluding the source to be passenger generated aerosol. Zhai, Li, and Zhao (2014) concluded the same based on measurements of the supply air for particles >0.3 mm in diameter from nine flights. In contrast, Li et al. (2014) reported that the main source of particles in the cabin air was from unfiltered bleed air based on nine flights. In all studies, the concentrations of ultrafine particles (UFP) and larger particles are variable and vary flight-to-flight.
Another possible source of UFP from within the cabin is from cooking activity from the galley. Spengler, Vallarino, and McNeely (2012) reported data from 85 commercial flights, during which UFP were monitored on 55 flights. The flights that had cooking vs those with no cooking showed more increases in UFP.
Sources of particles outside of the cabin are harder to quantify. Contamination of the bleed air via leaking seals allowing lubrication oil or decomposition byproducts to pass into the supply air have been suggested. Isomers of tricresyl phosphate (TCP) have long been thought to be a candidate for identifying a seal failure. Spengler, Vallarino, and McNeely (2012) reported that levels on the 85 flights were within safe limits and no fume events were reported based on TCP measurements, whilst Crump, Harrison, and Walton (2011) reported that TCP and all other contaminants fell within the UK domestic and occupational safe limits and that in over 95% of cabin samples, no TCP was found. Spengler, Vallarino, and McNeely (2012) report that during flights when no food was being served, increased levels of UFPs were associated with increased levels of ozone, which the authors attributed to a section of flights when the outdoor ozone was elevated. However, Spengler also noted that there were fluctuations in the UFP during no cooking flights and were unable to link those to bleed air events or aircraft activity. Li et al. (2014) noted increase in particles >0.3 mm in diameter with turbulence events and flying through clouds, implying ingress of outdoor aerosols.
On the ground, airports are a major source of particles, from ground handling activities and aircraft engine emissions. Work by Cao et al. (2017) compared mass concentrations on board with measurement taken at the airport 340 m from the runway, and concluded that the concentrations were higher on the ground when the ambient concentrations were higher, implying the external sources contribute to the pollutants on board.
In this article, simultaneous measurements of inboard and outboard UFP are reported from 12 flights on the UK's BAe-146-301 Atmospheric Research Aircraft, the FAAM BAe-146, between 2016 and 2017, to test the hypothesis that airframe generated particles can contribute to in-cabin concentrations and that during flight, outside aerosol can ingress into the cabin. In addition, three flights monitored the chemical composition and mass concentration of PM 1 particles, with one flight recording a significant increase in organic material during take-off. A theoretical cabin occupant exposure per flight will be compared with current UK and European workplace and air quality exposure limits. Whilst there are no current workplace exposure limits on UFP, due to a relatively low number of studies and a lack of standardization in methodologies (Viitanen et al. 2017), the average number of concentration exposure per flight for the transit style flights are reported.

FAAM BAe-146
The FAAM BAe-146 (www.faam.ac.uk) is a highly modified BAe-146-301, four-engine aircraft. The aircraft has a range of approximately 2000 nautical miles and can operate from $15 to 10,700 m. Typical missions are between 1 and 6 h, with the capability of carrying up to 18 scientists and 4 tonnes of scientific equipment.
The aircraft operates under British Civil Aviation rules and has extremely stringent procedures and protocols for installing and operating equipment on board. The procedures require close working between the instrument scientists, the facility manager (FAAM), the aircraft service providers (Avalon Aero), and the aircraft designers (BAe Systems), before final approval by BAe Systems on behalf of the UK Civil Aviation Authority (CAA). There are multiple safety considerations and checks that are required before any instrument can be approved for flight, one of which is that no equipment is allowed to vent to the cabin. Connections to sample and exhaust lines are normally made with Swagelok pressure fittings. Connections with flexible tubing have additional retaining clips to prevent them from dislodging. It is therefore extremely unlikely that any equipment will generate UFP into the cabin under normal operations.
The FAAM BAe-146 has reduced galley facilities as compared with commercial aircraft. It operates a water heater for tea/coffee and only serves pre-prepared, cold snacks (sandwiches, fruit, biscuits, etc.), with no re-heating, making the galley an unlikely source of UFP. This is important to note as cooking has been identified as a potential source of UFP during commercial flights and other indoor environments.

