We dream of go-

ing to the stars.

As young chil-

dren growing up

in the 1950s, my friends and I were

awestruck by the possibility of space

travel. As I have learned over the years,

our fascination was not unique to my

Costa Rican upbringing. Indeed, many

of my co-workers today, coming from

different parts of the globe, recount

similar childhood longings. In the past

50 years, I have had the opportunity to

witness the development of the first

ships that have transported humans be-

yond Earth. In the past 20, I have been

fortunate to ride on some of these

rockets and to get a firsthand glimpse

of wonders I could only imagine be-

fore. It would seem as if we are des-

tined to burst off our fragile planet and

move into the cosmos in a new human

odyssey. Such a mighty undertaking

will dwarf the westward European ex-

pansion of the 16th century.

Rby Franklin R. Chang Díaz

PAM

FRA

NC

IS

RADICALLY NEW ROCKET engine, just threemeters long, produces an exhaust of unprecedent-ed speed: 300 kilometers per second. The windowon the right reveals the glowing plasma inside. Fora cutaway view, see page 92.

TheVASIMR

ocketThere used to be two types of rocket:powerful but fuel-guzzling,or efficient but weak.

Now there is a third option that combines the advantages of both

Yet we lack the ships required to ven-ture far into the vastness of space. Withtoday’s chemical rockets, a trip to Marswould take up to 10 months in a vulner-able and limited spacecraft. There wouldbe little room for useful payload. Mostof the ship’s mass would be taken up bypropellant, which would be spent in afew short bursts, leaving the ship tocoast for most of the journey [see “Howto Go to Mars,” by George Musser andMark Alpert; Scientific American,March]. If people were to travel toMars under these conditions, their bod-ies and minds would suffer consider-ably. Months of exposure to weightless-ness would weaken their muscles andbones, and the persistent radiation ofouter space would damage their im-mune systems.

To be safe, human interplanetaryspacecraft must be fast, reliable and ableto abort in the event of malfunction.Their propulsion systems must be capa-ble of handling not just the cruise phaseof the journey but also the maneuveringnear the origin and destination planets.Whereas chemical propulsion can con-tinue to provide excellent surface-to-orbit transportation, new technologiesare required to send humans to the plan-ets and ultimately to the stars.

Plasma rockets are one such technolo-gy. Utilizing ionized gases accelerated byelectric and magnetic fields, they in-crease performance far beyond the lim-its of the chemical rocket. My researchteam has been developing one of theseconcepts, the Variable Specific ImpulseMagnetoplasma Rocket (VASIMR),since the early 1980s. Its genesis datesback to the late 1970s, when I was in-volved in the study of magnetic ductsand their application to controlled nu-clear fusion. In such ducts, a magneticfield insulates a hot plasma from itsnearest material surface, letting it reachtemperatures of hundreds of millions ofdegrees Celsius.

I theorized that a duct, properlyshaped, could form a magnetic nozzleand convert the plasma energy to rock-et thrust. Such a structure functions likea conventional rocket nozzle but canwithstand much higher temperatures.Further investigation suggested that thesystem could also generate a variableexhaust, adaptable to the conditions offlight, just as an automobile transmis-sion matches the power of the engine tothe needs of the road. Although theidea of variable exhaust dates to theearly rocket pioneers, its implementa-tion in chemical rockets with fixed ma-

terial nozzles has proved impractical. InVASIMR the concept is finally poisedto become a reality.

Newton’s Rocket

The principle of rocket propulsionstems from Newton’s law of action

and reaction. A rocket propels itself byexpelling material in the direction op-posite to its motion. The material isusually a gas heated by a chemical reac-tion, but the general principle appliesequally well to the motion of a simplegarden sprinkler.

Rocket thrust is measured in newtonsand is the product of exhaust velocity(relative to the ship) and the rate ofpropellant flow. Quite simply, the samethrust is obtained by ejecting eithermore material at low velocity or less athigh velocity. The latter approach saveson fuel but generally entails high ex-haust temperatures.

