An aerodynamic particle sizer (APS) can put a precise-looking size distribution on a screen within seconds, but interpreting that graph requires knowing what the instrument measures. In an APS time-of-flight measurement, aerosol particles accelerate through a controlled flow and pass an optical timing region. Their response is converted through a calibration to aerodynamic diameter. This is not the same as measuring each particle with a ruler, and it is not automatically a measurement of particulate mass, chemical composition or health hazard.
To translate an APS report into useful language, start with its horizontal-axis definition, particle-count denominator, chosen diameter range and quality-control flags. A peak at two micrometres aerodynamic diameter does not by itself prove that particles are two-micrometre solid spheres; a rising number-size spectrum does not automatically mean rising PM2.5 mass. Concentration, time window, inlet efficiency, dilution, coincidence and conversion assumptions have to travel with the result if a reader is to understand what changed.
This guide explains aerodynamic particle sizing from first principles, follows the flow through a time-of-flight spectrometer and checks numerical worked examples. We will calculate count concentrations, integrate logarithmic bins, examine tentative mass estimates and track uncertainties. The independent problems at the end test whether these skills transfer to a new monitoring situation. Examples are explicitly invented for teaching; they are not observations from a named site or assertions about any city’s pollution.
The first diagnostic: what question can the instrument answer?
An APS is built to describe airborne particles within its specified aerodynamic-size range, usually by number concentration as a function of size. Its strongest direct product is an operational aerodynamic diameter distribution built from timing signals and calibration. It can also record optical scattering information, depending on the model and mode. This helps characterize coarse-mode aerosols, ventilation test particles, pharmaceutical aerosols and environmental size distributions. It does not provide a self-contained identification of what each particle is made of.
A report may have labels for number, surface area, volume or mass. Those outputs are not equally direct. Counting and sizing are the basic measurements; derived surface, volume and mass distributions generally require additional geometric or density assumptions. Before a translator repeats a quantity as fact, the report should reveal whether it was independently measured or generated by a software conversion. If the density is an entered default rather than a measured property, the reader needs to know.
For the APS model 3321, the manufacturer describes aerodynamic sizing across approximately 0.5–20 micrometres and separate relative light-scattering measurements. That specification belongs to a particular model, not to every device using the abbreviation APS. Instruments differ in calibration, response, size bounds, software and operating conditions. A reliable report identifies the make, model, firmware or analysis version when relevant, instead of borrowing a specification from an unrelated instrument.
If a laboratory calls an exported curve “particle size,” ask one clarifying technical question of the file, not necessarily of the author: size by what physical principle? Optical, electrical mobility and aerodynamic sizing observe different equivalent properties. A comparison may still be useful, but numerical alignment is not guaranteed. The best first translation is often “number distribution by aerodynamic diameter,” followed by the instrument’s range and acquisition conditions.
Aerodynamic diameter begins with motion through air
A particle moving relative to air experiences drag. Under a fixed air-flow history, its acceleration depends on inertia and the resistance of surrounding gas. A dense particle and a lighter particle of the same geometric width can respond differently because they carry different mass. An irregular particle can also have more drag than a compact one. Aerodynamic diameter compresses these effects into an equivalent size that describes specified aerodynamic behaviour, making results comparable for transport and collection problems.
The conventional aerodynamic diameter is the diameter of a sphere of reference density, commonly 1 gram per cubic centimetre, which has the same settling behaviour as the particle under the relevant reference conditions. In practice a time-of-flight device infers aerodynamic equivalent size from the particle’s acceleration in an accelerating flow and a calibration. This equivalent diameter is useful without implying that the sampled particle actually possesses the density or shape of that reference sphere.
In a simplified continuum-drag comparison for a compact particle, aerodynamic diameter is roughly the volume-equivalent diameter multiplied by the square root of particle density divided by reference density and then modified by a dynamic-shape factor. That shorthand has important qualifications: gas slip corrections matter for small particles, shape factors are not always known and some particles are porous aggregates. It should be taught as a physical intuition, not an exact universal calculator for field samples.
Consider two ideal spheres with identical physical diameter but different densities. The denser sphere possesses greater inertia and generally has the larger aerodynamic diameter under otherwise comparable conditions. Now consider two particles with similar aerodynamic diameter but dissimilar chemistry and shape; they can behave similarly in the instrument’s flow even though their actual geometries differ. This is why size alone cannot identify pollen, dust, sea salt, smoke or biological material.
