Characterization and first results of an ice nucleating particle measurement system based on counterflow virtual impactor technique

Leibniz Institute for Tropospheric Research, Leipzig, Germany Institute for Atmospheric and Environmental Sciences, Goethe University of Frankfurt am Main, Frankfurt am Main, Germany Max Planck Institute for Chemistry, Particle Chemistry Department, Mainz, Germany Technical University of Darmstadt, Institute of Applied Geosciences, Darmstadt, Germany Laboratory of Atmospheric Chemistry, Paul Scherrer Institute (PSI), Villigen, Switzerland now at: Max Planck Institute for Chemistry, Particle Chemistry Department, Mainz, Germany now at: Julich Research Center, Institute of Energy and Climate Research, Troposphere (IEK-8), Jülich, Germany


Introduction
Ice crystals in clouds influence precipitation and the microphysical and thus the radiative properties of clouds. They play an important role for the cloud lifetime, the interactions of clouds with solar radiation, cloud electricity and cloud dynamics. Mixed- 25 phase clouds consisting of supercooled droplets and ice particles exist at altitudes 10587 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | where temperatures warmer than −38 • C occur. In such clouds, ice can only be formed heterogeneously, due to the presence of so-called ice nucleating particles (INP) (Vali et al., 2014;DeMott et al., 2011;Pruppacher and Klett, 1997). Known pathways of heterogeneous ice formation are deposition nucleation and condensation, immersion and contact freezing. Field data and modeling studies indicate for many situations that liq-5 uid water droplets are a prerequisite for ice formation (Murray et al., 2012;Westbrook and Illingworth, 2011;de Boer et al., 2011;Ansmann et al., 2009). This suggests an important role of immersion and contact freezing processes in atmospheric ice formation. But also sub-saturated nucleation processes (Sassen and Khvorostyanov, 2008) (e.g. deposition nucleation) were found to be an active freezing mechanism under at-10 mospheric conditions. However, the relative importance of the different pathways is not well known at present.
Many species have been identified to act as INP, e.g. mineral dust (e.g., illite, montmorillonite, kaolinite, feldspar), primary biological particles (PBAP), soot, and glassy organics. In laboratory experiments many of these models substances were investi- 15 gated and their ice forming efficiencies (nucleation temperatures and rates) have been quantified and parameterized (e.g. Murray et al., 2010;Hoose and Moehler, 2012;Atkinson et al., 2013;Wex et al., 2014). However, the type of ambient aerosol particles acting as INP, the importance of their size and the influence of anthropogenically emitted aerosol particles are not well understood. There is still a lack of field measurements 20 concerning the in-situ physio-chemical characterization of atmospheric INP.
In 2011 the Ice Nuclei Research Unit (INUIT) was established with the objective to achieve a more detailed understanding of heterogeneous ice forming processes. The central objective of INUIT is to obtain a better knowledge of ambient aerosol particles serving as INP. Therefore, an ice nucleus counter was connected by means of the counterflow virtual impactor (CVI) technique to online mass spectrometry, and other aerosol measurement techniques, similar to the coupling that had been originally presented by Cziczo et al. (2003) and some years later by Corbin et al. (2012). Corbin et al. (2012) presented measurements from a field study in downtown Toronto (SPORT 2011), where they sampled 196 particles that nucleated ice in a continuous flow diffusion chamber (CFDC) by a pumped CVI (PCVI), and analyzed them with the single particle mass spectrometer ATOFMS. Besides dust, also biomass burning and elemental carbon were identified in these particles. However, statistical limitations hampered the data interpretation. In a similar experimental approach a PCVI was combined 5 with a condensation nucleus counter (CCNC) to investigate cloud droplet residues (Hiranuma et al., 2011;Slowik et al., 2011).
The main components of the present study are the INP counter FINCH (Fast Ice Nucleus Chamber; Bundke et al., 2008Bundke et al., , 2010, the IN-PCVI (a pumped CVI (BMI, Model 8100, 2011) operated at new flow settings), and the single particle mass spectrometer 10 ALABAMA (Aircraft-based Laser Ablation Aerosol Mass Spectrometer; Brands et al., 2011). This combination of INP activation, ice particle selection, and INP characterization was first established and technically improved under laboratory conditions. Afterwards the combination was deployed for atmospheric measurements at the Jungfraujoch (JFJ) research station to sample and characterize ambient ice nucleating particles 15 at cloud level.