To gauge rocket performance, engi-neers use the term specific impulse (Isp),which is the exhaust velocity divided bythe acceleration of gravity at sea level(9.8 meters per second per second). Al-though thrust is directly proportional toIsp, the power needed to produce it isproportional to the square of the Isp.

92 Scientific American November 2000 The VASIMR Rocket

QUARTZ TUBE

Step 4ICRH ANTENNAheats gas to 10 million kelvins

Step 2HELICON ANTENNA ionizes gas

Step 5VACUUM CHAMBERcaptures hot gas as itescapes magneticconfinement

Step 3MAGNETIC COILSgenerate field that confines the ionized gas

HOW IT WORKSANTENNA FEEDS

Step1INJECTOR feeds hydrogen or helium gas

WINDOW

SAM

UEL

VEL

ASC

O

Therefore, the power required for a giv-en thrust increases linearly with Isp. Inchemical rockets this power originatesin the exothermic reaction of the fueland oxidizer. In others, it must be im-parted to the exhaust by a propellantheater or accelerator. Such systems de-pend on a power source elsewhere inthe ship. Solar panels are generallyused; the abundant power requirementsof human space exploration, however,will favor nuclear reactors [see box onpage 97]. This is especially true for mis-sions beyond Mars, where sunlight isrelatively feeble.

In our quest for high fuel efficiency,hence high Isp, my research team hasmoved away from chemical reactions,in which the temperature is only a fewthousand degrees, and entered the realmof plasma physics, in which the temper-ature is high enough to strip the atomsof some (if not all) of their electrons.The temperature of a plasma starts atabout 10,000 degrees Celsius, but pres-ent-day laboratory plasmas can be 1,000times hotter. The plasma is a soup of

charged particles: positive ions and neg-ative electrons. At these temperaturesthe ions, which have the bulk of themass, move at velocities of 300,000 me-ters per second—60 times faster than theparticles in the best chemical rockets.

Usually, by design, the power outputof the engine is kept at a maximum, sothrust and Isp are inversely related. In-creasing one always comes at the ex-pense of the other. Therefore, for thesame propellant, a high Isp rocket deliv-ers a greater payload than a low Isp one,but in a longer time. If a rocket couldvary thrust and Isp, it could optimizepropellant usage and deliver a maxi-mum payload in minimum time. I callthis technique constant power throttling(CPT). It is similar to the function of anautomobile transmission in climbing ahill or the feathering of a propeller en-gine in moving through the air.

One way to visualize CPT is by con-sidering the way in which the ship ac-quires kinetic energy from the exhaust.If this process were totally efficient, theexhaust particles (viewed by a ground

observer) would leave the ship at rest;the ship would be moving at the ex-haust speed. All the exhaust energywould have been given to the ship.Thus, for a slow ship, an appropriatelyslow exhaust better utilizes the powersource. As the ship speeds up, a faster(hotter) but leaner exhaust gives betterresults. Under CPT, the ship starts athigh thrust for rapid acceleration. As itsspeed increases, Isp gradually increasesand thrust decreases for greater fueleconomy. A car does exactly the samething when it starts in low gear andsteadily shifts up.

Bouncing Back and Forth

VASIMR embodies a class of mag-netic ducts called magnetic mirrors.

The simplest magnetic mirror is pro-duced by two ring electromagnets withcurrent flowing in the same direction.The magnetic field is constricted nearthe rings but bulges out in betweenthem. Charged particles move in a helixalong field lines, orbiting around themat a specific radius, the Larmor radius,and at the so-called cyclotron frequen-cy. As one might expect, for a field of agiven strength, the heavier particles (theions) have a lower cyclotron frequencyand larger Larmor radius than the lightones (the electrons) do. Also, strongfields lead to a high cyclotron frequencyand small Larmor radius. In VASIMR,the ion cyclotron frequency is a fewmegahertz (MHz), whereas its electronequivalent is in the gigahertz range.

The particles’ velocity has two com-ponents: one parallel to the field (corre-sponding to the forward motion alongthe field line) and the other perpendicu-lar (corresponding to the orbital mo-tion around the line). When a particleapproaches a constricted (hence strong-er) field, its perpendicular velocity in-creases, but its parallel one is reducedproportionately to keep the total energy

www.sciam.com Scientific American November 2000 93

LIKE A CAR GEARSHIFT, and unlike other rockets, VASIMR can adjust its output.By increasing its temperature, it boosts its specific impulse (blue) and reduces fuel con-sumption (yellow)—at the price of less thrust (red). The power is constant 10 megawatts.

