A hygroscopic tandem differential mobility analyser (HTDMA) translates water uptake by airborne particles into a size-resolved growth-factor measurement. It usually chooses a dry mobility diameter with a first differential mobility analyser, exposes the selected aerosol to a known relative humidity, and measures the resulting wet mobility-diameter distribution with a second analyser. The central question is not simply whether air is humid. It is how particles of a stated original size respond to a controlled water-vapour environment.
The quantity most readers encounter is a dimensionless growth factor, GF = D_wet/D_dry. A dry particle selected near 100 nanometres and measured at 150 nanometres after humidification has a growth factor of 1.50 at the reported relative humidity, if dry and wet sizing definitions, charge assumptions and corrections are compatible. GF is not a percentage increase in mass, and it is not the same thing as cloud condensation nuclei concentration. Each quantity answers a different laboratory question.
This article explains how an HTDMA experiment works, how to read its distribution, how to calculate and qualify water uptake, how phase transitions and mixtures complicate the interpretation, and how to repair common reading errors. All numerical cases are illustrative teaching exercises rather than claimed observations at an actual monitoring station. The guiding principle is to translate measurement conditions together with numbers, because the number loses meaning when its humidity, selected dry size and correction model disappear.
The physical meaning of hygroscopic growth
Hygroscopic particles draw water from surrounding vapour when the chemical and thermodynamic conditions permit. Dissolved salts can substantially increase particle water content at high relative humidity. Some organic materials absorb less, but organic aerosol is diverse: its oxidation state, solubility, phase and interactions with inorganic material matter. Soot may remain relatively weakly hygroscopic until coatings or mixing change its surface and bulk properties. Mineral dust may have different pathways again. A label such as pollution, smoke or marine aerosol cannot replace measurement of a specific population.
A particle may grow in mobility diameter because it gains condensed water, yet changes in morphology also matter. Fractal aggregates can restructure, coatings can redistribute, and a particle may change phase. Mobility diameter is the diameter of a reference sphere with equivalent electrical mobility under specified conditions; it is not necessarily a direct ruler measurement of a solid geometric edge. The safest reading says the instrument observed a humidity-dependent mobility-size shift, then discusses what water uptake and structure could explain it rather than assuming only liquid water volume changed.
Why two differential mobility analysers are used
The first DMA separates aerosol approximately by electrical mobility under a chosen flow and voltage. An upstream neutralizer establishes a charge distribution rather than magically giving every particle the same charge. The first DMA transmits a selected mobility interval. This population travels through controlled conditioning, including drying or humidification, before the second DMA scans mobility and an associated particle counter records transmitted counts. The resulting wet-size distribution can be inverted or fitted to characterize growth-factor modes at one selected dry mobility diameter.
The second DMA does not count water molecules. It classifies the electrical mobility of whatever particle population survives the tubing, conditioning and classifier. Data processing maps the measured wet mobility to a growth factor. A multiply charged larger particle may pass the first selection; limited DMA transfer width broadens measured modes; counting noise affects small populations. The SMPS mobility-diameter guide is a useful foundation for distinguishing selected mobility, charge state and inferred particle size. Those controls remain essential in a tandem system.
The difference between a growth factor and a growth distribution
An average GF is convenient, but the wet-size distribution may contain one mode, several modes, a shoulder or a long tail. Suppose a selected 100 nm dry population contains one component growing to about 110 nm and another growing to 155 nm at 90% RH. The corresponding approximate modes are GF 1.10 and 1.55. A single reported mean around 1.3 would conceal that two subpopulations may coexist. Such structure may indicate mixing-state differences, although instrument transfer functions and charge artefacts must first be evaluated.
A published growth-factor probability density may be weighted by particle number, fitted to parametric modes or recovered through an inversion. Peak height is therefore not automatically a particle fraction. Integrating the area under a normalized distribution provides a probability; reading the vertical height at one growth factor does not. Always find the x-axis quantity, the y-axis units or normalization, the inversion method, and the dry selection diameter before announcing the abundance of a less-hygroscopic or more-hygroscopic mode.
What relative humidity actually means in this measurement
Relative humidity is water-vapour pressure divided by saturation vapour pressure at the relevant temperature, commonly written RH as a percentage. At a nominal 90% RH, it is the actual controlled conditions of the wet-sizing section that matter. Temperature gradients, sensor placement, vapour equilibration and residence time may cause the particle environment to differ from a controller’s displayed setpoint. RH is not a dimensionless growth factor and 90% RH does not imply that the particle diameter grows by 90%.
