The CIC measures the number concentrations of positive and negative air ions simultaneously. It comprises two independent cylindrical differential mobility analyzers, each with three collecting electrodes (Figs.  and ). The ions in the air entering the analyzers are repelled by a central electrode, which is held at a fixed potential between -40 and 40 V. A switchable high-voltage electrical inlet filter is used to prevent ions from entering the analyzer during zero-point measurements. Charged particles deposit on the outer wall of the analyzer, which is divided into three independent collecting electrodes. The current flowing through each electrode is measured with a high-precision integrating electrometer that samples and digitizes the current at up to 30 Hz. The inlet of the analyzer has been designed to minimize ion losses from diffusion and stray electric fields. A flow meter and blower at the exit of each analyzer control the flow rate of the air drawn through the instrument.

The geometry of the analyzer has been chosen so that, for 10 L min⁻¹ flow rate and 6.5 V voltage on the central electrode, the limiting mobilities of the three collecting sections are 2.5, 0.5, and 0.25 cm² V⁻¹ s⁻¹. As can be seen from numerical modeling results performed using COMSOL Multiphysics^® (Fig. ), ideally, all ions with a mobility of z > 2.5 cm² V⁻¹ s⁻¹ (dp < 0.9 nm) are deposited onto the first electrode, those with a mobility of z > 0.5 cm² V⁻¹ s⁻¹ (dp < 2.0 nm) are deposited onto the first and second electrodes, and those with a mobility of z > 0.25 cm² V⁻¹ s⁻¹ (dp < 2.8 nm) are deposited onto the first, second, and third electrodes.

Figure  shows the theoretical transfer functions for the CIC analyzer based on a simple general model of a first-order differential mobility analyzer. This does not take into account diffusional and electrical losses, diffusional dispersion, and airflow characteristics caused by specific inlet configurations or mechanical imperfections. The data analysis uses experimentally determined transfer functions, which are broader and flatter than those predicted by the simple theoretical model. The electrodes do not have a sharp lower cut-off limiting mobility. Due to the absence of a sheath airflow, electrodes will always detect a small fraction of larger particles with low mobility that enter the analyzer close to the collecting electrode. The limiting mobilities are primarily determined by the geometry of the analyzer and the ratio between airflow rate and central electrode voltage. The airflow rate through the CIC can be set by software to values between 5 and 50 L min⁻¹ per analyzer. The central electrode voltages are automatically adjusted so that the ratio (and, consequently, the limiting mobilities of the collecting electrodes) remains constant. The first two channels of the CIC are designed to capture overlapping portions of the complete typical atmospheric cluster ion mobility band. This allows for a rough estimate of the average cluster ion mobility based on the balance of the signal between the two channels. The third channel is designed to capture only larger ions beyond the cluster band, i.e., above 2 nm mobility diameter. This allows the CIC to measure the cluster ion concentrations with a sharp cut-off mobility range between 0.5 and 0.25 cm² V⁻¹ s⁻¹ by estimating the signal contributed by larger particles using the signal from the third electrode and subtracting this from the summed signal of the first two electrodes.

The measurement of small currents (typically 1–10 fA) generated by air ions is demanding and causes many problems related to changes in temperature, relative humidity, turbulence, and electromagnetic noise in ambient conditions. The design of the electrometers and the data acquisition electronics, as well as the signal processing algorithms, of the CIC are identical to those of the Neutral cluster and Air Ion Spectrometer (NAIS; ). The design has proven its performance, robustness, and reliability in many instruments operating in a wide range of demanding environmental conditions, some continuously for over 2 decades.

Each collecting electrode of the CIC is connected to its own dedicated, low-noise, integrating electrometer. The electric current produced by the depositing ions is accumulated on a capacitor (typically 33 pF) in an integrator circuit using a low-input-current operational amplifier (LMC6042). The output voltage of the electrometer, which is proportional to the accumulated charge, is periodically measured by a 24-bit analog-to-digital converter. The average electric current is calculated from the rate of change of the voltage. Since ion current collection is continuous and independent of the analog-to-digital converter's sampling frequency, the integration time can be optimized based on the desired signal-to-noise ratio and time resolution. To prevent saturation, the capacitor is automatically discharged and reset to zero volts when the electrometer output voltage reaches a threshold, typically ±1 V. This reset causes a brief 5 to 10 s gap in the measurements, occurring usually once every hour to once every day. The low ion losses and high measurement rate of the electrometers allow the instrument to achieve a very fast time resolution, above 1 Hz.

