Haze production rates in super-Earth and mini-Neptune atmosphere experiments

Published online:


Numerous Solar System atmospheres possess photochemically generated hazes, including the characteristic organic hazes of Titan and Pluto. Haze particles substantially impact atmospheric temperature structures and may provide organic material to the surface of a world, potentially affecting its habitability. Observations of exoplanet atmospheres suggest the presence of aerosols, especially in cooler (<800 K), smaller (<0.3× Jupiter’s mass) exoplanets. It remains unclear whether the aerosols muting the spectroscopic features of exoplanet atmospheres are condensate clouds or photochemical hazes1,2,3, which is difficult to predict from theory alone4. Here, we present laboratory haze simulation experiments that probe a broad range of atmospheric parameters relevant to super-Earth- and mini-Neptune-type planets5, the most frequently occurring type of planet in our galaxy6. It is expected that photochemical haze will play a much greater role in the atmospheres of planets with average temperatures below 1,000 K (ref. 7), especially those planets that may have enhanced atmospheric metallicity and/or enhanced C/O ratios, such as super-Earths and Neptune-mass planets8,9,10,11,12. We explored temperatures from 300 to 600 K and a range of atmospheric metallicities (100×, 1,000× and 10,000× solar). All simulated atmospheres produced particles, and the cooler (300 and 400 K) 1,000× solar metallicity (‘H2O-dominated’ and CH4-rich) experiments exhibited haze production rates higher than our standard Titan simulation (~10 mg h–1 versus 7.4 mg h–1 for Titan13). However, the particle production rates varied greatly, with measured rates as low as 0.04 mg h–1 (for the case with 100× solar metallicity at 600 K). Here, we show that we should expect great diversity in haze production rates, as some—but not all—super-Earth and mini-Neptune atmospheres will possess photochemically generated haze.


For well-constrained Solar System atmospheres, experimental simulations of haze formation use an energy source (for example, ultraviolet or cold plasma) to initiate chemistry in gas mixtures that are consistent with measured atmospheric abundances and can therefore be used to study the chemical processes and resulting particles. However, for super-Earths and mini-Neptunes, there are not yet sufficiently accurate and precise atmospheric composition measurements to use as the basis for a reactant mixture. For the experiments presented here, we used a chemical equilibrium model to guide our initial gas mixtures10. Although photochemistry can drive the atmospheric composition away from chemical equilibrium, the major gas components tend to survive photochemical destruction, so chemical equilibrium provides a good first-order prediction of the dominant available constituents. Disequilibrium processes are clearly important, but modelling them requires even more knowledge of a planet and would require adding additional parameters (for example, the mass, eddy diffusion coefficient, and so on) to an already complex phase space. Atmospheres in chemical equilibrium under a variety of expected super-Earth and mini-Neptune conditions can contain abundant H2O, CO, CO2, N2, H2 and/or CH4 (refs 5,10), various combinations of which may have a distinct complement of photochemically produced hazes, such as ‘tholins’14and complex organics in the low-temperature, H2-rich cases, and sulphuric acid in the high-metallicity, CO2/H2O-rich cases. Warm atmospheres outgassed from a silicate composition can also be dominated by H2O and CO2 (refs 15,16). We therefore chose to focus on a representative sample of gas mixtures that are based on equilibrium compositions for 100×, 1,000× and 10,000× solar metallicity over a range of temperatures from 300–600 K at an atmospheric pressure of 1 mbar. This phase space is consistent with atmospheric temperatures and metallicities for planets with R p  < 4.0 REarth5,17,, where R p and REarth are radii of the planet and the Earth, respectively—the bulk of the predicted Transiting Exoplanet Survey Satellite planetary yield18. Table 1 gives the mole fractions used for each laboratory experiment. Although elements heavier than hydrogen and helium are not likely to maintain exact solar proportions due to various evolutionary processes, such as atmospheric escape, the compositions from this representative sample are probably common within the super-Earth and mini-Neptune population. Note that we only included gases with a calculated abundance of ≥1% to maintain a manageable level of experimental complexity; this resulted in the exclusion of sulphur-bearing species, which may be important for haze formation19 and will be an avenue of future work. The pressure, temperature and gas compositions used in the experiments are self-consistent based on the model calculations.

