A broadband cavity enhanced absorption spectrometer for aircraft measurements of glyoxal, methylglyoxal, nitrous acid, nitrogen dioxide, and water vapor

. We describe a two-channel broadband cavity enhanced absorption spectrometer (BBCEAS) for aircraft measurements of glyoxal (CHOCHO), methylglyoxal (CH 3 COCHO), nitrous acid (HONO), nitrogen dioxide (NO 2 ) and water (H 2 O). The instrument spans 361–389 and 438–468 nm, using two light-emitting diodes (LEDs) and a single grating spectrometer with a charge-coupled device (CCD) detector. Robust performance is achieved using a custom optical mounting system, high-power LEDs with electronic on/off modulation, high-reﬂectivity cavity mirrors, and materials that minimize analyte surface losses. We have this during two and two ﬁeld campaigns to date. The demonstrated HONO and (pptv) in accuracy is and by available absorption cross


Introduction
Broadband cavity enhanced absorption spectroscopy (BBCEAS) belongs to a class of techniques that use high-finesse optical cavities to achieve sensitive measurements of 15 optical extinction (Fiedler et al., 2003). BBCEAS is distinct from other techniques in this class, such as cavity ring-down spectroscopy (CRDS), because it employs a broadband light source and a multichannel detector. Broadband measurements enable the simultaneous detection of multiple absorbing species across a wide spectral region, and the quantification of species with significant spectral overlaps in their features. The basic Printer-friendly Version
Similar to glyoxal, there is a need for accurate, rapid, in situ measurements of HONO, which plays an important role in atmospheric radical budgets through photolytic production of OH radicals and NO (Platt et al., 1980;Alicke et al., 2002). Scientific questions remain about its sources, sinks, and vertical profile (Zhou et al., 1999;Su et al., 15 2008; Young et al., 2012;VandenBoer et al., 2013). Existing detection methods can be categorized as wet chemistry, mass spectrometry, and optical spectroscopy. The wet chemical detection methods are sensitive, but generally rely on conversion of HONO to nitrite ion (NO − 2 ) (Appel et al., 1990;Dibb et al., 2002;Kleffmann et al., 2006) and may be susceptible to chemical interferences and sampling artifacts (Stutz et al., 2010). Re- 20 cently, chemical ionization mass spectrometry (CIMS) with acetate ion (Roberts et al., 2010) or iodide ion (Veres et al., 2015) chemistry has been successfully used for sensitive HONO detection. Spectroscopic methods for remote sensing include long-path DOAS and MAX-DOAS (Platt et al., 1980;Hendrick et al., 2014). In situ, spectroscopic detection methods include cavity ring down spectroscopy (Wang and Zhang, 2000), 25 tunable diode laser spectroscopy (Li et al., 2008;Lee et al., 2011) and Fourier transform spectroscopy (Barney et al., 2000;Yokelson et al., 2007), although limited attempts have been made for field deployment. BBCEAS has been used successfully for ground-based measurements of HONO (Washenfelder et al., 2011a;Young et al., 2012).
Here, we present a new aircraft BBCEAS instrument, the Airborne Cavity Enhanced Spectrometer (ACES), and describe its use to measure CHOCHO, HONO, NO 2 , methylglyoxal (CH 3 COCHO), and water (H 2 O). This instrument follows the de-5 velopment of laboratory and ground-based BBCEAS field instruments by our group (Washenfelder et al., 2008(Washenfelder et al., , 2011aYoung et al., 2012), with significant improvements in engineering and data acquisition that allow rapid, precise aircraft sampling. This is the first instrument for in situ measurements of CHOCHO from an aircraft. The HONO measurement is lower in precision than that of CHOCHO due to lower cavity mirror re-10 flectivity and narrower LED spectral output, but it is sufficient for aircraft measurements under high signal conditions, such as in biomass burning plumes. The ACES instrument was successfully deployed during the SouthEast NEXus (SENEX) 2013 and the Shale Oil and Natural Gas Nexus (SONGNEX) 2015 aircraft studies, where it operated on the NOAA WP-3D aircraft at altitudes from 0-7 km and flight durations of 6-7 h. In The optical system is shown schematically in Fig. 1a and technical details for the components are given in Table 1. Two LEDs (NCSU033B, Nichia Corp., Tokyo, Japan; LZ1-00DB05, LedEngin Inc., San Jose, CA, USA) are separately temperature-controlled using thermoelectric coolers. Their output light is collimated using off-axis parabolic mirrors (50328AL, 2.0 cm effective focal length, Newport Corp., Irvine, CA, USA) prior 15 to entering each cell. The off-axis parabolic mirrors optimize both photon throughput and space efficiency. The light is coupled into a high-finesse cavity formed by highreflectivity mirrors (1 m radius of curvature, Advanced Thin Films, Boulder, CO, USA), which are separated by 48 cm. The difference in sensitivity between this design and a confocal mirror separation is negligible (Wild et al., 2014). 20 The light exiting the cavity is imaged by an off-axis parabolic mirror (50331AL, 15.2 cm effective focal length, Newport Corp., Irvine, CA, USA) through a band pass filter onto a 0.5 cm F/2 lens that couples the light into a 1 m UV-VIS fiber optic bundle. The bundle contains two groups of seven 200 µm-diameter fibers Princeton Instruments,Trenton,NJ,USA)  contains a 1200 groove mm −1 (500 nm blaze) grating centered at 418 nm, with spectral coverage of 119 nm. Two regions of interest (2048 × 128 pixels each) are defined based on the illumination of the CCD. In addition, a dark region of the CCD (2048 × 50 pixels) between the two regions of interest is recorded to monitor photon diffusion among adjacent pixels and light scattering inside of the spectrometer.

