Terraforming Mars and Venus Using Machine Self-Replicating Systems (SRS)

Robert A. Freitas, Jr.

Xenology Research Institute, 8256 Scottsdale Drive, Sacramento, California 95828, USA.
Journal of the British Interplanetary Society 36:139-142 (1985).

Note: This web version is derived from an earlier draft of the paper and may possibly differ in some substantial aspects from the final published paper.

As distinguished from previous proposals for large-scale planetary engineering projects employing passive, explosive, orbital mechanical or biological agents to terraform various planets of the Solar System, the present paper offers an alternative means: Machine self-replicating systems (SRS). Terraforming via SRS involves the deposition of a small "seed" unit near the surface of the body to be altered, which then self-replicates into a giant factory complex capable of undertaking permanent modification of the target environment. SRS terraforming methods are suggested for Mars and Venus.



Recently it has been suggested that highly complex factories capable of partial or complete self-replication might be sited somewhere in Earth-Moon orbit, on the lunar surface or elsewhere in space, or be used to facilitate the exploration of the Galaxy via self-replicating interstellar probes [1-9]. From the human standpoint, perhaps the most exciting consequence of self-replicating systems (SRS) is that they provide a means for organising vast quantities of matter to produce an ever-widening habitat for humankind throughout the Solar System. SRS provides such a large amplification of matter-manipulating capability that it is possible even to consider the "terraforming" of Mars, Venus and other worlds. Terraforming is a theoretical concept in which a planetary environment with otherwise inhospitable conditions for human life is purposefully and artificially altered so that people may live there with little or no life support equipment. This paper proposes that SRS technology should be added to the growing list of terraforming engineering tools. Traditional theoretical approaches to terraforming include passive [10], explosive [11], orbital mechanical [12], and biological [13-14] mechanisms, all of which are expected to require several hundred years or longer to run to completion. A largely unexplored alternative approach is to use non-biological machine self-replicating systems which may be far more durable and useful under extreme conditions. The following analysis demonstrates that self-replicating systems are highly competitive in this application, and terraforming times of the order of several centuries are conceivable for many celestial bodies using the SRS approach.



The preliminary engineering and design considerations for a lunar-based SRS factory were considered at length in a recent NASA study [5,9]. This study concluded that the development of a fully-automated self-replicating growing factory module with 100% "closure" (i.e., full independence in materials, energy, and information that might be supplied from Earth) could be achieved in twenty years if pursued on a crash-programme basis. The strawman system was keyed to element abundances in the lunar regolith, but for the purposes of the present work the specific composition of the planetary substrate is insignificant in terms of its effects on basic SRS performance characteristics, The minimum mass for a self-replicating machine system was not determined but the baseline design was thought to be near the region of nonlinear scaling, requiring drastic redesign for downsizing to lower masses. For purposes of the present work, the NASA figures for the baseline SRS "seed" configuration are adopted as follows: Mass, 100 tons; power, 1 MW; replication time, 1 year; circular radius, 50 metres; and operating temperatures, 173-373 K.

In the NASA study it was suggested, but not proven, that any self-replicating machine system having 100% closure can be used to upgrade itself to the level of a general product factory. A general product factory is one which can be instructed to manufacture anything which is physically possible to make. Such a system is the physical realisation of von Neumann's universal constructor automation, which can construct anything constructable, given an adequate substrate and the rules of operation of his artificial cell space universe. That any physical SRS with full closure is inherently a general product factory is implicit in von Neumann's work, but has not yet been proven rigorously by automata theorists. However, in the present work it is assumed that each SRS seed and offspring may be regarded as general product factories.

As for productivity, if T is elapsed time in years, N is the number of seed units or seed mass-equivalents generated during T, and P is the total productivity of the SRS field in tons/year, then for simple doubling growth T = 1 + log2 N if each unit works only on completing its own replica, assuming a 1-year replication time. The fastest theoretical growth rate, assumed throughout this paper, is achieved when all available machines contribute directly to the replication of the next replica, in which case T = 1 + 1/2 + ... + 1/N. In either case, P = 100 N tons/year. During the growth phase, this production is entirely diverted to the replication of factory facilities; during the subsequent production phase, the entire factory complex output is diverted to the specific terraforming tasks required.



