- October 30, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
A diesel generator emits a cloud of particles spanning a huge range of sizes, from visible soot flakes to particles so tiny that billions fit in a cubic centimetre of exhaust without any visible smoke. When a hospital, school or housing society asks about RECD ultrafine particles, the real question is what happens to that smallest end of the cloud.
The short answer: a well-designed wall-flow filter, with exhaust forced through porous ceramic walls, typically captures solid ultrafine particles very efficiently, often better than it captures particle mass. Other RECD technologies vary much more, from good to modest, depending on design and operating condition. And one category of ultrafine particle, the volatile droplets that form as exhaust cools and mixes with air, can appear after the device no matter how good the filter is.
This guide follows the particles from largest to smallest, comparing RECD technologies at each size band, then covers measurement, health evidence and buyer questions.
A size primer before we start the journey
Sizes are given in micrometres (µm) or nanometres (nm); one micrometre is 1,000 nanometres. A human hair is roughly 50 to 70 µm across.
The air-quality labels are cumulative cut-offs, not exclusive bins:
- PM10: particles up to about 10 µm. Includes everything smaller.
- PM2.5: up to about 2.5 µm. The fraction most Indian AQI discussion revolves around.
- PM1: up to about 1 µm.
- PM0.1, or ultrafine particles (UFP): below 100 nm.
If you want a fuller explanation of the first two labels, our guide on the difference between PM2.5 and PM10 covers them in everyday terms. For diesel exhaust, almost all of the particulate sits well inside PM2.5, and most of it inside PM1.
Three modes, two ways of counting
Researchers describe diesel exhaust particles in “modes”, groups that form by different routes:
- Nucleation mode: very small particles, typically below about 30 nm. Many are droplets of condensed sulphuric acid, water and heavy hydrocarbons (often from fuel and lubricating oil) that form as exhaust cools and dilutes. Some engines also emit a small solid core in this range.
- Accumulation mode: roughly 30 nm to 500 nm, peaking somewhere around 50 to 150 nm. These are mainly chains and clusters of carbon spheres (soot) with organic material and sulphate coated on them.
- Coarse mode: above about 1 µm. Usually soot flakes shed from exhaust pipe walls, rust and deposits, plus some ash.
Now the key point for this article. If you count particles, most of them, often the vast majority, are ultrafine: nucleation mode plus the smaller half of the accumulation mode. If you weigh particles, most of the mass sits in the accumulation mode, because one 200 nm particle weighs about as much as a thousand 20 nm particles of the same density (mass scales with diameter cubed, and 10 × 10 × 10 = 1,000).
So a test in milligrams per cubic metre and one in particles per cubic centimetre can tell two different stories about the same filter.
How a filter catches particles: the journey through the wall
A wall-flow filter is the clearest example of how capture works; the DieselNet overview of diesel particulate filters lists the same deposition mechanisms.
In a wall-flow design, alternate channels in a ceramic honeycomb are plugged at opposite ends. Exhaust enters an open channel, finds the far end blocked, and has to pass through the porous wall into a neighbouring channel to escape. Wall pores are typically 10 to 20 µm, far larger than the particles, so the filter is not a sieve. Three physical effects pull particles out of the gas instead.
Impaction and interception: the big particles’ problem
Inertial impaction: a heavy particle cannot follow the gas around a ceramic grain and hits it. Works best roughly above 1 µm.
Interception: a particle follows the streamline but passes close enough for its edge to touch a surface. Wider particles are intercepted more easily.
Diffusion: the small particles’ problem
Very small particles are jostled by gas molecules and wander off the streamline (Brownian motion), bumping into pore walls and sticking. This diffusion capture of small particles strengthens as size falls: a 10 nm particle diffuses far more than a 100 nm one.
A longer, more tortuous path through the pores helps diffusion, which is why forcing all exhaust through a wall works so well for nanoparticles.
The most penetrating particle size
Put the two trends together. Impaction and interception get weaker as particles shrink. Diffusion gets weaker as particles grow. Somewhere in between there is a size where neither mechanism is very strong, and particles of that size slip through most easily. Filtration scientists call it the most penetrating particle size.