Table 1 Summary of initial gas mixtures

Following refs 10,20, at low temperatures, the equilibrium composition transitioned from a Neptune-like H2-dominated atmosphere at low metallicities to a H2O-dominated atmosphere at intermediate metallicities, and to a more Venus-like CO2-dominated atmosphere at high metallicities, as shown in Fig. 1. CH4 is a major carbon component at low temperatures, but CO takes over at high temperatures. The CO2 relative abundance increases with increasing metallicity. As shown in Fig. 2, the gas mixtures were brought to temperature and flowed through the Planetary Haze Research (PHAZER) chamber13, where they were exposed to a cold plasma (alternating current (AC) glow) discharge that initiated chemistry resulting in the formation of gas- and solid-phase products. The experiment ran continuously for 72 h, at which time the chamber was placed in a dry, oxygen-free glove box, where the samples were removed and weighed. See Methods for additional information regarding the experimental setup and measurements.

Fig. 1: Composition of the initial gas mixtures used for our experiments.
Fig. 1

Our experimental phase space spanned 100 to 10,000× solar metallicity and temperatures ranging from 300 to 600 K. It was divided broadly into three categories: H2-dominated, H2O-dominated and CO2-dominated. The exact compositions of the gases used in the initial gas mixtures are shown in Table 1. The model values are for 1 mbar, which is consistent with the pressure used for the experiments.

Fig. 2: Schematic of the PHAZER Chamber at Johns Hopkins University.
Fig. 2

Due to the large variety of gases used for the experiments, this schematic provides a general idea of the setup. The details varied depending on the gases used, with attention paid to the solubility of gases in liquid water, condensation temperatures and gas purity.

As shown in Fig. 3 and Table 2, the H2-dominated experiments (100× at 400 and 600 K, and 1,000× at 600 K) had lower haze production rates (0.04, 0.25 and 0.15 mg h–1, respectively) than the other experiments with measurable production rates. Previous studies have shown that H2 decreases particle formation in gas mixtures that include CH4 and/or CO2 (refs 21,22,23), probably by termination of chains, which decreases the production of larger molecules; this mechanism may be responsible for the low production rates in these cases. The 100× at 300 K experiment produced particles, but did not have a sufficiently high production rate for measurements to be taken; further work is required to determine whether this was a result of the addition of NH3, the lower temperature, or both. In all cases, the production rates were non-zero, indicating that these atmospheres may still be capable of producing a tenuous photochemically generated haze. The H2-dominated giant planets in our own Solar System have optically thin stratospheric hazes that are produced photochemically.

Fig. 3: Production rate values.
Fig. 3

‘F’ indicates that there was enough solid produced by the experiment to result in a visible film on the substrates, but there was not enough to collect and weigh. Our standard Titan production rate is indicated by the grey line. All production rate values are listed in Table 2.

Table 2 Measured production rates (mg h–1)

The two highest production rates measured (~10 mg h–1) were for the 1,000× metallicity (H2O-dominated) gas mixtures at 300 and 400 K. This production rate is even higher than the production rate from our standard Titan experiment (5% CH4 in 95% N2), which produced ~7.4 mg h–1 using an identical setup13. The exoplanet experiments had a higher CH4 content than our standard Titan experiment; however, they had significantly less N2, and previous work has shown that N2 plays an important role in gas-to-particle conversion in N2/CH4 gas mixtures24,25. Additionally, previous work with plasma experiments has indicated that the addition of CO2 should decrease particle formation26. In contrast, previous work has shown that H2O (present as a liquid and gas) may promote the formation of organics27, so further work is necessary to understand how the complex interplay of these different chemical pathways results in such a high production rate.

The CO2-dominated experiments (10,000×) at 300 and 400 K did not produce sufficient sample to collect and weigh; however, inspection of the quartz substrates revealed that a thin film was present, indicating that solid was produced, albeit at a lower rate than the other experiments presented here. However, the 600 K experiment produced enough sample to collect. Aside from the increase in temperature, the main difference between this experiment and the 300 and 400 K experiments was the addition of CO and H2 to the gas mixture. Given that H2 is usually observed to decrease particle production, as discussed above, it seems unlikely that the addition of H2 to the gas mixture was responsible for the increase in the production rate. However, previous work has shown that the addition of CO can increase the production rate28; CO is a better source of carbon for haze formation than CO2, as dissociation of CO produces atomic carbon more efficiently, which can be used to build larger organic molecules. The combination of CO and H2O may be particularly efficient for increasing the production rate as photochemical reactions beginning with these two species produce a variety of organic compounds29.