5
The CCD must be darkened while the accumulated charge is moved across the semiconductor surface to a readout amplifier and analog-to-digital converter. This would typically be achieved using a mechanical shutter that requires ∼ 8 ms to open and close, and can fail after rapid, continuous operation. To improve our instrument duty cycle and reliability, we electronically modulate the optical output of the LEDs. We 10 further improve the duty cycle and signal-to-noise ratio with a low readout amplification (gain: 16 e − count −1 , digitization speed: 2 MHz) and a CCD temperature of −50 • C.
For automated wavelength and lineshape calibrations, we use a Hg lamp (HG-1, Ocean Optics, Dunedin, FL, USA) with a custom adapter located between the band pass filter and fiber optic bundle to introduce Hg light without obscuring the main beam 15 path. A fused silica window (Edmund Optics, Barrington, NJ, USA) mounted at 45 • inside the adaptor reflects Hg light to illuminate the fiber optic during wavelength calibrations. The spectrometer wavelength calibration and lineshape are calculated from four narrow Hg lines at 365. 02, 404.66, 407.78 and 435.83 nm (air wavelengths;Sansonetti et al., 1996), assuming a Gaussian lineshape.

Cage system and hardware
As shown in Fig. 1a, the optical components are mounted in a cage system, measuring 63 × 29 × 13 cm. The cage system consists of carbon fiber rods with 1.25 cm outside diameter (OD), and mounts for each optical component attached to the rods (Wild et al., 2014). Custom-designed aluminum plates aligned and locked parallel to each other on 25 the carbon rods provide robust alignment without the use of spring-loaded, commercial mirror mounts. This system provides stable optical alignment that does not require re-Introduction alignment and is insensitive to vibration, pressure, or temperature changes. The optical cavity, spectrometer, and CCD are mounted in a temperature-controlled box, which is maintained at 30 or 35 • C using thermoelectric coolers.

