An open-source particle screening tool for Mars warming research
Accelerating iteration on Mars-warming particle designs.
Our particle screening dashboard is online! Read the article below to learn how you can use it to calculate the warming effect of engineered aerosols for Mars Terraforming.
Background
One proposed approach to warm Mars is to alter Mars’ energy budget using engineered aerosols manufacturable at scale, and made from materials that are abundant on Mars (e.g metals from the soil, carbon from the atmosphere) (Ansari et al. 2024). To give mid-latitude summertime temperatures >273K, which is the minimum needed to start to melt ice, the goal is to warm the surface by at least +35K on average. Engineered aerosols interact with both incoming solar radiation (short wavelengths) and the thermal radiation emitted by both the surface and the atmosphere (longer wavelengths).
To understand how particles can warm the planet, we consider the top of the atmosphere (TOA) annual-average global energy budget. At the TOA there is no downward thermal infrared energy coming from space. The energy budget is:
The thermal energy lost to space is called the Outgoing Longwave Radiation (OLR).
Incoming Sunlight net of Reflected Sunlight is sometimes referred to as the Absorbed Solar Radiation (ASR).
Figure 1: Energy balance of the Mars system (adapted from DeBenedictis et al. 2025) Geothermal heat is negligible (<0.02 W/m²).
Therefore, a necessary (though not sufficient) condition for the engineered aerosols to warm the surface when added to Mars’ atmosphere is that—for a given surface temperature—the particles minimize the OLR (“trap” the surface’s heat) and do not greatly reduce the ASR.
The flux emitted through radiation at the surface (assuming unit emissivity) is:
with σ the Stefan-Boltzmann constant. As the atmosphere and the surface warm in response to the engineered aerosols (Tsfc increases), the radiation flux emitted by the surface in Eq. 2 also increases which tends to cool the ground - a negative feedback on the surface temperature. Eventually the OLR and the ASR reach a new balance (OLR = ASR), corresponding to a new, warmer average surface temperature.
The natural (e.g. mineral dust) or engineered aerosols interact with the thermal radiation and sunlight via:
Scattering, which conserves the total amount of energy in the radiation field, but alters the direction in which the radiation propagates. Depending on the particle type, scattering can be mostly isotropic (uniform in all directions), predominantly forward scattering, or predominantly backward scattering compared to the direction of incoming radiation.
Absorption, which removes energy from the electromagnetic radiation field, and converts it to heat.
The sum of the scattering and absorption is called extinction and corresponds to the total influence of the particles on the electromagnetic radiation. The amount of natural or engineered aerosol in the atmosphere is specified through a number called optical depth τ, with τ=0 meaning no aerosols are present and τ»1 implying the atmosphere is optically thick and light is readily absorbed (see details in the technical corner section at the end of the post). For reference, the optical depth τ for natural martian dust typically varies between 0.2 - 3 depending on dust conditions.
At intermediate levels within the atmospheric column, the budget described (for the top of the atmosphere) in Eq 1. has extra terms because each atmospheric layer now also receives visible and thermal radiation transmitted from other atmospheric layers and from the surface. When the temperature structure of the atmosphere is unstable, heat is redistributed through convection (vertical air currents). This radiative-convective problem cannot accurately be solved with analytical methods and must therefore be solved numerically for each nanoparticle design. Inspired in part by earlier radiative-convective work by Ramirez & Kasting (2017) and Turbet et al. (2020), we built a tool able to represent those physical processes with fast-enough runtimes (minutes) to quickly feed back warming-performance results for improving particle designs.
TerraScreen
Our screening tool is based on a 1-dimensional radiative-convective-equilibrium (RCE) model with fast (minutes) runtime to assess how different particle designs warm Mars’ surface. The screening tool is open-source (link) and based on the radiation code from the open-source NASA Ames Mars Global Climate Model. (https://github.com/nasa/legacy-mars-global-climate-model). It uses 180 spectral bands, 10✕ more than typically used in Mars GCMs. Gaseous absorption by CO2 is computed with a “correlated-k” approach (a numerical method that allows us to efficiently approximate line absorption, e.g. Mischna, et al. 2012 ) at various temperatures and pressures for each of the 180 spectral bands. The atmosphere is dry, with only trace amounts of water vapor (the water vapor mass mixing ratio MH2O =10-7[kg/kg]).