Aerodynamic diameter influences how particles travel through bends, impactors, collectors and parts of the respiratory tract. It does not determine individual health risk by itself. Exposure duration, chemistry, solubility, deposition, biological response and personal vulnerability also matter. A good report uses aerodynamic size as a transport variable. It resists the temptation to equate a plotted peak with a clinical conclusion about the people who might breathe the sampled aerosol.
Follow a particle through the time-of-flight measurement
A sampling line carries aerosol towards the spectrometer. In the instrument, a controlled flow accelerates particles through a nozzle. Their acceleration relative to the carrier gas varies with aerodynamic response. Optical detection produces timing information as the particle traverses the measurement zone. The APS converts this time-of-flight signature to a calibrated aerodynamic size bin. The mathematical chain is therefore flow physics to timing to calibration to diameter to distribution—not optical colour or brightness to chemical identity.
In the model 3321, a double-crest optical arrangement uses two partially overlapping laser beams. A suitable single particle produces a signal with two crests, and the time separation between those crests contributes to its aerodynamic size estimate. The same design helps detect coincidence events involving multiple particles in the viewing volume. This is a specific signal-processing method, and simplified explanations must not claim it is literally two widely separated cameras taking pictures of a single dust grain.
Time of flight is measured at fine temporal resolution, but a fine electronic time step is not the same thing as arbitrarily fine environmental-size accuracy. Nozzle geometry, gas flow, particle trajectory, calibration particles, instrument efficiency and signal classification all contribute. A useful translation says “timing-calibrated aerodynamic diameter,” not “the laser has measured the exact physical diameter.” Distinguishing resolution from accuracy prevents impressive-looking decimal places from becoming unsupported claims.
The optical scattering channel is an additional property. Scattered intensity depends on optical size, shape, refractive index and instrument geometry, and can be correlated with aerodynamic response. It is not a laboratory assay of composition. Two particle populations with different refractive indices may scatter differently without being reliably identifiable from that signal alone. If a report displays side-scatter classes, treat them as instrument response categories unless a validated identification method establishes something more.
The instrument is designed for a bounded aerosol and operating environment. Very small particles may be below aerodynamic detection or classification limits; very large particles may fail to reach the sensing volume because of settling or inertial losses. Saturation and coincidence can occur at high number concentrations. The spectrum represents the qualified particles that passed through the actual inlet and instrument system, not an exhaustive inventory of every airborne particle at the site.
Do not confuse three different size axes
Electrical mobility diameter tells us how particles drift in an electric field after charging. Optical-equivalent diameter depends on scattering and an optical model. Aerodynamic diameter tells us about a reference aerodynamic response. A soot aggregate can have one mobility diameter and a different aerodynamic or optical equivalent diameter without creating a contradiction. These measurements describe different properties and assumptions. Harmonising them requires more physics than simply converting nanometres to micrometres.
A familiar units conversion is necessary but not sufficient: one micrometre equals one thousand nanometres. Thus a 700-nanometre particle-size label represents 0.7 micrometre numerically. But an electrical mobility diameter of 700 nanometres does not automatically equal 0.7 micrometre aerodynamic diameter for a non-spherical particle. Preserve the equivalent-size adjective when converting units. That single phrase tells a scientist whether a cross-instrument comparison is physically sensible.
When instruments cover different size ranges, a combined distribution must be designed carefully. An SMPS may characterize a nanoparticle number distribution, while an APS characterizes larger particle aerodynamic sizes. Joining their curves at an arbitrary diameter can create an artificial discontinuity. A defensible merge considers overlapping regions, counting efficiencies, relative calibration, inlet conditioning, the differing size definitions and possible changes between instrument sampling times. It may be preferable to present two distinct distributions side by side.
A chart with one logarithmic axis and another linear axis may also conceal apparent differences. A broader size interval contributes a larger total even if its displayed density is not particularly high. The vertical-axis convention matters as much as the horizontal label. Readers should inspect whether the exported results are bin totals, concentration per unit diameter, concentration per base-ten logarithmic interval, percentage of measured counts or a normalised probability distribution.
The concentration denominator is sampled air, not any flow on the label
In an ideal counting calculation, number concentration equals detected particles divided by the actual sampled air volume, with corrections made only when justified. The model 3321 specification describes approximately 1.0 litre per minute of aerosol sample flow together with sheath flow. Those flows play different roles. The total instrument flow should not be substituted for the aerosol sample volume in a concentration denominator; doing so can understate results considerably.