MAGNETIC MIRRORtraps the particles sothey can be heated to 10million kelvins. It con-sists of two ring electro-magnets that set up abulging magnetic fieldbetween them.

0.1

1

10

103

104

105

THRUST

FUEL

Fuel

Cons

umpt

ion

(gra

ms p

er se

cond

)

T(k

li)

Plasma Temperature (kelvins)

Spec

ific I

mpu

lse (s

econ

ds)

Thru

st (n

ewto

ns)

SPECIFIC IMPULSE

107106105104

10

100

1,000

JOH

N D

ESA

NTI

S

Near each magnet, the field lines tilt. A compo-nent of the force now pushes the particles awayfrom the magnet. If the particles are moving to-ward the magnet slowly enough,they are stoppedand reversed. In that region, the helix tightens.

Near the center of the mirror, thefield lines are parallel, so the mag-netic force is radial. Particles travelat a constant speed along a helix ofnearly constant radius.

JUA

N V

ELA

SCO

constant. The reason has to do with thedirection of the force exerted by thefield on the particle. The force is alwaysperpendicular to both the particle’s ve-locity and the field direction. Near thecenter of the mirror, where the field linesare parallel, the force is radial and sohas no effect on the parallel velocity. Butas the particle enters the constriction,the force tilts away from the constric-tion, resulting in an imbalance that de-celerates the particle [see top illustrationon preceding page]. If the particle is exit-ing the constriction, the field has the op-posite effect and the particle accelerates.Because no energy has been added, the

acceleration comes at the expense of ro-tational motion. The magnetic field doesno work on the particle; it is simply avehicle enabling this energy transfer.

These simple arguments hold as longas the field constriction is slow and grad-ual compared with the particle motion—

a condition known as adiabaticity. Intheir curling motion around the fieldlines, the particles are guided by them,but like fast-moving vehicles on a slip-pery highway, they can follow only linesthat do not curve sharply.

A magnetic mirror can trap particles ifthey are sufficiently slow to be reflectedat the field constrictions. The particles

bounce between them until somethingdisrupts their parallel velocity such that itovercomes the trap or until one of theconstrictions is reduced. With enoughparallel velocity, the particle will pushthrough and accelerate on the otherside. Sudden changes in the velocity of atrapped particle, which may be enoughto untrap it, can be brought about byrandom events such as collisions withother particles, interaction with electro-magnetic waves, or plasma instabilitiesand turbulence. The magnetic field ofEarth is a natural mirror. Charged parti-cles from the ionosphere bounce backand forth between the North and South

94 Scientific American November 2000 The VASIMR Rocket

LOOKING DOWN THE BARREL of the rocket, you see the plasma coming straight at you. The window is 15 centimeters across.PA

M F

RAN

CIS

Poles. Some of them penetrate deep intothe upper atmosphere, creating the spec-tacular auroras seen at high latitudes.VASIMR uses three such magnetic struc-tures, linked together: a forward plas-ma injector, which ionizes the neutralgas; a central power amplifier, which en-ergizes the plasma; and an aft magneticnozzle, which finally ejects it into space.

Beam in the Power

Most plasma rockets require physi-cal electrodes, which erode quick-

ly in the harsh environment. In contrast,VASIMR uses radio antennas. The radio

waves heat the plasma just like a micro-wave oven heats food. Two wave pro-cesses come into play. First, neutral gasin the injector stage becomes a denseand comparatively cold (about 60,000kelvins) plasma through the action ofhelicon waves. These are electromagnet-ic oscillations at frequencies of 10 to 50MHz, which, in a magnetic field, ener-gize free electrons in a gas. The electronsquickly multiply by liberating other elec-trons from nearby atoms in a cascade ofionization. Although the details are poor-ly understood, helicons are widely usedin semiconductor manufacturing.