The primary challenge in terra-forming Mars is to generate a minimum breathable (150 mbar) oxygen atmosphere planetwide, requiring the release of 6 x 1017 kg of oxygen into the existing 6 mbar predominantly CO2 atmosphere. The most favourable source for this oxygen is the crustal silicates and mineral oxides, but polar/regolith or imported water-ice are also possible. The shortage of nitrogen on Mars, a severe problem for all terraforming schemes, is not addressed here.


3. 1 Preferred Scenario

The martian regolith apparently consists primarily of montmorillonite, a mineralogical mixture of oxides of silicon, iron, calcium, aluminium, and titanium in varying proportions. The minimum energy required to release the chemically bound oxygen is the heat of formation for each compound. These vary from 4-16 x 106 joules/kg, and a weighted average of 107 J/kg is supported by abundance data from the Viking mission lander soil analysis experiments. Thus generation of the required mass of oxygen requires 6 x 1024 joules.

In the most promising scenario, a single seed factory lands on the surface of Mars, unfurls and commences self-replication at the maximum rate. If standard units replicate until they just cover the entire surface area of Mars, then the maximum final number of SRS units is (1.44 x 1014 metre2 )/(7854 metre2) = 1.8 x 1010, requiring about 24 years for complete construction. The planetary replicant factory system then reconfigures to produce and release oxygen gas, a far simpler manufacturing task than the production of replicas. However, if the generation of O2 proceeds at the low replica-mass production rate of 100 tons/ year per SRS replica, then the maximum time required to release enough gas to produce a 150 mbar atmosphere is (6 x 1017 kg O2/[(105 kg O2/yr-SRS units) (1.8 x 1010 SRS units)] = 330 years. The power supplied by the SRS field is 1.8 x 1016 watts, well in excess of the minimum (6 x 1024 J)/(1.1 x 1010 sec) = 5 x 1014 watt requirement. Thus Mars may be given a breathable atmosphere in about 350 years by this method, or less than 100 years if the SRS O2/replica production rate ratio for output mass is greater than 4.4.

Oxygen production via humidification of surface soil superoxides and suboxides [15] is not expected to make a major contribution. Gas released when samples of martian surface soil were wetted during the Viking mission GEX experiments amounted to a maximum of 790 nanomoles O2/cm3. Even if this region of most active soil extended planetwide to a depth of 10 metres, which it does not, only 10-4 of the oxygen required for terraforming could be liberated from this source.


3.2 Alternative Scenarios

Semenov [16] has discussed the electrolysis of martian water to obtain oxygen. The mass of the remaining stores of frozen water trapped on Mars at the present time is a subject of much current debate [17-18], but Pollack [19] gives a rough estimate of 3 x 1018 kg H2O planetwide, about equally distributed between permanent polar icecaps and permafrost/ regolith. To obtain enough O2 to provide a 150 mbar breathable atmosphere. 6.8 x 1017 kg H2O (only 23% of the total available) must be electrolysed and released into the atmosphere. The methodology and terraforming timescales discussed in the previous section are equally applicable here - about 1-4 centuries should be required.

The main difficulty associated with this scenario is the problem of disposing of the 8 x 1016 kg H2 byproduct. Even if liquefied this will occupy a volume of 1015 m3, enough to fill a cube 100 km on an edge. It is difficult to see how local industry could absorb such a tremendous production. Outshipment to hydrogen-poor space factories or lunar industries is one possibility, though the steep martian gravity well may prove a serious impediment to this plan. Atmospheric liberation is too dangerous, since the exosphere escape time is 104 years for H2 on Mars.

Dyson [2] eloquently proposed using SRS devices to mine the Saturnian moon Enceladus for its water-ice and return the material to Mars as a supply of hydrogen and ready-made water. However, it may be unnecessary to use Enceladean water-ice for this purpose since the martian polar caps and regolith probably retain about an order of magnitude more oxygen than necessary to provide 3 breathable atmosphere planetwide. Water delivered to the surface is not breathable and must be either biologically processed or electrolysed to liberate atmospheric O2, so a planetary-scale biosphere or electrolysis factory must be replicated anyway in addition to the replicated mining operation on Enceladus - a needless complication.