Flow system
The flow system for ACES includes the inlet, aerosol filter, two cells, pressure sensors, 5 temperature sensors, mass flow controllers, and pump. It is shown schematically in Fig. 1b and described in further detail below. The inlet consists of two parts: a coaxial inlet designed for overflow by calibration gases and a filter to remove aerosols. All parts are made from Teflon perfluoroalkoxy polymer resin (PFA), which provided the best CHOCHO transmission efficiency in lab-10 oratory tests (see Sect. 2.4). The coaxial inlet consists of 0.95 cm OD tubing (0.35 m length, 0.79 cm ID), which contains 0.64 cm OD tubing (0.15 m length). The 0.64 cm OD tubing runs through the PFA tee to allow additions of calibration gases (Wagner et al., 2011). The calibration gas flow is 0.3-0.5 sL min −1 (standard liter per minute) greater than the total sample flow rate of 5.0 sL min −1 . During aircraft sampling, the 15 total length of the coaxial inlet is 1.5 m and it is contained in a winglet that extends outside the boundary layer of the aircraft. A PTFE membrane filter (25 µm thickness, 4.6 cm diameter, 1 µm pore size; Tisch Environmental, Cleves, OH, USA.) is installed downstream of the coaxial inlet to remove ambient aerosol because aerosol extinction limits the effective path length and adds 20 a variable background extinction. The filter is changed prior to each 8 h aircraft flight and once or twice per day for ground based measurements. Ground-based data were examined for discontinuities in CHOCHO or HONO before and after a filter change, and no discrepancies were observed within the measurement precision. Prior experience with NO 2 (Osthoff et al., 2006;Wagner et al., 2011) and CHOCHO (Washenfelder 25 et al., 2011a) suggest that aerosol filters transmit these compounds quantitatively.
After the filter, the flow is evenly divided to two sample cells constructed of Teflon PFA (2.5 cm OD, 1.5 cm ID, 40 cm length). Unlike in previous designs (Wagner et al., 11217 Introduction  Washenfelder et al., 2013), the ACES instrument does not include mirror purges of dry gas for mirror cleanliness. The mirror purges were found to be unnecessary, and their absence eliminates the small uncertainties from flow dilution and the relative sample length occupied by sample gas and purge gas over the length of the cavity. The cells are connected to separate mass flow controllers (MC-50SLPM-D-DB15B, 5 Alicat Scientific, Inc., Tucson, AZ, USA) and a scroll pump (IDP-3, Agilent Technologies, Inc., Santa Clara, CA, USA) to maintain a constant flow rate of 2.5 sL min −1 , resulting in a residence time less than 1.5 s in each cell. This flow rate and residence time were consistent with the data acquisition rate and minimized He and ZA consumption during reference measurements. The cell pressures (PPT, Honeywell International Inc., Plymouth, MN, USA) and temperatures (KMQSS-020U-6, Omega Engineering Inc., Stamford, CT, USA) are measured immediately downstream of the flow. The cavity loss (or effective optical path length) was measured by sequentially overflowing the inlet with He and zero air (ZA) from compressed gas cylinders or by acquiring zero air measurements at different pressures. Further details are given in Sect. 3.1.

Flow system materials
Flow system materials can potentially cause production or loss of target analytes, affecting the accuracy of in situ measurements. Previous tests have demonstrated that NO 2 has negligible losses on Teflon and metal surfaces (Osthoff et al., 2006;Fuchs et al., 2009). Prior measurements have shown that inlet length has negligible impact 20 on glyoxal losses for PTFE tubing (Huisman et al., 2008).
We measured glyoxal losses for ten materials to determine the best cell and tubing choices. For the tests, we configured both BBCEAS channels to measure CHOCHO, with the cells connected in series and a length of tubing between them. Losses between the two cells were attributed to the tubing material. Constant CHOCHO concen-  Table 2 lists the ten tubing materials and their measured losses, in units of fractional loss per unit tubing surface area per residence time (% cm −2 s −1 ). Teflon PFA showed the smallest loss (0.0001 ± 0.005 % cm −2 s −1 ), while glass tubing had greatest loss (0.211 ± 0.006 % cm −2 s −1 ). Finishing methods also affected the loss rate, with lower losses by polished metal surfaces compared to unpolished metal surfaces. We 5 calculate the total loss of CHOCHO from the surface area and residence time of individual components of the ACES sample system to be less than 0.1 %, even for inlet tubing of 10 m in length and 0.79 cm ID with 5 sL min −1 flow. The relative humidity during these experiments was low (< 20 %) and constant. At elevated relative humidity, the losses may be greater, although we expect this to be a small effect (Washenfelder et al., 2008) 10 and did not observe a dependence of the CHOCHO/HCHO ratio on relative humidity during SENEX 2013 that would indicate inlet effects. Actual inlet lengths were 1.5 m for aircraft sampling during SENEX and SONGNEX, 3-12 m for ground-based sampling during CARE Beijing and 6.5 m for ground-based sampling during UBWOS 2014. 15 During aircraft operation, the ACES instrument is typically turned on 1 h prior to flight, allowing the LEDs, optics box, and CCD to stabilize in temperature. The CCD dark background is characterized by acquiring 50 spectra under dark conditions with the same integration time as the subsequent measurements. This is necessary because the CCD produces non-zero signal under optically dark conditions. At the beginning 20 and end of each flight, we measure dark background spectra, Hg spectra for wavelength calibration and instrument lineshape, and standard additions of NO 2 and CHO-CHO. The inlet filter is changed before and after each flight. During flight, ZA is introduced for 30 s every 5 min, while He is introduced for 30 s every 15 min for calibrating mirror reflectivity. During ground-based operation, the gas additions are 30-60 s in duration, depending on the inlet length. Inlet filter change, dark background, and Hg wavelength and instrument lineshape measurements are performed once or twice per day.