Experimental setup -methods
A schematic of the measurement setup which was realized and operated within INUIT (research project RP 2) is shown in Fig. 1

Pumped counterflow virtual impactor -PCVI
The CVI used in this study is a pumped CVI (PCVI) from Brechtel Manufacturing In-corporated (Model 8100, 2011). A first characterization was described in Boulter et al. (2006) and another one was presented by Kulkarni et al. (2011). The CVI technique is used to separate large particles from smaller ones and from the gas that originally 5 surrounded the large particles. This is realized by a counterflow that is directed against the IN-PCVI input flow inside the IN-PCVI. Due to their larger inertia larger particles overcome the counterflow and are thus sampled by the IN-PCVI. In the present setup ( Fig. 1) large ice crystals are separated from smaller supercooled droplets and intersti-tial particles. The IN-PCVI input flow corresponds to the FINCH output flow (illustrated 10 in Fig. 1). The flow rates of the IN-PCVI input flow (F IF ), counterflow (F CF ) and sample flow (F SF ) determine the aerodynamic cut-off diameter (D p50 % ) of the IN-PCVI. This cut-off diameter represents the particle diameter at which particles are sampled with 50 % efficiency. Particles smaller than the D p50 % are sampled with less than 50 % ef-ficiency. Larger ones are collected with higher efficiency. F IF and F CF are adjusted by 15 the pumped flow (F PF ) and the add flow (F AF ) (Eqs. 1 and 2). The ice particles that penetrate the counterflow are then transferred into the IN-PCVI sample flow where the water is evaporated in particle-free and dry carrier air.
Due to the position of the IN-PCVI between FINCH and the aerosol analysis instrumen-tation ( Fig. 1), the IN-PCVI needs to be operated with different flow settings than the ones used previously (Boulter et al., 2006;Kulkarni et al., 2011). Here the PCVI is oper-ated with a sample flow slightly larger than 1 L min −1 and with an input flow in the range of 5 L min −1 requiring the determination of the cut-off-characteristics for these specific flow rates. For this purpose a calibration of the D p50 % was done with the setup shown in formula: (C 10 H 10 · C 8 H 8 )x, Duke Scientific Corporation, now Thermo Scientific) with diameters in a range of 2 to 120 µm (aerodynamic size: Fig. 3) were used as well-defined reference aerosol particles. The polystyrene DVB particles were dispersed manually by a powder dispenser (pushing the balloon which is shown in Fig. 2) and pumped into the IN-PCVI at a prescribed input flow rate of 4, 4.5, or 5 L min −1 . For several combinations 5 of input flow rate, counterflow rate (0, 2, 2.5, 3, 3.5 L min −1 ) and sample flow rate (1, 1.3, 2, and 3 L min −1 ), at which the IN-PCVI was operated (Table 1), the number size distribution of the test aerosol was measured with an APS. The number size distribution measured with the counterflow being switched off was taken as reference. It was normalized to a maximum of 1 and all cut-off size distributions (i.e., F CF switched on) 10 measured for the same inlet and sample flow were scaled to match the right branch of the reference size distribution. This approach was chosen to minimize the error due to size-dependent particle losses in the inlet region of the IN-PCVI. The calibration runs were repeated several times for averaging the aerodynamic size distribution and thus to minimize the error due to the manual dispersion of the calibration aerosol. 15 The calibration curves of the most common sample and input flow settings of the INUIT-JFJ/CLACE 2013 campaign (Run 4-7, Table 1) are illustrated in Figs. 3 and 4. Figure 3 shows the normalized APS measured aerodynamic size distribution, here for a fixed F IF and F SF , while F CF is varied. As expected, for larger counterflows the cut-off is shifted to larger sizes. The transmission efficiency (TE, sampling efficiency of particles 20 of a certain size passing the IN-PCVI) is defined by the ratio of the normalized sample flow concentration N out with counterflow switched on and the normalized reference concentration N in (F CF = 0) (Eq. 3).
(3) Figure 4 shows the transmission efficiency curves for the measurements shown in Fig. 3. The vertical dashed lines mark the resulting cut-off-diameter, D p50 % . It can be seen that with increasing F CF from 0.5 to 3.5 L min −1 the D p50 % increases from 5 to 10591 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 6.5 µm. The behavior of the D p50 % was also studied for different input and sample flows. The resulting D p50 % for all flow variations were in a range between 5.2 and 8.4 µm (Fig. 5). It is evident that a change of the D p50 % is most sensitiv to a change of F IF compared to a change of F SF or F CF (slopes of regressions for the chosen flow regime are shown in Fig. 5a-c). In Fig. 5a, D p50 % results determined by Kulkarni et al. (2011) 5 are displayed for comparison. Since F IF in our experiments is lower ( , the absolute D p50 % values are higher. Nevertheless, the same relative change in D p50 % (same slope) is found for the two F IF settings. The cut-off diameters for the new flow parameters (Table 1)

