Counting Efficiency (CPC & APS)
Recorded, not applied
AeroViz does not correct for any of this. A size distribution from
RawDataReader is what the instrument reported, uncorrected. This page exists
so the size-dependent under-counting is known and quantified when you use
the data — it is a real bias at the ends of both size ranges, not a rounding
error. Nothing here is wired into QC.
A size distribution is a counting measurement, and neither counter counts every particle that enters it. The efficiency is size-dependent and falls off sharply at one end of each instrument's range:
- SMPS — the DMA classifies, but a CPC does the counting. Below the CPC's cut-off, particles are too small to grow into detectable droplets, so the lowest channels under-report. Which CPC is attached decides where that happens.
- APS — under-counts at both ends: small particles fall below the optical detector's reliable response, large ones are lost to inlet aspiration and transmission before they reach the detector. Droplets fare far worse than solid particles.
1. CPC — which counter is attached
The SMPS raw file records this in its metadata block, above the data header:
Detector Model 3750 Detector S/N 3750234302 Nano Enhancer None
Detector Sample Flow (L/min) 1.00 Detector Inlet Flow (L/min) 1.00
AIM 11.x CSV exports carry the same fields (Detector Model, Detector S/N,
Nano Enhancer) one per line. Read it from your own files rather than
assuming — the CPC is a separate instrument that can be swapped, and the cut-off
moves with it.
Current TSI line
From TSI Application Note CPC-002, Choosing the Right CPC for Your Application (Table 1, rev. D 2024). D50 is the diameter at which counting efficiency reaches 50 %.
| Model | D50 (nm) | Working fluid | Sample flow (LPM) | Conc. accuracy | Pairs with SMPS 3082 |
|---|---|---|---|---|---|
| 3007 | 10 | isopropanol | 0.1 | ±20 % | no |
| 3750 | 7 | butanol | 1.0 | ±5 % | yes |
| 375010 | 10 | butanol | 1.0 | ±5 % | yes |
| 3752 | 4 | butanol | 0.3 | ±5 % | yes |
| 3756 | 2.5 | butanol | 0.05 | ±10 % | yes |
| 3757-50 | 1 1 | DEG + butanol | 0.15 | ±10 / ±15 % | yes |
| 3783 | 7 | water | 0.12 | ±20 % | no |
| 3789 | 2.2, 7, custom 2 | water | 0.3 | ±5 % | yes |
| 3790A | 23 | butanol | 1.0 | ±10 % | no |
| 3790A-10 | 10 | butanol | 1.0 | ±10 % | no |
The 375010 is the same hardware as the 3750 given a CEN-compliant 10 nm
calibration — same instrument, deliberately different cut-off. The Detector
Model field alone will not always tell you which calibration a unit carries.
The curve, not just the cut-off
D50 is one point on a curve, and the curve is what matters for the lowest SMPS channels. Three findings worth carrying:
Nominal D50 is not measured D50. Wiedensohler et al. (2012) calibrated ten CPCs against a reference electrometer with silver particles; the TSI 3772 units averaged a 50 % detection diameter of 7.52 ± 0.04 nm against a nominal 10 nm. Units at the same factory settings still differed from each other by a few nm.
The cut-off moves with the temperature difference. The saturator–condenser ΔT is an operating parameter, not a constant: running at 25 °C instead of the factory 17 °C shifts D50 to smaller sizes measurably (Wiedensohler et al. 2012, Fig. 3 — the same model appears at both settings with visibly different curves). If your ΔT is not the factory default, published curves for that model do not describe your instrument.
Output mode can bias the count. Hermann et al. (2007) found CPC pulse outputs under-reporting by 2–10 % relative to the serial output, requiring a ~10 % correction for the 3776.
Hence the ACTRIS/GAW practice: calibrate the counting-efficiency curve annually against a reference, and accept it only within 5 % on the plateau and 1 nm on D50 (Wiedensohler et al. 2012, §4.2 and the QC criteria in §7). The same paper lists "correction for CPC counting efficiency" as a required processing step for network-quality data — a correction AeroViz leaves to you.
Curve shape
Measured curves are conventionally fitted with a saturating exponential in diameter, of the form
where a is the plateau efficiency (ideally 1), D₀ the onset diameter and D₁
the width of the rise. Coefficients are specific to the unit and its ΔT, so
this page deliberately does not tabulate them: use the calibration certificate
for your own counter, or the curves in the reference below for the model class.
2. APS 3321 — efficiency at both ends
The APS under-counts differently: not a condensation threshold but a combination of aspiration losses, transmission losses and detector errors (Volckens & Peters, 2005).
| Particle type | Efficiency | Size dependence |
|---|---|---|
| Solid | 85–99 % | roughly flat across the range |
| Liquid droplets | 75 % → 25 % | falls steadily from 0.8 µm to 10 µm |
The droplet result is the one to remember: at 10 µm the APS counts roughly one droplet in four. Ambient aerosol in the coarse mode is often at least partly deliquesced, so which of these two rows applies to a given dataset is a judgement about the aerosol, not a property of the instrument.
Peters & Leith (2003) characterised concentration measurement and counting efficiency for the 3321 specifically; Pfeifer et al. (2016) intercompared 15 APS 3321 units and quantified the unit-to-unit spread in sizing and number concentration — the counterpart to the CPC unit-to-unit variability above.
3. What this means when reading AeroViz output
- The lowest SMPS channels are under-counted by an amount set by your CPC's cut-off. With a 3750 (D50 7 nm) against an SMPS range starting at 11.8 nm, the first channels sit on the rising part of the curve, not the plateau.
- The APS coarse end is under-counted, severely so for droplets.
- An SMPS–APS merge therefore joins two differently-biased measurements. The
merge in
AeroViz.merge_psdfits and blends the overlap region; it does not correct either instrument's counting efficiency, so a merge inherits both. df.attrsrecords the instrument and its native resolution, but not the CPC model — read it from the raw file header (§1) when it matters.
References
- TSI Inc., Choosing the Right CPC for Your Application, Application Note CPC-002 rev. D (2024) — PDF
- Wiedensohler, A. et al. (2012), Mobility particle size spectrometers: harmonization of technical standards and data structure…, Atmos. Meas. Tech. 5, 657–685 — open access
- Wiedensohler, A. et al. (2018), Mobility particle size spectrometers: Calibration procedures and measurement uncertainties, Aerosol Sci. Technol. 52(2) — DOI
- Hermann, M. et al. (2007), Particle counting efficiencies of new TSI condensation particle counters, J. Aerosol Sci. 38, 674–682 — DOI
- Volckens, J. & Peters, T. M. (2005), Counting and particle transmission efficiency of the aerodynamic particle sizer, J. Aerosol Sci. 36, 1400–1408 — DOI
- Peters, T. M. & Leith, D. (2003), Concentration measurement and counting efficiency of the aerodynamic particle sizer 3321, J. Aerosol Sci. 34 — DOI
- Pfeifer, S. et al. (2016), Intercomparison of 15 aerodynamic particle size spectrometers (APS 3321), Atmos. Meas. Tech. 9, 1545–1551 — open access