The two experiments with the highest production rates had the two highest CH4 concentrations, but the one with the third highest production rate (10,000× at 600 K) had no CH4 at all, demonstrating that there are multiple pathways for organic haze formation and that CH4 is not necessarily required. In the case of the experiment with no CH4, the gas mixture had CO, which provided a source of carbon in place of CH4. However, it is important to note that the production rates are not simply a function of carbon abundance, C/O, C/H or C/N ratios in the initial gas mixtures. This result also demonstrates the need for experimental investigations to develop a robust theory of haze formation in planetary atmospheres.

Finally, we note that visual inspection of the films on the quartz substrates indicated large variation in particle colour as a function of metallicity, and efforts are underway to measure optical constants of these analogue materials.

It is important to remember that the experimental matrix varied temperature and metallicity; thus, to maintain self-consistency, all nine gas mixtures investigated were compositionally distinct. This approach makes interpreting the chemical pathways responsible for the observed production rates challenging, but provides insights into which regions of temperature and metallicity phase space may result in photochemical haze production, thereby enabling future investigations focused on specific pathways. The complexities of atmospheres compared with in the laboratory makes it challenging to convert laboratory production rates into planetary haze production rates, but the relative comparisons between our experiments, in addition to the comparison with our nominal Titan rate, provide guidance on which gases are more likely to result in particle formation. Further work is necessary to elucidate the chemical pathways responsible for these differences, and particularly to understand the cases with very high production rates.

Although models of atmospheric photochemistry and haze optical properties provide good first estimates, they are incomplete and biased due to the relatively small phase space spanned by the Solar System atmospheres on which they are based. Laboratory production of exoplanet hazes is a crucial next step in our ability to properly characterize these planetary atmospheres. These experimental simulations of atmospheric chemistry and haze formation relevant to super-Earth and mini-Neptune atmospheres show that atmospheric characterization efforts for cool (T < 800 K) super-Earth- and mini-Neptune-type exoplanets will encounter planets with a wide variety of haze production rates. These findings provide an importance balance in both caution and optimism for the planning of near-term observations of super-Earths and mini-Neptunes with facilities such as the James Webb Space Telescope. Hazes also impact reflected light and therefore must be understood for future direct imaging efforts4,30.


We performed the experiments using the PHAZER Chamber at Johns Hopkins University13. A schematic of the experimental setup is shown in Fig. 2. We mixed the reactant gases (CH4 99.999%, CO 99.99%, N2 99.9997%, CO2 99.999%, H2 99.9999% and He 99.9995%; Airgas) in a custom-built mixing manifold designed to enable the mixing of multiple gases over a broad range of compositions. Since water is liquid at room temperature, it was introduced into the gas mixtures through the use of a dry-ice temperature bath surrounding a canister of HPLC Grade water (Fisher Chemical), which allowed for control of the water vapour pressure (and therefore the resulting mixing ratio) by variation of the temperature of the bath (by adjusting the relative proportions of dry ice, CH3OH and water). For the one mixture that contained NH3 (100× at 300 K), a solution of ammonium hydroxide (ACS Grade; Millipore) was used with the dry-ice bath to provide the NH3 and H2O required for the mixture. A liquid nitrogen cold trap was used to remove known contaminants from CO before it was mixed with the other gases13,28. The gas mixtures then flowed through a custom heating coil, which allowed temperature control from room temperature up to 800 K. The gases then flowed continuously at a rate of 10 standard cubic centimetres per minute through a stainless-steel chamber (a mass flow controller (MKS Instruments) maintained a constant pressure of a few mbar (depending on the temperature)), where the gases were exposed to the cold plasma generated by an AC glow discharge for approximately 3 s. The plasma produced by the AC glow discharge was used as a proxy for energetic processes occurring in planetary upper atmospheres. While this is not analogous to any specific process, we used it because it was sufficiently energetic to directly dissociate very stable molecules, such as N2 or CO. This is often accomplished in planetary atmospheres by extreme ultraviolet radiation photons; therefore, AC glow discharge plasma is typically used as an analogue for the relatively energetic environment of planetary upper atmospheres14. Note that the AC glow discharge is a cold plasma source and therefore the neutral gas temperature was not significantly altered by the plasma. The discharge dissociated and/or ionized the reactant gases initiating chemical reactions in the chamber. Product gases and remaining reactant gases flowed out of the chamber, while any solid produced in the experiment remained. Quartz substrates placed in the chamber collected thin films of particles and were used to check for particle production at very low production rates. The experiments ran continuously for 72 h and gas-phase measurements demonstrated that the gas phase composition remained stable throughout this period after the initial equilibration time (less than 30 min)13. After 72 h, the plasma was shut off and the chamber was slowly returned to ambient temperature, while pumping to remove the remaining gases. The chamber was placed in a dry (<0.1 ppm H2O), oxygen-free (<0.1 ppm O2) N2 glove box (I-Lab 2GB; Inert Technology), where the samples were removed from the chamber and weighed (Sartorius Entris 224-1S with a standard deviation of 0.1 mg), providing measurements of the production rates.