For a clean filter, this weak spot typically falls somewhere in the low hundreds of nanometres, often quoted as roughly 100 to 300 nm, although the exact position depends on pore structure, wall thickness and gas velocity. It is the reason a brand-new, perfectly clean filter can show its lowest efficiency in the middle of the accumulation mode, not at the ultrafine end.
The soot cake changes everything
A clean wall works by depth filtration: particles are caught inside the pore network. Within a short time of operation, soot starts to build up as a thin layer on the inlet face of the wall. This soot cake has very fine pores of its own and becomes the main filter. Cake filtration is very efficient across the whole size range, so the dip at the most penetrating size largely flattens out.
The practical consequences:
- Right after regeneration or cleaning, efficiency can dip until a cake re-forms.
- The trade-off is back-pressure: the cake that improves capture also raises resistance, which is why regeneration exists. Our guide to why exhaust back-pressure matters for a DG set explains the engine side of that trade-off.
Above 1 µm: the coarse fraction
What these particles are
Particles above 1 µm are a tiny share of the count and a modest share of the mass. They are mostly soot flakes shed from pipes and silencers, rust and scale, and ash fragments: the black specks that land on neighbouring cars and walls.
How each technology tends to perform
| Technology | Typical behaviour above 1 µm |
|---|---|
| Wall-flow filter | Very high capture, by impaction and interception; few coarse particles reach the outlet |
| Partial-flow / flow-through filter | Good capture of the share that enters the porous media; the bypass share passes, and stored deposits can sometimes be released |
| Electrostatic (ESP-type) | Generally good collection; large agglomerates can re-entrain if collected deposits break off during cleaning or flow surges |
| Wet scrubber | This is where wet scrubbing is strongest; impaction on droplets works well for coarse particles |
| DOC only (oxidation catalyst, no filter) | Little or no physical capture; may reduce some coarse deposits indirectly by burning soluble organics |
Coarse particles are rarely the deciding issue. One caution: an old stack that carried untreated soot for years can keep shedding flakes for a while after an RECD is fitted.
100 nm to 1 µm: the accumulation mode
What these particles are
This band dominates particle mass. Primary soot spheres of typically 15 to 40 nm stick together into chains and clusters, coated with condensed organics and sulphate. Most of what a gravimetric compliance test weighs lives here.
How each technology tends to perform
Wall-flow filters. With some soot cake present, capture is typically very high; on a freshly cleaned filter it can dip briefly. Cracks, failed plugs or a leaking gasket show up first as lost capture here, so integrity matters more than nominal pore size.
Partial-flow filters. Only part of the exhaust passes through porous media, so capture is lower and more variable, depending on load, soot state and flow. Our sibling article on wall-flow versus partial-flow filtration in RECD systems compares the two in more detail.
Electrostatic designs. Performance depends on particle charging, field strength, residence time and how collected material is shed; many electrostatic collectors are weakest somewhere in this submicron range. See filter-based vs electrostatic RECDs for a fair comparison of what to check.
Wet scrubbers. Low-energy scrubbers tend to struggle here: particles are too small for impaction and too large for strong diffusion. High-energy designs do better at a cost in pressure drop and water handling. The trade-offs are set out in wet scrubbing vs dry filtration for DG retrofits.
DOC only. No meaningful soot trapping. Above light-off it can oxidise part of the organic coating, trimming mass but leaving solid cores.
Below 100 nm: solid ultrafine particles
What these particles are
This is PM0.1 proper: small soot agglomerates and primary particles, plus some solid nucleation-mode cores often linked to lube-oil ash. They carry little mass but much of the count.
This band drives most concern about nanoparticles in diesel exhaust: solid nanoparticles do not evaporate, and persist until they coagulate or deposit, including in the lungs.
How each technology tends to perform
Wall-flow filters. This is where wall-flow designs excel. Diffusion grows stronger as size falls, so solid particles in the tens of nanometres are captured very efficiently even on a fairly clean wall, and the soot cake makes capture better still. DieselNet notes that wall-flow filters are more efficient for elemental carbon and solid particle number than for total particulate mass, and that mass-based efficiency for total particulate is typically in a lower range than solid-number efficiency (DieselNet on wall-flow monoliths). Very high solid-number reductions are widely reported for intact wall-flow filters, but treat these as typical of the class, not a guarantee for a given RECD, engine or duty.