Fast Ice Nucleus Chamber (FINCH)
FINCH was developed and built at the Goethe University of Frankfurt. It is an in situ counter for ice nucleating particles, which is operated by mixing air flows with different 15 humidity and temperature to achieve supersaturation (with respect to water (S wat F ) and with respect to ice (S ice F )) (Bundke et al., 2008). The main component of the device is a temperature-controlled, stainless steel growth chamber with a length of 0.8 m that can be cooled down to −65 • C. Sampled aerosol particles are activated and grow to macroscopic ice particles or supercooled droplets depending on the temperature (T F ) 20 and supersaturation (S ice F ) inside the chamber. Depending on the mixed air flow rate the aerosol particles typically need ∼ 10 s to pass the chamber. The (grown) particles are counted in an optical particle counter mounted directly below the chamber (BIO-IN-OPC; Bundke et al., 2010). Based on scattering properties a distinction between water droplets and ice particles is possi-25 ble (P44/P11 ratio of the scattering matrix; Hu et al., 2003). The current setup of the detector consists of a 405 nm laser, which stimulates the particles to fluoresce as well. In this study the outlet of the BIO-IN-OPC was directly connected to the IN-PCVI.
After several hours of operation the measurements with FINCH are interrupted for a short heating period to melt the ice that had built up in the inlet and thus resulted in a reduced or plugged flow. After every heating period the saturation ratio inside 5 FINCH is automatically re-established to the set value, which can take up to an hour. During operation S ice F is adjusted continuously as a consequence of changing ambient temperature and relative humidity.

Aircraft-based laser ablation aerosol mass spectrometer (ALABAMA)
ALABAMA is a single particle laser ablation instrument to detect the chemical compo-10 sition, mixing state and size of aerosol particles in a range from 150-900 nm (Brands et al., 2011). It was built and characterized at the Max Planck Institute for Chemistry (MPIC) Mainz. The particles are detected and sized by two continuous lasers at a wavelength of 405 nm (InGaN Blu-Ray laser). Thereafter, a 266 nm wavelength pulsed Nd-YAG-laser is used to evaporate the aerosol particles and ionize the compo-

Microphysical aerosol instrumentation
For a microphysical characterization of the INP a condensation particle counter (CPC TSI 3010) and an aerodynamic particle sizer (APS TSI 3321) were connected downstream the IN-PCVI. The CPC measures the number concentration of particles that are larger than 0.01 µm in diameter. The APS measures the aerodynamic number size distribution of aerosol particles in the size range between 0.5 and 20 µm. The sizing is done by a time-of-flight measurement between two laser beams. The IN-PCVI in combination with CPC and APS are called IN-PCVI system in the following. 5 Ice-nucleating particles were collected downstream the IN-PCVI by a two-stage impactor system (50 % cut-off aerodynamic diameters 1.0 and 0.1 µm, respectively) and were analyzed offline by scanning electron microscopy and energy-dispersive X-ray fluorescence (EDX). Particles were collected on transmission electron microscopy grids and elemental boron. The samples were analyzed in a scanning electron micro-10 scope (FEI Quanta 200 FEG, FEI, Eindhoven, the Netherlands) with attached energydispersive X-ray fluorescence microanalysis (EDX, EDAX, Tilburg, the Netherlands) to characterize the particles with regard to their chemical composition, morphology, size, internal mixing state and electron beam stability (volatility). Based on criteria of chemical composition, morphology mixing state and beam stability, particles were sorted 15 into ten singular classes (carbonaceous, sulfate and other secondary aerosol, soot, sea-salt, Ca-rich, metal oxide, silicate, Pb-bearing, droplets, and other particles). For details on the particle classes, the reader is referred to Worringen et al. (2014). Due to their low abundance, in the present work sea-salt and Pb-bearing particles are subsummarized with all other unclassified particles into the so-called "other" class.

