The Effect of Oxygen on Organic Haze Properties

Hazes were produced with 0.1% CO₂ (Airgas, 99.99%), 0.1% CH₄ (Airgas, 99.99%), and 0, 2, 20, and 200 ppmv O₂ (Airgas, 99.994%) in N₂ (Airgas, 99.999%). These gas mixtures are an approximation for the early Earth's atmosphere. A CO₂:CH₄ ratio of 1:1 was chosen because it produces sufficient aerosol signal for the instrumentation used in this study (Trainer et al. 2006; Hasenkopf et al. 2010). Moreover, while Hicks et al. (2016) showed that most of the oxygen incorporation for CO₂/CH₄/N₂ hazes comes from the oxygen in the precursor CO₂, some of the oxygen in these particles could result from oxygen contamination in the precursor gases.

The photochemical haze generation system has been described in detail previously (Trainer et al. 2006; Hasenkopf et al. 2010; Hörst & Tolbert 2013). Briefly, CO₂, CH₄, O₂, and N₂ flow into a mixing chamber and mix for at least 8 hr. Then, a mass flow controller (MFC; Mykrolis, FC-2900) flows the gas mixture into the UV reaction cell at a rate between 60–100 standard cubic centimeters per minute (sccm). One side of the reaction cell has a deuterium continuum lamp (Hamamatsu, L1835) that outputs light between 115 and 400 nm. Chemistry is initiated at ambient conditions (∼620 Torr and 20°C), and the resulting haze particles flow to the instrumentation used for analysis: a quadrupole aerosol mass spectrometer (Q-AMS) and a scanning mobility particle sizer (SMPS) for particle effective density measurements, and a photoacoustic spectrometer and a cavity ring-down spectrometer (PASCaRD) for complex refractive index measurements. Due to flow rate constraints, particles were flowed alternately to these instruments. An additional 200–240 sccm of prepurified N₂ dilution flow was added after the UV reaction cell for measurements with the SMPS and PASCaRD to satisfy instrument requirements.

The Q-AMS alternates between particle time-of-flight (PToF) mode and mass spectrum (MS) mode to measure the size distribution by mass of selected ion fragments and particle composition, respectively. Aerosol mass spectrometry has been described in detail previously (Jayne et al. 2000; Jimenez et al. 2003). Briefly, as particles and gas enter the AMS, an aerodynamic lens focuses the particles into a beam while differential pumping removes the gas. The particles then travel through a high vacuum time-of-flight chamber where, in PToF mode, their vacuum aerodynamic diameter (D_va; the diameter of a unit density sphere that will reach the same terminal velocity in vacuum as the particle of interest) is calculated from their size-dependent velocity through the chamber (Jimenez et al. 2003). In MS mode, the particle beam is flash vaporized at 600° C and ionized by electron ionization (70 eV). Quadrupole mass spectrometry (QMA 410, Balzers, Liechtenstein) is used to analyze the ions with unit mass resolution.

The SMPS is a combination of two instruments: a differential mobility analyzer (DMA; TSI, 3081), and a condensation particle counter (CPC; TSI, 3022A). Briefly, polydisperse particles flow into the DMA, where a constant electric field is applied to the particles. The particles are then size-selected based on their electrical mobility against the drag force from the sheath flow. The diameter measured by the DMA is the electrical mobility diameter (D_m; the diameter of a unit density sphere with the same electrical mobility in an electric field as the particle of interest; Flagan 2001; DeCarlo et al. 2004). The size-selected particles then enter the CPC where the number density (N, particles cm⁻³) at each D_m is measured by light scattering. The sheath flow used was 3.0 liter per minute.

Before haze particles enter PASCaRD, the polydisperse particles flow to the DMA to be size-selected. The size-selected particles then flow to the cavity ring-down spectrometer (CaRD) where particle extinction (b_ext, Mm⁻¹) is measured at 405 nm. The same particles then flow to the photoacoustic spectrometer (PAS) where particle absorption (b_abs, Mm⁻¹) is measured at 405 nm. After the PAS, the particles flow into the CPC where particle number density (N, particles cm⁻³) is measured. Extinction and absorption measurements are made for five to six different mobility diameters between 175–350 nm. The best fit n and k are found by minimizing the merit function described in Zarzana et al. (2014), which compares the experimentally retrieved extinction and absorption measurements to extinction and absorption measurements calculated using Mie theory and correcting for doubly-charged particles that can result from the DMA size selection process. For details on the PASCaRD instrumentation and the refractive index calculation, see Ugelow et al. (2017).

Size distributions measured by the Q-AMS and SMPS for the particles produced in this study are shown in Figure 1. It was not possible to measure the Q-AMS size distribution for the particles produced with 200 ppmv O₂ because of the particles' smaller diameters. The aerodynamic lens transmits particles with a d_va between 20 and 1000 nm, but the collection efficiency drops off steeply for d_va < 60 nm (Jayne et al. 2000). About 60% of the particles produced with 200 ppmv O₂ have diameters less than 60 nm. Moreover, the AMS y-axis is normalized to the peak area because the AMS measures the size distribution for the most prominent organic peaks, not every peak in the mass spectrum.

