Laboratory evaluation of organic aerosol relative ionization efficiencies in the aerodyne aerosol mass spectrometer and aerosol chemical speciation monitor

Abstract

Organic aerosol (OA) mass concentrations measured by the Aerodyne Aerosol Mass Spectrometer (AMS) and Aerosol Chemical Species Monitor (ACSM) depend on particle relative ionization efficiency (RIE_OA). Here, a series of laboratory experiments were conducted to investigate how RIE_OA differs for different classes of OA and between AMS and ACSM instruments. For OA surrogates with high oxidation state ($\overline{\mathit{OSc}}$), the measured RIE_OA is 1.60 ± 0.56 for all instruments. However, for OA surrogates with lower $\overline{\mathit{OSc}},$ the absolute magnitude and the variability of the measured RIE_OA increases. The increase in RIE_OA is different between the AMS and ACSMs due to differences in m/z-dependent ion transmission/detection efficiencies in the mass spectrometers. A new metric is introduced to explore RIE_OA—the fractional contribution of ions at less than or equal to m/z 50 (f_Σm/z50). When f_Σm/z50 is high (>64%, similar to ambient OA), the average RIE_OA is 1.23 ± 0.62. This supports the use of a default RIE_OA of 1.4 for typical ambient conditions, where secondary OA dominates the OA budget. When f_Σm/z50 is less than 64%, the RIE_OA increases as higher m/z ions contribute more to the total signal. These observations reflect the fact that the observed RIE_OA is a combination of multiple processes. The f_Σm/z50 may be used as a metric to determine if primary-like OA is contributing significantly to total OA. Overall, the results indicate that changes in RIE_OA are most important for studies measuring primary-like OA, and the magnitude of the changes will depend on the instrument being used to measure the OA.

Editor:

• Jim Smith

1. Introduction

Atmospheric particles, or aerosol, play important roles in the earth system, impacting human health, climate, and air quality (Cohen et al. 2017; Monks et al. 2009; Myhre et al. 2013). An important component of aerosol is organic aerosol, or OA (Chen et al. 2022; Jimenez et al. 2009; Zhang et al. 2007). The exact chemical structure and physicochemical properties of OA vary depending on source and whether it was primarily emitted or produced by secondary processes such as gas-phase photochemistry or condensed-phase aqueous chemistry (Hallquist et al. 2009; Moise, Flores, and Rudich 2015; Rudich, Donahue, and Mentel 2007). Thus, measurements to determine the OA sources and chemical transformations are important to improve our understanding of aerosol.

The Aerodyne Aerosol Mass Spectrometer (AMS; Canagaratna et al. 2007) and Aerosol Chemical Speciation Monitor (ACSM; Fröhlich et al. 2013; Ng et al. 2011b) are two widely used instruments to measure the non-refractory chemical composition and mass concentration of aerosol. Both use the same technique to sample non-refractory aerosol and a detailed description, along with their working principles, can be found elsewhere (Canagaratna et al. 2007; Fröhlich et al. 2013; Ng et al. 2011b). However, there are some key differences between the AMS and ACSM. The main differences between the AMS and ACSM include differences in how air background is subtracted (use of particle time-of-flight with AMS and use of particle filter with ACSM as there is no particle time-of-flight in ACSM), detector, and simplified electronics which reduces the overall complexity of the ACSM compared to the AMS (Fröhlich et al. 2013; Ng et al. 2011b). However, the general technique for sampling aerosol is similar with the AMS and ACSM. Briefly, ambient aerosol is sampled through an aerodynamic lens (Liu et al. 1995a, 1995b) and focused into a narrow beam. Then, the aerosol is evaporated using a heated surface (∼600 °C to vaporize non-refractory aerosol) and ionized by 70 eV electrons (Canagaratna et al. 2007; Jimenez et al. 2003; Jayne et al. 2000). The ions are then extracted into a mass spectrometer to be measured. The aerosol mass concentration of any given species, C_s (μg m⁻³), can be obtained from the measured mass spectrum signal using the following equation: (1) $${\mathrm{C}}_{\mathrm{s}}\text{=}\frac{1}{{\mathrm{CE}}_{\mathrm{s}}{\mathrm{RIE}}_{\mathrm{s}}}\frac{{10}^{12}{\text{MW}}_{\text{NO}3}}{{\text{IE}}_{\text{NO}3}\mathrm{Q}{\mathrm{N}}_{\mathrm{A}}}\sum _{\text{all}\mathrm{i}}{\mathrm{I}}_{\text{s,i}}$$(1)

Here, MW_NO3 is the molecular weight of species nitrate as nitrate (g mol⁻¹) is used as the main calibrant for AMS and ACSM (Jayne et al. 2000), Q is the volumetric flow rate (cm³ s⁻¹), N_A is Avogadro’s number (molecules mol⁻¹), I_s,i is the ion rate of each ion, i, for species, s (ions s⁻¹), 10¹² is for unit conversion (g to μg and cm⁻³ to m⁻³), IE_NO3 is the nitrate-dependent ionization efficiency (ion molec⁻¹), RIE_s is the species-dependent relative ionization efficiency (unitless), and CE_s is the collection efficiency, a unitless factor that accounts decreased focusing efficiency due to non-spherical shape, the shape factor of the aerosol (E_S), the transmission of the aerosol through the lens (E_L), and the bounce of the aerosol on the vaporizer (E_B) (Huffman et al. 2005; Middlebrook et al. 2012).

The IE used for conversion to mass concentration is based on using ammonium nitrate particles, as it is a well-characterized and well-behaved aerosol (Jayne et al. 2000). To convert from ammonium nitrate equivalent mass concentration to the species specific mass concentration, a relative ionization efficiency for each species (RIE_s) is used and defined as the ratio of the IE for each species, s, to that of nitrate (Alfrarra 2004; Canagaratna et al. 2007; Jimenez et al. 2003). For inorganic species, including ammonium, sulfate, and chloride, their individual RIE can be calibrated with lab-generated aerosol with no matrix effects (Nault et al. 2018; Xu et al. 2017). However, because OA found in the atmosphere is composed of 100s–1000s of different compounds with different physicochemical properties (Hallquist et al. 2009; Moise, Flores, and Rudich 2015; Rudich, Donahue, and Mentel 2007), a calibration for the OA RIE (RIE_OA) from a single OA standard is not possible. Instead, several standards would be necessary to explore the potential range of values for RIE_OA.

