Introduction

You know you’ve been around for a long time if you get asked to write a perspective on something. I started working in electric propulsion (EP) in 1978 and have been doing so ever since. This article gives my perspective on how electric propulsion went from a small, but enthusiastic band of technologists around the world to literally darkening the skies with EP satellites. The primary messages from the perspective that follow are that for new technologies to make the transition from development to wide spread application, the benefits have to be compelling and rock solid, perseverance is essential, and the exact timing of the adoption of any new technology can both be a matter of luck and dependent on developments outside of your control.

Fundamentals

Mission designers have long bemoaned the “tyranny of the rocked equation” (see [1] for example). For electric propulsion technologists, however, the rocket equation is your friend. Space is big and getting around in time scales of interest requires large velocity changes. This fundamental fact of nature inescapably drives the need for high exhaust velocities. But high exhaust velocities by themselves are not sufficient for a practical propulsion system.

The first EP conference I attended was the International Electric Propulsion Conference (IEPC) at Princeton, NJ in 1979. This was already the 14th IEPC, which at the time were held approximately every 18 months, indicating that dedicated EP conferences had been ongoing for over 20 years. My major impression from this conference was “holy cow” there are a lot of different electric propulsion concepts including: direct-current and radio frequency ion thrusters of various sizes and various propellants; solid-, liquid-, and gaseous-propellant magnetoplasmadynamic (MPD) thrusters; pulsed inductive thrusters (PIT); pulsed plasma thrusters (PPT); rail guns; and mass drivers. The market place would ultimately select a few of these concepts for wide-spread application. The technologies selected for application would change over time in response to new EP developments and changing market place needs. The technologies not selected would remain in the category of “advanced technologies” where “advanced” here means low on NASA’s Technology Readiness Level (TRL) scale [2]. NASA would continue to support research on many of these concepts, but as Earl Vanlandingham, who was an outstanding program manager for power and propulsion technologies at NASA Headquarters, once told me, “Advanced propulsion technology is like parsley, you want a little on your plate, but you don’t want too much.”

Power, lots of power

When I was doing my graduate research at Colorado State University in the late 1970’s, my advisor Professor Paul Wilbur would occasionally reflect on the fact that NASA had been supporting the development of EP technologies for 20 years but there still weren’t any real applications or even concrete plans to implement the technology. He used to tell me that, “we just need a mission.” But to “just get a mission” required power, lots of power, and for that the needs of electric propulsion would have to wait for available, affordable power in space catch up.

In the 1950’s, estimations of the power required for an EP mission took the following equation for the power per spacecraft mass, P/MS/C,

$$\frac{P}{M_{S/C}}=\frac{a{v}_{ex}}{2\eta }$$

and plugged in an acceleration of a = 0.01 g, an exhaust velocity of vex = 100 km/s, and an efficiency of η = 0.7 to get a required power per unit mass of 7 kW/kg. Half a century later, NASA’s Dawn spacecraft that used an ion propulsion system to rendezvous with the giant main belt asteroid (4) Vesta and the dwarf planet (1) Ceres, had an initial wet mass of 1218 kg. According to the above calculation the Dawn spacecraft would need an estimated power level of order 8.5 MW. Furthermore, in 1958, the first solar-powered spacecraft, Vanguard (https://en.wikipedia.org/wiki/Vanguard_1), was launched with an array of 6 solar cells that produced ~ 1 W of photovoltaic power. It is no wonder then that early expectations assumed practical EP systems would be nuclear powered.Footnote 1 It did not turn out that way.

The actual maximum input power to Dawn’s EP system was 2.5 kW provided by a photovoltaic solar array that produced slightly more than 10 kW of electrical power beginning of life at 1 AU [3]. This was possible because in the 1960’s, trajectory specialists determined that practical deep space missions could be accomplished with much lower vehicle accelerations than originally expected. Practical, missions could be performed with accelerations of order 10− 5 g instead of the 10− 2 g assumed above. In addition, exhaust velocities of roughly 30 km/s were found to be better suited to near-term missions of interest instead of 100 km/s. These two changes result in a factor of ~ 3300 reduction in the power required per unit mass, reducing 8.5 MW for a Dawn-sized vehicle to ~ 2.6 kW.

In parallel with improved low-thrust mission designs, the development of photovoltaic solar arrays for spacecraft advanced rapidly with the maximum solar array power doubling roughly every 4 years from the 1960’s through the early 2000’s. This resulted in the situation where by the 1990’s available, affordable power in space had caught up to the needs of electric propulsion resulting in the adoption of EP in various forms on commercial geosynchronous communication satellites. For example, beginning in 1997 the Hughes 601HP satellite used solar array power margin to operate gridded ion thrusters for station keeping. The use of EP on geosynchronous comsats significantly advanced the thruster technology in the US, Europe and Japan, and also increased customer acceptance of EP as a valuable cost-savings tool rather than a high-risk gadget.