3. 3 Advantages of SRS Terraforming

There are a number of advantages is using SRS machines instead of competing methods for terraforming Mars:
  1. Lower investment in R&D - Assuming basic SRS technology has already been developed for lunar or space manufacturing as supposed in Section 2. the only further R&D required is (a) development of a martian SRS growth architecture and (b) reprogramming an off-the-shelf SRS seed unit. On the other hand, microorganisms useful for terraforming must be genetically engineered from scratch (and produced in megaton quantities), since each planetary environment is unique and terraforming microbes are unlikely to have more general utility.
  2. Lower investment in transportation and material - Only a single 100-ton SRS seed unit need be built and transported to Mars. In most other schemes, millions or even billions [10] of tons of materials must be assembled and transported.
  3. SRS terraforming schemes may require about the same length of time (102-103 years) for execution as biological schemes [14], whereas other non-biological methods require 1-3 orders of magnitude more time.
  4. The end result of SRS terraforming is a well-ordered physical habitat for man. In essence the entire surface of the planet will have been excavated to a mean depth of about 4 m. During terraforming the machines can be ordered to excavate or to backfill in specific patterns which later may be used for artificial seas, lakes, canals, roadways, and- subterranean agricultural greenhouses or cities. For example, during the O2-production phase, the excavation of more than 1018 kg of martian regolith is enough to carve out 6 million kilometres of 1000-m wide, 100 m deep canals. (Such a system could hold about 20% of the estimated available water on Mars). The SRS mining fleet could even be used to redistribute the off-balance weight load of the Tharsis volcanic shield formation and thus stabilise the planetary axial tilt, permitting increased polar heating and enhancing the greenhouse effect to achieve global warming. By contrast, the sole end product of biological terraforming using bioengineered microorganisms is a planetwide green plant cover. Other non-biological schemes yield even less desirable end states.
  5. The ultimate benefit of SRS terraforming on Mars is a fully industrialised planet, complete with a 1012-ton/year reprogrammable general product factory manufacturing capability, 1018 kg of refined byproduct metals (A I, Fe, Ti) or enriched metal ores, and a 1010 megawatt self-repairing distributed solar power source for industrial use, or for further terraforming. Alternative terraforming methods provide no comparable benefits.



The main obstacles in terraforming Venus are high temperature and high pressure. Both problems may be solved through elimination of the atmosphere, either by removal to space or by physical/chemical burial in the crust. All schemes proposed to date [20] require at least 1028 joules for execution, but only using self-replicating machines can humanitýs investment be reduced to reasonable levels and payoff be maximised.


4. 1 Preferred Scenario

The mass of the Venusian atmosphere is 4.7 x1020 kg. To remove the gravitationally-bound atmosphere to Sun/Venus L1/L2, where it enters heliocentric orbit and disperses rapidly, Ei = 2.5 x1028 joules are required. The simplest way to provide this energy is by impact with an asteroid of mass Ei[GMo (Rv-1 Rj-1)]-1 = 2 x 1019 kg, roughly the size of 624 Hektor, a Trojan asteroid near Sun-Jupiter L4.  A gravitational assist manoeuvre at Jupiter to aim Hektor at Venus requires a delta-V of about 1 km/sec from Hektor's present location, with a 1-2 year post-encounter fall time to Venus. A single SRS seed would be sent to Hektor to replicate into a large factory able first to build thrusters adequate for moving Hektor to Jupiter, then later to break up the asteroid into shaped, autopiloted chunks to ensure maximum energy transfer to the Venusian atmosphere upon impact. Related impact methods have already been proposed [20-21].

With Venus' atmosphere gone, its surface begins to cool. A sunscreen twice the diameter of the solar disk positioned at Sun-Venus L1, constructed of 50 gm/m2 solar sail material [22], has a mass of 3 x 1013 kg which can most cheaply be obtained from the Moon. Hence, simultaneous with the launching of the Hektor seed, another standard SRS unit is dispatched to the Moon where in 18 years it replicates into a factory covering less than 2% of the lunar surface, able to manufacture the screens in seven more years and then to generate the launching energy to Venus orbit in ten years. The screen arrives just after Hektor's impact, perhaps 50 years from the start of the programme.

