Quantifying Light-Absorbing Aerosol Snow Darkening Using Cryogenic Snow Generation and Integrating Sphere Spectrophotometry

1. Introduction

The deposition of light-absorbing particles (LAPs) onto the cryosphere triggers a powerful snow-darkening effect by dominating visible broadband albedo reductions [1], exerting a baseline global annual mean surface radiative forcing of up to +0.054 W m⁻² [2,3]. Modern deposition profiles demonstrate an increasingly complex LAP burden on global snowpacks characterized by highly diverse aerosol mixtures. This includes black carbon concentrations in mid-latitude regions that exceed polar baselines by up to two orders of magnitude, driving nearly 20% of the albedo reduction across the Tibetan Plateau [4]. Biomass and fossil fuel burning produce plumes of smoke that can travel thousands of kilometers before depositing light-absorbing carbon on snow and ice [5]. Furthermore, this burden is compounded by mineral dust, which dominates high-altitude snow darkening over High Mountain Asia [6] as massive storms from the Taklamakan Desert persistently darken surrounding mountain snowpacks across multiple seasons [7]. This complexity is exacerbated by the emergent accumulation of microplastics across the Arctic, Antarctic, and alpine cryosphere [8]. Despite these substantial regional impacts, the Intergovernmental Panel on Climate Change (IPCC) [9] notes that most CMIP6 climate models do not explicitly represent the darkening of snow by the deposition of this complex aerosol mixture, leading to a systemic underestimation of global radiative forcing.

Beyond this broad mixture of aerosols, the specific parameterization of carbonaceous particles constitutes a major source of uncertainty, primarily due to the intricate brown-black continuum of light-absorbing combustion aerosols [10]. Along this continuum, the imaginary part of the refractive index (k) varies by orders of magnitude, progressing from weakly absorbing light brown carbon ($k=0.004$ at 550 nm) to strongly absorbing dark brown carbon (d-BrC; $k=0.25$ at 550 nm) depending on combustion conditions. Recent aircraft measurements and global assessments have revealed the ubiquity of d-BrC in wildfire plumes, demonstrating that its mass absorption cross-section (0.5 to 1.5 m² g⁻¹ at 500 nm) and overall shortwave absorption frequently rival or exceed that of black carbon, challenging the traditional paradigm of black carbon absorption dominance [11,12]. Consequently, the deposition of d-BrC has emerged as a potent, yet critically under-evaluated, radiative forcing agent in global snowpacks. A recent modeling study [13] established that d-BrC deposition alone can increase annual mean snow radiative forcing by 0.6 to 17.9 W m⁻², a 1.6- to 2.1-fold enhancement over scenarios considering only black carbon deposition. Given that current models struggle to capture these dynamics, He, 2022 [14] asserts that “more measurements of BrC optical properties and concentrations in snow are needed” to effectively constrain model estimates. This explicit gap necessitates the development of controlled empirical methodologies capable of precise laboratory deposition.

While robust methodologies have been developed for laboratory sea-ice generation [15,16], the controlled synthesis of nature-identical snow presents significantly greater logistical challenges. Conventional approaches to actual snow generation typically necessitate resource-intensive, walk-in cold rooms to house massive experimental chambers, such as those requiring 24-hour prolonged crystal growth times to produce merely a few kilograms of snow [17,18]. Conversely, techniques circumventing indoor facilities rely heavily on naturally freezing outdoor environments, tethering experimental feasibility to unstable ambient temperatures and weather conditions [19]. Consequently, attempts to study aerosol deposition on snow often suffer from a lack of strict microphysical control; for instance, recent efforts to quantify the effects of brown carbon deposition were constrained by their reliance on outdoor natural snowpacks, where the lack of environmental containment limits precise optical baselining [20].

To overcome these limitations, the field desperately requires a simple, low-footprint method to simultaneously generate snow and conduct precise, controlled aerosol deposition within a standard, ambient-temperature laboratory. The direct precursor to our apparatus was introduced by Hadley and Kirchstetter, 2012 [21], who utilized a liquid nitrogen-based methodology to synthesize snow doped with black carbon. By spraying an aqueous aerosol suspension into a 1.3-m-tall insulated chamber cooled to -100 °C, their technique inherently trapped the aerosols within the ice matrix during the phase change. This internal freezing mechanism fundamentally differs, both physically and optically, from true atmospherically relevant dry deposition, wherein aerosols are deposited onto the surfaces of existing snow grains. This is especially relevant as dry deposition flux has been found to be more than five times greater than wet deposition flux [22]. Building upon this foundation, our custom-built, low-footprint apparatus overcomes these spatial and logistical friction points. By facilitating true aerosol dry deposition onto laboratory-generated snow, the system enables highly controlled optical investigations into the radiative forcing of light-absorbing aerosols without the need for expansive cold rooms or outdoor environments.