Imagine a two-minute sample with actual aerosol intake of 1.0 litre per minute and 2,400 accepted particle events in a chosen diameter group. Sample volume is 2.0 litres, equal to 2,000 cubic centimetres. Ignoring all corrections, that bin contains 2,400 divided by 2,000, or 1.2 particles per cubic centimetre. If someone mistakenly divides by an assumed total instrument flow of 5 litres per minute, the reported result is five times too small. Flow labels are part of the scientific claim.
Now suppose a validated response correction indicates that only 80 percent of particles in this specified test category pass through the measurement chain and are accepted. A simplified corrected concentration would be 1.2 divided by 0.8, equal to 1.5 particles per cubic centimetre. Real correction functions can depend strongly on aerodynamic diameter, particle type, inlet condition and concentration; this single factor is deliberately illustrative. A correction without a stated basis can be worse than a transparent uncorrected value.
A sample collected for ten seconds and a sample collected for ten minutes do not ordinarily have the same random counting precision. If both report one particle per cubic centimetre, the longer sample usually contains more counted events, assuming flow and concentration remain steady. The longer averaging interval may, however, hide rapid concentration pulses. Include duration and sampling volume. A good translation can say that the longer result is statistically steadier yet less temporally detailed, instead of treating one duration as universally superior.
Logarithmic bins: convert plotted density into particles
Particle distributions often use logarithmic diameter because the range spans many multiples in size. A vertical axis labelled dN/dlog10(Da) means number concentration per unit base-ten logarithmic aerodynamic diameter. To approximate the number in a finite channel with bin-edge diameters a and b, multiply the local reported density by log10(b divided by a). The logarithm acts on a ratio, so its input has no physical unit. Using b minus a directly would mismatch a logarithmic-density label.
For an invented bin with edges 1.00 and 1.26 micrometres and a constant dN/dlog10(Da) of 400 particles per cubic centimetre, the logarithmic width is log10(1.26), approximately 0.10037. The integrated number concentration for that channel is therefore 400 times 0.10037, or 40.15 particles per cubic centimetre. The bin width is not 0.26 in the formula because the plotted ordinate is per logarithmic size interval, not per linear micrometre.
If neighbouring channels have distinct widths or have been rebinned, blindly adding their displayed y values is incorrect. Integrate each with its own edge ratio, then sum the resulting concentrations across the stated diameter limits. If the table instead provides actual channel concentrations already integrated, another multiplication would double-apply the width. Software exports can label similar columns in inconsistent ways, so consult the data dictionary and test whether channel totals reproduce the reported instrument total.
Natural-log spectra introduce another convention. Because ln(Da) equals ln(10) times log10(Da), a density per unit natural log equals a density per unit base-ten log divided by ln(10) for the same physical distribution. A density of 2,000 per log10 unit corresponds to approximately 869 per natural-log unit. It is possible for two correctly plotted curves to have different numerical heights solely because of this denominator. The integrals over matching physical bins should agree.
Number and mass are different questions
Each measured particle contributes one to a number count, but its mass depends on material volume and density. For compact spheres of the same density, mass scales with diameter cubed. A sphere five micrometres across contains one thousand times the volume of a sphere 0.5 micrometre across. Thus an aerosol with fewer large particles can dominate an assumed mass-weighted plot while an aerosol with many smaller particles dominates a number-weighted plot. Neither outcome is mathematically surprising.
An APS number spectrum can be converted to an estimated mass spectrum only under stated assumptions. A common simplified calculation treats each particle as a compact sphere with diameter equal to the size used in the formula and assigns a density. For environmental aerosols, aerodynamic size may not equal geometric size; shape, density and slip complicate the conversion. The software output may be useful for a deliberately chosen model, but it cannot automatically replace a gravimetric particulate-matter measurement.
For one hypothetical calculation, suppose ten spherical particles per cubic centimetre all have geometric diameter 2.0 micrometres and material density 1.5 grams per cubic centimetre. A 2-micrometre diameter equals 0.0002 centimetre. The volume of each ideal sphere is pi divided by six times (0.0002)^3, approximately 4.189 × 10⁻¹² cubic centimetres. Multiplying by 1.5 gives 6.283 × 10⁻¹² grams per particle. The assumption of a geometric diameter is crucial to this example.
Ten such ideal particles per cubic centimetre contribute 6.283 × 10⁻¹¹ grams per cubic centimetre of air. There are one million cubic centimetres in a cubic metre and one million micrograms in a gram. Multiplying by both conversions gives approximately 62.83 micrograms per cubic metre of air. This is a checked arithmetic exercise, not a direct APS reading. If the diameter were merely aerodynamic, the geometric-volume assumption would need independent justification before calling 62.83 an atmospheric mass result.