With its atmosphere gone and a >90% reflective screen in place, Venus's crust should cool to 373 K in about 300 years assuming weak mantle-crust thermal coupling. After that, a third SRS seed unit lands on Venus and replicates 5.9 x1010 copies of itself in 26 years. This planetary factory can then be reprogrammed to produce 7.9 x1017 kg O2, enough to provide a 150 mbar breathable atmosphere, in 134 years. As for the source of oxygen, Khodakovsky [23] has estimated that about 1020 kg H2O may be bound in hydrated minerals in the 33-km Venusian crust, and that oxygenated minerals may contain an additional 1020-1021 kg oxygen. The SRS field produces more than enough energy to release these resources. Although 650 years of cooling are nominally required for surface temperatures to reach a human comfortable 300 K, the liberation of only 10% of the crustal water produces an ocean of mean global depth 22 m which may hasten global cooling during its condensation. Hence the terraforming time for Venus via SRS is >= 500-600 years.

Other options may be added to this plan. If it is desired to save some of the nitrogen and carbon from the original atmosphere, biological concentrators might be released in the Venusian atmosphere prior to the arrival of Hektor to precipitate these elements on to the crust for later salvage. If Venus' rotation rate must be increased, another medium size Trojan asteroid could be deorbited, immediately following Hektor to prograde impact, adding 1028 joules to Venus' rotational energy producing a 100-hour day.


4.2 Additional Considerations

One option to eliminate the atmosphere is to garden the topmost 100 km of Venus to promote mineralogical oxidation once. biological methods have decarbonised the air [20]. While this scheme may or may not produce the desired result, the only hope of carrying it out in reasonable time is to use a fleet of SRS-generated burrowing robots. Another option is to bury rather than disperse the excess atmosphere of Venus. In this scenario, with the sunscreen in place, the planet cools to near 200 K in about 3600 years, at which point the CO2 atmosphere would first liquefy and later freeze, precipitating as snow. During this time SRS machines could prepare giant caverns deep in the crust, draining off and burying the CO2 liquid slurry as it condenses. These caverns could then be sealed off permanently, of be mined for volatiles as required. More speculative is the possibility of outfitting the crustal caverns with giant canted rock nozzles capable of generating an exhaust velocity (>= Venus escape velocity) of at least 10.3 km/sec as the planet is allowed to re-warm and the CO2 to re-vaporise underground. 4.7 x1020 kg of atmosphere ejected at escape velocity would impart a maximum rotation to the planet of 66 hours/day and would also rid Venus of the excess gas.

The many advantages cited earlier in regard to an SRS approach to terraforming Mars are equally applicable here. Human colonists would arrive to find a planet not only terraformed but also industiformed - teeming with industrial machinery and habitats in addition to air and water. To maintain an Earthlike temperature the solar insolation must be permanently reduced by 50%, more if greenhousing is taken into account, or else Venus will become unbearably hot. This reduction may be accomplished either by maintaining a sunscreen at L1 or a reflective cloud layer around the planet. This layer could either be of natural origin or may consist of a fleet of high-altitude reflective balloons manufactured, launched, and maintained by the planetary SRS factory system.



Self-replicating systems, whether machine, biological, or hybrid, offer the only feasible means for organising matter on a planetary and galactic scale [9]. Machine SRS are preferred because they can be designed to survive a wider range of environmental conditions than biological-based systems, can be efficiently and rapidly reprogrammed, and because of their far greater versatility yield. a wider range of output products.

In principle, the design and deployment of SRS by humanity will permit planetary engineering of new habitats for man of an entirely new order. For example, assuming the standard 100-ton, 1-year SRS could be redesigned for the task, replicating machines could be assigned to disassemble whole planets, converting their substance into factory machines, interstellar probes or arks, or O'Neill-style space colonies. Neglecting crowding effects, transport times, latent planetary heat, and a host of other complicating factors, SRS devices replicating at the maximum rate could theoretically convert all of the mass of the Moon or Belt asteroids to orbiting machinery in 42 years, all of Mercury in 43 years, all of Jupiter in 52 years, and even the mass of the Sun in 59 years. Perhaps these timescales cannot be realised by 1-2 orders of magnitude, but still this indicates the tremendous organising potential of SRS.