There are many processes controlling RIE_OA (Jimenez et al. 2003, 2016; Murphy 2016). In addition to the ionization process itself, the measured value of RIE accounts for multiple processes leading up to detection of the OA ions. Thermal decomposition of oxidized OA via dehydration and decarboxylation prior to ionization of the gases is one well known process (Canagaratna et al. 2015; Moldoveanu 2018): (R1) $$\text{RC}\text{OOH}\stackrel{\mathrm{\Delta }}{\to }{\text{CO}}_2\text{+}{\mathrm{H}}_2\text{O + CO+}{\mathrm{R}}^{\prime }$$(R1) (R2) $$\text{RCOH}\stackrel{\mathrm{\Delta }}{\to }{\mathrm{H}}_2\text{O+}{\mathrm{R}}^{\prime }$$(R2)

This leads to detection of CO₂, H₂O, CO, and R′. Carbon-Carbon (C-C) bond cleavage to form methyl or ethyl radical compounds plus an unsaturated hydrocarbon may also occur at these temperatures (Rice 1931, 1933; Rice, Johnston, and Evering 1932). For example, the following products from thermal decomposition of n-butane have been observed and calculated (Rice 1931, 1933): (R3) $$\mathrm{C}{\mathrm{H}}_3{\text{CH}}_2{\text{CH}}_2{\text{CH}}_3\stackrel{\mathrm{\Delta }}{\to }{\text{CH}}_3^{\text{-}}\text{+}{\text{CH}}_3{\text{CH}}_2{\text{CH}}_2^{\text{-}}$$(R3) (R4) $$\mathrm{C}{\mathrm{H}}_3{\text{CH}}_2{\text{CH}}_2{\text{CH}}_3\stackrel{\mathrm{\Delta }}{\to }{2{\text{CH}}_3\text{CH}}_2^{\text{-}}$$(R4) (R5) $${\text{CH}}_3{\text{CH}}_2{\text{CH}}_2^{\text{-}}\stackrel{}{\to }{\text{CH}}_2\text{=}{\text{CH}}_2\text{+}{\text{CH}}_3^{\text{-}}$$(R5) (R6) $$\mathrm{C}{\mathrm{H}}_3{\text{CH}}_2{\text{CH}}_2{\text{CH}}_3\text{+}{\mathrm{R}}_{}^{\text{-}}\stackrel{}{\to }\text{RH+}{\text{CH}}_3{\text{CH}}_2{\text{CH}}_2{\text{CH}}_2^{\text{-}}$$(R6) (R7) $$\text{RH+}{\text{CH}}_3{\text{CH}}_2{\text{CH}}_2{\text{CH}}_2^{\text{-}}\stackrel{}{\to }\text{RH+}{\text{CH}}_2\text{=}{\text{CH}}_2\text{+}{{\text{CH}}_3\text{CH}}_2^{\text{-}}$$(R7) (R8) $$\mathrm{C}{\mathrm{H}}_3{\text{CH}}_2{\text{CH}}_2{\text{CH}}_3\text{+}{\mathrm{R}}_{}^{\text{-}}\stackrel{}{\to }\text{RH+}{\text{CH}}_3\text{CH}\mathrm{(}{\text{CH}}_2{\text{CH}}_3^{}{\mathrm{)}}^{\text{-}}$$(R8) (R9) $$\text{RH+}{\text{CH}}_3\text{CH}\mathrm{(}{\text{CH}}_2{\text{CH}}_3^{}{\mathrm{)}}^{\text{-}}\stackrel{}{\to }\text{RH + C}{\mathrm{H}}_3\text{CH=}{\text{CH}}_2\text{+}{\text{CH}}_3^{\text{-}}$$(R9)

Here, R can be CH₃, CH₃CH₂, or H. For larger saturated hydrocarbons, including branched hydrocarbons, even more products were observed (Rice 1933). Note, though possible for the vaporization conditions of the AMS and ACSM, the thermal decomposition of saturated hydrocarbons (R3–R9) has not been investigated for the AMS and ACSM. After thermal decomposition, the molecules undergo ionization. As the ionization cross section is approximately proportional to the molecule’s molecular weight (Jimenez et al. 2003), the ionization efficiency will depend on the molecular weight of the thermally decomposed OA. Further, the ionization cross section is also related to the oxidation state ($\overline{\mathit{OSc}}$ = H/C − 2 × O/C; Kroll et al. 2011) of the thermally decomposed OA, with more reduced molecules having larger electron ionization cross sections than more oxidized molecules (Jimenez et al. 2003). However, as shown in Xu et al. (2018), there is not a clear trend in the RIE_OA with $\overline{\mathit{OSc}}.$ It has been hypothesized that the molecular weight of the thermally decomposed OA compounds will dictate how long the molecule will remain in the ionization region and thus affect RIE_OA (Murphy 2016). However, ambient observations of OA have shown minimal impact for high molecular weight OA impacting the quantification of total OA mass concentration (Jimenez et al. 2016). Finally, after ionization, other processes that may impact RIE include ion transmission and detection (Canagaratna et al. 2007; Jimenez et al. 2003). As demonstrated here, the complexity of processes that lead up to detection of OA species makes a priori predictions of RIE_OA difficult. Thus, direct measurements provide the best method to determine RIE_OA. Laboratory measurements of commercially available oxidized OA surrogates indicate an average RIE_OA of 1.4 (Canagaratna et al. 2007; Jimenez et al. 2003, 2016; Xu et al. 2018). This RIE_OA leads to generally good agreement between the AMS or ACSM and other measurements sampling ambient aerosol (e.g., DeCarlo et al. 2008; Hayes et al. 2013; Nault et al. 2018).

Recent observations of directly emitted OA, or primary OA (POA), have indicated higher RIE_OA than the default value of 1.4 (Katz et al. 2021; Reyes-Villegas et al. 2018). Xu et al. (2018) investigated OA surrogates that ranged in molecular weight and $\overline{\mathit{OSc}}$ and found that RIE_OA remained relatively constant for $\overline{\mathit{OSc}}$ representative of secondary OA (SOA) at 1.6 ± 0.5 and increased with decreasing $\overline{\mathit{OSc}}.$ This is in line with the studies that found higher RIE_OA was needed for POA (Katz et al. 2021; Reyes-Villegas et al. 2018). However, Xu et al. (2018) observed high variability in RIE_OA versus $\overline{\mathit{OSc}},$ reducing the ability to provide guidance in what RIE_OA to use with observations collected by AMS and ACSM. Further, Xu et al. (2018) only investigated RIE_OA with an AMS and did not determine if what was observed on the AMS is applicable to the ACSM, and what role the other processes that can impact IE and/or RIE may have. Field measurements, e.g., Katz et al. (2021), suggest that RIE_OA for POA may be quite different between the two types of instruments.

This study expands upon the work from Xu et al. (2018) to investigate RIE_OA measurements for both AMS and ACSM instruments and to provide guidance on how to calibrate or estimate RIE_OA. An important aspect of this study is the investigation of the ion transmission efficiency for different versions of AMS and ACSM that use different mass spectrometers, and how this impacts the measured RIE_OA. We propose a simplified method to calibrate RIE_OA that reduces the need for equipment used in Xu et al. (2018) while maintaining high reproducibility. We also investigate a new metric for estimating RIE_OA self-consistently for all AMS and ACSM systems when calibration is not possible.

2. Methods

2.1. Instrumentation

In this study, seven instrument versions with different mass spectrometers were investigated. These included the Aerodyne Quadrupole Aerosol Mass Spectrometer (Q-AMS; Jayne et al. 2000), the Aerodyne Compact Time-of-Flight AMS (CTOF-AMS; Drewnick et al. 2005), the Aerodyne High Resolution Time-of-Flight AMS (HTOF-AMS; DeCarlo et al. 2006), the Aerodyne Long Time-of-Flight AMS (LTOF-AMS), the Aerodyne Quadrupole Aerosol Chemical Speciation Monitor (Q-ACSM; Ng et al. 2011b), the Aerodyne Time-of-Flight ACSM (TOF-ACSM; Fröhlich et al. 2013), and the Aerodyne TOF-ACSM with eXtended resolution (TOF-ACSM-X). For all instruments, the ionization cage is nearly identical thus should not impact the interpretation of the RIE_OA and ion transmission. Also, the standard vaporizer was used for all instruments (Joo et al. 2021). The temperature for the AMS vaporizer was verified using sodium nitrate, as discussed in Williams (2010). All were used to measure ion transmission, and the Q-AMS, Q-ACSM, and TOF-ACSM were used to measure RIE_OA of single component OA surrogates and OA surrogates mixed with ammonium nitrate. All instruments were equipped with the standard vaporizer and PM₁ lens (Joo et al. 2021). A differential mobility analyzer (DMA, TSI 3080, Shoreview, MN, USA), DMA column (TSI 3081), (ultrafine) condensation particle counter (CPC TSI 3775 and UCPC TSI 3776), centrifugal particle mass analyzer (Cambustion CPMA; Olfert and Collings 2005), and an atomizer (TSI 3076) were used for the calibrations.