An alternate illustration of the advance in solar array power and affordability is the Dawn spacecraft that was selected for flight implementation in 2001. Even though Dawn was a relatively small, cost-capped deep space robotic science mission, it could afford a higher-power solar array than that which flew on the human-tended Skylab space station from the 1970’s.

Propellants: many are called, few will serve

The first propellants used in early ion thrusters were mercury and cesium. These propellants were attractive because they had a large atomic mass, were easy to ionize, and were relatively easy to vaporize. They could also be stored as a liquid at low pressure and room temperature resulting in small tankage fractions. In spite of these desirable features and despite decades of research, no actual operational electric propulsion systems ever used them as propellants. The best these propellants could manage was a few flight experiments.

In the 1960s and 70’s, technologists demonstrated that you could operate ion thrusters and Hall thrusters on pretty much anything you could turn into a vapor. But this did not automatically indicate that any of these materials would be useful as a propellant in an operational propulsion system. In fact, a good way to tell that a new electric propulsion technology is both immature and in “sell mode” is the claim that it can be used with a wide variety of propellants.

To actually be an attractive propellant for electric thrusters requires consideration of factors other than those mentioned above, i.e., large atomic mass, easy ionization, easy vaporization, and low tankage fraction, which may be considered as an initial filter or acceptance criteria. Electric propulsion systems, of course, are part of a spacecraft, and have to operate for typically thousands to tens of thousands of hours. Therefore, the potential contamination of the spacecraft by the propellant is a critical consideration. Not all of the propellant injected into an electric thruster gets ionized and exhausted at high velocity. Some fraction, of order 10% or so, “leaks” out of the thruster unionized. This low velocity exhaust can get partially ionized and envelop the host spacecraft. No surface of the spacecraft can be guaranteed to be free from propellant deposition. Furthermore, the long operating times for electric thrusters requires extensive life testing in ground-based vacuum test facilities. Contamination of these increasingly expensive facilities by hazardous propellants is also a major consideration.

In the 1980s, NASA’s Lewis Research Center (now the Glenn Research Center) switched from mercury to xenon as the propellant of choice for ion thruster technology development. The Soviet Union, of course, had been using xenon in their Hall thrusters (or Stationary Plasma Thrusters) since the 1970s and NASA had been investigating the operation of ion thrusters on noble-gas propellants for many years before the switch to mercury.

The use of xenon has several key benefits as a propellant for spacecraft. Aside from meeting the acceptance criteria listed above, xenon is inert for all intents and purposes. It poses no contamination hazard to the host spacecraft or to ground test facilities. The flow rate of xenon can be precisely controlled using feed system components (valves, pressure regulators, etc.) that were developed and flight qualified for other applications. This enabled the straight forward and affordable development of propellant feed systems for ion thrusters.

Of course, there are always exceptions. In some cases, condensable propellants may be indispensably fundamental to the operation of a particular type of thruster. If the benefits of the thruster technology are sufficiently compelling, then the issues of spacecraft contamination and ground test facility contamination would have to be dealt with.

Thruster life

The power-limited, low-thrust nature of electric propulsion typically requires thrusters capable of operating for several thousand to tens of thousands of hours. Efforts to demonstrate adequate thruster life date back to the very beginnings of electric propulsion technology development. While there are many, many examples of life testing efforts (see for example Ref [4]). An illustrative example is the development of NASA’s 30-cm-diameter, J-Series mercury ion thruster. This thruster development began in 1970 and ended with the demise of NASA’s Solar Electric Propulsion Stage (SEPS) program in 1981. The thruster development went through ten design iterations beginning with the 100-series thruster, then 200-, 300-, and so on. When the program got to the tenth iteration, they did not want to call it the 10-hundred series, so instead called it the J-series, with J being the tenth letter in the alphabet. While this development program identified and solved many technical issues, its key challenge was demonstrating adequate thruster life. The goal to demonstrate thruster life drove many of the design changes generally resulting in the next thruster generation and corresponding higher thruster-series number. Typically, multiple thrusters of each generation were fabricated and tested. These thrusters were serialized, i.e., 101, 102 or 201, 202, etc. So important was the life demonstration goal that as soon as the latest series thruster was fabricated and acceptance tested, it was placed into a long-duration life test, i.e., thruster #101, or 201, 301, etc. This led Bob Bechtel—who worked at NASA LeRC (now GRC) and finished his career at MSFC—to state the unwritten rule of EP thruster development, “you always life test serial #001.”