Condensation particle counter (CPC) comparisons
UFP concentrations for this study were measured with two modified TSI model 3786 ultrafine water-based condensation particle counters (UWCPC; Hering et al. 2005), counting total particle number concentrations from 2.5 nm to $2-3 mm. These two units differ from the standard UWCPC in their pneumatic design, which has been modified to allow operation at subambient pressures, limited to 180 hPa. These two CPCs, serial numbers 183 and 86 have been flown on the FAAM BAe-146 since 2008. During this time they have periodically been subjected to comparison with each other and various laboratory-based CPCs to determine any possible bias; the methodology of these experiments is laid out here.
Both instruments have been subjected to multiple comparisons between 2013 and 2017. There are two forms of comparison: a direct comparison between serial numbers 183 and 86 and larger, inter-laboratory comparisons. In the direct comparison, a nebulized polydisperse salt, normally ammonium sulfate, is sized between 0.01 and 0.1 mm using an electrostatic classifier and Differential Mobility Analyser (DMA). This monodisperse aerosol is sampled via a y-piece and instrument inlets of the same length, as shown in Figure  1. This experiment is repeated at multiple particle diameters and multiple concentration points. An example of the data from an individual diameter, 0.05 mm ammonium sulfate aerosol can be seen in Figure 2.
All of the data for these comparisons are size segregated and shown in Figure S1 (see the online supplementary information [SI]). Across the 4 years of comparisons, the CPCs showed very good agreement with a maximum of 12% discrepancy in the slope across a concentration range of 4 orders of magnitude in all diameters investigated.
The two CPCs used in this article have also been part of larger comparison experiments. In total five comparisons were undertaken in the 4 years these data span.

Sampling setup
CPC S/N 183 sampled the ambient aerosol particles (referred to as outboard), whilst CPC S/N 86 sampled the cabin air (referred to as inboard). Both CPCs are mounted on separate instrument racks, where minimizing length and bends in samples lines on aircraft is not a primary driver, safety is. As such sampling pneumatics are not always the shortest or straightest possible in an aircraft cabin. The outboard CPC has a sample line that is 2.6 m long with a total curvature of 4.56 radians connected to a modified Rosemount housing inlet. The inlet is a combination of stainless steel and conductive silicone tubing. Using a laminar diffusional loss model, with a flow rate of 0.6 l/min and a temperature and pressure of 293 K and 101.2 hPa, yields a D 50 of 5.75 nm, and a D 90 of 22.2 nm for this sampling architecture. The inboard CPC had no inlet connected, and therefore the D 50 is that of the instrument, 2.5 nm. The difference in D 50 means that for any new particle formation at the smallest sizes (<10 nm), the outboard CPC would record slightly less than the inboard CPC.
On flights where the inlet pressure of the outboard CPC was saved, the concentration was corrected to the inboard cabin pressure so absolute number concentrations comparison can be made to the inboard CPC. The correction factor was highest at high levels, of the order of a factor of 2, reducing to 1 at low levels/on stand. The absolute difference in number concentration is not used in this article.