Besides, the measurements conducted in this study, measurements of RIE_OA from other instruments from different years (Xu et al. 2018) and groups (Katz et al. 2021; Slowik et al. 2004) are also included. Comparisons of measurements in this study with those from prior studies highlight minimal role in potential slight differences in vaporizer temperature and instrument tuning in the measured RIE_OA.

2.2. OA surrogates

A list of OA surrogates used for calibrations is given in Table 1. The surrogates were used without further purification, as they were ≥90% purity. All compounds except, anthracene and palmitic acid, are from Sigma-Aldrich; anthracene and palmitic acid are from Fluka. The solvents used for the solutions were either deionized water or acetonitrile (HPLC Plus, ≥99.9%, Sigma-Aldrich). Acetonitrile was selected as the organic solvent as it consistently showed minimal signal (<0.1 µg m⁻³ OA) when atomized compared to other solvents. No corrections were made for the OA standards as there was minimal change in the mass spectra for standards atomized in water versus atomized in acetonitrile (Figure S1). Ammonium nitrate was used for IE calibrations and as an internal standard. Similar to recommendations for ammonium nitrate calibrations, for the atomizer used (TSI 3076), the solutions used here were at 0.01 M or less, or less than 1000 particles cm⁻³ at 300 nm. This ensures that there were minimal doubly-charged particles, which impacts the comparison in the mass calculated from the CPC and mass observed by the AMS. An example particle time-of-flight measurement for two of the solutions is shown in Figure S2 to show minimal doubly-charged particles.

Table 1. List of organic standards used to measure species-dependent RIE. For RIE_AMS, the values are from standards in acetonitrile. RIE_AMS from other solvents and/or mixtures can be found in Table S1.

Class/Name	$\overline{\mathit{OSc}}$	MW (g mol^-1)	This work RIEAMS	This work CE	This work RIEQ-ACSM	This work RIETOF-ACSM	Prior RIE
Saturated hydrocarbon
Squalane	−2.07	422.81	11.46	1.00	2.10	4.87	7.20^2, 6.98^3
Octadecane	−2.11	254.50	10.68	1.00
Polycyclic aromatic hydrocarbon
Anthracene	−0.71	178.23	1.94	0.21
Pyrene	−0.63	202.26	2.46	0.23
Acid
Oleic acid	−1.67	282.26	5.22	1.00	1.75	3.94	3.00^2, 5.10^2, 3.18^3
Sebacic acid	−1.00	202.25	4.16	0.69
Succinic acid	0.50	118.09	0.65	0.60			1.40^2
Azelaic acid	−0.89	188.22	2.68	1.00
Pentadecanoic acid	−1.73	242.40	0.93	1.00	0.31
2-butyloctanoic acid	−1.67	200.32	1.29	1.00
2-hexyldecanoic acid	−1.75	256.42	5.14	1.00
Palmitic acid	−1.75	256.40	1.93	0.97
Alcohol
Levoglucosan	0.00	162.05	1.80	1.00	1.27	1.40	1.40^2
Xylitol	−0.40	152.15	1.86		1.40	2.05	1.70^2
1,2,6-hexanetriol	−1.33	134.17	2.42	1.00	1.84		2.60^2
Mannitol^1	−0.33	182.17	0.76	1.00	1.27
Threitol^1	−0.50	122.12	1.93	1.00	1.61		1.90^2
Esters
Dioctyl phthalate	−1.25	390.56	6.77	0.65

¹ Mannitol and threitol were only sampled in water and not acetonitrile.

² Xu et al. (2018).

³ Katz et al. (2021).

Some of the solutions used here were mixtures of OA and ammonium nitrate. This provided a means to eliminate the need to measure CE, reducing the complexity of the experiment. Similar to the pure compounds, the total concentration of the solution should be 0.01 M or less, or less than 1000 particles cm⁻³ at 300 nm, for TSI 3076 atomizer. Approximately equal molar amount of OA standard and ammonium nitrate were dissolved in either water or acetonitrile, depending on OA standard’s ability to dissolve in water. Note, only non-acidic compounds could be mixed with ammonium nitrate, as discussed in Xu et al. (2018).

2.3. Laboratory measurement of OA RIE

The procedure for the RIE calibration has been discussed in prior studies (e.g., Xu et al. 2018). The general premise is that a known concentration of the calibrant (M_in) is introduced into the AMS or ACSM. The known concentration can either be derived from a CPC, using known mobility diameter from a DMA and density from the calibrant, or by mass selection with a CPMA (Olfert and Collings 2005; Xu et al. 2018). For simplicity and reproducibility for most laboratory environments, only the CPC and DMA were used for these measurements. Furthermore, as discussed in SI (Section S1 and Figure S3), the use of the CPMA to quantify volatile aerosol mass can lead to uncertainty. The CPMA was only used to determine whether the mixture of ammonium nitrate with OA surrogates were internally or externally mixed. To derive M_in (µg m⁻³), the following equation was used: (2) $$M_{in}=\frac{4}{3}\pi (0.5\times d_m\times {10}^{-7}{)}^3{\rho }_s\times {10}^{12}\times N$$(2) where d_m is the mobility diameter from the DMA (nm), 10⁻⁷ is the conversion factor from nm to cm, ρ is the density of the OA standard (g cm⁻³), N is the particle concentration (cm⁻³), and 10¹² is the conversion factor between g to µg and cm⁻³ to m⁻³. M_in is then used as to determine the RIE for OA standard as follows: (3) $${\mathit{RIE}}_s{CE}_s=\frac{{10}^{12}{MW}_{NO3}{\mathit{RIE}}_{NO3}}{M_{in}{IE}_{NO3}Q{N}_A}\sum _{\mathit{all}\mathrm{ }i}{I}_{s,i}$$(3) where all other variables are the same as Equation (1)(1) $${\mathrm{C}}_{\mathrm{s}}\text{=}\frac{1}{{\mathrm{CE}}_{\mathrm{s}}{\mathrm{RIE}}_{\mathrm{s}}}\frac{{10}^{12}{\text{MW}}_{\text{NO}3}}{{\text{IE}}_{\text{NO}3}\mathrm{Q}{\mathrm{N}}_{\mathrm{A}}}\sum _{\text{all}\mathrm{i}}{\mathrm{I}}_{\text{s,i}}$$(1) . RIE_NO3 is 1.1 for AMS and 1.05 for ACSM (Canagaratna et al. 2007; Middlebrook, Bahreini, and Jimenez 2007). Here, we assume that the particles are spherical and that the particle density is equal to the material density.