The particles’ scattering and absorption optical data in each of the 180 spectral bands are computed offline for various geometries (e.g. nanorings, nanorods) and materials (e.g. aluminium, graphene) prior to use in the particle screening tool. Tools used to calculate these can include electromagnetic solvers such as finite-difference time-domain (FDTD), discrete dipole approximation (DDA), or a Mie theory semi-analytic approaches. Our colleagues at Northwestern University (Mohseni Lab) and UCLA (A. Raman Lab) provided optical properties for some of the particles considered (e.g. Ansari et al. 2024), which are also shared at our Github link. Using the optical properties, the screening tool computes the radiative fluxes and temperature structures for a single atmospheric column representative of the Martian atmosphere: the incoming solar radiation at the top of the the column is set to the average solar insolation received by Mars, assuming the sun is 30° over the horizon (a zenith angle of 60°, the daytime average value for a rapidly-rotating planet). Therefore the air and surface temperatures predicted by the screening tool are representative of average values. TerraScreen diagnoses radiative-convective equilibrium when the net radiative fluxes at the top and at the bottom of the atmosphere are zero, meaning that the upwelling and downwelling fluxes are balanced.
By default TerraScreen is initialized with a temperature profile that is close to equilibrium with a ‘clear’ atmosphere that is free of aerosols, intended to be representative of the present-day climate state. TerraScreen can be used in two modes:
Static mode: After an instantaneous injection of aerosols into the atmosphere, TerraScreen diagnoses the radiative forcings (now out of equilibrium) without updating the temperature profile. This is used to assess the relative radiative forcings of various engineered aerosols (for example on the OLR), while holding the main source of thermal infrared radiation constant (~220 K surface).
Dynamic mode: TerraScreen runs the model to radiative-convective equilibrium. The main outputs of interest are the resulting surface temperature and atmospheric temperature profiles for various concentrations of aerosol and particle types.
Use cases
TerraScreen static mode output are available in the Mars Terraforming Research dashboard (link) and an example is shown in the top panel of Figure 2. The orange bars show the clear sky (no aerosol) ASR at visible wavelengths (the ASR maximizes around the Sun’s ~500 nm peak emission) and the OLR lost to space is shown at thermal wavelengths (the OLR curve follows the 220 K blackbody emission from Mars’ surface, with the exception of the strong CO2 absorption feature centered at 15 μm). When summing the energy in all the spectral bands, the sum of the ASR - OLR is zero, and the system is in equilibrium. The cyan and red curves show the same metric when aerosols are present. In Figure 2 we used an aerosol visible optical depth τ=0.5 for a distribution of natural dust (spherical particles with an effective radius Re =1.5 μm) and a monodisperse distribution of 8 μm-long aluminum nanorods, respectively. Each run has one or the other aerosol (not both). Both aerosols are vertically distributed using a parameterization (“Conrath-nu”) that provides a fairly uniform distribution of aerosols from the ground to ~20 km altitude that gently tapers-off above 20km.
Figure 2: Top: TerraScreen output shows the absorbed solar radiation (shown as positive values at visible wavelengths) and the outgoing longwave radiation (shown as negative values at thermal wavelengths) for a clear atmosphere (orange), and an aerosol optical depth τ = 0.5 for natural dust (cyan) and aluminum nanorods (red). Bottom: the extinction (solid), and scattering contribution (dashed) efficiencies for the natural and engineered aerosols.
While natural dust (cyan) only slightly reduces OLR, the aluminum rods (red) significantly reduce the OLR, effectively blocking thermal radiation that would be otherwise lost to space. As a consequence, the net radiative budget ASR - OLR becomes positive, and subsequent large-amplitude warming of the Martian surface follows. This is due to the strong extinction for the engineered nanorods which are 1-2 orders of magnitude higher than those for the natural dust at thermal wavelengths (Figure 2, bottom panel and ‘technical corner’ section at the end of this document). (Ansari et al. 2024).