Zoom In Zoom Out Reset image size

Using the mode D_va and D_m from the size distributions for each haze, the particle effective densities were calculated by the relationship shown in DeCarlo et al. (2004):

$$\begin{eqnarray}&&{\rho }_{\mathrm{eff}}={\rho }_{0}\displaystyle \frac{{{\rm{D}}}_{\mathrm{va}}}{{{\rm{D}}}_{{\rm{m}}}}\end{eqnarray} \tag{ 1 }$$

where ρ₀ is unit density (1 g cm⁻³). The calculated effective densities are included in Table 1 and Figure 2(a) shows the particle effective densities as a function of precursor O₂. Because the Q-AMS could not measure the size distribution for the particles produced with 200 ppmv O₂, the particle effective density for this haze could not be calculated. For comparison to hazes synthesized without any oxygenated species present, the effective density of photochemical haze particles produced with 0.1% CH₄ in N₂ from Hörst & Tolbert (2013) is included (blue point).

Zoom In Zoom Out Reset image size

Table 1.  Summary of Particle Effective Density, the Ratio of m/z 44 to Total Organics, and the Refractive Index at 405 nm

			λ = 405 nm
Mixture	ρeff (g cm−3)	m/z 44/Total Organics	n	k
0.1% CH4/N2	0.65 ± 0.11(1)	⋯	1.40 ± 0.02	0.002 ± 0.001
0.1% CH4/0.1% CO2/N2	0.94 ± 0.03	0.081 ± 0.004	1.58 ± 0.04	${0.001}_{-0.001}^{+0.002}$
2 ppmv O2/0.1% CH4/0.1% CO2/N2	1.03 ± 0.05	0.084 ± 0.005	1.53 ± 0.03	0.002 ± 0.002
20 ppmv O2/0.1% CH4/0.1%CO2/N2	1.12 ± 0.05	0.094 ± 0.001	1.67 ± 0.03	${0.002}_{-0.002}^{+0.003}$

Note. Uncertainties in particle effective density and m/z 44 to total organics ratio are the 1σ standard deviation of the reproducibility of each experiment. Uncertainties in n and k result from 1σ standard deviations in the extinction, absorption, and particle number density measurements at each particle mobility diameter, along with the 1σ standard deviation of the reproducibility of each retrieval.

References. (1) Hörst & Tolbert (2013).

Download table as:  ASCIITypeset image

As seen in Figure 2(a), as the oxygen content in the precursor gas mixture increases (by adding CO₂ and then O₂), the particle effective density increases. Increasing particle density with increasing particle O:C ratio has been observed for Earth aerosol (Pang et al. 2006) and planetary haze simulations (He et al. 2017; Hörst et al. 2018), suggesting that the oxygen content in the hazes produced in this study is increasing with precursor O₂. To confirm this, a measure of oxygen content as a function of precursor O₂ was made using the Q-AMS, where the peak at m/z = 44 (COO⁺) was compared to the total organic signal and is included in Table 1. As particles are produced with increasing amounts of precursor O₂ from 0, 2, and then 20 ppmv, the ratio of m/z = 44 to the total organic signal increases as 0.081 ± 0.004, 0.084 ± 0.005, and 0.094 ±0.001, respectively, confirming that as precursor oxygen increases to 20 ppmv, oxygen incorporation into the particles increases as well. As the particles produced with 0 and 2 ppmv O₂ have ratios that are not statistically different from each other, it is possible that these two hazes have similar oxygen contents. Due to possible particle morphology differences, these two hazes can be compositionally similar but still have statistically different effective densities.

Using the haze particles' calculated effective densities and the SMPS measured particle number densities, haze particle mass loadings (μg m⁻³) were calculated. Figure 2(b) displays the mass loading for each haze, including the mass loading of photochemical haze produced with 0.1% CH₄ and N₂ calculated using the reported values in Hörst & Tolbert (2013). The effective density measured for the 20 ppmv O₂ haze particles (1.12 g cm⁻³) was used to calculate the mass loading for the 200 ppmv O₂ haze. Because effective density increases with oxygen content, the 200 ppmv O₂ haze mass loading represents a lower limit. As Figure 2(b) shows, haze mass loading decreases nonlinearly with precursor O₂ concentration. Additionally, the mass loading of haze produced with 1:1 CO₂:CH₄ in N₂ is slightly greater than without CO₂, which Trainer et al. (2006) also observed. Moreover, the total mass loading of the haze produced with trace amounts of molecular oxygen is substantial. Even with the addition of 200 ppmv precursor O₂, the mass loading of haze in our experiments is comparable to current mass loadings of organic aerosol in the Earth's atmosphere (Junker et al. 2004; Gupta et al. 2007; Jimenez et al. 2009; Levy et al. 2014; Wang et al. 2016). Therefore, while not being a thick atmospheric haze, this amount of aerosol could still impact the radiative budget and surface temperature of the early Earth, just as the current aerosol in Earth's atmosphere do today.