Many HTDMA comparisons use approximately 90% RH and specified dry sizes, but not every instrument or experiment does so. A humidogram deliberately changes RH while retaining a selected dry size; a fixed-RH experiment may vary dry diameter. These protocols answer different questions. The University of Manchester’s instrument description distinguishes constant-humidity and varying-humidity modes. To compare curves fairly, translate the protocol, temperature and RH measurement method, not just the headline diameter ratio.
Worked example: diameter change and linear growth
Consider a hypothetical size-selected aerosol at D_dry = 80 nm. At a reported 90% RH, the measured wet mode is D_wet = 116 nm. The growth factor is 116/80 = 1.45. The diameter increase is 36 nm, or (116−80)/80 × 100 = 45%. Both statements are correct: GF 1.45 and a 45% diameter increase. They are not claims that the particle has gained 45% water by mass or volume. Units cancel in the ratio because both diameters are expressed in nanometres.
If a second sample has D_dry = 120 nm and D_wet = 174 nm, its GF is also 1.45. The absolute growth is 54 nm rather than 36 nm, while relative growth matches. That example shows why GF facilitates cross-size comparison but is not a guarantee of identical composition or water content. Curvature effects and size-dependent chemistry can still yield different interpretations, and both comparisons require confidence that each dry mobility selection and humidification protocol were correct.
Converting a diameter ratio into a volume ratio
For idealized spherical particles that retain a simple shape, volume is proportional to the cube of diameter. A GF of 1.50 implies a wet-to-dry volume ratio of 1.50³ = 3.375, so the added liquid-equivalent volume is 2.375 times the dry-equivalent volume. A GF of 1.10 yields a volume ratio of about 1.331, only 0.331 extra dry-equivalent volume. This cubic relation explains why a moderate diameter shift can correspond to a substantial change in volume.
It would be premature to equate that added volume to a directly measured mass of water. Different densities of dry material and liquid water, changes in particle shape, nonideal internal structure and possible semivolatile material exchange complicate the mass balance. To infer water mass one needs assumptions about dry density, wet solution density, phase and possible dry-material loss. A careful translation says wet-to-dry volume ratio under a spherical approximation, rather than asserting an experimentally measured water mass from a diameter measurement alone.
The supersaturated and subsaturated worlds are related but distinct
An HTDMA usually studies particle growth below water saturation, while a CCN counter studies activation at small supersaturations above saturation. Both illuminate water affinity, but they probe different regimes. An aerosol may show subsaturated growth that supports an effective hygroscopicity estimate, yet its supersaturated behaviour can also depend on surface activity, phase separation, mixing state and kinetic limitations. Extrapolating a single subsaturated point directly to cloud activation requires a model and uncertainty discussion.
For the complementary measurement, see the CCN, supersaturation and activation-fraction guide. It explains why cloud activation thresholds depend on dry diameter and effective hygroscopicity and why a controlled counter does not directly forecast rain. When reading both datasets together, be explicit about whether the quoted κ comes from subsaturated growth, cloud activation or chemical composition. They can be compared, but they are not automatically equal observations.
A simplified κ estimate from growth
The κ hygroscopicity parameter can be estimated from subsaturated growth within a specified thermodynamic model. For a simplified ideal relation that neglects the Kelvin term, κ ≈ (GF³−1)(1−a_w)/a_w, taking water activity a_w approximately equal to fractional ambient RH at equilibrium. At 90% RH, (1−a_w)/a_w = 0.1/0.9 = 1/9. Thus a GF of 1.50 gives κ ≈ (3.375−1)/9 ≈ 0.264.
This calculation is a deliberately simplified classroom estimate. For small particles the Kelvin curvature term can matter; accurate retrievals consider particle size, temperature and water-surface properties. If a paper uses a full κ-Köhler inversion, do not reproduce its κ by applying this shortcut and call the difference an error. Compare the equations and inputs. Likewise, a single κ cannot identify sodium chloride, ammonium sulfate, one organic compound or a particular emission source. It condenses assumed equilibrium water uptake rather than measuring a molecule.
Worked example: compare two growth factors
Suppose at the same nominal RH of 90%, two hypothetical internally treated populations have GF_A = 1.20 and GF_B = 1.60. Under the simplified no-Kelvin equation, κ_A ≈ (1.728−1)/9 = 0.0809 and κ_B ≈ (4.096−1)/9 = 0.344. The growth-factor difference is 0.40, but the approximate κ ratio is 0.344/0.0809 ≈ 4.25. This contrast is another reminder that the inferred relationship is nonlinear.