Data availability

All data generated or analysed during this study are included in this published article.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Change history

  • Correction 16 March 2018

    In the version of this Letter originally published Table 2, which tabulates the production rates shown in Fig. 3, was mistakenly omitted. It has now been included in all versions of the Letter.


  1. 1.

    Knutson, H. A., Benneke, B., Deming, D. & Homeier, D. A featureless transmission spectrum for the Neptune-mass exoplanet GJ436b. Nature 505, 66–68 (2014).

  2. 2.

    Kreidberg, L. et al. Clouds in the atmosphere of the super-Earth exoplanet GJ1214b. Nature 505, 69–72 (2014).

  3. 3.

    Sing, D. K. et al. A continuum from clear to cloudy hot-Jupiter exoplanets without primordial water depletion. Nature 529, 59–62 (2016).

  4. 4.

    Morley, C. V. et al. Thermal emission and reflected light spectra of super Earths with flat transmission spectra. Astrophys. J. 815, 110 (2015).

  5. 5.

    Hu, R. & Seager, S. Photochemistry in terrestrial exoplanet atmospheres. III. Photochemistry and thermochemistry in thick atmospheres on super Earths and mini Neptunes. Astrophys. J. 784, 63 (2014).

  6. 6.

    Fressin, F. et al. The false positive rate of Kepler and the occurrence of planets. Astrophys. J. 766, 81 (2013).

  7. 7.

    Moses, J. I. et al. Disequilibrium carbon, oxygen, and nitrogen chemistry in the atmospheres of HD 189733b and HD 209458b. Astrophys. J. 737, 15 (2011).

  8. 8.

    Line, M. R., Vasisht, G., Chen, P., Angerhausen, D. & Yung, Y. L. Thermochemical and photochemical kinetics in cooler hydrogen-dominated extrasolar planets: a methane-poor GJ436b? Astrophys. J. 738, 32 (2011).

  9. 9.

    Miller-Ricci Kempton, E., Zahnle, K. & Fortney, J. J. The atmospheric chemistry of GJ 1214b: photochemistry and clouds. Astrophys. J. 745, 3 (2012).

  10. 10.

    Moses, J. I. et al. Compositional diversity in the atmospheres of hot Neptunes, with application to GJ 436b. Astrophys. J. 777, 34 (2013).

  11. 11.

    Venot, O., Agúndez, M., Selsis, F., Tessenyi, M. & Iro, N. The atmospheric chemistry of the warm Neptune GJ 3470b: influence of metallicity and temperature on the CH4 /CO ratio. Astron. Astrophys. 562, A51 (2014).

  12. 12.

    Morley, C. V. et al. Forward and inverse modeling of the emission and transmission spectrum of GJ 436b: investigating metal enrichment, tidal heating, and clouds. Astronom. J. 153, 86 (2017).

  13. 13.

    He, C. et al. Carbon monoxide affecting planetary atmospheric chemistry. Astrophys. J. Lett. 841, L31 (2017).

  14. 14.

    Cable, M. L. et al. Titan tholins: simulating titan organic chemistry in the Cassini–Huygens era. Chem. Rev. 112, 1882–1909 (2012).

  15. 15.

    Elkins-Tanton, L. T. & Seager, S. Ranges of atmospheric mass and composition of super-Earth exoplanets. Astrophys. J. 685, 1237–1246 (2008).

  16. 16.

    Schaefer, L., Lodders, K. & Fegley, B. Vaporization of the Earth: application to exoplanet atmospheres. Astrophys. J. Lett. 755, 41 (2012).

  17. 17.

    Fortney, J. J. et al. A framework for characterizing the atmospheres of low-mass low-density transiting planets. Astrophys. J. 775, 80 (2013).

  18. 18.

    Sullivan, P. W. et al. The Transiting Exoplanet Survey Satellite: simulations of planet detections and astrophysical false positives. Astrophys. J. 809, 77 (2015).