In Equation (3)(3) $${\mathit{RIE}}_s{CE}_s=\frac{{10}^{12}{MW}_{NO3}{\mathit{RIE}}_{NO3}}{M_{in}{IE}_{NO3}Q{N}_A}\sum _{\mathit{all}\mathrm{ }i}{I}_{s,i}$$(3) , it is not possible to separate RIE_S from CE_S without additional information. We measured CE directly, using the Q-AMS in particle time-of-flight operation mode. The single particles can be directly counted using a “single particle” detection mode. In this mode, the number of single particle events for which AMS ion signal exceeded a threshold signal were counted and ratioed to the number of particles observed by the CPC. This ratio corresponds to a direct measure of the collection efficiency due to the particle bounce: (4) $$\text{CE=}{\mathrm{N}}_{\text{PToF}}\text{/}{\mathrm{N}}_{\text{CPC}}$$(4)

Here, N_PToF and N_CPC are the particle number concentrations observed from the AMS and CPC, respectively. Compounds found to have CE ∼1 for the Q-AMS were used as single components on other instruments that cannot measure single particles (ACSMs).

Some OA surrogates (non-acidic ones; Xu et al. 2018) were mixed with ammonium nitrate to reduce particle bounce, as particles with high ammonium nitrate content have an E_B ∼ 1. Ammonium nitrate also provides an internal standard to directly determine the RIE_OA. Using the CPMA, it was confirmed that ammonium nitrate was internally mixed with most of the OA surrogates (Figure S4). These mixtures were used on ACSMs to minimize uncertainty in CE. To determine the RIE_S for the OA standard in mixture, first, the ammonium nitrate mass concentration measured on the AMS/ACSM is converted to the particle concentration (cm⁻³) using Equation (2)(2) $$M_{in}=\frac{4}{3}\pi (0.5\times d_m\times {10}^{-7}{)}^3{\rho }_s\times {10}^{12}\times N$$(2) and the known density (1.72 g cm⁻³) and ammonium nitrate correction factor (0.8) (Jayne et al. 2000). Then, the ammonium nitrate particle concentration is subtracted from the total CPC particle concentration to derive the OA standard particle concentration. Finally, Equations (2)(2) $$M_{in}=\frac{4}{3}\pi (0.5\times d_m\times {10}^{-7}{)}^3{\rho }_s\times {10}^{12}\times N$$(2) and (3)(3) $${\mathit{RIE}}_s{CE}_s=\frac{{10}^{12}{MW}_{NO3}{\mathit{RIE}}_{NO3}}{M_{in}{IE}_{NO3}Q{N}_A}\sum _{\mathit{all}\mathrm{ }i}{I}_{s,i}$$(3) are used with the OA standard particle concentration to derive the RIE_S for the OA standard, with CE_S = 1.

2.4. Ion transmission measurements

To measure the dependency of ion transmission on m/z, an effusive source of naphthalene was introduced into the detection chamber, similar to the Q-ACSM (Ng et al. 2011b). The naphthalene was contained in a stainless-steel bulb with a 1–2 µm pin hole leak. The naphthalene source was located in the ionization chamber for all instruments. The naphthalene spectra were obtained on all the instruments with the inlet closed, ensuring that no external aerosol was influencing the signals at the naphthalene ion fragments. The instruments were operated with inlet closed for at least 24 h prior to sampling naphthalene to ensure OA background had been reduced.

The spectra were analyzed by normalizing the signal at each m/z to the signal at m/z 51. Then, the normalized signal was compared to the normalized NIST spectra for naphthalene (Wallace 2022). Finally, any signal that was equal to less than 1% of the parent ion (m/z 128) from the NIST spectra was removed, and groups of closely spaced ions within 5 amu were averaged together (Ng et al. 2011b).

3. Investigating the processes governing RIE_OA

3.1. RIE_OA versus oxidation state

The analysis from Xu et al. (2018) was replicated and expanded with a range of laboratory-generated OA standards (Figure 1 and S5). This included examining standards in different solvents (water versus acetonitrile), mixtures (pure versus with ammonium nitrate), instruments, and lower $\overline{\mathit{OSc}}$ compounds. As in Xu et al. (2018), for compounds with $\overline{\mathit{OSc}}$ similar to more oxidized organic aerosol (OOA) and more background-like OA (rural and remote), the RIE_OA shows less variability (mean and one standard deviation is 1.60 ± 0.56 and median is 1.73). However, as $\overline{\mathit{OSc}}$ decreases to values less than −0.75, similar to more urban-like OOA and hydrocarbon-like primary OA (cooking OA and hydrocarbon-like OA, COA and HOA, respectively), there is large variability with RIE_OA versus $\overline{\mathit{OSc}}.$ This strongly suggests that though $\overline{\mathit{OSc}}$ explains some of the variability in RIE_OA, there are numerous other processes impacting RIE_OA. Even for the same compound, there are large differences in measured RIE_OA, indicating the difficulty and potentially higher uncertainty associated with measuring the RIE_OA for the most reduced OA. Further, measurements of the $\overline{\mathit{OSc}}$ of ambient aerosol require high-resolution mass spectrometers, which the Q-ACSM, TOF-ACSM, and older AMS measurements (e.g., Q-AMS and CTOF-AMS) lack. Thus, the use of other metrics is desirable to determine the RIE_OA.

Figure 1. (Top) Range of oxidation state ($\overline{\mathit{OSc}}$) reported in Canagaratna et al. (2015) for oxidized organic aerosol (OOA), biomass burning organic aerosol (BBOA), cooking organic aerosol (COA), and hydrocarbon-like organic aerosol (HOA) and in Chen et al. (2015) for locations labeled as urban and urban downwind (urban), rural, and remote. (Bottom) Measured RIE_OA from this study for compounds in acetonitrile, from Xu et al. (2018), and from Katz et al. (2021), along with OA RIE = 1.4 and 1.6 ± 0.5 from Xu et al. (2018). Compounds measured in this study that are repeated RIE_OA from Xu et al. (2018) are highlighted in red. For all measurements, see Figure S5. The data from this study is from Q-AMS.

3.2. RIE_OA comparison between different methods and instruments

Xu et al. (2018) showed that measurements of internally mixed ammonium nitrate/OA particles were useful for accurately determining the RIE_OA of non-acidic OA standards. In these measurements, mixtures were used to explore simpler methods of measuring RIE_OA that do not require the use of a CPMA. Ammonium nitrate is used as an internal standard and the measured OA loading is compared to the ratio of OA/NO₃ expected from the calibration solution. This method eliminates the need for the CPMA as the mass does not need to be known a priori to measure the RIE_OA. Moreover, internal mixing with ammonium nitrate leads to the aerosol having minimal bounce, reducing the uncertainty in CE and need to measure CE. Figure 2 compares the RIE_OA for the five SOA-like standards—levoglucosan, xylitol, 1,2,6-hexanetriol, mannitol, and threitol—both pure and mixed with ammonium nitrate. The generally low variability (2–13%, Figure 2) observed between these measurements confirms the utility of using internally mixed non-acidic ammonium nitrate/OA particles for these measurements.

Figure 2. Comparison of measured RIE, from Q-AMS, for OA standards levoglucosan (Levo), xylitol (Xyl), 1,2,6-hexanetriol (Hxntrl), mannitol (Mann), threitol (Threit), oleic acid (Olc Acid), squalane (Squal), and pentadecanoic acid (PDA) in different solutions (water vs acetonitrile), mixtures (with or without ammonium nitrate, NH₄NO₃), with the Q-ACSM or ToF-ACSM, and against results from Xu et al. (2018).