Neither sample is proved to be a unique material. One could contain soluble salts mixed with organics, the other different inorganic fractions, or the growth patterns could reflect changes in particle phase. The number of meaningful decimal places should reflect size and humidity uncertainties; quoting κ_B = 0.34400000 falsely implies a precision not supported by the simplified model. State the temperature, dry diameter and inversion method if the estimate will be used beyond a teaching exercise.
Growth-factor uncertainty begins with two diameters
For GF = D_wet/D_dry, independent small relative diameter uncertainties combine approximately in quadrature. Imagine D_dry = 100 ± 1 nm and D_wet = 150 ± 2 nm, with independent standard uncertainties for this example. GF = 1.50. Relative uncertainty is sqrt[(2/150)² + (1/100)²] ≈ 0.0167, so the standard uncertainty in GF is about 0.025. A defensible simplified report is GF = 1.50 ± 0.03 after sensible rounding.
This excludes RH accuracy, DMA voltage and flow calibration, transmission width, charge-state correction, instrument inversion and non-spherical shape effects. The more complete budget may be much larger. If dry and wet sizing share calibration standards, errors can be correlated, invalidating the naive independence assumption. The right way to translate a published ± symbol is to identify whether it represents a standard deviation across a population, the standard error of a fitted mean, an expanded calibration uncertainty or model spread.
Why humidity uncertainty can dominate interpretation
At high RH the relation between particle water uptake and water activity may be steep. A shift from 89% to 91% RH can cause a meaningful difference in GF for strongly hygroscopic particles, even if the displayed controller appears stable. The relevant uncertainty is not just sensor accuracy but spatial RH uniformity, temperature, residence time and equilibration. Water activity is temperature-dependent through saturation vapour pressure, so a small thermal gradient can be important.
Suppose two teams report GF 1.47 and 1.52 but one conditions at 88% RH and the other at 92%, using different residence times. Calling the higher-GF sample more soluble is premature. One should compare measured humidograms at matched RH, propagate instrument and sample uncertainty, and assess whether the difference survives correction. If the original paper does not provide RH conditions, do not silently assume 90% because that setting is common. Mark the comparison as incomplete and seek methods.
Deliquescence is a transition, not a synonym for all growth
For some crystalline soluble particles, deliquescence is the transition to an aqueous solution at a characteristic humidity under relevant conditions. Below that transition, a solid particle may take up relatively little water compared with its post-deliquescence state. Above it, the particle can grow markedly. However, real mixed ambient aerosol can show broadened or suppressed transitions, partial dissolution, coatings and metastable behaviour. A sharp step in a laboratory humidogram should not be mechanically expected in every field aerosol mixture.
The reading procedure is to inspect the humidity path and any discontinuity or change in slope, then consider whether the material and equilibrium conditions support a deliquescence interpretation. A sudden change could also result from a humidifier-control step, a calibration issue, instrument lag or artefact. The fact that a steep segment exists in a plot is not itself a chemical identification. Pair it with controlled standards, chemical evidence and repeated up-and-down scans whenever possible.
Efflorescence and hysteresis need the scan direction
Efflorescence refers to crystallization or loss of the aqueous state as humidity falls under suitable conditions. Because nucleation can be delayed, a particle may remain liquid at a lower RH than the humidity where it first deliquesced on the upward scan. The resulting hysteresis means that growth at 70% RH, for example, may differ depending on whether the aerosol arrived there from a drier or wetter history. A chart that combines the two directions without labelling them can erase the mechanism.
Consider a hypothetical particle showing GF 1.30 at 70% RH on humidification and GF 1.45 at the same RH on dehumidification. It would be wrong to average the values as though the RH instrument were merely noisy. They may represent different phase trajectories. This is a diagnostic hypothesis, not proof: repeatability, temperature, residence time and possible sample aging must be checked. The explanatory text should preserve scan direction and particle conditioning so another team could reproduce the comparison.
Why one dry size is not necessarily representative
Selecting 50 nm dry particles and selecting 200 nm dry particles may probe different chemical populations. Atmospheric composition can depend strongly on size because primary emissions, nucleation, condensation and coagulation populate different modes. Surface-to-volume ratio and Kelvin effects also change with diameter. Thus an HTDMA mean measured at one size is not automatically a bulk hygroscopicity of all aerosol, and a site comparison at different dry selection diameters may produce a misleading trend.