The cubed diameter also magnifies uncertainty. Under the simplistic fixed-density sphere model, a ten-percent upward diameter error produces a mass estimate larger by a factor of 1.1 cubed, approximately 1.331, or 33.1 percent. A ten-percent downward error produces 0.9 cubed, or 0.729, approximately 27.1 percent lower. These examples show why modest sizing or shape differences can produce substantial changes in derived mass, without implying that actual instrument uncertainties equal ten percent.
Surface-weighted plots occupy a different middle ground. For ideal spheres, surface scales with diameter squared, number has no diameter multiplier and mass scales with diameter cubed at fixed density. A single report can therefore show three different modal maxima using the same underlying counts. Translate their vertical axes by what was weighted and what shape assumptions were used. The tallest mass-mode peak is not evidence that the number distribution was measured incorrectly.
A comparison with PM2.5 and PM10 needs more than a threshold
Regulatory PM2.5 and PM10 are mass-concentration concepts tied to defined sampling inlets, reference methods, conditioning and aerodynamic collection characteristics. The names do not mean that any device reporting particles smaller than 2.5 or 10 micrometres automatically produces equivalent mass data. An APS measures a bounded size distribution of the particles it successfully samples; a size-limited estimated mass curve depends on conversion assumptions. Treat a comparison with certified gravimetric results as a validation exercise, not an identity.
Imagine an APS reporting many two-micrometre particles after a cleaning activity. It might be tempting to infer a PM2.5 exceedance from the number count alone. That inference skips density, morphology, particle water, inlet transmission, below-range particles and measurement conventions. A more accurate statement describes the observed aerodynamic number concentration and notes which additional measurements would be needed for regulatory-style mass interpretation. Honesty about the gap strengthens rather than weakens the analysis.
Similarly, an instrument with a low-size cutoff near 0.5 micrometre excludes some smaller particles that may contribute modest mass but very large number. A difference between APS-integrated particle number and a broader-range counter can be expected even when both are working correctly. Before calling the instruments inconsistent, restrict the comparison to a common size range and evaluate their independent sampling and detection efficiencies. The missing lower sizes are a measurement domain issue, not something to invent in a graph.
A coincidence flag is a measurement warning, not a second particle species
Single-particle instruments perform best when particles are sufficiently separated in the sensing region. Coincidence occurs when more than one particle occupies the optical measurement volume during an event. Overlapping signals can corrupt timing, make false sizes or prevent individual events from being accepted. The APS model 3321 uses its double-crest signal characteristics to identify certain coincidence events. Detection reduces some sizing artefacts, but it does not magically remove the underlying high-concentration condition or recover all missed particles.
An exported “coincidence count” should not be described as the number of agglomerated particles. It refers to an instrument signal condition involving overlapping particles, not proof that particles physically stuck together. The total number of simultaneously present particles may not be known, and the effect on apparent concentrations can depend on event-processing rules. A separate event log can signal that sample dilution or another valid concentration-control procedure would improve reliability.
Suppose 3,000 valid-size events are recorded while 150 additional events are flagged as coincident. The flagged-to-valid ratio is 150 divided by 3,000, equal to five percent. But adding 150 to 3,000 or multiplying the result by 1.05 is not automatically a valid particle-number correction. Some flagged events contain multiple particles and others may be misclassified; detector dead time and transmission can also be affected. Report the flag ratio as a diagnostic until a validated correction is specified.
The manufacturer specifies operating limits and expected coincidence characteristics under specified particle conditions. Those thresholds are a guide to instrument operation, not a guarantee that every aerosol at the stated concentration is unbiased. Concentration, size and optical response influence signal crowding. When a report claims unusually high particle number, check the coincidence channel, any diluter installation, actual sample flow and test protocol. A curve drawn from saturated measurements should not be rescued by arbitrary smoothing.
Dilution makes sense only when its factor is known and stable
A high concentration sample may be diluted with clean air before measurement, lowering coincidence and expanding the useful measurement range. The upstream concentration is then estimated from downstream observations and a measured dilution ratio, with any separate transmission correction applied consistently. A label saying “diluter fitted” is not enough. The dilution factor’s definition, verification, leakage control, humidity and possible size-dependent losses affect the meaning of reported concentrations.