Once made, the plasma flows into the

central stage, where it is heated by fur-ther wave action; the waves of choicehere, however, are slightly lower-fre-quency ion cyclotron oscillations, sonamed because they resonate with thenatural rotational motion of the ions.The wave’s electric field is perpendicularto the external magnetic field and ro-tates at the ion cyclotron frequency. Theresonance energizes the perpendicularmotion of the particles. This effect,known as ion cyclotron resonance heat-ing (ICRH), is widely used in fusion re-search. The central stage is ultimately re-sponsible for the high Isp of the rocket.

One might wonder how a boost in

www.sciam.com Scientific American November 2000 95

PAM

FRA

NC

IS

BILLOWING CLOUDS OF WATER VAPOR pour off the magnets, which are kept at liquid-nitrogen temperature.

the ions’ perpendicular motion couldimpart any useful momentum to therocket exhaust. The answer lies in thephysics of the magnetic nozzle, the finalstage of VASIMR. The diverging fieldhere transfers energy from the perpendi-cular motion to the parallel motion, ac-celerating the ions along the exhaust.Being much more massive, the ionsdrag the electrons along, so the plasmaexits the rocket as a neutral fluid. InVASIMR, this nozzle expansion occursover a distance of about 50 centimeters.

Once the expansion is complete, theplasma must detach from the rocket.Recent studies by Roald Sagdeev of theUniversity of Maryland and Boris Breiz-man of the University of Texas highlightthe basic physics. The model involvesthe Alfvén speed, named after Swedishphysicist Hannes Alfvén, who first de-scribed it. Disturbances in a magnetizedplasma propagate along the field at thisspeed. In a magnetic nozzle, the Alfvénspeed plays a role similar to that of thesound speed in a conventional nozzle.

The transition from sub-Alfvénic tosuper-Alfvénic flow delineates a bound-ary beyond which the flow downstreamhas no effect upstream—which ensuresthat the detachment exerts no drag onthe rocket. The VASIMR nozzle is de-signed to expand the plasma past thisboundary, to a point where the energycontent of the field is small comparedwith that of the plasma flow. The plas-ma then breaks free, carrying with it asmall amount of the field. A similar be-havior is thought to occur in naturewhen solar flares detach from the mag-netic field of the sun. The energy expen-diture in the field distortion only mini-mally taxes the performance of therocket. Our studies show that plasmadetachment occurs one to two metersaway from the nozzle throat.

Partitioning the Power

The throttling that makes VASIMRdistinctive is done mainly by chang-

ing the relative fraction of power goingto the helicon and ICRH systems. Forhigh thrust, power is routed predomi-nantly to the helicon, producing moreions at lower velocity. For high Isp, morepower is diverted to the ICRH, withconcomitant reductions in thrust. Weare also studying two other exhaustvariation techniques, including a mag-netic choke at the nozzle throat for highIsp and a plasma afterburner for highthrust at very low Isp.

A key consideration is the efficiencyof the engine over its operating range.Creating a hydrogen plasma costs about40 electron volts per electron-ion pair.(An electron volt, or eV, is a unit of en-ergy commonly used in particle physics.)This energy expenditure is not availablefor propulsion; most of it is frozen inthe creation of the plasma. The parti-cles’ kinetic energy, which is what ulti-mately generates the thrust, must beadded to this initial investment. In theearly prototype of VASIMR, at high Isp,the kinetic energy is about 100 eV perion. Thus, a total energy expenditure of140 eV yields 100 eV of useful energy,or about 70 percent efficiency. LaterVASIMR designs will reach exhaust en-ergies of 800 to 1,000 eV for the sameinitial investment, leading to greater effi-ciency. At low Isp, as more plasma is gen-erated for higher thrust, the kinetic ener-gy per particle gets uncomfortably closeto the ionization energy, with consequentreductions in efficiency. In the end, how-ever, efficiency must be evaluated in thecontext of the overall mission. Some-times brief bursts of high thrust may infact be the most efficient approach.