A responsible presentation shows dry-size bins, their corresponding number fractions where relevant, and which growth modes occur at each size. If a paper reports one representative size because of time or instrument limitations, its conclusion should remain appropriately scoped. A single 100 nm result might still be valuable for a question about accumulation-mode particles, but it cannot replace an entire size distribution. The aerodynamic particle sizing guide adds the related caution that aerodynamic and electrical mobility diameters are different metrics.
Multiple-charge artefacts can imitate a low-growth population
A DMA selects mobility, not charge-independent geometric diameter. A larger multiply charged particle may be transmitted at the mobility setting nominally associated with a smaller singly charged particle. When humidified, such particles may appear in unexpected wet-diameter locations. Some reported growth-factor spectra contain peaks below one or other physically puzzling features because of charge contamination, inversion choices or morphology changes rather than evaporation of every particle. This is why an apparent GF less than one should be inspected, not immediately given a chemical explanation.
The remedy starts with a measured or modelled charge distribution and DMA transfer functions, followed by inversion or correction appropriate to the instrument. Compare the raw mobility spectrum, classification setting, assumed particle size distribution and corrected growth-factor distribution. A minor peak with only a few counts deserves a different confidence statement from a reproducible high-count mode. An annotated interpretation may say possible multiple-charge contribution while leaving room for genuine physical shrinkage or volatility when independent measurements support it.
Particle losses and changing concentration
Diffusion losses in tubing can remove small particles, while wall deposition, leaks, electrical effects and transport inefficiencies can change what reaches the second DMA. If a humidified branch loses more particles than the dry-reference branch, apparent modal proportions may shift. Losses may be size-dependent, so they cannot always be corrected by multiplying every bin by one constant. A rapid change in ambient aerosol concentration during a scan can also distort a distribution assembled sequentially across wet mobility settings.
An honest study documents line lengths, flows, residence times, dryers, humidifiers, neutralizers, CPC detection limits and correction assumptions. A lab note that merely states HTDMA GF = 1.4 without describing these elements may still contain a useful preliminary observation, but it is not enough for a quantitative cross-instrument intercomparison. Quality assurance is part of the evidence. Conversely, one small unresolved loss correction does not make an otherwise robust effect meaningless; the task is to estimate whether it affects the conclusion materially.
Worked example: a bimodal population
Imagine a hypothetical growth-factor distribution at 90% RH with 300 particles assigned to a weak-growth mode at GF 1.10 and 700 to a high-growth mode at GF 1.50 after valid inversion and equal detection efficiency. The number fractions are 300/(300+700) = 0.30 and 700/1,000 = 0.70. The number-weighted arithmetic mean GF would be 0.30 × 1.10 + 0.70 × 1.50 = 1.38. That number summarizes a distribution but should not replace the two-mode description.
If a reader now cubes 1.38 to infer the average individual volume growth, a subtle error arises. The mean of GF³ is 0.30 × 1.10³ + 0.70 × 1.50³ = 0.3993 + 2.3625 = 2.7618. In contrast 1.38³ ≈ 2.628. The difference reflects nonlinearity: the cube of a mean is not generally the mean of cubes. This is a broadly transferable mathematical lesson in environmental data translation. State what averaging operation was used, since different derived means support different physical interpretations.
A humidity scan is also a time series
A humidogram often changes RH gradually or in steps, and the instrument repeatedly records the growth distribution. That curve can reveal transitions, gradual uptake, hysteresis and instrumental drift. But the aerosol population may also evolve while the humidity sweep proceeds. For an outdoor sample, wind direction or emissions may change between the beginning and end. For laboratory-generated particles, generator concentration, drying efficiency or chemical aging may drift. Separating RH response from time variability requires controls.
The simplest diagnostic is to display ambient number concentration and, where feasible, a repeated reference-RH point during the sequence. If the first and last measurements at the same RH differ substantially, the reader should check whether the sample changed or the instrument drifted. Randomizing RH steps is not always practical, but repeating part of the sweep or documenting time alignment can reveal ambiguities. Do not treat every smooth humidogram as incontrovertible equilibrium thermodynamics if the underlying experiment was not stationary.