In an invented example, the instrument reads 3 particles per cubic centimetre after a dilution factor of 20, defined as upstream concentration divided by downstream concentration. Ignoring other losses, upstream concentration is 3 times 20, or 60 particles per cubic centimetre. If a separately measured upstream penetration for the particles of interest is 0.80, an illustrative pre-loss estimate becomes 60 divided by 0.80, or 75 per cubic centimetre. These two corrections describe different physical effects and should be documented independently.
A nominal factor printed on a dial is not necessarily an accurate effective dilution factor in every operating situation. Changing relative humidity can alter hygroscopic particle sizes; tubing and mixing may lose larger or very small particles; the clean-air source could introduce background. Validate the factor with an appropriate tracer or controlled test when accuracy matters. In comparisons, report whether the provided table is raw downstream, dilution corrected or fully inlet corrected to avoid applying the same correction twice.
Flows, inlets and sampling geometry define the observed population
Airborne particles do not necessarily follow every bend in a sampling tube. Larger particles carry inertia and may strike walls; particles in long horizontal tubing can settle, and extremely small particles can be lost diffusively. Inlet orientation relative to wind and the velocity at which air enters may change aspiration efficiency. The sampling system can therefore favour some sizes over others before particles reach the instrument. The exported distribution is a measurement of the resulting sample path, not an unconditional census of outdoor air.
The appropriate inlet design depends on whether one wants ambient outdoor aerosol, a ventilation duct, an industrial stack or an experimental aerosol chamber. A representative outdoor inlet is not automatically suitable for an accelerating duct with nonuniform velocity. Flow direction, isokinetic conditions and tubing deposition become central in some applications. A careful author describes the sampling environment and geometry rather than asserting representativeness merely because the instrument itself has been calibrated.
For a simple flow error example, imagine a five-minute run nominally at 1.00 litre per minute. If a valid calibrated reading shows 0.95 litre per minute throughout, the true aerosol sample volume is 4.75 litres rather than 5.00. Dividing the same number of accepted counts by 4.75 instead of 5.00 gives a concentration about 5.26 percent higher. This arithmetic isolates flow volume; in an actual APS, flow deviation may also affect aerodynamic sizing and transport, requiring additional checks.
Manufacturer documentation for the 3321 distinguishes approximately 1.0 litre per minute sample flow and 4.0 litres per minute sheath flow, with a combined total of about 5.0. These are specification figures and have associated tolerances. The aerosol concentration denominator belongs to the sample stream, not a simple sum of instrument air movements. Changing or miscalibrating flow can influence both volume and the particle acceleration that underlies time-of-flight sizing. Log actual operating values.
Relative humidity, evaporation and aerosol integrity
Water and volatile compounds complicate aerosol measurements. A hygroscopic salt aerosol can swell as humidity rises; a volatile droplet can shrink while travelling through warm or dry tubing. When composition changes inside the sampling system, an aerodynamic-size distribution at the detector need not represent the original ambient state. The model’s published size range does not protect against these physical transformations. Temperature, humidity, dryer conditions and sample residence time are relevant parts of the result.
Imagine two identical humidification experiments, one sampled through a short conditioned line and one through a long dry tube. Their particle modes might shift because water content differs when the instrument measures them. It would be mistaken to infer changed emission composition immediately from the difference. Compare sampling conditioning and equilibrium assumptions, and where possible record both wet and dry observations. In some experiments, conditioning is deliberate and valuable; the key is to describe exactly which aerosol state was measured.
An environmental report may show “mass” estimated from particles that have lost water or semi-volatile material before detection. Even a mathematically correct spherical mass conversion then describes the conditioned particles under those assumptions, not necessarily the ambient hydrated aerosol mass. A reliable translation carries the qualifier and avoids claiming perfect equivalence to filters sampled under another protocol. The distinction matters most where a small change in water or volatility could explain much of the apparent trend.
Check counting statistics without treating them as the whole uncertainty
Particle arrivals often display randomness, and an idealized Poisson counting approximation provides a useful first check. With 900 accepted events, square root of 900 is 30, giving about 3.33 percent relative counting standard uncertainty. With nine events, square root is three, yielding roughly 33.3 percent. The difference explains why weak channels can look ragged, especially with short averaging periods. These figures assume independent events and do not account for all real measurement errors.
Flow calibration, size classification, instrument-to-instrument variation, coincidence, inlet losses and software settings are distinct sources of uncertainty. If three approximately independent relative uncertainties were five percent, eight percent and three percent, an illustrative root-sum-of-squares combination would be square root of 25 plus 64 plus 9, about 9.90 percent. Independence must be justified; common calibration errors and systematic losses may be correlated. A report should not claim a 9.90-percent certified uncertainty from this toy calculation.