Determination of cavity loss and trace gas extinction
BBCEAS instruments measure steady-state light intensity transmitted through an optical cavity. The light attenuation inside the cavity includes (1) absorption, scattering, and transmission losses by the cavity mirrors, (2) Rayleigh scattering by gas within the 5 cavity, (3) absorption by trace gases, and (4) Mie scattering by particles. A general expression describing the sum of the extinction is where α i (λ) is extinction by the i th component, λ is the wavelength of light, R L is the ratio of the total cell length to the sample length, d is the cavity length, R(λ) is the mirror reflectivity, I 0 (λ) is the reference spectrum (without absorbing trace gases or aerosols and at the same temperature and pressure as the sample spectrum), and I(λ) is the measured spectrum of ambient air, including trace gas absorptions, as a function of wavelength (Washenfelder et al., 2008).
The ACES instrument has no mirror purges, so R L is unity. We define I 0 (λ) as the 15 reference zero air spectrum so that the term α(λ) Rayleigh explicitly includes its Rayleigh scattering. For aircraft measurements, we must also explicitly account for the difference between the Rayleigh scattering of the reference zero air spectrum, I ZA (λ), and sample spectrum, I sample (λ), due to pressure differences during sampling. Incorporating these changes gives where ∆α Ray (λ) = α Ray,ZA (λ)−α Ray,sample (λ). The α i ,sample (λ) summation on the left hand side of Eq. (2) where σ i and N i are the absorption cross section and number density for the i th trace absorber. Accurate measurements of the trace gas absorption by BBCEAS require calibra-5 tion of mirror reflectivity, R(λ), or the cavity loss, which represents the inverse of the effective path length and is defined as α cavity (λ) = (1 − R(λ))/d . This quantity can be determined by introducing a species with well-known extinction into the cavity. For example, previous studies have used known Rayleigh scattering cross sections of He and N 2 or ZA (Washenfelder et al., 2008;Thalman and Volkamer, 2010) or N 2 and N 2 /NO 2 10 mixtures (Langridge et al., 2006;Venables et al., 2006). We have tested and compared two methods to determine α cavity (λ). First, we have compared Rayleigh scattering extinction in ZA relative to that in He (referred to as He/ZA in the following text). Second, we have used ZA spectra acquired at different known pressures (referred to as ZA/ZA in the following text). The second method is possible because the ACES cage system 15 maintains its optical alignment during pressure changes, and has the advantage of eliminating the need for He gas cylinders. To our knowledge, this method has not been reported previously. Empirical expressions for Rayleigh scattering cross section were determined using fits to Bodhaine et al. (1999) (Greenblatt et al., 1990), and HONO . Literature reference spectra were convolved with a Gaussian lineshape of full-width at half-maximum (FWHM) determined from Hg calibration lines, which was 0.82 nm for Ch 368 and 1.02 nm for Ch 455. The convolved literature reference spectra are shown in Fig. 2 for the relevant wavelength regions. Measured reference spectra were 10 determined from NO 2 and CHOCHO additions as described in Sect. 4.4, and these were used for the Ch 455 spectral retrievals.
The number density of trace gases was determined using least-squares, DOAS-style fit retrievals (Platt and Stutz, 2008) with DOASIS fitting software (Kraus, 2006). The fitted absorbers were HONO, NO 2 , and O 4 for Ch 368, and CHOCHO, CH 3 COCHO, 15 NO 2 , H 2 O and O 4 for Ch 455. In addition, a 3rd-or 4th-order polynomial was included in each fit to account for drift in the light intensity and cavity throughput of the measurement. Stretch and shift of the reference spectra wavelength were included as fit parameters to minimize discrepancies with the wavelength calibration of the grating spectrometer, which may be due to physical shifts in the spectrometer optics. For Ch 368, the Introduction

