New Particle Formation Surges in Summer Over the Indo-Gangetic Plain, but Pollution Keeps It Hard to See

New particle formation, the atmospheric process by which vapours condense into molecular clusters that grow into measurable aerosol particles, turns out to be strikingly frequent during the summer months over the Indo-Gangetic Plain, one of the most aerosol-loaded regions on Earth. A year-round-style observational campaign at a suburban site in Sonipat, Haryana, roughly 50 kilometres upwind of Delhi, found that new particle formation events occurred on 30 of 35 valid measurement days in May and June 2023, an 86 percent event frequency. By contrast, during the winter and post-monsoon period only 20 of 73 valid days, or 28 percent, showed such events. The study, published in Atmospheric Chemistry and Physics, is the first systematic characterisation of size-resolved atmospheric ions and sub-10-nanometre particles at this site, and it reveals both vigorous particle production and serious obstacles to detecting it reliably.

The research team, led by scientists at the Indian Institute of Technology Delhi together with collaborators at the University of Helsinki, deployed a Neutral Cluster and Air Ion Spectrometer, known as NAIS, at the Centre for Atmospheric Sciences Atmospheric Observatory in Sonipat. The instrument measures naturally charged ions from 0.8 to 42 nanometres and, after unipolar charging, total particles from about 2.5 to 42 nanometres. A Scanning Mobility Particle Sizer extended coverage to larger diameters up to roughly 570 nanometres, and an automatic weather station recorded temperature, humidity, wind, pressure and radiation at five-minute resolution. The measurement campaign ran from May to December 2023, although instrument interruptions and precipitation-affected periods meant that 108 valid days were available for event classification, and only 26 days had concurrent NAIS and SMPS data suitable for combined analyses.

The choice of site matters enormously for what the data can show. The observatory sits amid agricultural land, with National Highways 44 and 34 crossing about 500 metres from the sampling inlet, industrial chimneys beyond the roads, residential areas with diesel generators and household emissions 300 to 400 metres away, and the Yamuna River within about 10 kilometres. This mix of traffic, agricultural, industrial and regional sources produces a condensation sink, the rate at which pre-existing particles scavenge condensable vapours and freshly formed clusters, that ranks among the highest found anywhere. In such an environment, whether a nascent molecular cluster survives long enough to grow into a climate-relevant particle depends on a delicate balance between vapour supply and loss to the enormous background aerosol population.

The seasonal signal in the particle data was unambiguous. During May and June, daytime median concentrations of particles in the 7 to 12 nanometre range reached roughly 3,000 to 5,000 per cubic centimetre, two to three times higher than nighttime levels, with upper percentiles exceeding 8,000 per cubic centimetre. Particles in the 3 to 5 nanometre range, the freshest reliably measurable fraction, showed daytime medians near 3,000 per cubic centimetre, exceeding nighttime values by a factor of about three to three and a half. Growth typically began early, around 06:00 local time, consistent with strong photochemical activity and effective dilution in the hot, dry pre-monsoon boundary layer. Through August and September the daytime enhancement weakened and became more episodic, and by October through December the distributions flattened, with November and December sometimes showing slightly higher nighttime than daytime concentrations, a signature of background accumulation dominating over local production.

Ion concentrations told a different story. Naturally charged particles in the 2 to 4 nanometre range remained low in absolute terms, generally between 5 and 12 per cubic centimetre, with only modest daytime enhancement during the warm months, day-to-night ratios of about 1.5 to 2.5. Unlike the nucleation-mode particles, ions varied weakly across seasons and occasionally showed higher nighttime medians in late autumn and winter. This decoupling between ion abundance and particle growth suggests that the availability of charged clusters is not the primary constraint on new particle formation in this region. Instead, the authors argue, precursor vapour abundance and the condensation sink control whether freshly formed particles survive and grow, an interpretation supported by the presence of transported event days and the complete absence of clear ion bursts in the dataset.

Quantifying growth rates proved to be a challenge in itself. Using two independent methods, a Gaussian-fitting maximum-concentration approach and a sigmoid-fitting appearance-time method, the team estimated how fast the growing particle mode swept through successive diameter bins. In the nominal 3 to 7 nanometre range, median growth rates were 10.6 nanometres per hour by the appearance-time method and 14.7 nanometres per hour by the maximum-concentration method. In the 7 to 20 nanometre range, the corresponding medians rose to 16.8 and 19.0 nanometres per hour. This clear size dependence, with faster growth at larger diameters, points to a rich supply of condensable vapours in the anthropogenically influenced boundary layer. Compared against a global compilation of growth rates from remote, marine, rural and urban environments, the Indo-Gangetic Plain values sit at the upper end of urban and polluted regional sites, well above most background locations, and are comparable to the 5.7 to 30 nanometres per hour previously reported for Delhi.