Some uncertainty changes the vertical axis, other uncertainty changes which size bin receives a particle. A systematic diameter shift can alter both the apparent location of a mode and any diameter-cubed mass conversion. Counting noise mainly affects limited-event precision. Inlet loss can selectively reduce one diameter range. These mechanisms should not be collapsed into one vague claim that the instrument is “accurate to ten percent.” Separating them helps readers identify what experiment would actually resolve a disagreement.
When comparing two days, show which differences exceed plausible instrument variation and which are only suggestive. A two-percent change in a channel with an unresolved ten-percent bias is not robust evidence of an environmental trend. Larger changes can still be artefacts if a sampling hose or conditioning stage changed between days. Keep a simple provenance record containing time, instrument settings, quality flags, calibration status, inlet geometry and sample conditioning before telling a narrative about causes.
Quality-control checks before believing the most interesting peak
Start by confirming that the instrument warmed up and operated within its specified conditions, the sampling line was unobstructed and the actual aerosol and sheath flows were checked. Review alarms, concentration and coincidence limits, zero or background records, and the interval over which the plot was averaged. The aim is not ritual compliance with a checklist; each check eliminates a way that a measurement could look plausible while being physically wrong. Save the full settings alongside exported results.
Challenge particles of characterized size are useful for checking a sizing response, but calibration material is not a magic proxy for every irregular field aerosol. Polystyrene latex spheres are often used as stable test objects; their properties must be considered when comparing aerodynamic and geometric size. Verify the manufacturer’s appropriate procedure and calibration certificate. A well-calibrated instrument can still sample the environment unrepresentatively because a separate inlet or tubing problem remains.
If two otherwise similar instruments disagree, do not immediately average their outputs. Start with shared-size-range comparisons and verify the same time basis, flow, inlet, concentration corrections and bin convention. Controlled intercomparisons have documented unit-to-unit variability in aerodynamic particle sizers. A mismatch could be due to counting efficiency in one size interval rather than a physical aerosol change. Report disagreement directly while establishing whether it is constant, size dependent or associated with transient concentrations.
A clean-air blank checks background signals, but it does not guarantee that a loaded instrument will classify correctly under heavy event rates. A flow check verifies volume and some timing conditions, but not particle transmission through every tube. A known-size aerosol challenges calibration, but may not reveal humidity-induced evaporation in field particles. Each quality test addresses a specific failure mode. A thoughtful quality narrative explains what was verified and what remained outside the test’s reach.
Translate source hypotheses without inventing chemical measurements
A broad coarse-particle mode may be compatible with resuspended dust, sea spray, agricultural activity, indoor cleaning or mechanical processing. The APS cannot choose among these sources by diameter alone. A change coinciding with wind direction is suggestive, not conclusive. Source analysis can be strengthened by chemical tracers, spatial comparisons, controlled activities, meteorology and repeated observations. Use “consistent with” only when those alternative possibilities have been considered and explain which were not excluded.
A person reading a health-related study should separate three claims: how many particles were measured, where a particle of a given aerodynamic response might deposit and whether the particular aerosol caused an effect. Aerodynamic size helps with the second question through deposition physics, but the answer depends on breathing patterns and modelling assumptions. Dose and toxicity also require composition and biology. The instrument output should never be rewritten as proof of illness, nor as proof of safety.
In a workplace test, a burst in the 2–5 micrometre aerodynamic range during sweeping may reasonably motivate further measurement of dust controls. The defensible observation is the size-resolved increase under the monitored activity. Attribution to sweeping is stronger if the activity timing and control periods align, but occupational exposure conclusions require relevant sampling methods and standards. In a classroom or home, similar caution applies: instrument evidence is a useful prompt for ventilation and cleaning questions, not a standalone medical diagnosis.
Three case studies that expose interpretation mistakes
Case study one: a chart shows a large number peak near 1 micrometre during a dry windy period. The researcher writes, “The wind carried one-micrometre sand grains that caused dangerous exposure.” This sentence overclaims particle identity, physical geometry and health effect. A repaired statement is: “The monitor observed an increased number concentration near 1 micrometre aerodynamic diameter during the stated windy interval; dust transport is a possible contributor requiring independent compositional and sampling checks.” That version protects the measurable finding.
Case study two: a high-concentration chamber experiment suddenly shows fewer large particles when the particle generator is turned up. The author claims that particles fragmented into smaller sizes. Before accepting that, inspect coincidence-event rates, detector saturation, dilution, inlet loading and time alignment. High counts can distort single-particle timing and cause rejection of events. A controlled dilution series that preserves source conditions can help distinguish instrument limits from a true physical change. The original report should record whichever explanation survives testing.