Measured cavity loss
The pressure response of the ACES instrument was tested by measuring N 2 extinction at sample pressures between 300-600 hPa in the laboratory, as shown in Fig. 3a. The measured optical extinction from Eq.
(1) at 455 nm is plotted against the number den-5 sity, N, determined from temperature and pressure. I 0 (λ) is defined to be the spectra acquired at 400 mb. Similarly, the slope at each wavelength element can be used to determine the Rayleigh scattering cross section of N 2 at that wavelength, and the results for 438-468 nm are shown in Fig  LED has a maximum near 368 nm, while the mirror reflectivity is optimal near 376 nm. Ch 455 has maximum LED intensity much closer to the best mirror reflectivity near 460 nm, and close to the maximum glyoxal absorption feature at 455 nm. Intensity in Fig. 5 is plotted as CCD detector counts for a 0.5 s exposure, and the actual photon count rate is 16× greater due to the detector gain. Rayleigh scattering loss is compa-5 rable to cavity loss for Ch 368, but is much smaller than cavity loss for Ch 455. The effective path length, neglecting Rayleigh scattering, is 3.0 km at 368 nm and 17.8 km at 455 nm. The lower panels show the minimum detectable extinction as a function of wavelength on each channel as a function of wavelength, calculated from the photon shot noise limit. See caption for further details. burning plume with large absorption by CHOCHO and HONO, as well as measurable CH 3 COCHO.

Precision and accuracy
The instrumental precision and stability can be determined from optical extinction during continuous zero air measurements. Figure 8 shows an Allan deviation plot (Allan, Introduction at constant pressure. The Allan deviation (2σ) follows the square root of the averaging time up to few hundred seconds. The Allan deviation in Ch 455 is roughly four times smaller than Ch 368 up to 1 min averaging time due to the longer effective light path length of Ch 455. The extinction values shown in Fig. 8 represent an upper limit for the precision of trace gas retrievals because they are calculated for single pixels, rather 5 than the range of pixels used in spectral fitting. Figure 9 shows spectral retrievals of NO 2 , CHOCHO, and HONO concentrations fit to zero air spectra acquired during ground measurements in the CARE Beijing-NCP 2014 study. The acquisition time for these data is 0.478 s point  The instrumental precision (2σ) for each trace absorber is estimated from two different methods: (1) standard deviation from dry zero air retrievals and (2) fit errors derived from the DOAS retrieval (Platt and Stutz, 2008). For both methods, the data sets for the Allan deviation analysis and regular zero air measurements during the field mission (30 s injection at 5 min intervals) were used, as described above. The estimated 15 precision (2σ) for each retrieved trace gas is summarized in Table 3. Precisions are adequate for ambient measurement of CHOCHO, HONO, and NO 2 , although the current precision for HONO is insufficient to quantify the small ambient mixing ratios reported during daytime or in low NO x (NO + NO 2 ) environments (e.g. Ren et al., 2010).  (Greenblatt et al., 1990). The esti-25 mated uncertainty for the Rayleigh scattering cross sections of zero air is 2 % and the uncertainty for He makes a negligible contribution (Washenfelder et al., 2008). The uncertainties in pressure and temperature measurements are 0.5 and 0.7 %. CHOCHO loss along the inlet line is negligible, as described in Sect. 2.4 (< 0.1 % for any sam- pling conditions used for aircraft-and ground-based sampling). HONO inlet artifacts are more difficult to characterize, since HONO may undergo loss due to adsorption or production due to heterogeneous reaction of NO 2 and H 2 O (Finlayson-Pitts et al., 2003). The latter is of particular concern in assessing the accuracy low-level HONO mixing ratios during daytime (Li et al., 2014). For this work, we have minimized HONO 5 sampling artifacts by using a short inlet with short residence time, but have not characterized the inlet behaviour under different atmospheric conditions. The propagated errors (summed in quadrature) are ±4.6/±5.0 % for NO 2 and ±5. respectively.