Partial-flow filters. Particles that enter the porous medium are captured by diffusion as well, but the share that stays in open passages is not. Number reduction is therefore lower and less predictable, and it can fall as the media loads up and more flow takes the open path.
Electrostatic designs. Very small particles are harder to charge; a fraction of particles below a few tens of nanometres may carry no charge at all at any moment, and uncharged particles are not collected. Some designs use agglomeration or diffusion-charging stages to help. Ask for measured particle number data rather than assuming.
Wet scrubbers. Diffusion to droplets does increase for the smallest particles, so a scrubber can perform better at 20 nm than at 300 nm. But overall solid number capture is usually well below that of a wall-flow filter unless the design is specifically engineered for it.
DOC only. No meaningful capture of solid ultrafines.
Why “particle number reduction filter” is a fair label for wall-flow designs
Wall-flow filters often reduce particle number by a higher percentage than mass, the opposite of what many assume. A buyer who sees only a mass result, such as the at least 70% PM capture the CPCB procedure requires, sees part of the picture; a device may do much better, or worse, on number.
Below 30 nm: volatile nucleation particles
What these particles are
Many nucleation mode particles are not emitted as particles at all. They leave as hot vapour (sulphuric acid, fuel and lube-oil hydrocarbons, water) and condense into new droplets of a few nanometres as exhaust cools and dilutes.
Whether nucleation happens, and how many particles form, depends on:
- How much condensable vapour is present (fuel sulphur, oil consumption, unburned hydrocarbons).
- How much existing particle surface is available. Soot particles act as a “sponge” that vapours condense onto instead of forming new droplets.
- Temperature, humidity and dilution ratio during cooling.
The filter paradox
An efficient filter removes almost all soot, and with it the surface vapours would condense onto. Remaining vapour then forms new droplets, so a well-filtered engine can sometimes show more volatile nanoparticles than before, even as solid emissions fall sharply.
Catalyst coatings can push this either way. An oxidation stage can burn off much of the hydrocarbon vapour above its light-off temperature, which reduces organic nucleation. At high exhaust temperatures, the same catalyst can oxidise some sulphur dioxide to sulphur trioxide, which forms sulphuric acid and can increase sulphate nucleation. That is one reason low-sulphur fuel matters for aftertreatment; see how ultra-low sulphur diesel improves emission control. The interplay is covered from the operating side in catalysed vs uncatalysed RECD filters.
How each technology tends to perform
| Technology | Volatile nucleation particles below 30 nm |
|---|---|
| Wall-flow filter | Cannot capture vapour; may raise nucleation by removing soot surface; catalysed versions reduce hydrocarbon precursors when hot, but can add sulphate at high temperature with sulphur-bearing fuel |
| Partial-flow filter | Similar vapour behaviour; remaining soot gives some surface for condensation |
| Electrostatic | Vapour passes through; droplets that form downstream are not collected |
| Wet scrubber | Can remove some soluble vapours, and cooling inside the scrubber can trigger condensation within the device; outcome depends heavily on design |
| DOC only | Reduces hydrocarbon vapour above light-off; may increase sulphate at high temperature |
The biggest levers on volatile nucleation lie outside the filter: low-sulphur fuel, an engine not burning oil, and stable warm operation. If you see blue haze, look at the engine first; our guide on how engine oil consumption affects RECD suitability explains why.
Measuring ultrafines: why the method decides the answer
For RECD ultrafine particles, weighing particulate on filter paper tells you almost nothing. To see ultrafines, you have to count them.
Particle number counters
A condensation particle counter (CPC) grows tiny particles into droplets large enough to count optically. Each counter has a lower cut-off size below which it misses most particles. Sizing instruments such as mobility spectrometers show the full distribution but are mainly research tools.
Cut-off sizes: 23 nm and 10 nm
European vehicle rules have long counted solid particles above roughly 23 nm, partly for repeatability; newer methods move toward about 10 nm. A “PN” result means little until you know the cut-off.