  1. Boyce, Chris, Extraterrestrial Encounter: A Personal Perspective, David and Charles, Newton Abbot, London, 1979.
  2. Dyson, Freeman, Disturbing the Universe, Harper and Row Publishers, New York, 1979.
  3. Freitas, Robert A. Jr., "A Self-Reproducing Interstellar Probe," JBIS, 33, 251-264 (1980).
  4. Von Tiesenhausen, Georg and Darbro, Wesley A., Self-Replicating Systems - A Systems Engineering Approach, NASA TM-78304, July 1980.
  5. Long, James E. and Healy, Timothy J., Advanced Automation for Space Missions: Technical Summary, A Report of the 1980 NASA/ASEE Summer Study on the Feasibility of Using Machine Intelligence in Space Applications, University of Santa Clara, 15 September 1980, NASA CR-163827.
  6. O'Neill, Gerard K., Driggers, G. and O'Leary, B., "New Routes to Manufacturing in Space," Astro. ?Aero., 18, (10). 46-51 (1990).
  7. Tipler, F. J., "Extraterrestrial intelligent beings do not exist," Phys.s Today, 34, (9), 70-71 (1981).
  8. Freitas, Robert A. Jr. and Zachary, W., "A Self-Replicating Growing Lunar Factory," in 5th Princeton/AIAA/SSI Conference on Space Manufacturing, 18-21 May 1981, Princeton University, AIAA, NY, 1982.
  9. Freitas, Robert A. Jr., Gilbreath, W., Healy, Timothy J. and Long, James E. (eds.), Advanced Automation for Space Missions: Final Report of the 1980 NASA/ASEE Summer Study on the Feasibility of Using Machine Intelligence in Space. Applications, University of Santa Clara, NASA Conf. Publ. 2255, December 1982.
  10. Sagan, Carl, "Planetary Engineering on Mars," Icarus, 20, 513-514 (1973).
  11. Vondrak, Richard R., "Creation of an Artificial Lunar Atmosphere," Nature, 248, 657-659 (1974).
  12. Burns, Joseph A. and Harwit, Martin, "Towards a More Habitable Mars - or - The Coming Martian Spring," Icarus, 19,126-130 (1973).
  13. Sagan, Carl, "The Planet Venus," Science, 133, 849-858 (1961).
  14. Averner, M. M. and MacElroy, R. D., On the Habitability of Mars: An Approach to Planetary Ecosynthesis, NASA SP-414, 1976.
  15. Oyama, Vance I. et al, "The Chemical Activities of the Viking Biology Experiments and the Arguments for the Presence of Superoxides, Peroxides, g-Fe2O3 and Carbon Suboxide Polymer in the Martian Soil," In R. Holmquist, ed., Life Sciences and Space Research XVI, (COSPAR), pp. 3-8, Pergamon Press, NY, 1978.
  16. Semenov, N. N., "Changes in the Martian Atmosphere," in B. P. Konstantinov, V. D. Pekelis, eds., Inhabited Space, Part 1, p. 192, NASA TT-F-819, February 1975.
  17. Levine, Joel S., "A New Estimate of Volatile Outgassing of Mars," Icarus, 28,165-169 (1976).
  18. Mutch, T. A., Arvidson, R. E., Head, J. W. III, Jones, K. L. and Saunders, R. S., The Geology of Mars, Princeton University Press, Princeton, NJ, 1977.
  19. Pollack, James B., "Mars," Sci. Amer., 233, 106-117 (1975).
  20. Oberg, James E., New Earths, Stackpole Books, Harrisburg, 1981.
  21. Adelman, Saul, "Terraforming Mars," Spaceflight, 24, 50-53 (1982). Also "Can Venus be Transformed into an Earth-Like Planet," JBIS, 35, 3-8 (1982).
  22. Brereton, Roy G. (ed.), Discussion Meeting on Gossamer Spacecraft (Ultralightweight Spacecraft): Final Report, JPL Publication 80-26, 15 May 1980.
  23. Khodakovsky, I. L. et al, "Venus: Preliminary Prediction of the Mineral Composition of Surface Rocks," Icarus, 39:352-363 (1979),  [ADS Abstract]; Erratum Icarus 41:471 (1980) [ADS Abstract]

Created: July 20, 1998
Original HTML Editor: Robert J. Bradbury
Last Modified by Freitas: 14 March 2008