Organic composition
In addition to the number concentration, chemical composition measurements were made during three flights. The measurements were taken using an Aerodyne compact aerosol time-of-flight mass spectrometer (C-Tof-AMS, herein referred to as AMS). The instrument is described by Drewnick et al. (2005). In brief, the AMS samples gas and aerosol through an aerodynamic inlet and is passed through three differentially pumped chambers, which maximizes aerosol transmission whilst removing as much gas as possible. In the third chamber, the particles impact on a heated surface held at 600 C. Particles with a vaporization temperature of 600 C or less, flash vaporize and produce a kinetic gas. This gas is ionized by standard electron impaction (70 eV) and the ions are extracted into the time-of-flight mass spectrometer. This configuration means the AMS will quantify the non-refractory components of the aerosol (e.g., sulfates, nitrates, and organics), but not soot and metals, for example.
The AMS is a unit mass resolution instrument, which has been shown to produce chemically resolved, quantifiable data over many years (Canagaratna et al. 2007) from locations all over the world (Jimenez et al. 2009). The AMS converts arrival times of ions to m/z ratio, which is calibrated on each scan. The ensemble gas and aerosol phase mass spectrum is deconvolved using what is known as a fragmentation table to produce chemically resolved data as a function of m/z (Allan et al. 2004). Identification of organic peaks and their ratios (Canagaratna et al. 2015), allows source profiles of different aerosol types to be generated and matched to measured data, with changes in patterns indicating changes in composition. Summation of different species across all m/z produces total mass concentrations. The electron impaction at 70 eV is a standard ionization technique, which produces repeatable fragmentation patterns for a given molecule, and allows direct comparison between similar measurement techniques and, for example, the National Institute of Standards and Technology (NIST; Linstrom and Mallard 2001) 70 eV electron ionization spectral database. The ionization (and to a lesser degree the thermal decomposition) fragment parent molecules, meaning original compositional data is lost, but functional groups are retained. The AMS has a 100% transmission window between approximately 50-600 nm (vacuum aerodynamic diameter), with a steep drop off in efficiency on either side of this window.
Total exposure, average exposure, and average concentration exposure In the discussion, the total exposure, average exposure per flight, and average concentration exposure are calculated. Total exposure is calculated by integrating either the total number or mass concentration over time, starting before take-off and ending after landing to simulate boarding and deboarding. The total exposure is then divided by the total time to yield the average exposure per flight. This allows comparison to workplace exposure limits. The average concentration exposure calculates the average number or mass concentration for the flight, which is compared with ambient air quality metrics.

Results
The results presented are from 12 flights, covering opportunistic sampling during 3 different research campaigns. Seven of the flights involved multiple altitude changes and low level flying, whilst the remaining five were transit style flights, which are more representative of commercial passenger flights. The campaigns were: Volcanic and Atmospheric Near-to far-field Analysis of plumes Helping Interpretation and Modelling (VANAHEIMbased at Keflavik, Iceland); Methane Observations and Yearly Assessments (MOYA -Dakar, Senegal); Cloud and Aerosols Radiative Impact and Forcing (CLARIFY -Ascension Island). The configuration of the aircraft was changed between each campaign to meet the different science requirements, but the CPCs used in this article sampled from the same ports and from the same locations. A summary of all flights and available data is presented in Table 1, and the time series of all transit style flights are presented either in the main body or in the SI. Figure 3 shows a time series from flight B989. The vertical dashed lines highlight the take-off and landing times. The data show that the outside concentrations pre-take-off and post-landing, appear to be dominated by outside sources, with the outboard CPC saturating at the maximum detectable concentration, 1 Â 10 6 cm À3 (note the left axis has been restricted to 6 Â 10 4 cm À3 for clarity). These sources include the ground handling vehicles and the Ground Power Unit (GPU) used to power the FAAM BAe-146 science systems before main engine start. The FAAM BAe-146 operates at remote stands and often has the forward door open for access, so allows ingress of air. Despite this, whilst the concentrations in the cabin pre-takeoff and post-landing are high, they are not as high as the outside concentrations. In addition, the concentrations in the cabin are less variable than the outdoor concentrations, attributable to the lower ventilation within the airframe. However, the peak inboard concentration whilst on stand and during taxi occurs after the main doors have closed and engines start in 9 of the 12 flights. After take-off, the concentration of the inboard CPC decreased, reducing to low levels in the order of 15-20 min, as the ECS recirculated the air.