Volatile particle remover
To count only solid particles, standardised methods pass the sample through a volatile particle remover: hot dilution followed by an evaporation stage that drives off volatile droplets before counting. Without it, the count includes whatever nucleated in the sampling line, which can swing by orders of magnitude with small changes in conditions.
Dilution sensitivity
Nucleation depends on dilution ratio, temperature and humidity, so two labs can report very different total counts from one engine. Mass and opacity measurements have their own dilution pitfalls, covered in our article on how dilution can mislead before-and-after RECD measurements. Sampling point matters too.
What Indian RECD compliance measures
At the time of writing, Indian RECD type approval under the CPCB procedure for in-use DG engines up to 800 kW is based on particulate matter mass, with at least 70% capture efficiency required. There is no particle number or black carbon requirement for RECDs in that procedure. Check the Central Pollution Control Board website and your SPCB consent conditions for any changes. If you are reading a type approval report, our guide on how to read an RECD test certificate shows which figures to look for, and how opacity differs from particulate mass explains why “no visible smoke” is not the same as low particle emissions.
Black carbon is another separate metric, usually measured optically. Our companion piece on RECDs and black carbon explains what a particulate reduction claim does and does not include.
Worked example: number versus mass through a wall-flow filter
The figures below are illustrative only. They are chosen to show how mass and number percentages can diverge. They are not measurements from any specific RECD, engine or DG set.
Scenario A: solid particles, counted properly
A DG set at steady load is measured upstream and downstream of an intact, partly soot-loaded wall-flow filter. Particle number is measured with a volatile particle remover (solids only). Particulate mass is collected on filter paper and includes condensed volatile material.
| Metric | Upstream | Downstream | Reduction |
|---|---|---|---|
| Solid particle number (#/cm³) | 2.0 × 10⁸ | 2.0 × 10⁶ | 99% |
| Total particulate mass (mg/Nm³) | 80 | 16 | 80% |
| of which solid (soot and ash) mass | 64 | 1.0 | 98.4% |
| of which condensable / volatile mass | 16 | 15.0 | 6.3% |
Arithmetic, shown in full:
- Number reduction: (2.0 × 10⁸ − 2.0 × 10⁶) ÷ 2.0 × 10⁸ = 1.98 × 10⁸ ÷ 2.0 × 10⁸ = 0.99, or 99%.
- Total mass reduction: (80 − 16) ÷ 80 = 64 ÷ 80 = 0.80, or 80%.
- Solid mass reduction: (64 − 1.0) ÷ 64 = 63 ÷ 64 = 0.984, or 98.4%.
- Condensable mass reduction: (16 − 15.0) ÷ 16 = 1 ÷ 16 = 0.0625, or about 6.3%.
- Check: downstream solid plus condensable = 1.0 + 15.0 = 16 mg/Nm³, which matches the total.
The filter removed almost all solids by number and mass. The 80% total-mass figure is dragged down by condensable material that passes as vapour and condenses later on the cooler filter paper, so it understates solid ultrafine removal.
Scenario B: the same exhaust, counted without removing volatiles
Now suppose the downstream sample is counted without a volatile particle remover, through a cool dilution line on a humid day. Vapour that the soot used to soak up now nucleates into new droplets.
| Downstream count component | #/cm³ |
|---|---|
| Solid particles (unchanged from Scenario A) | 2.0 × 10⁶ |
| Volatile nucleation droplets formed in the sampling line | 3.0 × 10⁷ |
| Total counted | 3.2 × 10⁷ |
Arithmetic:
- Total counted: 2.0 × 10⁶ + 3.0 × 10⁷ = 0.2 × 10⁷ + 3.0 × 10⁷ = 3.2 × 10⁷ #/cm³.
- Apparent reduction against the solid upstream count: (2.0 × 10⁸ − 3.2 × 10⁷) ÷ 2.0 × 10⁸ = 1.68 × 10⁸ ÷ 2.0 × 10⁸ = 0.84, or 84%.
The filter has not changed; only the measurement has. Volatile droplets formed after the device now sit in the count, and the upstream and downstream figures are no longer like for like. On a different day the nucleation term could be ten times smaller or larger.