2. Materials and Methods

2.1. Laboratory Snow Generation and Aerosol Deposition

To precisely study the radiative impacts of light-absorbing aerosols on snowpack in a controlled setting, we developed a custom-built laboratory snow generation and deposition apparatus. The system simulates natural snow accumulation by flash-freezing individually atomized water droplets during gravitational settling. The primary component is a 4-foot-tall, thermally insulated styrofoam snow generation chamber. Liquid nitrogen is introduced into the bottom of the chamber to establish cryogenic ambient temperatures through evaporation, which are tunable down to $-120$ ^∘C. Water is injected into the top of the chamber from a pressurized liquid dispensing tank through an ultra-fine fogging nozzle (ITW Vortec, 60-200 $\mu$m droplet size). The tank pressure is regulated by a step-down pressure regulator, allowing the operator to adjust the liquid feed pressure (typically maintained around 10 psi). To further control the atomized droplet size, a flow control valve is used between the tank and the nozzle. Inside the air-powered nozzle, the pressurized water flow is sheared apart into fine droplets by compressed air (40 psi) at a maximum spray angle of 21°. The compressed air flow through the nozzle also serves to circulate the air within the chamber and increase the liquid nitrogen evaporation rate by preventing pooling at the bottom of the chamber. As the droplets descend through the cryogenic environment, they rapidly freeze into granular ice crystals that accumulate at the base of the chamber, closely mimicking the physical properties of fresh natural snow. A schematic overview of the entire snow generation apparatus is presented in Figure 1a.

Following snow synthesis, snow is collected into multiple sample holders. A flat surfaced putty knife is used to compact the snow in the sample holder and spread the surface evenly. Aerosols are deposited onto the snow surface in a controlled manner. For this study, we focused on Cabojet 200 (Cabot Corporation) and dark brown carbon (d-BrC) aerosols. Cabojet 200 is a BC-like material that has been used for BC analysis in ice and snow [23,24], and is deposited on snow samples for system validation and limit of detection calculations. Cabojet 200 suspensions in water of varying dilution ratios are aerosolized using a Collison nebulizer. D-BrC is deposited on snow samples as a case study to demonstrate an application of the system to a less well-known particle. D-BrC aerosol was generated via the piezoelectric nebulization of liquid tar suspension followed by a thermal heat shock in a tube furnace [25]. The aerosol stream is subsequently cooled to prevent localized melting of the snowpack upon impaction. Dry deposition of the d-BrC tarballs onto the snow surface is achieved using an inverted funnel enclosure that ensures spatially homogeneous deposition (Figure 1b). An SMPS was inserted in parallel to measure particle size distribution. D-BrC aerosol was nebulized at a flow rate of 0.2 LPM and diluted by a factor of 20 using nitrogen gas. The total deposition time (ranging from 2 to 4 minutes) is tuned to vary the surface mass loading of the light-absorbing particles across different samples. During deposition, the snow sample holder is held in a bath of evaporating liquid nitrogen to ensure the snow remains frozen, and a slight air gap is maintained between the inverted funnel and sample holder to allow residual air to flow out.

2.2. Integrating Sphere Reflectance Spectrophotometer

We utilized a spectrophotometric technique based on an integrating sphere to quantify the spectral albedo reduction of the contaminated snow. The optical system centers on a 6-inch diameter integrating sphere (RT-060-SF, Labsphere Inc.) lined with Spectraflect, a highly reflective Lambertian coating (Figure 1c). The sphere is illuminated by a collimated halogen light source capable of a maximum 10 W output. The source is driven by a programmable DC power supply controller, which allows for precise regulation of the output current and target set points.

Snow samples are collected in custom-designed 3D-printed cylindrical holders with a depth of approximately 5 cm, which exceeds the ultraviolet and visible light penetration depth for snow, thereby ensuring semi-infinite optical thickness. The sample holder is mounted flush against the measurement port of the integrating sphere. The measurement port is located opposite to the collimated light source, with the rays reflecting off the surface of the sample undergoing multiple reflections within the integrating sphere and being distributed uniformly over the interior surface of the sphere. The input light is spatially integrated, hence the name “integrating” sphere. A portion of this spatially integrated light enters a detector port where a spectrometer via fiber optic cable records the intensity of the reflected light. The ratio of the spectral intensity measured for a sample to that measured for a known reference can be used to calculate the spectral albedo for the sample [26]. Spectral reflectance is measured over the 350–1000 nm wavelength range using a high-resolution spectrometer (Flame-S-VIS-NIR, Ocean Optics). Measurements were made under the single beam measurement condition with the reference and sample physically swapped at the same port. Typically, this results in single-beam substitution errors due to the variation in the sphere’s throughput between the reference and the sample. However, because the samples have a high albedo (>0.85) that are close to the reference albedo, this error is negligible, allowing the spectral albedo of the snow to be derived directly from the ratio of the measured intensities.