NO 2
Standard NO 2 concentrations were generated using a custom calibration system (Washenfelder et al., 2011b), which generates and measures O 3 using a commer- 15 cial ozone monitor (49i, Thermo Fisher Scientific Inc., Waltham, MA, USA) and subsequently titrates O 3 with excess NO (2040 ppm ± 2 %, NO in N 2 , Scott-Marrin Inc., Riverside, CA, USA) to produce a known concentration of NO 2 . During SENEX 2013, we used this system daily to generate standard NO 2 concentrations for validation, with typical NO 2 mixing ratios between 3-100 ppbv (parts per billion).
20 Figure 10a shows one set of standard additions from the NO 2 calibration system. The agreement between retrieved NO 2 from ACES and the NO 2 mixing ratio from the calibration system is shown in Fig. 10d. The relationship is linear over the entire range (r 2 = 1.000), with slopes (intercepts) of 1.031 ± 0.003 (0.16 ± 0.14 ppbv) and 1.015 ± 0.005 (0.23 ± 0.17 ppbv) for Ch 368 and Ch 455, respectively. The non-zero intercept 25 is due to small concentrations of NO 2 which are present in the NO titration gas of the calibration system (Washenfelder et al., 2011b). from the calibration source (3.1 and 1.5 %) are well within the absolute measurement accuracy of ±5.0 % for NO 2 given in Sect. 4.2. The NO 2 standard additions demonstrate the accuracy of the ACES instrument response. In addition, we performed standard additions of HONO and CHOCHO. For these additions, we compared the ACES measurement to the calculated flow dilution 5 because the HONO and CHOCHO concentrations were not independently determined. Figure 10b shows five additions from the CHOCHO source, in the range 2-8 ppbv. Each addition was allowed to stabilize for 1 min, and the subsequent 2 min of data are shown in bold. Figure 10e shows the average and standard deviation for these addi-10 tions. The correlation coefficient (r 2 = 0.999) indicates the linearity of the CHOCHO measurements from ACES and the CHOCHO calibration system. The retrieved fit errors from DOASIS were 30 pptv (parts per trillion; Table 3).

HONO
Constant HONO concentrations were generated using a calibration source based on 15 the design in Roberts et al. (2010). Briefly, humidified air mixed with HCl from a permeation tube passes through a sodium nitrite bed (NaNO 2 mixed with glass beads) to generate HONO via acid displacement. The entire system (i.e. permeation tube, water vessel and NaNO 2 bed) is temperature-controlled for stable HONO generation. The output from the HONO generator may contain small amounts of NO and NO 2 . 20 Figure 10c shows three additions from the HONO source with different flow dilutions, in the range of 4-7 ppbv. Each addition was allowed to stabilize for 1 min, and the subsequent 2 min of data are shown in bold. The observed delay in measurement response after a change in concentration is due to the addition source and not the inlet, which responded rapidly (< 5 s) during aircraft sampling. Figure 10f shows the average and standard deviation of these additions, with high linearity (r 2 = 0.999). The DOASIS fit errors for the 5 s spectra were 314 pptv (Table 3).