Running TerraScreen in static mode enables faster iteration. We can assess the effect of changing particles’ sizes, shapes (e.g. ribbon, cylinder, ring, aggregates), materials (e.g aluminum, iron, silica, graphite, salts, porous aerogel nanospheres), on the net radiative budget for the atmosphere. For shortlisted particles, we can further quantify the warming from the particles by running TerraScreen in dynamic mode for various aerosol loadings.
Figure 3 shows the equilibrium surface temperature for four types of engineered aerosols: Al nanorods, Al nanoribbon, Al nanorings, and a mix of 0.25 μm - 1 μm graphene disks as a function of aerosol loading. For this example, we converted the aerosol optical depths on the x-axis to mass loadings. We show the mass of aerosols in a unit column in [mg/m2], and the steady-state total mass of aerosols required (in million tonnes) which is obtained by multiplying the column amount by the surface area of Mars.
Figure 3. TerraScreen ran to radiative-convective equilibrium to give estimated surface temperature for various engineered aerosols. Shaded green area shows target warming temperature range of interest.
We see in Figure 3 that several particle designs reach the target average warming or +35 K for a few MT of aerosols: aluminum nanorods, graphene disks and aluminum nano-rings. The nano-ribbons in their current design fall short of providing adequate warming, even at large (>10 MT) mass loadings.
Because nanoribbons remain attractive from a full-scale deployment perspective due to their expected relative ease of manufacturing, we are investigating with TerraScreen if further optimization of their geometry—and perhaps, use of coatings—could raise their efficiency per unit mass.
Figure 4: First batches of 3 μm nanoribbon (bottom panels) and nanorings (top right panel) manufactured by collaborators at NorthWestern University (link). A human hair (typical width of ~70 μm) is added to the background to give a sense of scale.
For the graphene disk and aluminum nanorods, we have validated warming predictions from TerraScreen against full 3D GCM simulations from Richardson et al. 2025 https://arxiv.org/abs/2504.01455, giving us confidence in the tool’s robustness. To find out how to set-up and run TerraScreen locally, check out the Mars Terraforming Research Github page: github.com/mars-terraforming-research/TerraScreen
The Dashboard
We have set-up a dashboard to reference particles design tested in TerraScreen. The dashboard provides a streamlined way to assess how different particles interact with the electromagnetic radiation at Mars and calculate the average warming effects on the surface. If you are working on a particle type, you can test it with TerraScreen (code available here) - and please reach out so we can add it to the dashboard.
mars-terraforming-research.github.io/TerraScreen
Alex Kling
Technical corner
Applications of TerraScreen
The radiative transfer code is flexible enough to be adapted for other climate scenarios: for example, early Mars [Steakley et al. 2019, 2023] or a Mars-like exoplanet Hartwick et al. [2022]. The vertical distribution of aerosols, solar flux, zenith angle, and albedo can be easily modified within the code. Other aerosols and gases can be included by providing new optical properties and correlated-k tables, respectively. Supported pressures in the table provided are 10-6 Pa to 10 bar and supported temperatures are up to 350 K (which should be sufficient for a 1D globally-averaged model). CO2-CO2 collision induced absorption, which kicks in at a few 100's mbar of surface pressure, is currently not included.
Definition of the optical depth τ
The optical depth τ defines the column-integrated extinction at a reference wavelength (we use 670nm). In the simplifying case where the aerosol is a pure absorber that does not scatter in the visible (e.g. graphene), the aerosol optical depth τ is related to the reduction in the solar energy at the surface by the Beer–Lambert law:
where Fsfc and Ftoa are the solar energy at 670nm at surface and at the top of the atmosphere, respectively, and the cosine of the zenith angle cos(θₒ) accounts for the longer optical path due to the sun being low on the horizon.
While Eq. 3 can be useful to interpret the physical meaning for the reference optical depth τ, in the general case, scattering effects by the aerosols complicate the calculation of Fsfc, and the balance of fluxes at the surface must be considered for all visible and thermal infrared wavelengths (not just 670nm). This is why we need a radiative-transfer solver like TerraScreen.