Test measurements in the laboratory
Prior to field operation, the volume flow management of the combination of FINCH, the IN-PCVI and several aerosol particle instruments was set up and tested during two technical laboratory campaigns in Frankfurt and Leipzig. Since each component has its own standalone flow system, the main concern was to adapt and couple the flows 25 and to ensure a smooth and efficient transition of the sample from one component to the following one. As a main result of these tests, the FINCH closed loop had to be switched off and the FINCH aerosol flow had to be directly controlled by the IN-PCVI pump flow. This led to stable pressure conditions inside the coupled FINCH + IN-PCVI flow system. Additionally, it was found that changes in the IN-PCVI -FINCH flows caused fluctuations in the supersaturation of FINCH, which sometimes resulted 5 in a frozen inlet and thus blocking of the growth chamber. Therefore, the FINCH -IN-PCVI flows were was kept constant during single measurement runs.

Atmospheric measurements at the high Alpine research station Jungfraujoch
The combination of INP activation and detection (FINCH), separation/preparation particles inside mixed-phase or, sometimes, even entirely glaciated clouds, where supercooled drops and small ice particles were evaporated during collection by heated total aerosol inlet (operated by Paul Scherrer Institute, Villigen; Weingartner et al., 1999). Also ambient background aerosol particles, which are present during cloud free time periods, are sampled at these altitude that are realistic for the mixed-phase cloud for-20 mation at mid latitudes.

10595
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Proof of principle for the FINCH + IN-PCVI coupling
The main focus of this work was to verify the feasibility of the FINCH + IN-PCVI coupling for the physical and chemical characterization of atmospheric INP by different aerosol sensors attached to the IN-PCVI. 5 A stability criterion of the adjusted ice saturation ratio inside FINCH needed to be specified to ensure the data analysis under constant measurement conditions. Fluctuations in the FINCH saturation ratio may lead to different conditions under which freezing occurs and thus to differences in the final sizes to which the ice particles grow. Therefore, only measurement periods in which the ice saturation ratio varied by 10 less than 1 % (relatively) within 300 s were used for further analysis. This criterion is somewhat arbitrary, but appeared to effectively eliminate the poorly defined activation periods while leaving sufficient amount of data points for the proof-of-principle study.
During the INUIT-JFJ/CLACE-2013 campaign the complete coupled system of FINCH, IN-PCVI and aerosol instruments was connected to the heated total aerosol 15 inlet sampling line to investigate the ice activation ability of ambient aerosol particles and during cloud periods also of cloud ice particle residuals and cloud droplet residuals. For this purpose FINCH was operated in the temperature range between −18 and −26 • C and at ice saturation ratios between 1.05 and 1.5. Figure 6 shows the T F − S ice F pairs (10 min averages) from the campaign that remain after applying the stability cri-20 terion. Moreover, the corresponding lines of the calculated water saturation ratio (0.90 to 1.25 in 0.05 steps) are indicated. The majority of the 10 min periods used in the following discussion are located in the water supersaturated region (S wat F > 1.0) when immersion freezing is the dominant heterogeneous ice nucleation mechanism inside FINCH. Only a few data points are below S wat F = 1.0 when ice activation in FINCH is limited to deposition nucleation and immersion freezing in highly concentrated solution droplets (Wex et al., 2014) only. Figure 7 shows a log-log scatterplot of the number concentration of ice nucleating particles measured by the FINCH optics (N INP FINCH ) and measured by the IN-PCVI CPC (N INP IN-PCVI , same 10 min averages). The data points are subdivided into 3 classes with regard to the prevailing FINCH water saturation ratio: S wat F < 1.0 (squares), 1.0 < S wat F < 1.1 (circles) and 1.1 < S wat F < 1.2 (crosses).
The 1 : 1 line indicated in Fig. 7 shows that there exists a very close correlation be-5 tween N INP FINCH and N INP IN-PCVI for S wat F < 1.0 (squares), which implies that the residue of each ice particle grown in FINCH is transferred with high efficiency through the IN-PCVI and all other particles exiting FINCH (supercooled droplets, unactivated particles) are diverted into the F PF of the IN-PCVI. With increasing water saturation ratio, however, N INP IN-PCVI concentration exceeds N INP FINCH . Since it is not likely that more ice nucleating 10 particles exist after the IN-PCVI than there were ice particles counted after FINCH, the most probable explanation is that some liquid supercooled droplets that are formed in FINCH reach the same size as the ice particles at higher saturation. Thus, they cannot be pre-segregated by the IN-PCVI and become erroneously counted and sampled as INP. In order to estimate the magnitude of this effect, power regressions (with an 15 exponent equal 1) were calculated for each class and plotted in Fig. 7. These indicate contribution of large droplet residuals 0, 45, and 63 % for the intended INP sampling with respect to the three S wat F classes. Table 2 gives a more detailed verification of the droplet contamination effect. Again power regressions are derived from the N INP FINCH to N INP IN-PCVI relationship but now sep-20 arately for a S wat F increment of 0.01. The desired regression close to 1 is only achieved for S wat F ≤ 1.01, which denotes an ice saturation between 1.21 and 1.3 for the adjusted temperatures. However, the uncertainty of the regression slope includes 1 still for a S wat F ≤ 1.06 (1.21 < S ice F < 1.34). Accepting a ratio of one drop residual out of three INP IN-PCVI would allow to take into account all measurements up to S wat F ≤ 1.08 25 (1.21 < S ice F < 1.34). For all INP results obtained by the coupled system these limitations and uncertainties have to be taken into account.