It wasn’t until the NSTAR (NASA’s SEP Technology Applications Readiness) project in the 1990’s that an ion thruster intended for primary propulsion demonstrated adequate life to be useful for missions of interest.Footnote 2 Demonstration of the NSTAR thruster life began with a scheduled 2000-hr test at NASA LeRC. This test did not go well, revealing several accelerated erosion mechanisms, causing the test to be terminated after just 1000 hrs [5]. Straight forward fixes to the thruster were made and a second 1000-hr wear test was performed, this time at JPL [6]. This wear test demonstrated that the changes made after the first 1000-hr wear test successfully addressed the accelerated erosion issues setting the stage for an 8000-hr life demonstration test. The duration of 8000 hours was selected as the minimum time (and propellant throughput) that would enable useful deep space science missions with the NSTAR technology.

A vacuum test facility at JPL was created with the specific objective of being used for long-duration gridded ion thruster testing. At the time, demonstrating thruster life was one of the major outstanding technical issues impeding adoption of the ion propulsion technology and there were no facilities at NASA dedicated to addressing this issue. There were lots of test facilities at NASA GRC and JPL that were used for electric thruster testing, but there were also lots of demands on these facilities and none of them could be tied up for the multiple years necessary to qualify a single thruster for life. Somewhat unexpectedly, as soon as the facility at JPL was operational, it was used to demonstrate the life not of a gridded ion thruster, but of the SPT-100 thruster [7]. At the completion of the SPT-100 life test, the facility was dedicated to wear testing of the NSTAR gridded ion thruster for the next 8 years.

During the 1000-hr NSTAR thruster wear test at JPL, a flow rate calibration error led to discovery that efficient discharge chamber operation could be achieved at lower discharge voltages than previously believed [6]. Specifically, discharge voltages of around 24 V instead of the more typical 28 V were found to yield performance nearly as good as at 28 V. Operation at 24 V resulted in significant lower internal erosion rates making the trade off between thruster performance and thruster life lean heavily in favor of thruster life. This operating mode was carried over to the 8000-hr NSTAR life demonstration test (LDT) [8]. Not everyone involved in the NSTAR project, however, was convinced that the right choice was made in this tradeoff between performance and life. Consequently, for 100 hours nearly right in the middle of the LDT, the operating point was changed to the 28-V condition. This is indicated in Fig. 1 as the “100 Hour PAT Test.” These data indicate that the 28-V condition results in an increase in the double ion fraction in the ion beam from about 13% to 22%. Since most of the internal erosion in a gridded ion thruster is known to be caused by doubly-charged ions, this increase along with the higher discharge voltage would have had a disastrous effect on the thruster life. After 100 hours, the operating point was returned to the low discharge voltage condition to finish out the LDT. It was subsequently used in the 5-year-long extended life test (ELT) and in both the DS1 and Dawn missions.

Fig. 1
figure 1

Discharge current and voltage (top) and double ion fraction (bottom) as a function of time during the 8000-hr Life Demonstration Test (LDT). The “100 Hour PAT Test” indicates the period of time in which the thruster flow rates were changed to what was perceived by some to be a more efficient operating point. The problem, as these figures clearly indicate, is that this operating point results in a significant increase in discharge voltage and double ion content, to the extent that the result would be a much shorter thruster life capability

Full size image

The key to EP thruster life qualification is to identify all of the important wear-out failure modes and to understand their behavior. Unexpected results must be thoroughly investigated and understood. As Jay Polk from JPL would say, “if it seems weird, it is,” meaning it is not acceptable to discount weird results as “anomalous” and then ignore them.

The extended life test (ELT) was the final test in the NSTAR life qualification effort [9]. This test used the flight spare thruster from DS1 and served as the thruster life qualification test for the Dawn mission. The test spanned a duration of 5 years (from 1999 through 2003). At this same time, NASA began the NEXT (NASA’s Evolutionary Xenon Thruster) program to develop a bigger, better, higher-power, higher specific impulse, ion thruster. This set up a resource battle that lasted throughout the duration of the ELT. Every year of the ELT there was a serious threat to cancel the life test and redirect the resources to NEXT. These threats finally succeeded in the 5th year even though the test was still producing useful information about long-duration operation at low power. Cancellation of the test increased the risk for the already-selected Dawn mission whose mission design required extensive operation at lower. The ELT was cancelled in favor of supporting the NEXT technology, a technology which, as it turned out, would not fly for another 20 years, and even then only as an extremely limited technology demonstration (https://en.wikipedia.org/wiki/Double_Asteroid_Redirection_Test).

The ELT demonstrated over 30,000 hours of thruster operation, covered the full range of power throttling and provided sufficient data to enable the life qualification of the NSTAR thruster for the Dawn mission. It provided high confidence that the NSTAR thruster could perform the Dawn mission with a low risk of wear-out failure.