The CIC employs a switchable high-voltage electrical inlet filter to block ions from entering the analyzer during calibrations of zero-level currents and noise levels of the electrometers. While the data acquisition electronics are designed for minimal temperature sensitivity, some thermal drift is unavoidable and needs to be corrected. An additional and significant factor influencing the zero-level currents, especially during long-term measurements, is the gradual contamination of the collecting electrodes and insulators. This contamination can lead to leakage currents, which can be exacerbated by high relative humidity. Performing frequent zero measurements compensates for these combined drift and leakage currents. This ensures the reliability of the measurements, extends the intervals between necessary physical maintenance (cleaning of the analyzer), and provides a clear diagnostic indicator when such maintenance is required. This periodic zero-level calibration, or so-called “offset” measurement, is carried out automatically by the device. The user-selectable settings are typically 1–5 min of signal measurements followed by 20–30 s of offset measurements.

The inlet filter is a brass sphere of 15 mm diameter, supported by two small, diametrically placed spokes and set to a voltage of around -300 or +300 V for the positive or negative channel, respectively. Although the spokes and the sphere itself inevitably result in flow separation and circulation behind the sphere, the computational fluid dynamics (CFD) model suggests that the effect is localized and does not extend downstream into the mobility analysis region, where the flow remains laminar. Alternative inlet filter designs, such as a mesh with an applied voltage, were evaluated, but the spherical inlet filter was selected because it demonstrated the lowest particle losses.

The CIC uses internal Venturi flow meters and blowers independently for both analyzers in a software-controlled feedback loop to maintain constant sample airflow rates. Temperature, relative humidity, and absolute air pressure are also monitored for both airflows. The central electrode voltages are supplied by a bipolar software-controlled voltage source, which allows each analyzer to operate at a freely chosen voltage and polarity. To confirm correct operation of the device, all relevant parameters from the different subsystems are monitored and logged alongside the measurement data. A graphical measurement software (Spectops) is provided for monitoring, controlling, and storing measurement and diagnostic data from the CIC. A quick-view program (Retrospect) can be used to review the stored data files.

Compared with earlier instruments developed at the University of Tartu, such as the NAIS or the Balanced Scanning Mobility Analyzer (BSMA; ), the CIC features a simpler design. While it does not match the sensitivity of the BSMA or provide the detailed sizing information of the NAIS, the CIC offers several advantages over these more complex instruments. Notably, the CIC can operate at low sample flow rates that are unachievable by the BSMA. Compared with the NAIS, the CIC has a higher detection efficiency and a shorter inlet tract, allowing for more reliable measurement of small ions at a higher time resolution and with a lower mobility diameter cut-off.

The external dimensions of the CIC are 200 mm width, 130 mm height, and 400 mm length. The weight of the instrument is 5 kg. The power consumption is 4–5 W at 30 L min⁻¹ sample flow rate per analyzer.

Experimental evaluations of the CIC have been carried out in several aerosol laboratories: at the Institute for Atmospheric and Earth System Research (INAR), University of Helsinki; at the CERN CLOUD chamber, Switzerland; at the Laboratory of Atmospheric and Environmental Sciences, University of Tartu; and at Airel OÜ, Tartu.

At INAR, the University of Tartu, and Airel, the calibration aerosol was generated and size-selected in fixed, ultra-dry conditions. Maintaining a dry environment during calibration experiments is crucial for stable aerosol generation. At CERN, the calibration aerosols were nucleated and grown from various atmospherically relevant vapors in the CLOUD chamber , the contents of which were continuously sampled by a suite of instruments that included the Airel CIC and the Neutral cluster and Air Ion Spectrometer. The calibrations at INAR supported ion cluster experiments performed at CLOUD. Thus, in both locations, the sampling line was the same as that used at CLOUD, which included a Y splitter for connection to the CLOUD sampling probe. On the other hand, at the University of Tartu and at Airel, the CIC was calibrated without a CLOUD-specific inlet, thereby providing measurements that are more applicable to general operation.