On stand concentrations: The indoor-outdoor interface
In-flight, in-cabin UFP Figure 3 also shows periods of UFP increases during the flights. Two of the periods of increased UFP activity coincide with increases in the outboard CPC. Given the similarity in trends, especially around the 15:30-16:00 events, it is likely these particles are  caused by the ingress of external particles into the cabin. However, the sharp increase around 13:50 is not associated with any increase in the outboard CPC. The clean out time for this event is also around 15-20 min. It is noted that the sharp increase occurred during low-level maneuvers, when there are lots of changes in thrust/engine conditions. This occurrence of UFP during low-level maneuvers is seen in other flights reported here. Figure 4 shows the time series from flight P153. This was a transit flight to relocate the aircraft, and the majority of the science equipment was not operated. This flight has a profile more representative of a commercial passenger flight. The data show the UFP dominating around take-off, followed by cabin venting. There is an inboard UFP event around 14:00 not seen in the outboard UFP and shows the low levels of UFPs when no events occur during high altitude transits. In addition, a UFP event can be seen on approach, not associated with external UFPs. Of the 12 reported flights, 11 had both the inboard and outboard CPC operational for approach, and of those 11, 7 recorded inboard UFP events on approach which cannot be attributable to increases in external particles. Figure S5 shows the organic aerosol composition when the aircraft was on stand and during taxi when the AMS was sampling outboard (ambient) condition during normal operation in the MOYA campaign. The data show the composition within the airport environment is similar across the campaign, dominated by ions produce from saturated aliphatic hydrocarbons, evidenced by two homologous chains starting at m/z 27 and 29, increasing in steps of m/z 14, plus smaller contributions from aromatics. The hydrocarbon fragments are indicative of fresh emissions, most likely from aircraft activity. Smith (2021) looking at emission from a gas turbine engine attributed a similar alkane fraction (AlkOA) to contributions from unburnt fuel. Figure 5 shows the mass spectra from C003 at take-off, and during landing when both the inboard and outboard CPCs were high. Whilst the alkane fragmentation is seen in both, during take-off, there is a dominant peak at m/z 119 and an increase in the contribution from m/z 85 and 97. Similarly, during landing, whilst the same peaks are prevalent, there is a change in the ratios of 41 and 43, and 55 and 57, and an increased contribution from m/z 85, indicating a change in composition. Therefore, there is evidence that in addition to the contribution from the external combustion sources, other airframe sources are influencing the cabin air.

C003
C008 -fume event Figure 6 shows a time series of the CPCs and AMS from C008. During take-off, the AMS measures a significant increase in organic mass. The mass concentration averaged $142 mg m À3 during this event, which is over an order of magnitude larger than the average ambient concentrations at the airfield. The mass spectra ( Figure S6) show a significant increase in the alkane chains, the presence of aromatic compounds, and noticeable (and as yet unidentified) peaks at m/z 107 and 123.

Evidence of non-combustion products in cabin
C040 was a transit style flight, with the AMS measuring cabin air for the majority of the flight, except for a profile in the middle section of the flight (see Figure  S3 for time series). During the up-down profile, there is an increase in the cabin number, no corresponding increase in the outboard CPC, and an increase in the AMS (now outboard) reported mass. The average mass spectra of the ambient during the profile and the cabin air is displayed in Figure 7, showing that the chemical composition inboard and outboard is different.
The cabin MS in Figure 7 has some distinct differences from the ambient MS from MOYA and the aircraft AlkOA as described by Smith (2021), both attributed to direct engine emissions. The fragments at m/z 27, 29, 41, 43, 55, and 57 are still present, but with distinct markers at m/z 85 and 113 and to a lesser extent m/z 99 and 127. This MS, with the strong signal at m/z 85 and 113 is similar to that reported by Yu et al. (2012) on the chemical composition of oil used in aircraft engines. Yu et al. also report that an 85:71 ratio greater than 0.66 is required to distinguish oil from primary engine emissions, a criterion that is met with these data. There are some small differences; for example, Yu et al. found a peak at m/z 98 rather than m/z 99. Similar analysis of a different engine oil (BP engine oil, Mobile 2192) at the University of Manchester using a high-resolution AMS found a similar MS, with the strong oil marker at m/z 85, 85:71 greater than 0.66 and a peak at m/z 98 rather than 99. The MS reported in all three cases have markers at m/z 113 and 127, but in addition, the Mobil oil had a peak at m/z 125. Therefore, the slight difference in MS (m/z 98 vs m/z 99) could be attributed to different oil being used in the FAAM BAe-146, compared with both the Yu work and Mobile 2192 oil. However, it must be acknowledged that other aerosol types are contributing as the source profile is not an exact match, but that the oil signature is prominent in the reported MS.
The MS shown in Figure 7 has some of the same markers seen during landings for C003 ( Figure 5), C008, and C040 ( Figure S7), all with the same distinct markers. The composition of the approach/landing organic aerosol for C003 can be analyzed for when the UFP are increasing inboard only (around 19:04), as well as at the very end, when both inboard and outboard CPCs are high on land as in Figure 5. The MS from both are more representative of the oil MS, albeit with different relative intensities. Flight C008 was flagged as a flight that did not have cabin only UFP on approach (Figure 6) because increases in the inboard and outboard CPC were seen (albeit with different temporal trends), as shown in the expanded insert in Figure 6. However, the MS for this section on approach of elevated organic has the same markers as the oil and not the ambient aerosol at the airport, suggesting this could be an airframe event as well as possible ingress from outside. This highlights that both internal and external sources can impact the UFP concentration and the particle composition, but the mass is potentially influenced more by internal sources on approach, and that the three measurements combined can distinguish the different sources.
Taxi and take-off data highlight influences of both oil and ambient aerosol. C040 has a MS with the distinctive oil markers, whereas C003 and C008 (pre the fume event), both have a MS similar to ambient with the additional peaks at m/z 97 and 119 for C003 ( Figure 5). The large fume event during C008 highlights yet another, unidentified, source of aerosol.