False starts

From the 1970s through the early 1990s, there were many attempts by NASA to establish flight projects that took advantage of the significant benefits enabled by electric propulsion. Some notable examples are the Comet Haley Rendezvous mission [10, 11], the SEPS (Solar Electric Propulsion System) program [12], and the Space Exploration Initiative (https://en.wikipedia.org/wiki/Space_Exploration_Initiative).

The Comet Haley Rendezvous mission study included a shoot-out between high-power solar electric propulsion and large solar sails. The SEP system proposed the use of an array of ten mercury ion thrusters with an Isp of 4750 s and a 60-kW concentrator solar array [10]. The solar sail system proposed the use of multikilometer-long blades in a “heliogyro sail configuration [11]. The SEP system won the competition, but it was a hollow victory, since the mission was not funded. In hindsight, both sides of the competition were significantly exaggerating their technology readiness and flight implementation of either system would have been extremely high risk.

Around 1979, NASA initiated the SEPS program [12]. The objective of this program was to develop a high-energy upper stage that could be used for multiple deep space science missions. The candidate missions included a comet rendezvous mission, a Saturn orbiter with probe mission, and a multiple main-belt asteroid rendezvous mission. The stage was to be based on the use of the 2.5-kW, J-Series, mercury ion thruster and a 15-kW solar array. The project funded two Phase-B contractor studies. The results of these studies were presented at the Marshall Space Flight Center (MSFC) at the end of 1980. The project was cancelled as part of federal budget considerations early in 1981 before the winning contractor was selected. This effectively ended all work at NASA on mercury-fueled ion thrusters.

In the early 1990’s, NASA briefly flirted with what became known as the Space Exploration Initiative (SEI) (https://en.wikipedia.org/wiki/Space_Exploration_Initiative)—a plan to send humans back to the Moon, this time to stay and then go on to do human exploration of Mars. This stimulated a lot of studies into high-power nuclear electric propulsion (NEP) systems and technologies since NEP was one of the options, albeit not the leading option, for human missions to Mars. As EP technologists, we embraced this initiative wholeheartedly. The initiative, however, was short-lived and effectively cancelled in 1993 due mostly to the estimated price tag. This, of course, wasn’t the first round of NEP studies for human missions to Mars and it wouldn’t be the last.

Stationary plasma thrusters

At the start of the 1990s, the Soviet Union began marketing selected space technologies to the west. One of these technologies was the Stationary Plasma Thruster (SPT). The claimed performance for the SPT-100 was an 50% efficiency at an Isp of 1500 s. This was simultaneously attractive and the subject of some skepticism. In fact, the skepticism was sufficiently high that Len Caveny of the Ballistic Missile Defense Organization (BMDO) funded a small team that I led to travel to the Soviet Union in 1991 and experimentally verify the performance of the SPT-100. This team included: John Barnett, a former supervisor of JPL’s electric propulsion group; John Sankovic, an electric propulsion technologist at the Glenn Research Center; and David Barnhart, a propulsion technologist for the U.S. Air Force.

Doubts about the performance claims largely centered on the following (flawed) logic. The SPT-100 efficiency was higher that that of a gridded ion thruster. In gridded ion thrusters, the acceleration process is in practice greater than 99% efficient (depending, of course, on how you define this efficiency). In this case, it’s defined as the ratio of the ion beam current to the beam current plus the ion current to the accelerator grid). With such a high efficiency, the argument went, how could a Hall thruster be more efficient?

Performance verification in the Soviet Union

We brought three large containers of instruments from JPL to the Soviet Union. We used this equipment to verify calibration of the thrust stand, calibration of the propellant flow meters, the vacuum chamber pressure, and the makeup of residual gasses in the vacuum chamber. The engineers and technicians at both the Scientific-Research Institute of Thermal Processes in Moscow (NIITP, now the Keldysh Research Center) and Design Bureau Fakel in Kaliningrad, Russia were extremely helpful and efficient in adapting our equipment to theirs to enable these measurements. The result of the evaluation performed using a combination of U.S. and Soviet instrumentation was complete confirmation of the performance claims within the limitations of the instrumentation and facilities. The explanation for why the efficiency was higher than that of a gridded ion thruster is that the SPT-100 produces and extracts into the ion beam ions more efficiently than gridded ion thrusters. That is, the energy cost to produce ions (typically in eV/ion) that make it into the exhaust is lower in the SPT-100, and this becomes an increasingly important term in the efficiency calculation as the specific impulse is reduced to 1500 s where the SPT-100 is designed to operate.