Aircraft measurements during SENEX 2013
The ACES instrument was deployed on the NOAA WP-3D research aircraft during the SENEX project from 1 June to 15 July 2013, with flights based out of the 5 Smyrna/Rutherford County Airport in Smyrna, Tennessee. In addition to the ACES instrument, many instruments to characterize gas-and aerosol-phase species were deployed (see partial list in de Gouw et al., 2015), including a CRDS instrument to measure NO, NO 2 , NO 3 , N 2 O 5 , and O 3 (Wagner et al., 2011). One goal of the field study was to understand the interactions between natural and anthropogenic emis-10 sions in the southeastern US and to evaluate their impact on air quality and climate. Figure 11a shows CHOCHO measurements acquired on 5 July 2013 during a flight that sampled an isoprene hot spot (Ozark Mountains, MO), an urban area (St. Louis, MO), and an ethanol refinery (Decatur, IL). CHOCHO mixing ratios greater than 140 pptv were observed over St. Louis, with NO 2 mixing ratios greater than 14 ppbv. 15 Outside of the urban area, measured CHOCHO concentrations were 53-79 pptv (interquartile range). The time series of CHOCHO is shown in Fig. 11c.
Time series data for NO 2 measured by ACES and CRDS are shown in Fig. 11b, with the CRDS data offset by 1 ppbv for clarity. The CRDS instrument reported NO 2 concentrations at 1 s time resolution with precision and accuracy of 200 pptv and 5 %, 20 respectively. A scatter plot comparing the ACES Ch 455 and CRDS NO 2 data in shown in Fig. 11d, with the data averaged to 10 s. The instruments agree well, with slope of 0.983 ± 0.013 with r 2 = 0.937, which is consistent with the uncertainty of the measurements.
CH 3 COCHO has been retrieved only for pyrogenic plumes during SENEX 2013, Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | than CHOCHO (see Fig. 2). As shown in Fig. 7, biomass burning plumes can represent large sources of CH 3 COCHO, with measured mixing ratios up to 5.7 ppbv.

Ground-based measurements during CARE Beijing-NCP 2014
The ACES instrument was subsequently deployed during the CARE Beijing 2014 study from 2 June-6 July 2014. Figure 12a shows a map of the North China Plain field site, 5 which was located 198 km southwest of Beijing in Wangdu, Hebei Province. The overall goals of the field study focused on HO x radical chemistry, new particle formation, and the impacts of air pollution. The scientific goals for the ACES instrument included examining the importance of anthropogenic precursors in CHOCHO formation and characterizing the importance of CHOCHO and HONO as radical sources. Figure 12 shows time series for NO 2 , HONO, CHOCHO, and H 2 O acquired during six days in June 2014. Peak NO 2 and HONO concentrations were 65 and 3.5 ppbv at night, with peak concentrations of 240 pptv CHOCHO during the day. NO 2 and HONO mixing ratios were low during daytime and higher during nighttime, consistent with accumulation of NO x emissions in a shallow nocturnal boundary layer and heterogeneous 15 conversion of NO 2 to HONO on the ground surface. In contrast, CHOCHO concentrations had a morning maximum, which is not typical of its diurnal pattern in other locations (Volkamer et al., 2005a;Sinreich et al., 2007;Huisman et al., 2011;Washenfelder et al., 2011a). Previous studies have shown that CHOCHO concentrations track the OH oxidation of precursors, with a maximum in the afternoon, similar to other photochemi-20 cally produced species. The unusual CHOCHO diurnal profile at this site suggests an interaction between emission, chemistry, and transport at the NCP site, such as a mixing of residual layer air masses with higher CHOCHO concentrations into the nighttime boundary layer during morning.
A number of other instruments were deployed, including a commercial instrument 25 to measure water (G2301, Picarro Inc., Santa Clara, CA, USA). Figure 12 g  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | ment between the two instruments is better than 5 % (slope = 0.957±0.008, r 2 = 0.927) and demonstrates that ACES accurately retrieves H 2 O, despite its weak cross section (2.5 × 10 −26 cm 2 at the instrument resolution, see Fig. 2).

Conclusions
The ACES instrument is engineered to provide robust, highly sensitive, and fast 5 time-response measurements under aircraft sampling conditions. The design includes a temperature-controlled, custom-designed optical system with electronic LED on-off modulation and low-noise CCD cooled to −50 • C. The instrument is also appropriate for field campaigns on other platforms, as well as for laboratory studies of trace gases and aerosols.

10
At 438-468 nm, the demonstrated precision (2σ) for field measurements of CHO-CHO and NO 2 is 34 and 80 pptv in 5 s, respectively. The calculated accuracies for these measurements are 5.8 and 5.0 %, which is consistent with standard additions and comparison to an independent NO 2 measurement.