To further investigate the utility of this method, a mass scan with a CPMA at a constant DMA diameter was used to evaluate whether the aerosol particles generated from the mixed OA/ammonium nitrate solution were internally mixed or not. Four different OA standards mixed with ammonium nitrate were measured by CPC and LTOF-AMS. As shown in Figure S4, for SOA-like standards (mannitol and 1,2,6-hexanetriol), the scans show that the OA and ammonium nitrate are internally mixed. This provides further evidence that ammonium nitrate can be mixed with SOA-like, non-acidic, OA standards to measure their RIE without a CPMA. However, for POA-like standards (tetradecane and anthracene), the results are less clear as tetradecane is internally mixed, but anthracene is not internally mixed. Thus, the use of internally mixed particles containing ammonium nitrate does not reduce uncertainties in the measurements of RIE_OA for reduced OA. The scans with POA-like standards and ammonium nitrate may partially be impacted by partial evaporation of either component that may occur in the CPMA due to their volatility (e.g., Huffman et al. 2009) (see Section S1 and Figure S3).

The effects of different solvents were also investigated since water does not dissolve many OA standards. Using a high purity solvent, e.g., acetonitrile, that fully dissolves different OA standards reduces the need for complex systems, such as custom-built evaporation-condensation generators (Sinclair and La Mer 1949; Xu et al. 2018). High purity acetonitrile was used as the solvent for standards that are immiscible in water. Figure 2 shows that there was minimal variability across the pure compounds dissolved in water versus compounds dissolved in acetonitrile (8–10% variability). This indicates that there were minimal differences in the interactions between the very polar water and somewhat polar acetonitrile as solvents. Thus, it is expected that there would be similarly minimal interactions with the POA-like OA standards with acetonitrile. Furthermore, Figure 2 shows that there was minimal variability between OA standards mixed with ammonium nitrate, no matter solvent (9–15% variability). This provides more evidence that there was minimal solvent interaction with the OA standard. Taken together, these observations indicate that internal mixtures with ammonium nitrate and different solvents (acetonitrile or H₂O) can be used as alternative and simple methods for accurately measuring RIE_OA across a wide range of SOA-like standards.

Comparisons of the RIE_OA measured using the Q-AMS versus the Q-ACSM and TOF-ACSM show two different trends that depend on OA composition (Table 1 and Figure 2). First, for the SOA-like standards (levoglucosan, xylitol, 1,2,6-hexantriol, mannitol, and threitol), the three instruments (Q-AMS, Q-ACSM, and TOF-ACSM) agreed within 25%. However, for POA-like standards, such as oleic acid, squalane, and pentadecanoic acid, the Q-ACSM and TOF-ACSM show large differences in RIE_OA when compared with the AMS. The largest differences in RIE_OA between both versions ACSM and Q-AMS instruments occur for low $\overline{\mathit{OSc}},$ which is generally associated with measurements of direct and/or fresh emissions of primary particles, and not background OA (e.g., urban background, rural, and remote), where the OA has higher $\overline{\mathit{OSc}}.$

3.3. Differences in ion transmission across AMSs and ACSMs impacts RIE_OA

The mass spectra of POA-like standards generally have a larger fraction of the total signal at higher m/z than those of SOA-like standards. Thus, how sensitive the instruments are to the higher m/z may play an important role in the RIE_OA. This can be explicitly seen in Equation (3)(3) $${\mathit{RIE}}_s{CE}_s=\frac{{10}^{12}{MW}_{NO3}{\mathit{RIE}}_{NO3}}{M_{in}{IE}_{NO3}Q{N}_A}\sum _{\mathit{all}\mathrm{ }i}{I}_{s,i}$$(3) , as lower sum of total ions would lead to lower RIE_s. Since the largest differences between AMS and ACSM RIE_OA occur with OA standards that have higher fractional total ion signals at higher m/z, the ion transmission, T_m/z, was investigated across instruments. The T_m/z has been measured for the Q-ACSM (Ng et al. 2011b); however, it has not been for other ACSMs and AMSs. As discussed in Jimenez et al. (2003), T_m/z affects RIE_OA, which makes a priori calculations of RIE_OA difficult. However, T_m/z was not considered once the HTOF was introduced, as discussed in DeCarlo et al. (2006).

In Figure 3, the ion transmission of naphthalene fragments is shown for the Q-ACSM, TOF-ACSM, TOF-ACSM-X, CTOF-AMS/m-AMS, HTOF-AMS, and LTOF-AMS. For this analysis, a linear fit is currently assumed for T_m/z and that the ion transmission goes to 0 at the x-intercept for each individual instrument, as shown in Figure 3. As there is no data for m/z > 128 (parent ion of naphthalene), it cannot be ruled out that the T_m/z is non-linear. Further, there is plenty of evidence from previous publications that non-zero intensities for ions greater than the x-intercept for the HTOF- and LTOF-AMS have been observed (e.g., Dzepina et al. 2007; Malmborg et al. 2017; Ulbrich et al. 2022a). Thus, the results presented for the remainder should be considered upper limits in T_m/z corrections. The m/z value at which the transmission starts decreasing is lowest for the Q-ACSM (starts at ∼ m/z 51) and highest for the HTOF- and LTOF-AMSs (starts at ∼ m/z 80). Further, how steeply the ion transmission decays show similar values for the ACSMs and CTOF-AMS but a slower decay for the HTOF- and LTOF-AMSs.

Figure 3. Comparison of effective ion transmission (T_m/z) measured on CTOF-AMS, HTOF-AMS, LTOF-AMS, Q-ACSM, TOF-ACSM, and TOF-ACSM-X. The measurements are naphthalene ion fragments measured in each instruments and ratioed to the NIST naphthalene ion fragments reported by Wallace (2022). Each m/z was normalized to the signal at m/z 51 and clusters of ions were averaged. Lines shown are the fit through the points. Note, only signals greater than 1% of the parent ion signal are shown (Ng et al. 2011b).

One potential complication interpreting the observed differences in ion transmissions measured across the different instrument versions is the ion extractor and the ion detector. The ion extractor, which is ∼46 mm long, is used in Tofwerk TOFs and is used to transfer the ions from the ionization region into the TOF (Drewnick et al. 2005). The ions are guided and made into a parallel beam into the extractor. The ions drift through the extractor at ∼50 eV before the ions are orthogonally extracted into the TOF by a pulsed voltage. Both the TOF-ACSM-X and HTOF-AMS have similar ion extractor and TOF technology; however, the detector is different (multi-channel plate, MCP, for AMS and ETP MagneTOF detector for TOF-ACSM-X). The different ion detectors used in the range of mass spectrometers in this study can exhibit different mass dependent ion detection efficiencies. The apparent lower ion transmission may be associated with the ETP MagneTOF detector detecting high m/z ions less efficiently than the MCP. This, however, does not explain the differences between the CTOF- and HTOF-AMS, as both use MCPs. Though the ion extractor is similar between the CTOF- and HTOF-AMS, the size of the drift tube is different and will affect ion transmission. Thus, a combination of factors that are associated with the TOF and detector impact the apparent T_m/z and complicate the ability to predict RIE_OA. The following discussion about T_m/z results should be interpreted as an effective ion transmission, which convolves all potential processes leading to ion detection efficiency.