The gas ionization involves other in-efficiencies. Neutral atoms lingering inthis initial plasma cause unwanted pow-er losses if they remain mixed with en-ergetic ions. In an effect known as chargeexchange, a cold neutral atom gives anelectron to a hot ion. The resulting hotneutral is oblivious to the magneticfield and escapes, depositing its energyon nearby structures. The cold ion leftbehind is virtually useless.

To avoid this, we are studying a radi-al-pumping technique, in which thecold neutrals are siphoned out beforethey wander into the power-amplifica-tion stage. They may be reinjected down-stream of the nozzle throat, where theions are already moving in the right di-rection and charge exchange actuallyhelps the plasma to detach from therocket. Charge exchange is a seriousproblem in fusion research today.

Although the helicon is able to ionizenearly any gas, practical considerationsfavor light elements such as hydrogenand helium. For example, the ICRHprocess is easiest in light gases, whosecyclotron frequencies at reasonable fields(about one tesla) are compatible withexisting high-power radio technology.Fortunately, because hydrogen is themost abundant element in our universe,our ships are likely to find an ample sup-ply of propellant almost everywhere. An-

96 Scientific American November 2000 The VASIMR Rocket

MARS TRAJECTORY

EMERGENCY ABORT

MARS ORBIT INSERTION

INTERPLANETARY CRUISE

25 Earth Radii

Orbit Insertion(t + 246 days)

5 Mars Radii

Returrn(t + 169 days)

Abort(t + 44 days)

Final Orbit(t + 250 days)

25,000 Earth Radii

EARTH ESCAPE

Escape(t + 30 days)

Low Orbit(t + 0 days)

VASIMR can turn around if somethinggoes wrong early in the cruise.

After flying by Mars to drop off the as-tronauts,the ship lowers itself into orbit.

Unlike ordinary rockets VASIMR nevercoasts.It shifts to higher gears,reachingan Isp of 30,000 seconds 75 days into theflight, and then starts to decelerate.

To get to Mars,the ship must first breakfree of Earth’s gravity. Using low gear(maximum thrust and an Isp of 3,000seconds),VASIMR builds up speed.

JOH

N D

ESA

NTI

S

Crew Landing(t + 115 days)

Mar

s Orbit

Earth

Orbit

other important engineering challenge isthe generation of strong magnetic fields.We are investigating new high-tempera-ture superconductors, based on bismuthstrontium calcium copper oxide com-pounds. The magnets will use the cryo-genic hydrogen propellant to cool them.

Early VASIMR experiments, which Ibegan in the 1980s with Tien-Fang Yangand others at the Massachusetts Insti-tute of Technology, have led to the pres-ent research program. The centerpiecetoday is the VX-10 prototype at theNational Aeronautics and Space Ad-ministration Johnson Space Center inHouston. Two smaller experiments atOak Ridge National Laboratory andthe University of Texas support the in-vestigations. We also have partnershipswith Rice University, the Princeton Plas-ma Physics Laboratory, the Universityof Michigan, the University of Mary-land and the University of Houston, aswell as with private industry and withNASA centers in Huntsville, Ala.; Cleve-land; Greenbelt, Md.; and Norfolk, Va.

We are now formulating plans to testVASIMR in space. A proposed mid-2004demonstration involves a 10-kilowattsolar-powered spacecraft, which willalso study Earth’s radiation belts. In an-other test, a VASIMR engine will try toneutralize the atmospheric drag on theInternational Space Station. Recent ex-perimental results and rapid progress inthe miniaturization of radio-frequencyequipment bode well for such space tests.

As a natural progression, a possibleVASIMR human Mars mission involvesa 12-megawatt system. The ship wouldclimb on a 30-day outward spiral fromEarth and cruise through interplanetaryspace for 85 more days, acceleratingmuch of the way and then deceleratingfor arrival at Mars. The trip would betwice as fast as one involving chemicalrockets. A crew module would detachand land using chemical rockets, whilethe mother ship would fly by the planetin a fuel-efficient trajectory to rejoin itfour months later. To protect the hu-man crew, the Mars vehicle would be

provided with a robust abort capabilityby virtue of its variable exhaust. Its mag-netic field and the hydrogen propellantwould act as a radiation shield.