2.3. Optical Microscopy and Microphysical Characterization

In parallel with the bulk optical measurements, we performed microphysical characterization of the synthesized snow grains and deposited aerosols. A thin layer of the freshly generated snow was collected on a pre-chilled, improvised aluminum block designed to prevent rapid melting. The snow grains were imaged using a high-magnification digital light microscope. These microscopic observations confirmed that the laboratory-generated snow features dendritic structures and grain diameters of approximately 100 $\mu$m, consistent with observations of natural fresh snow. Furthermore, the microscopy provided visual confirmation of the homogeneous distribution and mixing state of the dry-deposited d-BrC tarballs across the snow surface.

3. Results and Discussion

3.1. Optical Validation Using Cabojet

To rigorously validate the snow synthesis, aerosol deposition, and optical measurement system, initial experiments were conducted using a standardized, highly absorbing BC proxy (Cabojet). Spectral reflectance was measured for pristine laboratory-synthesized snow and snow doped with sequentially diluted aerosol suspensions (100x dilution all the way up to 100000x dilution). To isolate the physical optical signature from high-frequency instrumental noise, the raw, uncut directional reflectance arrays were processed using a Savitzky-Golay filter (window length = 151, polynomial order = 3) and truncated strictly to the 400–800 nm visible spectrum to avoid near-infrared artifacts from the halogen light source.

Because the spectrometer measures directional reflectance, which can naturally exceed 1.0 in forward-scattering geometries due to the faceted nature of ice crystals, the experimental spectra were dynamically scaled by a constant factor of 0.98. This scaling successfully mapped the directional observations onto the hemispheric albedo bounds (maximum 1.0) utilized by radiative transfer models.

To achieve optical closure, the experimental curves were compared against theoretical hemispheric albedos simulated by the Snow, Ice, and Aerosol Radiative (SNICAR-ADv4) model [27,28] (Figure 2a). Snow samples, which were 5 cm thick, were assumed to consist of two layers: a top layer with aerosol homogenously mixed, and an underlying layer of pristine snow. In the model, the physical structure of the snowpack was locked across all samples to a realistic compacted matrix of snow grains (density 200 kg m⁻³, effective grain radius 50 µm, top layer thickness 10 mm), allowing only the equivalent black carbon (BC) mass concentration to vary.

The model successfully reproduced the empirical spectral shapes with high fidelity. The 1000x and 100x dilution curves matched SNICAR outputs for 100 ppb and 1000 ppb of BC, respectively, yielding an average joint root-mean-square error (RMSE) of under 0.022. The distinct visible-spectrum curvature (400–600 nm) indicative of carbonaceous absorption is perfectly captured. Furthermore, the tenfold increase in modeled mass loading (100 ppb to 1000 ppb) directly mirrors the tenfold sequential dilution of the source suspension. This optical closure confirms that the deposition system operates with high predictability and linearity, and that the laboratory-generated snowpack responds to radiative forcing in accordance with established physical theory.

3.2. Limit of Detection (LOD) and System Boundaries

Having established optical closure, the boundaries of the measurement system were quantified. For atmospheric and cryospheric applications, integrating spectral data into a single broadband albedo metric is necessary to evaluate total shortwave energy balances. Consequently, the individual measured spectral albedos were collapsed into visible broadband albedos (VBA) by weighting the spectra by the incident irradiance fraction for a standard mid-latitude summer solar spectrum.

To calculate the Limit of Detection (LOD) of the optical measurement system, the $3\sigma$ uncertainty band (noise floor) of the pristine snow VBA was established. A regression of VBA versus logarithmic aerosol concentration was then performed using the standardized Cabojet-doped samples. As shown in Figure 2b, the LOD was determined at the intersection of the concentration regression and the pristine $3\sigma$ noise floor. The resulting BC-equivalent LOD is 164.5 ppb. This threshold indicates that the presented methodology is highly capable of detecting and quantifying the particulate darkening effect in moderately to heavily polluted snowpacks, establishing clear operational boundaries for future field-mimicking experiments.