The effect of differences in T_m/z on measured mass is explored with example POA spectra found in the AMS database (Ulbrich et al. 2009, 2022a, 2022b). POA mass spectra generally have more ions at higher m/z (Figure 4) than SOA mass spectra (Figure 5) due to the presence of molecular species that have not yet been photochemically processed (Donahue et al. 2012; Jimenez et al. 2009) and reduced by thermal decomposition (Canagaratna et al. 2015). The cumulative distributions of observed ions, using the T_m/z shown in Figure 3, show that 60–70% of the total ion signal are observed in the Q-ACSM while 80–90% of the ions are observed in the TOF-ACSM, TOF-ACSM-X, and CTOF-AMS (Figure 4). Since the Q-ACSM T_m/z is routinely measured with an internal naphthalene standard, reported Q-ACSM mass concentrations typically include a correction for T_m/z (Ng et al. 2011b). Thus, for cases where most of the ion signal is < m/z 150, this correction results in Q-ACSM mass concentration are nearly identical to the AMS. For POA dominated cases, however, Q-ACSM mass concentrations are systematically lower than AMS because unmeasured Q-ACSM ion signal above m/z 150 (due to near-zero ion transmission) cannot be corrected for with T_m/z.

Figure 4. Example mass spectra of (a) HOA, (c) BBOA, (c) COA, and (g) CCOA from PMF analysis of ambient data. Spectra taken from the AMS Database (Ulbrich et al. 2022b, 2022a). The different lines in (b), (d), (f), and (h) represent the cumulative ions observed for Q-ACSM (short-dashed), TOF-ACSM, TOF-ACSM-X, and CTOF-AMS (long-dashed), and HTOF- and LTOF-AMS (solid line). Note, this is an extreme example as ambient aerosol is rarely composed mostly of POA. Further, this is calculated without Q-ACSM T_m/z accounted for, which leads to lower total signal observed. Final processed observations from the Q-ACSM will have total cumulative signal similar to the HTOF and LTOF AMS with T_m/z included.

This is further explored in Figure S6. The mass spectra of oleic acid from the Q-AMS and Q-ACSM are compared, as well as the uncorrected and corrected T_m/z for the Q-ACSM is shown. There is less ion signal in the uncorrected than the corrected (e.g., uncorrected has 6% less ion signal without including T_m/z). However, both uncorrected and corrected T_m/z do not fully account for all the ion signal of oleic acid. The uncorrected signal of oleic acid from the Q-ACSM is 73% of the signal observed on the Q-AMS, and the T_m/z corrected signal from the Q-ACMS is 78% of the signal observed on the Q-AMS. Thus, the RIE_OA of oleic acid measured on the Q-ACSM is lower than measured on the Q-AMS Yet, the difference in oleic acid RIE_OA is larger between the Q-AMS and Q-ACSM than the 22% difference in T_m/z (Table 1), indicating other unknown processes are impacting RIE_OA for POA-like OA surrogates measured on the Q-ACSM.

For the SOA, all instruments generally observe a larger fraction of the total ions (>75%, Figure 5). Note, the number of ions differs between fresher SOA (less oxidized, oxidized OA, LO-OOA) and more aged SOA (more oxidized, oxidized OA, MO-OOA). A combination of factors could be influencing the larger number of high m/z ions in the LO-OOA spectrum, including the fact that the data is from the average of 15 urban studies (Ng et al. 2011a) in which LO-OOA may not be fully separated from POA in the positive matrix factorization (PMF), and/or the fact that the extent of aging for the LO-OOA may be different between campaigns.

Figure 5. Example mass spectra of (a) LO-OOA and (c) MO-OOA from PMF analysis of ambient datasets. Spectra taken from the AMS Database (Ulbrich et al. 2022b). Similar to Figure 4, for (b) and (d), the different lines represent the cumulative ions observed for Q-ACSM (short-dashed), TOF-ACSM, TOF-ACSM-X, and CTOF-AMS (long-dashed), and HTOF- and LTOF-AMS (solid line). Note, this is calculated without Q-ACSM T_m/z accounted for, which leads to lower total signal observed. Final processed observations from the Q-ACSM will have total cumulative signal similar to the HTOF and LTOF AMS with T_m/z included.

For studies that occur in sites away from direct emissions (e.g., urban background sites, rural sites, and remote sites), the differences in ion transmission are expected to have minimal impact on the mass measured between different instrument versions, leading to good agreement between AMS and ACSM results (Crenn et al. 2015; Fröhlich et al. 2015). However, for studies investigating point sources (e.g., indoor emissions; Katz et al. 2021) or less aged SOA, differences in the apparent T_m/z and how it impacts RIE_OA may need to be accounted for in comparisons between different AMS and ACSM instruments.

4. Improved Constraints on RIE_OA

4.1. Evaluation of laboratory measurements

Because a priori knowledge and/or measurements for molecular weight and $\overline{\mathit{OSc}}$ are not feasible for the AMSs and ACSMs, another way to constrain RIE_OA would be of value. Further, as discussed above, $\overline{\mathit{OSc}}$ does not appear to fully explain the measured RIE_OA. Since, as discussed above, RIE_OA can differ between AMS and ACSM instruments due to T_m/z dependence, we investigated the use of the fractional contributions of ions at high m/z’s compared to total ions observed to parameterize RIE_OA. RIE_OA as a function of the fractional contribution of ions less than m/z 50, hereafter called f_Σm/z50, is shown in Figure 6. We use m/z 50 because ion transmission is near 100% for all instruments (Figure 3). Further, as shown in Figure S7, when parameterizing RIE_OA as a function of the fractional contribution of ions, a cutoff of m/z 50 provides the best fit (maximum in R² and a minimum in chi squared) to observations.

Figure 6. (Top) Fraction of observed ions at m/z 50 (f_Σm/z50) or less to total observed ions for the PMF factors in Figures 4 and 5. (Middle) f_Σm/z50 for typical urban and rural/remote observations (Ulbrich et al. 2022a). (Bottom) Measured RIE of OA (RIE_OA) from this study for different instruments versus f_Σm/z50. The fit for the observations with f_Σm/z50 < 0.64, does not include PAHs and the saturated hydrocarbons, and is only for the AMS. The shaded area is the 95% confidence interval to the fit. Data from Katz et al. (2021) is included to show that the derived fit explains the measured RIE observed in that study while the lower RIE_OA from Slowik et al. (2004) is due to the PAHs in the lubricating oil. The slope for the fit is –14.46, intercept is 10.55, and R² is 0.53.

For the AMS data, there were two sets of populations. For standards with f_Σm/z50 ≥ 64%, the mean (median) RIE_OA was 1.23 ± 0.62 (1.18). The standards sampled with the ACSMs for these compounds fall within the spread in the RIE_OA measured with the AMS, in agreement with Figure 2. As the standards studied here have similar functionalities to SOA, it is expected that they will undergo thermal decomposition to CO₂, H₂O, CO, and/or smaller molecules (R1–R2). This thermal decomposition prior to electron ionization appears to lead to the RIE_OA of 1.23 ± 0.62 for high f_Σm/z50 standards.