The condensation sink emerged as the decisive discriminator between event and non-event days. On the 9 event days and 15 non-event days with concurrent NAIS and SMPS measurements, the median morning condensation sink, calculated over the 06:00 to 12:00 window when nucleation typically begins, was 0.0095 per second on event days against 0.0292 per second on non-event days, a difference the authors report as statistically significant with a Mann-Whitney U test p value of 0.00435. Event-day values fall near the cleaner end of the urban spectrum, comparable to springtime daytime conditions in Delhi, while non-event values approach the moderately polluted regime. For context, previous work in Delhi found events typically occur when the sink stays below about 0.06 per second, Beijing studies report a threshold near 0.03 per second, and clean boreal forest sites sit orders of magnitude lower, around 0.001 to 0.01 per second. A detailed case study from 5 to 10 October 2023 illustrated the mechanism: the well-defined event on 7 October coincided with north-westerly flow, moderate winds, low relative humidity near 35 percent, a temperature near 30 degrees Celsius, and a morning condensation sink of 0.0093 per second, while surrounding days with persistently elevated sinks showed no nucleation-mode growth.

Perhaps the most consequential finding concerns methodology. The team applied three complementary classification frameworks: the widely used visual scheme of Dal Maso and colleagues from 2005, the automated ion-and-particle-based method of Dada and colleagues from 2018, and the recent nanoparticle ranking analysis of Aliaga and colleagues from 2023, which assigns each day a continuous intensity metric based on the daytime-minus-nighttime difference in 3 to 5 nanometre particle concentration. The visual scheme yielded 10 Class I events, 40 Class II events and 58 non-events across the 108 valid days, with a striking 37 percent of days falling into the ambiguous Class II category. The automated Dada method consistently assigned more days to active categories, capturing transported and partially developed events that visual inspection placed elsewhere, and it flagged a notable fraction of transported events in August through October, suggesting advection of particles or precursors from upwind regions rather than purely local formation.

Each framework carries weaknesses that pollution amplifies. Visual classification depends on observer judgment, colour scales and plotting choices, producing lower agreement in murky environments. The automated method relies on 2 to 4 nanometre ion signals that high condensation sinks and anthropogenic loading can mask entirely. The ranking approach avoids categorical ambiguity but cannot distinguish a genuine nucleation burst from a traffic-related spike in sub-5-nanometre particles, a serious concern at a site beside two highways. Pollution plumes added extra modes to the size distributions that overlapped with, shouldered, or completely suppressed the growing nucleation mode, and wind shifts fragmented the signals. On some days, such as 17 August 2023, the new-particle signal was strong enough that a growth rate could still be extracted; on others, like 15 October 2023, wind shifts and pollution spikes distorted the distribution so badly that no reliable growth rate was possible. The authors conclude that none of the three methods alone is sufficient and that careful day-by-day screening remains necessary even alongside automated tools.

The broader implications reach toward climate and air quality. New particle formation followed by growth is a major global source of atmospheric aerosol number and, when particles grow large enough, of cloud condensation nuclei. The rapid growth rates observed here could raise the survival probability of freshly formed particles and, if the growing modes reach larger diameters, contribute to particle volume and mass. However, the authors caution that robust quantification of any contribution to particle mass or cloud condensation nuclei requires longer concurrent NAIS-SMPS measurements, supporting observations of precursor gases such as ammonia, sulfur dioxide and organic vapours, and verification that the measured growth rates persist over the hours needed for particles to reach cloud-relevant sizes. What the study establishes firmly is that in the Indo-Gangetic Plain, the atmosphere produces new particles vigorously when the scavenging environment briefly relaxes, and that seeing this process clearly demands as much methodological care as measurement hardware.

Subject of Research: Seasonal new particle formation and sub-10 nm ion and particle dynamics in the Indo-Gangetic Plain

Article Title: Ion and particle size distribution measurements in the Indo-Gangetic Plain: seasonal insights and challenges in new particle formation event characterization

Article References: Srivastav, G. K., Wagay, A. A., Lampilahti, J., Rathore, J., Kunchala, R. K., Ganguly, D., Kulmala, M., Petäjä, T., Paasonen, P., Thakur, R. C., & Gani, S. (2026). Ion and particle size distribution measurements in the Indo-Gangetic Plain: seasonal insights and challenges in new particle formation event characterization. Atmospheric Chemistry and Physics, 26(20), 14917-14938. https://doi.org/10.5194/acp-26-14917-2026

Image Credits: AI Generated

DOI: 10.5194/acp-26-14917-2026

Keywords: new particle formation, Indo-Gangetic Plain, aerosols, condensation sink, NAIS, atmospheric ions, particle growth rate, air pollution, Delhi, cloud condensation nuclei, NAIS classification methods, suburban observatory