We measured the detection efficiencies of the CIC for 1.2–5.0 nm positive and negative ions for each of the three collecting electrodes, as well as a sum of all three electrodes (Fig. ). The total ion detection efficiency exceeds 80 % between 1.2 and 2.0 nm mobility diameter and 90 % between 2.0 and 3.0 nm mobility diameter. The detection efficiency decreases above 3.0 nm, which provides a natural cut-off for the CIC outside the range of small ions. The close agreement between the experimental transfer functions and those derived from a simple theoretical model, which neglects the specific flow field inside the instrument, indicates that the effects of flow disturbance from the inlet filter sphere and other mechanical imperfections are not critical to the device's operation.

We also investigated the influence on ion detection efficiency of various CIC sample flow rates at 10, 30, and 50 L min⁻¹ (Fig. ). The higher sample flow rates produce mild increases in the detection efficiency below 3.0 nm and slight decreases for larger particles. Since higher flow rates provide an improved signal-to-noise ratio, it is generally preferable to use the maximum sample flow rate within the constraints of the experimental setup. If the sampling inlet includes bends or a Y splitter then the choice of optimum flow rate also needs to take into account turbulence at high flow rates, which leads to additional losses.

Figure  shows calibration measurements made with charged tungsten particles and the HDMA setup in the sub-5 nm size range. We compared the ion concentrations measured by the CIC with those of a reference electrometer. The linearity was measured for 0.9, 1.5, and 2.3 nm positive ions and for 0.9, 2.1, and 2.9 nm negative ions. In all cases, the response was linear for ion concentrations between 102 and 105 cm⁻³. These laboratory calibrations included the same Y splitter on the inlet as used at the CLOUD experiment, which introduces additional losses. Therefore, the CIC efficiency cannot be directly inferred from these data. In addition to demonstrating excellent linearity, these measurements show that the CIC can be operated reliably in environments ranging from pristine to highly polluted, i.e., aerosol number concentrations of up to 105 cm⁻³.

When the CIC measures in ion-free conditions, a small background ion concentration is always recorded due to unavoidable electronic noise in the electrometers. This sets a threshold ion concentration below which the CIC is unable to provide useful measurements. Figure  and Table  summarize CIC measurements at CLOUD when the electric clearing field is present, i.e., when all ions are swept from the chamber in less than 1 s. The CIC sampled the chamber air with a flow rate of 7.0 L min⁻¹ per analyzer, with an uncertainty of ±5 %. Under these conditions and for 1 s integration time, the median value of the instrument background for the ion concentration is below ±5 cm⁻³ (±0.1 fA). The 25th and 75th percentiles are below ±20 cm⁻³ (±0.4 fA). The 5th and 95th percentiles are between ±20 and ±50 cm⁻³ (±0.4 … ±0.9 fA). The standard deviation ranges from 18 to 35 cm⁻³ (0.35 to 0.65 fA), with a median at 26 cm⁻³ (0.48 fA). Lower noise levels can be reached by averaging signals over times longer than 1 s or by using higher sample flow rates. This instrument background is excellent and substantially smaller than most previous air ion spectrometers or electrometers that are commonly used as reference instruments.

The time response of the CIC has been measured in the CLOUD chamber using the CERN particle beam (Fig. ). The pion beam spill from the CERN Proton Synchrotron arrives in short bursts of 0.4 s duration, which, for these data, are spaced by about 30 s and have a flux of 1.3×106 particles per burst (Fig. b). The CIC ion number concentration responds to each beam pulse after a 5 s delay (Fig. a). This delay is due to the circulation time at the CLOUD experiment from the internal mixing fans for the beam-exposed air parcel to reach the tip of the CIC sampling probe. The CIC then reacts instantly (< 1 s) to the peak ionization pulse from the beam, followed by a gradual decay as the ions are mixed and diluted with non-irradiated air in the chamber. On longer timescales, ions are lost by ion–ion recombination and by collision with the walls of the CLOUD chamber. The response of the CIC to the beam bursts is mainly seen in channel 1 since only small ions are created in the chamber on these short timescales (note the absence of any signal in channel 3, which has zero response for small ions below 2 nm). The gaps in the CIC data correspond to periods when the instrument performed automatic internal zero-level measurements, with the inlet filter having been switched to 120 V. These data demonstrate an excellent response time for the CIC of around 1 s.