For OA standards measured on the Q-AMS with f_Σm/z50 < 64%, there was an increasing trend between RIE_OA and f_Σm/z50 upwards. The trend indicates that for each 1% decrease in f_Σm/z50, the RIE_OA increases by ∼0.07 (absolute value). Compounds in this region generally have long-chain saturated hydrocarbon backbones with lower thermal decomposition than the standards more similar to SOA. However, as discussed above, some thermal decomposition may occur for saturated hydrocarbon backbones (Rice 1931, 1933; Rice, Johnston, and Evering 1932) during thermal decomposition of attached oxygenated functional groups (R1–R2). Measured RIE_OA across different AMS instruments and for most standards fall within or very near the 95% interval of the fit, with important exceptions of PAHs and saturated hydrocarbons. Q-ACSM measurements also fall outside this range as discussed below.

PAHs are generally resistant to fragmentation after electron ionization due to the stabilizing effects of the delocalized π-electrons (Dzepina et al. 2007; McLafferty and Tureček 2021); thus, it is not surprising that the PAHs show the lowest f_Σm/z50 values. Also, unlike non-aromatic hydrocarbon that may undergo thermal decomposition at ∼600 °C, PAHs undergo thermal decomposition at much higher temperatures (>1000 °C) (e.g., Tsai et al. 2002). Similar to other studies (Dzepina et al. 2007; Jimenez et al. 2016; Slowik et al. 2004), a low RIE_OA was observed for the PAH standards, similar to the low RIE_OA for lubricating oil with PAHs. PAHs generally have lower cross sections than compounds of similar carbon number (Bose and Westmoreland 2020; Harrison et al. 1966), potentially suggesting a lower RIE_OA compared to other compounds. However, unlike other compounds, the cross sections for PAHs can vary widely for the same number of carbon atoms or same number of fused rings (Bose and Westmoreland 2020), again indicating that predicting the RIE_OA for these species is challenging. Furthermore, minimal fragmentation for the high molecular weight PAHs may lead to lower overall observation due to the T_m/z being <100% at the m/z of the parent compounds (Figure 3). Thus, the combination of these processes is thought to lead to the deviation of RIE_OA versus f_Σm/z50. However, PAHs are generally a small fraction of the total OA signal (less than 1%) (e.g., Dzepina et al. 2007); so, this deviation from the observed trend of RIE_OA versus f_Σm/z50 will not impact interpretation of most ambient measurements. Finally, this deviation is also observed for aromatic-containing standards, as the one point from Xu et al. (2018) well below the uncertainty in the trend of RIE_OA versus f_Σm/z50 is phthalic acid. The contribution of pure aromatic backbone SOA compounds to ambient OA is not as well established, to date; thus, the impact of aromatic containing SOA compounds to RIE_OA is unknown.

Lubricating oil was also investigated in this study because it may be another potentially important contribution to HOA (e.g., Worton et al. 2014). For non-PAH containing lubricating oil, RIE_OA versus f_Σm/z50 follows the same trend in RIE_OA (point labeled Engine Oil in Figure 6). Slowik et al. (2004) and Jimenez et al. (2016) reported lubricating oil to have an RIE_OA ∼1.4, but the lubricating oil studied in Slowik et al. (2004) was composed of PAHs, which generally has lower RIE_OA (Figure 6). Thus, while different from the non-PAH lubricating oil measurements, the RIE_OA reported in Slowik et al. (2004) is consistent with the PAH standard results reported here and do not follow the trend of RIE_OA versus f_Σm/z50.

Saturated hydrocarbons RIE_OA show the largest differences between studies and measurements (Table 1). This may reflect a combination of processes, including how the standards were atomized (including partial evaporation with CPMA), the amount of thermal decomposition prior to ionization (R3–R9), and T_m/z. RIE_OA is highly variable for these standards. Since the volatility of saturated hydrocarbons is high (e.g., Lu, Zhao, and Robinson 2018), the contribution of saturated hydrocarbons to the HOA PMF factor is expected to be low. As prior studies have indicated that HOA and other POA contains oxygen and other functional groups, and the compounds measured in this study that fall within the trend in Figure 6 also have oxygen functional groups, the deviation from the saturated hydrocarbons will have minimal influence on the interpretation of ambient measurements.

Similar to the AMS, the ACSMs measured the RIE_OA for standards similar to SOA of ∼1.4 (e.g., f_Σm/z50 > 0.64). However, for f_Σm/z50 < 0.64, two different populations occur for the Q-ACSM and TOF-ACSM; Q-ACSM RIE_OA remains constant between 1 and 2 while the TOF-ACSM shows a slower increase in RIE_OA versus decreasing f_Σm/z50. As has been discussed throughout, the higher m/z ions appear to be contributing to the higher RIE_OA. However, as discussed in Section 3.3 and shown in Figure 3, the ACSMs have lower T_m/z than the AMS. Note, the Q-ACSM corrects for the fraction of ions not observed above m/z 50 with the measured T_m/z (Ng et al. 2011b) and thus leads to similar total ions as observed with the AMS for cases where all ion signal is observed at m/z < 150. As shown in Figure 4, for many types of POA, 4–13% of the ions are found at m/z > 150. Since the transmission efficiency for the Q-ACSM is near zero for these higher m/z ions, these ions are not measured and thus not easily corrected for (Figure 2). However, as the T_m/z correction should only lead to a smaller difference than observed, other unknown processes are likely also impacting the differences between the observed RIE_OA measured on the Q-ACSM and AMS. This could lead to the relatively constant RIE_OA for the Q-ACSM at low f_Σm/z50. Since the TOF-ACSM detects a higher fraction of ions with m/z > 150 (Figure 3), the trend between the RIE_OA versus f_Σm/z50 is more apparent. The TOF-ACSM RIE_OA is generally lower than that of the AMS due to the lower transmission efficiency of the TOF-ACSM compared to the AMS (Figures 3 and 6).

4.2. Comparison of the Improved Constraints on RIE_OA with external dataset

Consistency between these measurements and an external dataset from Katz et al. (2021) further verifies the relationship between RIE_OA and f_Σm/z50, as shown in Figure 6. As Katz et al. (2021) measured the RIE_OA on a different AMS instrument and a separate technique, this provides confidence in the relationship to better constrain the RIE_OA. Using the f_Σm/z50 observed by Katz et al. (2021) for the HOMEChem COA observations (34%), the derived relationship shown in Figure 6 would suggest an RIE_OA of 5.63 ± 1.17. This agrees with the average RIE_OA during HOMEChem of 5.57 ± 3.71 at the 95% confidence interval. Also, using the f_Σm/z50 for HOA and COA, the predicted RIE_OA for the ambient measurements from Philadelphia (Avery, Waring, and DeCarlo 2019), reanalyzed in Katz et al. (2021), would be between 4.44 ± 1.17 and 5.49 ± 1.70, which is in agreement with the RIE_OA calculated for the data set of 4.16.

Similar to the trends observed in the analysis of OA standards with Q-ACSM and TOF-ACSM (e.g., Figure 2), Katz et al. (2021) observed differences in RIE_OA agreement between the Q- and TOF-ACSMs and HTOF-AMS. The ACSMs were about a factor 2 lower than the HTOF-AMS and SMPSs. This is a similar factor difference for the measured oleic acid RIE_OA observed between the AMS and Q-ACSM (Figure 2). Due to T_m/z corrections, though, it is expected that the Q-ACSM would only be lower than the factor of 2 observed. At this point, it is not apparent what is leading to the higher difference between the Q-ACSM and HTOF-AMS. Therefore, caution is recommended when measuring direct reduced OA sources with the Q-ACSM.