The two lessons for buyers: insist that before and after measurements use the same method, and always ask whether a particle number figure is “solid only” and at what cut-off size.
Ultrafine particle health: what is known, briefly
Ultrafine particles raise concern because they are numerous, have a large surface area for their mass and reach deep into the lungs. The evidence is still developing, and their effects are hard to separate from PM2.5 as a whole.
The World Health Organization’s 2021 global air quality guidelines did not set a numerical guideline value for ultrafine particles. Instead, they include qualitative good practice statements for managing ultrafine particles, black carbon and elemental carbon, as described in the WHO Q&A on its global air quality guidelines. Separately, the International Agency for Research on Cancer classified diesel engine exhaust as carcinogenic to humans (Group 1) in 2012.
For a hospital, school or housing society, a DG set often runs close to air intakes, windows and play areas, so cutting solid particles at source is worthwhile even when compliance reports only mass. Our pieces on planning RECDs around critical hospital backup loads and RECD purchase planning for RWAs apply this to specific settings.
Questions to ask about RECD ultrafine particles before you buy
If ultrafine particles matter to your site, these questions separate a well-grounded answer from a sales line:
- What filtration principle does the device use? Wall-flow, partial-flow, electrostatic, wet, or a combination? Each has a different size-band profile.
- Is all exhaust forced through the filtration medium, or does some pass through open channels or a bypass?
- Is there any particle number data? If so, was it measured on a DG engine, at what load, and with or without a volatile particle remover?
- What was the counter’s cut-off size? 23 nm, 10 nm or something else?
- Were upstream and downstream measured the same way, at the same load and dilution conditions?
- How is filter integrity checked over time? Ask about back-pressure trends and inspection intervals.
- Is there an efficiency dip right after regeneration or cleaning, and how long does it last?
- Is there a catalyst stage? If so, what fuel sulphur level is assumed, and what about sulphate at high load?
- What engine condition is assumed? High oil consumption adds volatile nanoparticles no filter can trap.
- Can performance claims be written into the order? Our guide to writing measurable RECD tender specifications shows how to phrase them.
Be wary of a single “ultrafine removal” percentage with no method attached, but do not dismiss a supplier who has only mass-based CPCB data; that is what Indian approval requires at present.
FAQ
Can RECDs capture ultrafine particles from a DG set?
Many do, but it depends on the technology. Intact wall-flow filters typically capture solid ultrafine particles very efficiently through diffusion. Partial-flow, electrostatic and wet designs vary more. Volatile droplets that form after the device are not captured by any filter.
Why can particle number reduction be higher than mass reduction?
Mass results include condensable organic and sulphate material that passes a filter as vapour. Solid particles, which dominate the count, are captured very efficiently. So a filter can show, for example, 99% number reduction and only 80% mass reduction in the same test.
What is the most penetrating particle size?
The size a filter captures least efficiently: too small for strong impaction, too large for strong diffusion. For a clean filter it typically lies in the low hundreds of nanometres; a soot cake reduces the weakness.
Does Indian RECD approval test ultrafine particles?
At the time of writing, no. Type approval under the CPCB procedure for DG engines up to 800 kW is based on particulate mass, with at least 70% capture. There is no particle number requirement for RECDs. Check CPCB and your SPCB for updates.
Can an RECD increase ultrafine particles?
It can increase volatile nucleation particles under some conditions, because removing soot leaves less surface for vapours to condense on. Solid ultrafine particles, the persistent kind, are reduced sharply by a good filter. Low-sulphur fuel and a healthy engine minimise the volatile effect.
Planning an RECD with Small Particles in Mind?
If your DG set runs near a ward, classroom or residential tower, ask how a retrofit handles RECD ultrafine particles and the whole size range, not just the certificate’s mass figure. Aceget’s product page describes its RECD as using a silicon carbide ceramic filter with a precious-metal coating and an oxidation catalyst stage, with up to about 96% particulate reduction depending on DG condition and fuel. Explore the Aceget RECD range, review our type approval and certification documents, or talk to our engineering team with your DG rating, run hours and site layout, and we will walk you through what to expect and how to verify it.