SPT-100 life testing at JPL

As with gridded ion thrusters, the major remaining technical issue with Hall thruster technology in the early 1990s was thruster life. The higher plasma densities enabled by the non-space-charge-limited ion acceleration in Hall thrusters were expected to result in increased erosion rates and shorter thruster lifetimes. For this reason, BMDO funded JPL to perform a life test of the SPT-100 thruster. This test successfully demonstrated 5730 hours of operation and also identified a previously unknown wear-out failure mechanism—erosion of the non-operating cathode [7]. The combination of good performance and acceptable thruster life for the SPT-100 led to the widespread adoption of Hall thruster technology around the world.

Deep Space 1

By 1990, NASA had been supporting the development of 30-cm diameter gridded ion thruster technology for 20 years. This began with the mercury-fueled, divergent-field, 100-series thruster in 1970 [13] evolving eventually into a 5-kW, xenon-fueled, ring-cusp thruster by 1990 [14]. It was clear by 1990 that a flight test was needed to push the technology to completion. In a meeting at NASA HQ, I asserted that since they had been funding the development of 30-cm gridded ion thruster technology for 20 years, they should either finish it by developing flight hardware for a flight test on a U.S. Air Force satellite called ELITE (Electric Insertion Transfer Experiment) [15] or terminate the development. Fortunately, NASA’s response was to start the NSTAR project with the objective of developing the flight hardware.

The ELITE satellite was intended to flight-test a high-power, ammonia-fueled arc jet. As a result, it had the all-important power resources available that could be used to operate the NASA-provided ion thruster. The teaming arrangement between the Air Force and NASA was expected to improve the political prospects for ELITE, making it more likely to be funded to completion. This was not to be the case. In efforts to keep ELITE alive it was renamed SSTAR (Space Surveillance Track and Autonomous Reposition) and the corresponding NASA ion thruster development program was named NSTAR in a show of support, but ELITE/SSTAR was eventually cancelled. This left NASA with a program to develop the NSTAR flight hardware with nothing to fly it on.

At the beginning of the NSTAR program, the maximum thruster power level had to be selected. At the time, NASA GRC had been developing a 30-cm gridded ion thruster to operate at 5 kW. However, to make it easier to fly, engineers at JPL made a convincing case to reduce the maximum input power to 2.5 kW. This would reduce the size and cost of the solar array required to power the ion propulsion system. It also had critical side benefit of significantly improving the thruster life. This follows the well-known approach of trading thruster performance for improved thruster life. The next step along this path was the simplification of the thruster throttle capability.

The NSTAR ion thruster is capable of operating over a broad range of ion beam currents and voltages represented by an area on a graph of beam current vs beam voltage, with input power as a parameter as indicated in Fig. 2. Roy Kakuda, a systems engineer at JPL, looked at the envelope of possible throttle points and said, “We cannot fly this. It will be impossible to flight qualify the thruster for operation at all these operating points.” We eventually settled on a much simpler throttling approach that maintained the specific impulse as high as possible for as long as possible as the input power was reduce. This was obtained by operating at a fixed beam voltage and reducing the beam current. When the limit of this approach was reached, the beam voltage was then reduced at a fixed low beam current. This throttling approach gave good performance over a range of deep space science missions of interest. Mission designers and thruster technologists quickly got used to it and the full throttling envelop was quietly forgotten.

Fig. 2
figure 2

Throttling envelope capability for the NSTAR ion thruster and the simplified throttling curve used for flight

Full size image

As luck would have it, just as SSTAR was being cancellation, NASA was initiating the New Millennium program designed specifically to test new technologies in space. The first of the New Millennium missions was designated Deep Space 1 (DS1Footnote 3). Also, at this time the NASA Director Dan Goldin instituted what I refer to as the “Goldin Rule.” This rule states that NASA deep space science missions must account for the full life-cycle cost of the mission including the launch vehicle. Since one of the primary benefits of electric propulsion is to enable the use of smaller, less expensive launch vehicles, whether intended or not, the Goldin Rule immediately made electric propulsion an important technology at NASA. Prior to this, when I would talk to experienced project managers at JPL about the use of electric propulsion, the response was typically along the lines of, “Why would I want to risk my mission on your wacky new propulsion technology. I don’t care about it’s ability to enable the use of a smaller launch vehicle because I don’t pay for the launch vehicle anyway.”

With the Goldin Rule in place, Joel Sercel (who was at JPL at the time) and I approached the DS1 management team and informed them of the ongoing NSTAR project and how it could provide an ion propulsion system that DS1 could flight test. Since it was a separately funded project, DS1 would not have to pay for its development. We asserted that this would be the most important technology that DS1 could demonstrate. After some deliberation, the DS1 management came back to me and said they agreed that, “Ion propulsion is the most important technology we could demonstrate on DS1. Unfortunately, we can’t fly it because we can’t afford the large solar array it requires. However, if you give us a free solar array, we’ll fly the NSTAR ion propulsion system.”