4.3. Atmospheric implications for f_Σm/z50 for the AMS and ACSM

Figure 6 also shows the range of f_Σm/z50 observed for bulk OA measurements from urban and rural/remote (Ulbrich et al. 2009, 2022a, 2022b) sites. Various urban studies show f_Σm/z50 ranges from 0.56 to 0.68. The f_Σm/z50 for urban sites is higher than pure POA and between LO-OOA and MO-OOA because urban sites are typically dominated by SOA (Jimenez et al. 2009). Though some urban sites fall below f_Σm/z50 < 64%, the uncertainty in the fit of RIE_OA versus f_Σm/z50 is still consistent with the use of an RIE_OA of 1.4 for bulk OA. Away from urban regions, except for biomass burning events, the contribution of POA is minimal (Jimenez et al. 2009), leading to the rural and remote mass spectra being dominated by ions less than m/z 50 (Figure 6). Thus, the results in Figure 6 indicate that the RIE_OA of 1.4 is applicable for urban background measurements that are not strongly impacted by fresh/local emissions.

The f_Σm/z50 for different POA factors fall in the same region where RIE_OA increases for the OA standards. As POA factors contribute a small fraction of the bulk OA in urban environments (Jimenez et al. 2009), their contribution is minor for RIE_OA. However, when conducting source apportionment with PMF and/or directly sampling source aerosol, higher RIE_OA will be necessary (Katz et al. 2021; Reyes-Villegas et al. 2018). This indicates primary-like OA reported with PMF may be an upper-estimate due to the higher RIE_OA; however, due to the potential competing effects stated before (particle lens transmission, internal versus external mixing), there may be offsetting effects that may reduce the overall amount the primary-OA is over-estimated. As shown in Jimenez et al. (2016), within the stated uncertainty of the AMS and Sunset OC instrument, there was minimal systematic difference between the two instruments even at high primary OA fractional contribution to total OA in an urban environment. An important caveat for implementing the observed trend in RIE_OA versus f_Σm/z50 at this time, though, for ambient data, is that the trend was derived with lab-generated OA standards. This means that there are no competing effects with particle lens transmission and/or internally versus externally mixed aerosol. The aerosol was size-selected to ensure 100% transmission, which may not be possible for primary aerosol in ambient conditions (Zhang et al. 2005a, 2005b). Also, the E_B was well characterized either with counting of single particles on the Q-AMS or mixtures of ammonium nitrate and OA surrogates. Fresh POA generally is not well-mixed with rest of the non-refractory aerosol, resulting in not well characterized E_B (Canagaratna et al. 2004; Zhang et al. 2005a, 2005b). As shown in Equation (1)(1) $${\mathrm{C}}_{\mathrm{s}}\text{=}\frac{1}{{\mathrm{CE}}_{\mathrm{s}}{\mathrm{RIE}}_{\mathrm{s}}}\frac{{10}^{12}{\text{MW}}_{\text{NO}3}}{{\text{IE}}_{\text{NO}3}\mathrm{Q}{\mathrm{N}}_{\mathrm{A}}}\sum _{\text{all}\mathrm{i}}{\mathrm{I}}_{\text{s,i}}$$(1) , OA mass concentration is a function of the product CE and RIE_OA. POA CE and RIE_OA may have off-setting effects on mass concentration. Thus, the parameterization shown in Figure 6 may result in incorrect RIE_OA as this parameterization does not take E_L and E_B into account. However, f_Σm/z50 will provide a metric to determine if a higher RIE_OA will be needed for the OA. If the observed f_Σm/z50 is < 0.64, this indicates that the OA with RIE_OA = 1.4 would be an upper limit of the OA mass concentration.

5. Conclusions

As organic aerosol, OA, is an important fraction of particulate matter, accurate quantification of its mass concentration is important. Here, the organic aerosol relative ionization efficiency, RIE_OA, was measured on the Aerodyne Aerosol Mass Spectrometer (AMS) and Aerosol Chemical Speciation Monitor (ACSM) (involving a total of 7 different mass spectrometers) to further evaluate what controls the RIE_OA and how the different instruments respond to OA standards. Although the OA oxidation state explains some of the differences in measured RIE_OA for different standards, there is also large variability in the RIE_OA for a given OA oxidation state. In this study, a new metric was developed to investigate RIE_OA for the AMS and ACSM, and the results from this study are summarized below.

1. The RIE_OA for different OA surrogates were measured on the Q-AMS, Q-ACSM, and TOF-ACSM. It was found that for SOA-like standards, the RIE_OA agreed across the three instruments; however, for POA-like standards, there were differences in the measured RIE_OA with the Q-AMS having higher RIE_OA than the TOF-ACSM (intermediate) and Q-ACSM (lowest).

2. Further, there were minimal differences in the measured RIE_OA when different solvents were used (water versus acetonitrile) and in comparing pure OA-standard to a mixture of an OA-standard with ammonium nitrate. The use of ammonium nitrate alleviates the need to measure the collection efficiency and to directly measure the RIE_OA.

3. Some of the differences in the RIE_OA can be partially explained due to differences in the dependency of the ion transmission on m/z (T_m/z). This dependency is most important for the ACSMs measuring fresh POA emissions.

4. The fractional contribution of the total ion signal for ions less than m/z 50, which is denoted f_Σm/z50, provides a more robust method to predict how RIE_OA varies across OA standards.

5. The metric f_Σm/z50 indicates that for bulk urban aerosol away from fresh, direct POA emissions, rural OA, and remote OA, the RIE_OA is 1.23 ± 0.62, in agreement with the value of 1.4. This metric can be used to either determine if bulk OA or OA factors from positive matrix factorization may need a higher RIE_OA and may be over-estimated.

6. The metric and measured RIE_OA found in this study is applicable to prior measurements using different AMSs, providing overall confidence in this metric’s sensitivity to the tuning and vaporizer temperature.

7. The metric also assists in explaining the differences in RIE_OA observed between polycyclic aromatic hydrocarbons and other POA-like OA.

8. Finally, the metric shows how the ACSMs measured RIE_OA behaves differently than the AMS measured RIE_OA.

9. However, it is important to keep the caveat in mind that there may be competing factors impacting the quantification of POA aerosol, including particle lens transmission and/or internally versus externally mixed aerosol.

10. At this time, it is still encouraged to report POA mass concentration from PMF and source apportionment studies with RIE_OA = 1.4 as an upper-limit, the f_Σm/z50 for the PMF or direct OA measurement, and the potentially lower limit of the OA mass concentration with the caveat about particle lens transmission and mixing state of OA.

As the newly proposed behavior of RIE_OA versus f_Σm/z50 has only been investigated for laboratory standards and a subset of published observations, further evaluation of this trend needs to be conducted on ambient measurements before it can be widely used to correct AMS and ACSM measurements.

Acknowledgement

We acknowledge Wen Xu for use of data in this analysis and discussions with the AMS and ACSM Users’.

Data availability statement

Mass spectra for Figures 4 and 5 are available at https://cires1.colorado.edu/jimenez-group/AMSsd/ and https://cires1.colorado.edu/jimenez-group/HRAMSsd/. The rest of data is available upon request.

Additional information

Funding

The work was supported by NASA Grant 80NSSC19K0065 and DOE SBIR DE-SC0017041.