3.3. Application to d-BrC Deposition

Following system validation, the methodology was applied to a complex, novel aerosol: nebulized liquid tar dark brown carbon (d-BrC). The d-BrC deposition was conducted in parallel with a Scanning Mobility Particle Sizer (SMPS) to empirically measure the aerosol size distribution and number concentration. By coupling the SMPS data with the deposition time and flow rate, precise surface mass loadings were calculated (468.4 ppb, 527.3 ppb, and 1028.1 ppb) assuming homogeneous mixing in the upper 10 mm of the snowpack. For the heaviest loading of 1028.1 ppb, the experimental visible broadband albedo dropped to 0.93 from a pristine albedo of 0.99, a decrease of 0.06.

Using these mass loadings, forward SNICAR simulations were performeds. As shown in Figure 3, while the heaviest deposition case (1028.1 ppb) showed reasonable agreement in the short-visible wavelengths, at lower deposition masses close to the limit of detection, modeled albedos diverged from experimental albedos, particularly at longer wavelengths ($>650$ nm).

This divergence suggests the necessity of the empirical methodology presented in this study. SNICAR and similar radiative transfer models assume ideal, spherical particles with neat external or internal mixing states embedded uniformly within an infinite ice matrix. The divergence of the model from the SMPS-derived d-BrC curves suggests that the morphological, mixing, and refractive properties of d-BrC are more complex than what standard parameterizations accurately predict.

3.4. Microphysical Evidence of Aerosol Deposition

Optical microscopy was performed on the snow samples (Figure 4) to observe snow grain shape and size, as well as aerosol deposition. The microscopy aligns with the structural assumptions used in the SNICAR modeling, including that pristine snow grains (Figure 4b) exhibit roughly spherical geometries on the order of 100 $\mu$m in diameter, supporting the $\mu$m effective radius parameter utilized in the optical closure.

However, images of the d-BrC deposited snow (Figure 4c and Figure 4d) reveal complex aerosol-snow interactions. The d-BrC particles do not deposit as independent, uniformly distributed spheres; rather, they exhibit clumping, pooling, and localized aggregation along ice crystal boundaries and air bubbles. These non-uniform microphysical changes during the dry deposition of d-BrC may not be adequately captured by standard radiative transfer models. Consequently, these images help underscore the critical need for robust, physical laboratory simulation chambers to empirically measure the true radiative forcing of complex carbonaceous aerosols.

4. Conclusions

In this study, we presented a novel, low-footprint laboratory apparatus designed for the cryogenic synthesis of nature-identical snow and the subsequent controlled dry deposition of aerosols. By coupling this generation chamber with an integrating sphere spectrophotometer, we achieved high-fidelity optical closure (average joint RMSE < 0.022) using a standardized carbonaceous proxy, establishing a visible broadband limit of detection of 165 ppb equivalent black carbon. Furthermore, the application of this system to nebulized d-BrC tarballs demonstrated that morphologically complex aerosols form localized aggregates and pools along ice crystal boundaries. These complex microphysical interactions drive nuanced spectral albedo reductions that deviate from the idealized spherical mixing states typically utilized in standard radiative transfer models, underscoring the necessity of empirical methodologies to accurately constrain aerosol radiative forcing.

Beyond static optical validation, the modular, multifunctional, and customizable design of this apparatus provides a robust platform for developing a process-based understanding of aerosol-cryosphere interactions. Future methodological refinements may focus on optimizing deposition flow rates and exposure times to mitigate surface cratering and localized melting during heavy dosing. Additionally, the controlled nature of the chamber enables high-resolution studies on the temporal evolution of snowpack optics, facilitating experiments on snow metamorphosis and grain evolution under varying relative humidity (RH) conditions and thermal cycling. Further microphysical investigations are also warranted to understand the precise structural and optical implications of LN²-soaked snow resulting from this specific cryogenic synthesis process.

This laboratory set up offers broad utility for the next generation of cryospheric research. As recent literature highlights the expanding diversity of light-absorbing impurities, this system is uniquely positioned to investigate the radiative impacts of established as well as emerging contaminants, including microplastics [29], snow algae [30], and other visible-light-absorbing pigments. Snow albedo geoengineering, which has recently attracted attention [31], may also be tested using this system before wider deployment. Furthermore, the lifecycle of these deposited particles can be comprehensively tracked beyond initial deposition through simulated meltwater analysis and subsequent re-aerosolization studies utilizing a Single Particle Soot Photometer (SP2). By providing an empirical bridge between complex aerosol morphologies and theoretical radiative transfer, this methodology supports a more accurate, holistic understanding of contaminant-driven cryospheric melt.