What’s bad luck for some can be good luck for others. Around this time, BMDO launched the METEOR (https://ntrs.nasa.gov/api/citations/20050177149/downloads/20050177149.pdf) satellite to flight-test the linear concentrator solar array called SCARLET (Solar Concentrator Array with Refractive Linear Element Technology) they had been developing. Failure of the Conestoga launch vehicle on October 23, 1995 wound up plunging the satellite into the ocean. Recognizing that BMDO still had an interest in flight testing the SCARLET array, I approached Len Caveny and said they could get a free ride into space for the SCARLET array if they provided it to NASA’s DS1 mission. Len was skeptical that NASA would actually complete the NSTAR project and that failure to do so could result in the delay or cancellation of DS1. I suggested that if NASA runs into trouble with the NSTAR development, BMDO could step in and provide an SPT-100 system in its place. Len Caveny agreed to provide the SCARLET array for DS1. With a free solar array in hand, DS1 agreed to fly the NSTAR ion propulsion system. NASA successfully developed the NSTAR system on time and it ultimately operated for over 16,000 hours, providing a record-shattering ΔV to the spacecraft of 4.5 km/s.

Shortly after launch, however, it wasn’t clear that this would be the case. The first attempts to turn on the thruster failed due to a grid-to-grid short. Upon hearing this, Dan Goldin said, “Just turn on the other one.” A reasonable suggestion except there was only one NSTAR ion thruster on the spacecraft. The grid-to-grid short was cleared by changing the spacecraft attitude to alternately have the thruster face into and away from the sun, thermally cycling the grids.

DS1 demonstrated how to fly an SEP mission in deep space where the normal state of the spacecraft is thrusting with the ion propulsion system. This is very different from conventional spacecraft that coast the vast majority of the time. It identified the need for missed thrust margin and how to manage it. It identified and took advantage of the ability to exchange margins between mass, power, and missed thrust to minimize overall risk. It demonstrated how to navigate a low-thrust trajectory and it demonstrated that interactions between the IPS and the rest of the spacecraft including EMI, solar array current collection, and contamination from non-propellant efflux from the spacecraft were all manageable. Most importantly, it moved the perception of solar electric propulsion from an “advanced concept” to a legitimate option for deep space science missions.

Dawn

NASA’s DawnFootnote 4 mission was selected in the first Discovery Mission opportunity after the completion of the DS1 mission. It got off to a rocky start. As part of NASA’s competitive procurement process, each proposal team hosts a “Site Visit” for the review team to visit the proposer’s facility and have their outstanding questions addressed. For Dawn the site visit was to take place at Orbital Science’s (now Northup Grumman’s) facility in Dulles, WA. The site visit was scheduled on September 12, 2001. We were just starting a dress rehearsal on September 11, 2001 when the historic events of that day terminated the rehearsal and postponed the site visit for a few months. Since air traffic was grounded nationwide on September 11th, most of the JPL personnel at Orbital Sciences gathered together their largest rental cars and carpooled back to JPL in a 48-hr, non-stop trek across the country.

Ultimately, NASA informed Chris Russel, the Dawn Mission PI from UCLA, telling him, “The good news is that you’ve been selected. That bad news is that you have to wait a year before getting started.” This delayed the planned launch date from the summer of 2005 to the summer of 2006. Over the course of its implementation, Dawn would ultimately have three different project managers. The second manager, facing cost and schedule pressures, desperately wanted to eliminate the relatively expensive, complicated ion propulsion subsystem. In addition, Joe Makowski, the Orbital Sciences chief systems engineer would lament, “This propulsion system is so complicated, why do you want to go through all the trouble?” But, the rocket equation says that a dual main belt asteroid rendezvous mission is not practical without ion propulsion, and so it remained on the spacecraft.

Dawn was canceled twice due to projected cost overruns. The first time was at the confirmation review following the project’s preliminary design review (PDR). In acquiescence to the recommendation by NASA’s standing review board, the Dawn management team changed the mission design from a dual asteroid rendezvous mission to a single asteroid rendezvous plus a flyby of a second main belt asteroid in order to fit within the cost cap. At the confirmation review, the head of NASA’s Science Mission Directorate cancelled the mission saying, “You were selected because of the ability to rendezvous with two main belt asteroids. If you can’t do that, then the mission will not go forward.” To save the mission, Orbital Sciences agreed to give up their fee for the spacecraft development. This saved enough money that the Project could show that accomplishment of the dual asteroid rendezvous mission could be done within the available resources. The mission was then uncancelled 4 days after it was cancelled.