10597
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

First measurements of INP properties
Having proven the feasibility and discussed the operational limitations of the FINCH + IN-PCVI coupling, now first exemplary results of INP properties obtained with this experimental setup during the INUIT-JFJ/CLACE-2013 campaign are presented. Identical to the experimental setup described in Sect. 3.1 ambient particles as well as residues 5 of supercooled drops or ice particles in the presence of clouds were sampled by the heated total aerosol inlet, to which the combination of FINCH, IN-PCVI, and aerosol sensors was connected to. During all measurements at the JFJ the IN-PCVI cut-off diameter was adjusted in the characterized range, which is shown in Fig. 5a-c. Figure 8 shows an INP time series from a 4 h case study on 9 February 2013. The data points 10 are again 10 min mean values and meet the above described criterion of the FINCH saturation ratio varying less than 1 % in 300 s. Within this period T F was adjusted in a range from −21 to −23.5 • C and S ice F to ∼ 1.1, respectively. Thus, FINCH was operated at water sub-saturated conditions, i.e. deposition nucleation and/or immersion freezing of highly concentrated solutions are the prevailing heterogeneous nucleation For the examined time period the ambient background particle concentration was roughly between 1100-2900 cm −3 (measured also at the heated inlet sample line by a PSI operated CPC, TSI 3772), which leads to an ambient INP fraction of approx. 5 2 × 10 −5 . Over the whole campaign the INP fraction was found to be between 1 × 10 −6 and 4 × 10 −5 (for ice saturation ratios between 1.05 and 1.35 and FINCH temperature between −18 and −25 • C). This agrees well to other atmospheric INP number concentration studies (Pruppacher and Klett, 1997;Mason, 1971;Rogers et al., 1998;Eidhammer et al., 2009;DeMott et al., 2010). Nevertheless, the INP concentration is most of high and remote location of the measurement site with low ambient aerosol particle number concentration. Figure 9 presents a frequency distribution of INP sizes measured by the APS (dashed line) downstream of the IN-PCVI. Only APS measurements are included which are carried out at S wat F ≤ 1.01 in order to be sure that only INP are sampled with-20 out contamination by large drop residuals. The frequency distribution decreases with a shallow slope towards super-micrometer diameters, which confirms previous measurements showing that large aerosol particles are more effective INP (Mertes et al., 2007). This finding is supported by the comparison with a frequency distribution of ambient aerosol particle sizes (solid line in Fig. 9), which decreases more steeply with 25 increasing particle size. The latter distribution was measured with an Optical Particle Sizer (OPS, TSI 3330) and normalized to the APS distribution at 0.55 µm.
The combination of FINCH + IN-PCVI with the single particle mass spectrometer ALABAMA for online chemical analysis of INP was one of the main objectives of this 10599 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | study. However, it turned out to be the most ambitious goal due to the limited measurement time at low and constant water saturation conditions in FINCH, and due to an insufficient particle detection efficiency of ALABAMA. Only seven single particle mass spectra of INP were recorded with ALABAMA during the INUIT-JFJ/CLACE-2013 campaign, which makes a statistical or cloud event differentiated analysis impossible. The 5 few ALABAMA INP mass spectra show mainly organic material but also other elements as Li, Na, Al, Fe, and K, that might be related to biological particles, biomass burning, or mineral dust particles coated with organic matter as the main INP components (see example in Fig. 10). These compositions are similar to those detected by ALABAMA in ice particle residuals during the same project (Schmidt et al., 2014).
Furthermore, impactor samples were taken downstream of the IN-PCVI and analyzed offline by electron microscopy. A detailed discussion of the SEM results is presented by Worringen et al. (2014).