Case study three: a mass-weighted APS plot shifts strongly towards a larger diameter while total particle number remains nearly constant. This is not inherently contradictory. At fixed assumed density, a modest increase in diameter may amplify an estimated mass because the calculation involves diameter cubed. Examine whether the software used a fixed density, whether particles were spherical and whether sizing calibration changed. The correct translation says the estimated model mass changed, not necessarily that an independent filter weighed more material.
Independent transfer problems
Problem one: an APS accepts 6,000 events during one minute when its actual aerosol sample flow is 1.0 litre per minute. Ignore losses and coincidences. What is the count-derived concentration in particles per cubic centimetre? Explain why 5.0 litres per minute of total air movement would be an inappropriate divisor if only 1.0 litre per minute is the aerosol sample.
Problem two: a result gives dN/dlog10(Da) = 500 particles per cubic centimetre in a bin running from 2.00 to 2.52 micrometres. Calculate the integrated bin concentration. State which logarithm you used and why direct multiplication by 0.52 micrometre would be inconsistent with the supplied vertical-axis units.
Problem three: a hypothetical sample is diluted by factor 12 before measurement. The corrected downstream size-bin concentration, with other losses ignored, is 4 particles per cubic centimetre. What is the estimated upstream concentration? If a separate verified inlet penetration is 75 percent, what is the adjusted pre-inlet estimate under the simplified independent-factor model? Name one condition that would invalidate a constant correction.
Problem four: the detector accepts only 16 counts in a narrow size channel. Estimate the relative Poisson counting standard uncertainty. Compare it with 1,600 accepted counts. Could either counting number alone prove that no flow or size bias is present? Problem five: at equal density and ideal compact spherical shape, how does mass per particle change if geometric diameter doubles?
Problem six: an APS-derived distribution and a separate gravimetric PM2.5 measurement disagree. List at least four possible measurement or modelling causes before deciding that one instrument is broken. Problem seven: a spectrum reports a shoulder at 0.55 micrometre near the lower instrument size bound. Why should a writer be careful about counting efficiency and range limits when naming that shoulder a real environmental mode?
Problem eight: an instrument log records 150 coincidence flags alongside 3,000 accepted events. Calculate the simple flagged-to-accepted event ratio. Explain why that ratio does not define a universal correction for true particle count. Problem nine: a report contains a relative optical-scatter channel associated with each aerodynamic bin. What additional independent evidence would you request before identifying those particles as salt rather than dust or biological material?
Independent problem solutions, with checked working
Answer one: 1.0 litre equals 1,000 cubic centimetres, so 6,000 divided by 1,000 is 6 particles per cubic centimetre. The larger total-flow figure includes sheath flow and is not the aerosol volume corresponding to sampled particle events. Answer two: log10(2.52 divided by 2.00) = log10(1.26), approximately 0.10037. Multiplying 500 by that dimensionless width gives approximately 50.19 particles per cubic centimetre in the bin.
Answer three: 4 times 12 gives 48 particles per cubic centimetre upstream of the diluter under the defined factor. Correcting separately for 0.75 inlet penetration gives 48 divided by 0.75, or 64 particles per cubic centimetre before that loss. A fixed correction would be questionable if penetration changes markedly with size or if dilution alters particle water content. Each number must identify which physical location in the sample pathway it describes.
Answer four: square root of 16 divided by 16 gives 4 divided by 16, or 25 percent relative counting standard uncertainty. For 1,600 counts, square root is 40, so the corresponding figure is 40 divided by 1,600, or 2.5 percent. Neither calculation addresses flow, sizing calibration, coincidence or sampling losses. Answer five: for constant density and ideal spherical geometry, mass scales with diameter cubed. Doubling diameter therefore multiplies mass per particle by eight.
Answer six: plausible reasons include unknown effective density or shape, particles smaller than the APS measurement range, inlet size-dependent losses, wet versus dry conditions, coincidence bias, different integration periods, flow errors and the difference between a model-converted number spectrum and a reference mass method. Answer seven: near a lower size boundary, classification and counting efficiency can fall or vary rapidly; a small apparent peak may be shaped by response correction rather than the environment.