Dawn’s second cancellation was during Phase C/D, again due to projected cost overruns. JPL’s laboratory director, Charles Elachi, objected to the cancellation saying NASA did not follow its own process for cancelling the mission. A four-month investigation into the technical and financial status of the mission revealed that the mission had a high probability of successfully completing its development. Based on the findings from this review, NASA uncancelled Dawn. The delay, however, slipped the launch date from the summer of 2006 to June of 2007.

By June 2007, the Dawn spacecraft was fully fueled and sitting on top of its Delta II Heavy launch vehicle at Space Launch Complex 17-B (SLC 17-B) at the Cape Canaveral Air Force Station (now the Cape Canaveral Space Force Station). At the same time, NASA’s Phoenix spacecraft was on top of its Delta II launch vehicle at SLC 17-A (Fig. 3). The two launch complexes are so close together that NASA rules did not permit launching from one complex with a high-value payload on the other. Phoenix was headed for Mars and, consequently, had a tight constraint on its launch period. Dawn’s highly-capable ion propulsion system on the other hand provided much more flexibility in launch dates even though Dawn had to rendezvous with two main belt objects. The decision was made to take Dawn off the launch vehicle, launch Phoenix, then re-integrate Dawn with the launch vehicle. This slipped Dawn’s launch date to the end of September 2007.

Fig. 3
figure 3

The Dawn spacecraft (SLC 17-B) and Phoenix spacecraft (SLC 17-A) at the Cape Canaveral Air Force Station in June 2007

Full size image

One impact of this slip was that we had to do the final close-out of the ion propulsion system on top of the rocket twice, once in June 2007 and again in September 2007. In both cases, we performed these close-out activities from midnight to 3 am. One of the coolest things I have ever gotten to do was, instead of taking the elevator down from the white room (the room at the top that provides access to the spacecraft), we took the stairs down around the Delta-II Heavy rocket in the middle of the night.

The ion propulsion subsystem on Dawn was extremely successful. It operated for over 50,000 hours and provided a ΔV to the spacecraft of 11.5 km/s. It demonstrated without a doubt the long-awaited promise that solar electric propulsion could enable missions with characteristic velocities much higher than practical with chemical propulsion systems. Dawn demonstrated the ability to do orbit insertion with electric propulsion. It demonstrated orbit transfer using EP at bodies with initially poorly known gravity fields, and it was the first spacecraft ever to orbit to different target bodies.

SEP as a shiny new hammer

The successes of Deep Space 1, Dawn, SMART-1 (https://en.wikipedia.org/wiki/SMART-1), and Hayabusa 1 and 2 (https://en.wikipedia.org/wiki/Hayabusa, https://en.wikipedia.org/wiki/Hayabusa2), resulted in SEP being regarded as a “shiny new hammer” and mission designers went about looking for other nails to pound in. In parallel, over the last 20 years at JPL there was a concerted effort to develop models and simulations of the important performance and life-limiting phenomena taking place in gridded ion and Hall thrusters. This work produced codes like CEX [16], ORCA2D [17], Hall2de [18], etc. These models not only explained how thrusters work but are also used to guide development and scaling to other power levels and specific impulses that could then enable new mission concepts.

Many of these mission concepts, as is typical of most new mission concepts, didn’t make it to flight. This included the nuclear-powered Jupiter Icy Moons Orbiter (JIMO) mission [19], another round of human missions to Mars studies [20], and the Asteroid Redirect Mission (ARM) [21]. One of the interesting new results that came out of this latest round of human missions to Mars studies was the development of SEP/Chem hybrid architectures that could provide the mass savings of all-SEP systems with trip times comparable to all chemical propulsion systems [20]. Given the ongoing rapid development of solar array technology that is continuing the process of making solar arrays larger, higher power, less massive, and less expensive, my expectation is that when people do actually go to Mars they will use a SEP/Chem architecture.

The Asteroid Redirect Mission, which NASA came to call the Asteroid Redirect Robotic Mission (ARRM), would have used a high-power SEP system to catch up to, capture, and return to lunar orbit an entire small near-Earth asteroid (NEA). Small, in this case, refers to an asteroid roughly 8 m in diameter with a mass of several hundred metric tons. Difficulty in identifying and adequately characterizing such small objects led NASA to redefine the mission to pick a roughly 2-m diameter boulder off the surface of a larger NEA and return it to lunar orbit. The original ARM concept would have been humanity’s first attempt at altering the solar system for our benefit. ARM went from a $50 K, NASA-sponsored study in 2011 to part way through Phase B of a flagship-class directed mission by 2016 before beging cancelled due to lack of congressional support. Future asteroid mining activities may find the ARM approach attractive, in which an asteroid of several hundred tons is delivered to the Earth-Moon system where material extraction and benefaction processes can be developed and tested relatively close to home.