Summary and conclusions
The combination of an ice nucleus counter (FINCH), a counterflow virtual impactor (IN-15 PCVI), and different instruments for physical and chemical aerosol particle characterization was set up, tested, characterized, and deployed for the first time to investigate the properties of ice nucleating particles under free tropospheric conditions. This included operation of a single particle mass spectrometer, ALABAMA, for chemical INP analysis.

20
The cut-off behavior of the IN-PCVI was successfully determined for the specific flow settings required for the instrument combination. Feasibility and functionality of the instrument combination were proven in the framework of the INUIT-JFJ/CLACE 2013 field measurements. It was found that an important prerequisite for a successful coupling of FINCH and the IN-PCVI are stable ice saturation ratios inside the IN counter. A relative change of less than 1 % in S ice F within 300 s was therefore chosen as a criterion for the data analysis. INP FINCH and INP IN-PCVI number concentration measurements showed an excellent agreement for water saturation ratios ≤ 1.01 (1.21 < S ice F < 1.3). This implies that ice particles nucleated in FINCH are efficiently separated by the IN-PCVI from the interstitial particles and supercooled droplets that are also present at the FINCH outlet, and that consequently only INP are selected by the IN-PCVI for further analysis. At higher water saturation ratios higher INP concentrations were measured 5 with the IN-PCVI CPC than with the FINCH optics, which may be attributed to supercooled drops growing into the same size range as the ice particles and can therefore not be rejected by the counterflow. This effect is estimated to result in an overestimation of INP concentrations by about 25 % for FINCH water saturation ratios up to 1.08 (1.21 < S ice F < 1.34).
Ambient INP properties were inferred from the measurements carried out during the INUIT-JFJ/CLACE 2013 field campaign. Restricting the considered measurements to water saturation ratios ≤ 1.0, i.e., considering deposition nucleation and/or immersion freezing of highly concentrated solutions, INP concentrations of up to 40 L −1 were measured (S ice F between 1.0 and 1.3, temperatures between −15 and −28 • C) over 15 the campaign duration. In a 4 h case study 87 % of the INP were found to be smaller than 0.55 µm. From the shape of the frequency distribution of INP diameters it can be concluded that the ice nucleating efficiency increases with particle size. The combination of FINCH + IN-PCVI + ALABAMA resulted in 7 mass spectra of INP, which cannot be statistically evaluated, but represent a successful proof-of-concept for 20 the system. A clear need to improve the counting efficiency of the FINCH + IN-PCVI + ALABAMA combination has to be stated when deploying the system in environments with low INP concentrations. However, for laboratory applications and use in environments with high INP concentrations, the FINCH + IN-PCVI + ALABAMA combination can already by applied in the current development stage. 25 ner, U., Frank, F., Nillius, B., Bundke, U., Rose, D., Curtius, J., Kupiszewski, P., Weingartner, E., Schneider, J., Schmidt, S., and Ebert, M.: Single-particle characterization of icenucleating particles and ice particle residuals sampled by three different techniques, Atmos. Chem. Phys. Discuss., 14, 23027-23073, doi:10.5194/acpd-14-23027-2014 Figure 6. Ice activation conditions in FINCH cover a range of combinations of temperatures (T F ) and ice saturation ratios (S ice F ). The majority of measurements was conducted at conditions supersaturated with respect to water (lines of constant S wat F are plotted in grey (0.05 steps)).

10613
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |   Figure 7. Scatter plot of the INP 10 min averaged number concentrations measured by the FINCH optics and the CPC of the IN-PCVI system, respectively. The concentrations are subdivided into three different water saturation ranges: S wat F < 1.0 (squares), 1.0 < S wat F < 1.1 (circles) and 1.1 < S wat F < 1.2 (crosses). The power regression curves (exponent equal 1) for these three classes are plotted as lines: S wat F < 1.0 (dotted line), 1.0 < S wat F < 1.1 (dashed line) and 1.1 < S wat F < 1.2 (solid line). For every number concentration range an example of the standard deviation is indicated by error bars.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |  . Examples for INP mass spectra measured online with the single particle mass spectrometer ALABAMA behind the FINCH + IN-PCVI. Both particle spectra are dominated by organic ions, but elements like Li and Al might also indicate that these particles may be mineral dust coated with organic matter.