  19. 19.

    Gao, P., Marley, M. S., Zahnle, K., Robinson, T. D. & Lewis, N. K. Sulfur hazes in giant exoplanet atmospheres: impacts on reflected light spectra. Astronom. J. 153, 139 (2017).

  20. 20.

    Moses, J. I., Madhusudhan, N., Visscher, C. & Freedman, R. S. Chemical consequences of the C/O ratio on hot Jupiters: examples from WASP-12b, CoRoT-2b, XO-1b, and HD 189733b. Astrophys. J. 763, 25 (2013).

  21. 21.

    Raulin, F., Mourey, D. & Toupance, G. Organic syntheses from CH4–N2 atmospheres: implications for Titan. Orig. Life 12, 267–279 (1982).

  22. 22.

    DeWitt, H. L. et al. Reduction in haze formation rate on prebiotic Earth in the presence of hydrogen. Astrobiology 9, 447–453 (2009).

  23. 23.

    Sciamma-O’Brien, E., Carrasco, N., Szopa, C., Buch, A. & Cernogora, G. Titan’s atmosphere: an optimal gas mixture for aerosol production? Icarus 209, 704–714 (2010).

  24. 24.

    Imanaka, H. & Smith, M. A. Formation of nitrogenated organic aerosols in the Titan upper atmosphere. Proc. Natl Acad. Sci. USA 28, 12423–12428 (2010).

  25. 25.

    Trainer, M. G., Jimenez, J. L., Yung, Y. L., Toon, O. B. & Tolbert, M. A. Nitrogen incorporation in CH4–N2 photochemical aerosol produced by far ultraviolet irradiation. Astrobiology 12, 315–326 (2012).

  26. 26.

    Trainer, M. G. et al. Haze aerosols in the atmosphere of early Earth: manna from heaven. Astrobiology 4, 409–419 (2004).

  27. 27.

    Miller, S. L. A production of amino acids under possible primitive Earth conditions. Science 117, 528–529 (1953).

  28. 28.

    Hörst, S. M. & Tolbert, M. A. The effect of carbon monoxide on planetary haze formation. Astrophys. J. 781, 53 (2014).

  29. 29.

    Bar-Nun, A. & Chang, S. Photochemical reactions of water and carbon monoxide in Earth’s primitive atmosphere. J. Geophys. Res. Oceans 88, 6662–6672 (1983).

  30. 30.

    Marley, M. S., Gelino, C., Stephens, D., Lunine, J. I. & Freedman, R. Reflected spectra and albedos of extrasolar giant planets. I. Clear and cloudy atmospheres. Astrophys. J. 513, 879–893 (1999).

Download references


This work was supported by National Aeronautics and Space Administration Exoplanets Research Program grant NNX16AB45G. C.H. was supported by the Morton K. and Jane Blaustein Foundation.

Author information


  1. Johns Hopkins University, Baltimore, MD, USA

    • Sarah M. Hörst
    • , Chao He
    •  & Nikole K. Lewis
  2. Space Telescope Science Institute, Baltimore, MD, USA

    • Nikole K. Lewis
    •  & Jeff A. Valenti
  3. Grinnell College, Grinnell, IA, USA

    • Eliza M.-R. Kempton
  4. National Aeronautics and Space Administration Ames Research Center, Mountain View, CA, USA

    • Mark S. Marley
  5. Harvard University, Cambridge, MA, USA

    • Caroline V. Morley
  6. Space Science Institute, Boulder, CO, USA

    • Julianne I. Moses
  7. Université Grenoble Alpes, Grenoble, France

    • Véronique Vuitton


  1. Search for Sarah M. Hörst in:

  2. Search for Chao He in:

  3. Search for Nikole K. Lewis in:

  4. Search for Eliza M.-R. Kempton in:

  5. Search for Mark S. Marley in:

  6. Search for Caroline V. Morley in:

  7. Search for Julianne I. Moses in:

  8. Search for Jeff A. Valenti in:

  9. Search for Véronique Vuitton in:


S.M.H., N.K.L., C.H., M.S.M. and J.I.M. conceived the study. J.I.M. calculated the starting gas mixtures. C.H. performed the experiments and measurements. S.M.H. prepared the manuscript. All authors participated in discussions regarding interpretation of the results and edited the manuscript.

Competing interests

The authors declare no competing interests.

Corresponding author

Correspondence to Sarah M. Hörst.