From subsaturated GF to optical scattering growth
A particle that absorbs water may scatter more light, but optical changes depend on particle size, refractive index, shape and wavelength. The size-resolved diameter growth recorded by an HTDMA is not the same as the humidified-nephelometer quantity f(RH), commonly a ratio of wet to dry aerosol scattering coefficients. A particle population’s optical scattering can change nonlinearly as it grows, and a small number of larger particles can contribute disproportionately to light scattering. Therefore GF and f(RH) should not be reported as interchangeable numbers.
NOAA’s aerosol humidifier description explains a system that conditions aerosol and compares scattering at different humidities. To link HTDMA measurements to optical response quantitatively, a study may combine size distributions, refractive indices and Mie or other appropriate optical calculations. If such modelling has not been performed, the honest statement is that enhanced water uptake may influence scattering, with the actual optical factor still to be determined. This avoids turning one laboratory property into an unverified visibility prediction.
A defensible κ mixing approximation
For an idealized internally mixed particle, one may approximate effective κ as a dry-volume-weighted combination of component κ values. Suppose a hypothetical particle comprises 60% dry volume of a constituent with κ = 0.50 and 40% of one with κ = 0.05. The idealized average is 0.60 × 0.50 + 0.40 × 0.05 = 0.32. This does not imply that 60% of particles are of the first type. It describes the components within the same particle under the mixing assumption.
In an externally mixed population, where 60% of particles are each mostly the first constituent and 40% each mostly the second, averaging κ before calculating an activation curve can hide two modes. The correct procedure depends on the question and state of mixing. Number fractions, dry-volume fractions and mass fractions are different denominators. A formula written without the type of fraction invites an error that can propagate into growth predictions and CCN closure. Always ask what the weights represent.
Independent problem: a GF and its actual physical statement
Problem A. A dry selected particle class is centred at 75 nm, while the corresponding wet mode is 99 nm at controlled 85% RH. Calculate GF and diameter percentage increase. Solution: GF = 99/75 = 1.32; diameter increase = 24/75 × 100 = 32%. The exact statement is a growth factor of 1.32 at 85% RH for that dry selection, not a 32% increase in mass. If asked for an idealized spherical volume ratio, 1.32³ ≈ 2.30.
If a labelling error states the wet diameter as 0.099 micrometres, that is in fact 99 nm and does not change the result when converted correctly. But dividing 0.099 directly by 75 would produce a false growth factor. Record the original units, normalize them, compute the ratio and then restore the correct descriptive label. Such unit reconciliation is particularly important when a translated figure uses nanometres but a spreadsheet column uses micrometres.
Independent problem: simplifying κ cautiously
Problem B. At an assumed 90% RH and with Kelvin effects ignored for a teaching approximation, GF = 1.35. Estimate κ. Solution: 1.35³ = 2.460375, so κ ≈ (2.460375−1)/9 ≈ 0.1623. A sensible rounded approximate report is κ ≈ 0.16 under the stated simplified assumptions. If the particle is very small, the full Kelvin term and measurement uncertainties may shift the inferred result, so it should not be treated as an exact chemical constant.
Now change RH to 80% while keeping GF = 1.35 as a hypothetical measured value. The simplified relation uses (1−0.80)/0.80 = 0.25, yielding κ ≈ 1.460375 × 0.25 ≈ 0.365. The difference is not arithmetic inconsistency. It shows that GF must be interpreted together with the RH at which it was measured. If a single material produced these two combinations under equilibrium conditions, a more complete model check would be appropriate rather than accepting both inferred κ estimates uncritically.
Independent problem: averages versus modes
Problem C. A valid inverted growth distribution contains 250 particles at GF 1.00, 500 at GF 1.30 and 250 at GF 1.60. The number-weighted mean GF is (250 × 1.00 + 500 × 1.30 + 250 × 1.60)/1,000 = 1.30. But the distribution is trimodal, not uniform. It would be inaccurate to claim every particle grew by 30%, and it would be misleading to identify a single material using only the average.
For idealized spherical volume comparisons, calculate mean GF³: [250 × 1³ + 500 × 1.3³ + 250 × 1.6³]/1,000 = (250 + 1098.5 + 1024)/1,000 = 2.3725. The cube of the mean is 1.3³ = 2.197, a different quantity. Both calculations are reproducible and answer different questions. The denominator, weighting basis and nonlinear transformation belong in any concise technical caption.