Aerosol Parameters
TerraScreen natively uses wavelength-dependent efficiency factors: Qscat(λ) (scattering), Qabs(λ) (absorption) and Qext(λ) (extinction), as well as an asymmetry factor g to characterize the aerosols. The efficiency factors are the optical cross-sections for the aerosol (i.e. Cscat, Cabs, Cext), divided by the geometric cross section πRₑ²:
The effective radius Rₑ for non-spherical particles (e.g. cylinders) is taken as the radius for the sphere of equivalent volume:
where Vₚ is the volume for the non-spherical particle in m³
Conversion from optical depth to mass of aerosols
The aerosol optical depth τ (defined at λ=670 nm) can be used to convert opacity (useful in radiative transfer calculation) to column mass loading for the aerosol (relevant to engineering applications).
The number of particles per unit atmospheric column is:
with τ the optical depth at the reference wavelength and Cext(670 nm) the extinction cross section at the reference wavelength . The mass of a single particle is:
with ρₚ the density for the particle in [kg/m³]. We can calculate the mass of aerosol per unit column:
This number can be written as a function of coefficients readily used in TerraScreen by combining Eq. (4-8) :
The steady-state total mass of aerosol is estimated by multiplying the column mass loading in Eq. 9 by Mars’ surface area.
with Rₘ = 3.4 × 106 [m] the radius of the planet. 1
REFERENCES
Ansari S., Kite E. S., Ramirez R., Steele L.J. and Mohseni H. Feasibility of keeping Mars warm with nanoparticles. Science Advances 10 (2024). Open access at: DOI:10.1126/sciadv.adn4650
Haberle R.M., Kahre M.A., Hollingsworth J.L., Montmessin F., Wilson R.J., Urata R.A., Brecht A.S., Wolff M.J., Kling A.M. and Schaeffer J.R. (2019) Documentation of the NASA/Ames Legacy Mars Global Climate Model: simulations of the present seasonal water cycle. Icarus 333, 130–164 Open access at: https://doi.org/10.1016/j.icarus.2019.03.026 Github: https://github.com/nasa/AmesGCM
Hartwick V.L., Haberle R.M, Kahre M.A., and Wilson R.J. (2022) The Dust Cycle on Mars at Different Orbital Distances from the Sun: An Investigation of the Impact of Radiatively Active Dust on Land Planet Climate The Astrophysical Journal, Volume 941, ApJ 941 54 DOI 10.3847/1538-4357/ac9481
Mischna, M. A., Lee C., and Richardson M. (2012) Development of a fast, accurate radiative transfer model for the Martian atmosphere, past and present, J. Geophys. Res., 117, E10009, doi:10.1029/2012JE004110.
Ramirez R.M. (2017) A warmer and wetter solution for early Mars and the challenges with transient warming. Icarus, 297,2017,Pages 71-82,ISSN 0019-1035,https://doi.org/10.1016/j.icarus.2017.06.025.
Richardson M.I., Ansari S., Fan B., Ramirez R., Mohseni, H., Mischna, M.A., Hecht, M.H., Steele L.J., Sharipov F. and Kite E.S. (2025) Atmospheric dynamics of first steps toward terraforming Mars https://doi.org/10.48550/arXiv.2504.01455
Steakley K.E., Murphy J., Kahre M.A, Haberle R.M. and Kling A.M. (2019) Testing the impact heating hypothesis for early Mars with a 3-D global climate model, Icarus 330 ,169-188, https://doi.org/10.1016/j.icarus.2019.04.005.
Steakley K.E., Kahre M.A., Haberle R.M. and Zahnle K.J.(2023) Impact induced H2-rich climates on early Mars explored with a global climate model, Icarus 394, 115401 https://doi.org/10.1016/j.icarus.2022.115401.
Turbet M., Gillmann C., Forget F., Baudin B., Palumbo A., Head J., Karatekin O. (2020) The environmental effects of very large bolide impacts on early Mars explored with a hierarchy of numerical models. Icarus 335, 113419, ISSN 0019-1035, https://doi.org/10.1016/j.icarus.2019.113419.
There are a number of other potential sources of energy on Mars that could be utilized by more exotic organisms, but these are not realistic at the moment for high-yield bioprocesses. For example, chemolithotrophs grow very slowly and methanotrophs would require gas-exchange bioreactors that would be very challenging to use in a space mission context.
Cite as: Mars Terraforming Research Report (2025). An open-source particle screening tool for Mars warming research figshare. Preprint. https://doi.org/10.6084/m9.figshare.30132643