Answer eight: 150 divided by 3,000 equals 0.05, or five percent. It indicates the frequency of flagged events relative to accepted ones in this invented example, not the number of particles lost. Different flagged events may contain different particle multiplicities, and detector dead time affects detection. Answer nine: request relevant chemical analysis, morphology, humidity and site context, such as ion composition for sea salt or elemental/mineral markers for dust, with appropriate controls.
Questions readers commonly ask
Does an APS measure absolute particle mass? It principally sizes and counts particles by an aerodynamic response. Mass estimates may be computed under assumed density and shape or compared with independent mass methods. Describe the modelling assumptions openly. A volume-equivalent geometric diameter, particle density and the actual aerosol state cannot all be recovered unambiguously from aerodynamic time of flight alone. A mass-like axis therefore needs a stated method rather than a label that implies direct weighing.
Why might an APS and an optical particle counter disagree? They generally convert different signal properties into equivalent sizes, and detection ranges and efficiencies differ. Relative refractive index, particle shape, sample conditioning, counting corrections and flow can produce different results. Agreement requires a planned comparison over common particle conditions, not matching graph colours or simply converting micrometres to nanometres. One method may be more appropriate than the other for a particular scientific question.
What does coincidence mean in ordinary language? It means more than one particle affected a sensing event closely enough in time and space to interfere with clean single-particle classification. It is an instrumental event category, not a chemical reaction, not proof of aggregation and not an aerosol source marker. The correct response is to inspect quality flags and concentration limits and use an appropriate validated correction or dilution procedure where supported.
Can an APS identify pathogens, pollen or toxic metals? Aerodynamic size alone cannot identify these materials. Some populations occupy familiar size ranges, but distributions overlap, and optical scattering still does not provide a chemical or biological assay. Suitable microbiological, microscopic or chemical methods may be required. If the question concerns biological safety, avoid presenting a particle-size spectrum as pathogen detection or absence, especially when the instrument’s measurement range excludes some potentially relevant particles.
What if the reported diameter is very precise? Distinguish output formatting from demonstrated accuracy. A digital channel centre of 1.234 micrometres does not automatically mean the aerosol was measured to the nearest nanometre. Calibration accuracy, response width, count sufficiency and size-related biases remain. Prefer a meaningful rounded estimate and include the valid size domain and quality evidence, especially for a peak near a lower or upper limit.
Why do some studies report particle number and others report mass? Number is sensitive to how many particles occur, including abundant small particles. Mass, under appropriate measurement methods, weights larger particle volumes much more strongly. These answer different environmental or engineering questions. A responsible explanation names the metric and avoids implying that one is a substitute for the other. Converting between them needs a justified relationship, not an arbitrary universal constant.
Primary sources and methodological reading
TSI Incorporated, Aerodynamic Particle Sizer Model 3321 specification, 2022 revision: manufacturer specifications. This source documents the 0.5–20 micrometre aerodynamic range, double-crest time-of-flight arrangement, coincidence concept, sample versus sheath flow and separate light-scattering measurement. Model-specific values in this guide are attributed to that publication.
TSI Incorporated, Aerodynamic Particle Sizer operation brochure: time-of-flight instrumentation overview. The manufacturer’s notes help distinguish aerodynamic timing from optical-scattering intensity. Any particular installed system still requires its own current manual, configuration and calibration.
Pfeifer and colleagues, “Intercomparison of 15 aerodynamic particle size spectrometers (APS 3321): uncertainties in particle sizing and number size distribution,” Atmospheric Measurement Techniques 9 (2016), 1545–1551: European Commission JRC research record. The intercomparison addresses the importance of flow checks, counting efficiency and between-instrument variability.
University of Cambridge research repository, “Measuring Aerosol Size Distributions with the Aerodynamic Particle Sizer”: methodological study. Read the original conditions and methods when transferring aerosol results to another sampling geometry; demonstration systems are not interchangeable with all field setups.
Google Search Central, Creating Helpful, Reliable, People-First Content, urges substantive original explanation, explicit methods and transparent limitations. This specialist guide aims to help readers audit numbers rather than treat a word count or a keyword as an assurance of correctness.
Related eduKate technical guides and learning connections
For the smaller-particle electrical mobility technique, compare How to Read SMPS Mobility-Diameter Spectra. For single-particle detection and counting efficiency, see How to Read Condensation Particle Counters. These methods answer related but different questions about aerosol particle size and number.
For mass measurement and calibration context, compare How to Read TEOM PM2.5 Mass and How to Read Beta Attenuation PM2.5. A step-by-step explanation and independent transfer problems follow the educational approach in the unchanged Clementi Mathematics reference. No measurement should be forced into a different method’s vocabulary.