Figure 3 shows an example of the extinction and absorption efficiency (Q_ext and Q_abs, respectively) as a function of particle size for the CO₂/CH₄/N₂ haze. The points are experimentally retrieved data and the lines are fits of the data using the procedure in Zarzana et al. (2014) and Ugelow et al. (2017) to determine the real and imaginary refractive indices using Mie theory. The figure shows strong agreement between the measured and calculated extinction and absorption efficiencies, and is representative of the agreement between the calculated and experimental data for all of the hazes studied. Table 1 includes the complex refractive indices at 405 nm retrieved for the haze particles produced in this study, including a non-oxygen-containing CH₄/N₂ haze. Using the standard deviation for the error bars of the retrieved k values results in k < 0 for certain hazes. However, k cannot be negative so the lower bound error bar for these hazes ends at k = 0.

Zoom In Zoom Out Reset image size

As oxygen is added via CO₂ and O₂, the real refractive index of the haze increases. This could mean that greater oxygen incorporation into haze particles results in increasing n values. The n values for the CO₂-containing mixtures without O₂ and with 2 ppmv O₂ are similar within error bars, and, based on their relative oxygen content measured by the Q-AMS, we believe that these two hazes are compositionally similar.

Whereas haze particles produced with just CH₄ and N₂ have a slight absorption (k = 0.002), particles produced with oxygenated species are non-absorbing. Although the measured imaginary refractive indices of the oxygenated particles are slightly greater than zero, the large error in the retrieved values cannot exclude k = 0, especially when considering the raw PAS signals. The raw PAS signal for the haze produced with 0.1% CH₄ in N₂ is the only haze with a raw absorption signal greater than zero, whereas the other hazes' raw absorption signals average to about zero (or less). Hasenkopf et al. (2010) measured a larger k value (k = 0.055) for hazes produced with 0.1% CH₄ and 0.1% CO₂ in N₂ at 532 nm and found that producing particles with CO₂ enhances particle absorption at this wavelength. However, that experiment was done with CaRD alone and Zarzana et al. (2014) demonstrated that coupling extinction to absorption measurements, as in the present study, greatly improves the complex refractive index retrieval, particularly resulting in enhanced accuracy and precision in k. Gavilan et al. (2017) observed a similar trend to the present study where particles produced with CH₄ and N₂ absorb light more strongly than particles produced with a 1:1 CO₂:CH₄ ratio in N₂. Additionally, for the portion of the visible light spectrum that they were able to obtain valid imaginary refractive indices, low k values (≤10⁻³) were reported.

The k value reported for the CH₄/N₂ haze at 405 nm is lower than the commonly applied value by Khare et al. (1984) where, using linear interpolation, at 405 nm k = 0.081. However, using lognormal interpolation Gavilan et al. (2017) reported a lower k value as well at 405 nm, where k = 0.0099. Possible reasons for differences in the observed k values include different energy sources, different initial CH₄ concentrations used to make the hazes, and therefore different haze particle compositions.

To further confirm that the oxygen-containing hazes produced in this study are non-absorbing, extinction (b_ext, Mm⁻¹) and scattering (b_sca, Mm⁻¹) were measured by a cavity-attenuated phase-shift spectrometer (CAPS-PM_ssa; Onasch et al. 2015; De Haan et al. 2017) at 450 nm for the total aerosol size distribution. The ratio of particle scattering to particle extinction describes the wavelength-dependent single-scattering albedo (ω), thus the particle's radiative impact, where ω = b_sca/b_ext. Particles with ω < 1 are light absorbing, whereas particles with ω = 1 are non-absorbing. Figure 4 displays the single-scattering albedo calculated from the CAPS measured extinction and scattering by the CH₄/N₂ and CO₂/CH₄/N₂ haze particles, with error bars reflecting the measurement reproducibility.

Zoom In Zoom Out Reset image size

The CAPS single-scattering albedo at 450 nm rises from 0.947 ± 0.008 to 0.995 ± 0.007 when CO₂ is added to the haze mixture, confirming that the particles produced without CO₂ are light absorbing, whereas the particles produced with CO₂ are essentially non-absorbing. Included in Figure 4 is the single-scattering albedo for the particles produced in this study at 405 nm calculated using the retrieved refractive index, mode mobility diameter, and Mie theory. Agreeing with what is observed at 450 nm, the particles produced with CO₂ are non-absorbing, whereas the particles produced with only CH₄ and N₂ are light absorbing, and absorb more light at 405 nm than at 450 nm. This strong wavelength dependence is typical for "brown carbon" oligomerized aerosol, and is also observed by Gavilan et al. (2017). A haze that is non-absorbing would form a light-scattering layer in the atmosphere. This light-scattering haze would prevent certain wavelengths of sunlight from reaching a planet's surface. Sagan & Chyba (1997) and Wolf & Toon (2010) calculate that an Archean organic haze could act as an ultraviolet radiation shield to greenhouse gases below the haze. While these studies assume a light-absorbing haze, a light-scattering haze could act as a shield as well.