Independent problem: reading a phase-transition graph
Problem D. During humidification a selected laboratory particle remains near GF 1.02 from 45% to 65% RH, then rises toward GF 1.42 by 80% RH. During dehumidification it stays near GF 1.40 at 70% RH and then shrinks markedly at lower RH. What is a plausible physical hypothesis? The answer is possible deliquescence on the upward path and delayed efflorescence on the downward path, with hysteresis. This is a hypothesis requiring material identification, repeat scans and RH calibration.
It would be wrong to assert the specific salt species merely from those numbers. Mixed salts, organics, aqueous metastability and instrument residence effects can produce complicated patterns. A strong answer also checks whether both scans were conducted at similar temperature and particle aging state, and whether an apparent jump could correspond to a change in humidity-controller response. Explain what the graph supports, what it suggests and what remains untested.
Independent problem: a suspicious sub-unity peak
Problem E. A growth-factor histogram shows a narrow peak near 0.78 and a broader peak near 1.35 for a supposedly nonvolatile salt mixture at 90% RH. List three investigations before interpreting the 0.78 peak as physical shrinkage. One should examine multiply charged particle transmission, wet-DMA transfer and inversion artefacts, and the original dry and wet calibrations. A fourth useful check is whether a volatile or metastable component really could have been lost during conditioning.
The number of counts and reproducibility matter. A few points close to the detection limit may not deserve the same interpretation as a stable mode. The report should show raw versus inverted spectra, instrument settings and any background subtraction. Translation should preserve the distinction between a corrected apparent distribution and an established atmospheric material process. In unfamiliar datasets, surprising physical claims require stronger, not weaker, quality control.
Reporting a reliable HTDMA result
A clear figure caption includes selected dry mobility diameter, the actual RH and temperature of wet conditioning, whether the scan was at fixed RH or varying RH, count or probability normalization, charge and transfer-function assumptions, and whether any modes were excluded. Describe the observed GF distribution rather than only a mean if multimodality is important. Where possible, provide a calibration check using known laboratory aerosol and report the uncertainty in RH and sizing. Name any model used to transform GF into κ.
Keep the science separate from a source story. An HTDMA can show that a class of aerosol has different water-uptake behaviour at different sizes or times. It does not by itself prove a particular emissions source or uniquely diagnose toxicity. Source attribution requires composition, meteorology, transport and other evidence. The strongest explanation is specific enough to support decisions yet modest enough to respect the measurement’s limits.
Preserving the scientific vocabulary across languages
The terms electrical mobility diameter, dry selected diameter, wet mobility diameter, growth factor, hygroscopicity, deliquescence, efflorescence, hysteresis, internal mixing, external mixing, number weighting and inversion are not stylistic substitutes for each other. A translation that replaces growth factor with water content turns a diameter ratio into a different observable. A translation that calls every humidity-driven change deliquescence overlooks gradual uptake and hysteresis. A translation that calls a mobility diameter aerodynamic diameter confuses instruments.
A useful method is to define each symbol once, preserve its unit and denominator, then test whether a translated sentence would still tell a scientist which variable to plot. The semantic crosswalk article provides a general framework for mapping technical meanings. In HTDMA work this is especially helpful when source papers switch between GF, g(RH), κ, a histogram mode and a volume-derived quantity in adjacent paragraphs.
A final practical audit
When handed an HTDMA table, first find dry selection size and humidity. Then reconstruct GF from wet and dry diameters and check that the ratio is dimensionless. Read the full distribution and look for multiple modes before accepting one average. Inspect charge treatment, DMA transfer-function inversion, RH calibration and conditioning time. If the report infers κ, locate its equation, Kelvin assumptions and temperature. If it relates HTDMA growth to CCN or optical scattering, identify the additional modelling step.
Finally ask what observation would refute the proposed explanation. If the claim is a real hysteresis loop, repeat humidity ramps should reveal a directional difference. If the claim is an externally mixed aerosol, size-resolved growth distributions and independent chemical evidence should support it. If the claim is that two sites differ in hygroscopicity, the comparison should survive matching RH, dry size, charge corrections and uncertainties. That is how a growth-factor result becomes an informative, reproducible statement rather than an impressive but ambiguous number.
Sources and additional reading
University of Manchester: Hygroscopic Tandem Differential Mobility Analyser
NOAA Global Monitoring Laboratory: humidity system and hygroscopic growth
Atmospheric Chemistry and Physics: κ-Köhler theory and effective supersaturation
Atmospheric Chemistry and Physics: subsaturated uptake and CCN potential