VASIMR could serve as a precursorto the great dream of those of us in thespace program: a fusion rocket. Such aship would have 10 to 100 gigawatts atits disposal. Although controlled fusionremains elusive, the efforts to achieve ithave been relentless and the progresssteady. Future generations will use it forrapid access to the planets and beyond.We now find ourselves preparing thegroundwork for achieving that vision.

www.sciam.com Scientific American November 2000 97

The Author

FRANKLIN R. CHANG DÍAZ was scheduled to make his first spaceflight on the ill-starredChallenger in January 1986, but at the last minute his team was moved up one flight. Since then,he has flown five missions, highlights of which include the deployment of the Galileo probe toJupiter, the first two tests of space tethers and the final shuttle Mir docking. In addition to his as-tronaut career and his plasma research (in which he has a Ph.D. from the Massachusetts Insti-tute of Technology), he has worked in mental health and drug rehabilitation programs.

Further Information

ELECTRICAL PROPULSION IN SPACE. Gabriel Giannini in Scientific American, Vol. 204, No. 3,pages 57–65; March 1961.

ION PROPULSION FOR SPACE FLIGHT. Ernst Stuhlinger. McGraw-Hill, 1964.THE DEVELOPMENT OF THE VASIMR ENGINE. F. R. Chang Díaz et al. in Proceedings of theInternational Conference on Electromagnetics in Advanced Applications, paper 99-102.Torino, Italy, Sept. 13–17, 1999. For copies of this paper and the next, visit spaceflight.nasa.gov/mars/technology/propulsion/aspl on the World Wide Web.

THE PHYSICS AND ENGINEERING OF THE VASIMR ENGINE. F. R. Chang Díaz, J. P. Squire, R.D. Bengtson, B. N. Breizman, F. W. Baity and M. D. Carter. American Institute of Aeronau-tics and Astronautics, conference paper 2000-3756; July 17–19, 2000.

POWER RICH

In space, power is life. In the spring of 1970 the Apollo 13astronauts managed to stay alive by judicious use of theirprecious battery power. Had their return flight taken

longer, they would have met with disaster.The electrical re-quirements of human spaceflight are set by basic survivalneeds:rapid transportation and life support.The space shut-tle consumes about 15 kilowatts in orbit; the InternationalSpace Station, 75 kilowatts. Estimates for a Mars habitatrange between 20 and 60 kilowatts—not including propul-sion.For a baseline Mars mission,a VASIMR engine would re-quire about 10 megawatts.Higher power means faster tran-sits.A 200-megawatt VASIMR would get to Mars in 39 days.

For human forays into near-Earth space, chemical fuelsand solar panels provide sufficient power. But for Mars andbeyond, chemical fuels are too bulky and the sun’s rays tooweak. A 10-megawatt solar array, for example, would beabout 68,000 square meters at Mars and 760,000 at Jupiter.Such gigantic panels are impractical; in comparison, the so-lar panels on the International space station are 2,500square meters.There is only one source: nuclear.

In the past,nuclear electricity has generally been obtained

from “nuclear batteries”—radioisotope thermoelectric gen-erators (RTGs), which rely on the heat generated by the nat-ural radioactive decay of plutonium. Such devices haveproved crucial to robotic space missions but are too ineffi-cient for human flight.Far better would be a nuclear reactor,which relies on the fission of uranium in a chain reaction.For each kilogram of fuel, a reactor produces up to 10 mil-lion times more power than an RTG does.

To measure the efficiency of power sources, space engi-neers use a parameter called alpha, which is the ratio ofpower plant mass to electrical output. Low alphas corre-spond to high efficiency and high power. Present solar ar-rays, operating near Earth, have alpha values of about 100.RTGs manage an alpha of 200. But for uranium reactors, al-phas can be as low at 0.5.

Reactors are inherently safer than RTGs, because the reac-tor and fuel can be launched separately and assembled in anorbit far from Earth.Even critics who call for a ban on nuclearpower in Earth orbit have acknowledged its importance fordeep-space missions [see “Nuclear Power in Space,”by StevenAftergood et al.;SCIENTIFIC AMERICAN,June 1991]. —F.C.D.

SA