Discussion
Pilots, aircrew, and passengers are exposed to both external and aircraft sources of aerosol. At the airport, the outboard sources dominate the UFP number particle concentration, which is consistent with other studies but has shown that in the majority of cases, the inboard concentration peaks after engines start (doors closed). Particle concentrations are seen to decrease to low levels in the order of 15-20 min (in the absence of an in-cabin UFP event). For the first time, this study has shown direct evidence of in-cabin, in-flight sources attributed to ingress of external particles.
Inboard only UFP events are generally sporadic and have seen to occur most frequently when the aircraft engines are changing conditions, with over 50% of approach/landing seeing such events. This includes the transit flights when very little equipment is operated. It is speculated that these events could be from the bleed air systems, as seal integrity may change under these conditions. The chemical composition data have been shown to help identify airframe sources from external ingress, when both inboard and outboard measurements are high. Indeed, the work presented shows that at least three parameters are needed to have confidence as to what is contributing to in-cabin events. The data suggest that there are non-exhaust products present in the cabin, and that with the strong marker at m/z 85 and 113, it is suggested this could be an oil product. However, the fume event of C008 and the additional markers at m/z 97 and 119 from C003 show the system is complex and more research is needed to elucidate the source of the particles.
The data presented here, in terms of number of flights, are typical of this under-represented environment, but to which a very high number of people are exposed per annum (>4 billion). Based on transit style flights P153, C003, C008, C040, and C041, the total exposure (N cm À3 h) to particle number concentration is between $23,000 and 60,000, which translates to an average exposure of between $6800 and 10,800 N cm À3 per flight. Aircraft produce high numbers of nonvolatile PM (nvPM), mostly soot. The distribution of soot is typically log-normal, with a width of 1.8, mode diameters between $20 and 50 nm (depending on thrust), and with an effective density of around 1 g cm À3 . Assuming all the CPC number is nvPM, a theoretical maximum exposure to black carbon can be estimated. Constraining a log-normal with these parameters gives the theoretical maximum total exposure and average exposure per flight as 0.46-18.58 mg m À3 h and 0.13-3.34 mg m À3 . For the fume event for flight C008, whilst the peak mass loading is $160 mg m À3 , the total and per flight organic mass concentration exposures are 19.0 mg m À3 h and 5.85 mg m À3 (none fume events are significantly lower). The UK 8-h workplace exposure limits for several non-hazardous dusts are 10 mg m À3 and 4 mg m À3 for inhalation and respiratory particles, respectively (Health and Safety Executive [HSE]), and for carbon black (the closest to black carbon), 3.5 mg m À3 . Within the United Kingdom and European Union, the 24 h average concentration limit for PM 10 is 40 mg m À3 , and in the USA 35mg m À3 , for ambient air quality; the average organic for C008 is 9.98 mg m À3 and theoretical maximum average BC of 0.1 mg m À3 for the short flight. The values reported here for these short flights are well below those limits. Therefore, if these particles are a source of health issues, the toxicology and doses of the acute exposures and events need to be determined.

Funding
This work was supported by the Natural Environment Research Council through the National Centre for Atmospheric Science (NCAS) and the Atmospheric Measurement and Observation Facility (AMOF). In Figure 7. Mass spectra from in cabin and external air during flight C040. Note that the AMS was switched from cabin to ambient and back to cabin during a profile in the middle of the flight. particular, the authors would like to acknowledge the support of the NERC-funded MOYA (NE/N016548/1) and CLARIFY (NE/L013584/1) projects and the NCAS-funded VANAHEIM project.