On a personal note, ARRM marked the third SEP mission I worked on in my career that was cancelled in Phase B: SEPS in 1981, Dawn in 2004, and ARRM in 2016. Fortunately, Dawn, the only competitively-selected mission in this group, was uncancelled as discussed above.

Present and near future

Solar electric propulsion is now firmly in the mainstream of propulsion options for both commercial satellites and deep space science missions. As of this writing there are more than 190 GeoComsats using Hall or ion thrusters [22] and over 2500 LEO satellites, most of which make up the Starlink constellation (that use krypton-fueled Hall thrusters) (https://en.wikipedia.org/wiki/Starlink). BepiColombo (https://www.esa.int/Science_Exploration/Space_Science/BepiColombo_overview2) launched in 2018 is using its ion propulsion system to rendezvous with Mercury in 2025. The Psyche mission (https://www.nasa.gov/psyche), with a possible launch in 2023, will use a 20-kW solar array and SPT-140 thrusters to rendezvous with the main belt asteroid (16) Psyche. Two Janus spacecraft (https://nssdc.gsfc.nasa.gov/nmc/spacecraft/display.action?id=JANUS) originally planned to be launched with Psyche would use Field Emission Electric Propulsion thrusters to fly by two binary asteroids: 1996 FG3 and 1991 VH. NASA’s Power and Propulsion Element (PPE) of the Lunar Gateway [23], scheduled for launch in late 2024, will use a 60-kW solar array and a combination of 13-kW Hall thrusters [24] and 6-kW Hall thrusters [25] to deliver the PPE along with the Habitation and Logistics Outpost (HALO), to lunar orbit. Since the Lunar Gateway will be a human tended spacecraft, the PPE will represent the first use of EP in a system with astronauts onboard. With PPE/Gateway, electric propulsion applications will have expanded to impact human exploration of space. Finally, the Mars Sample Return (https://www.airbus.com/en/newsroom/press-releases/2021-06-earth-return-orbiters-first-step-to-mars) program plans to use ion propulsion for the ESA-provided Earth Return Orbiter to return the Mars samples to Earth. This would represent electric propulsion’s entry into multibillion-dollar flagship robotic missions.

On the horizon

NASA’s latest decadal survey [26] identifies the following missions to be included in the New Frontiers 6 announcement of opportunity: Centaur orbiter and lander; Ceres sample return; Comet surface sample return; Enceladus multiple flyby; Lunar Geophysical Network; Saturn probe; Titan orbiter; and Venus In Situ Explorer. Of these missions, Ceres sample return and Comet surface sample return are known to benefit significantly from the use solar electric propulsion (see [27] for example for the Ceres Sample Return). SEP may provide benefits to other missions in this list, but that remains to be determined. The decadal survey also identifies planetary defense as an area of national importance and specifically identifies ion beam deflection [28] as a planetary defense technique that should be developed.

The decadal survey also identifies the next flagship missions to follow the current Mars Sample Return program. Topping the list are an Enceladus orbiter and lander mission, and a Uranus orbiter and probe. Typically, missions to these solar ranges of roughly 10 and 20 AU, respectively, would use plutonium-powered, radioisotope thermoelectric generators (RTGs). However, solar arrays for space applications have continuously gotten, larger, lighter, less expensive, and able to operate farther from the Sun. Solar-powered missions to 5 AU are no longer unusual [29,30,31] and solar cells have been successfully tested under conditions simulating solar ranges as far as 30 AU (equivalent to Neptune’s orbit). Advances in solar array technology may make solar-powered missions to 10 and 20 AU feasible and more affordable than RTG-powered missions. Coupling very large, lightweight solar arrays with electric propulsion may ultimately enable non-nuclear exploration throughout the solar system [32].

Farther in the future, asteroid mining has long been known to largely be a transportation problem [33]. Cost-effective transportation using some form of high-power solar electric propulsion will likely be required for viable asteroid mining operations.

Conclusions

The long-awaited promise of electric propulsion and its ability to reduce overall mission costs and enable high ΔV missions is being fulfilled. Electric propulsion is in the process of infiltrating every aspect of space exploration and exploitation from cubesats and constellations of LEO satellites to Geo comsats, competed and flagship robotic deep space science missions, and even human space exploration missions. This feat is the result of three key developments. First and foremost is the development of affordable solar arrays that provide the power needed by EP systems. Second is the development of EP thruster technologies that have sufficient life to accomplish missions of interest. Third is the demonstration that EP systems are compatible with the rest of the spacecraft in terms of EMI/EMC, contamination, and guidance, navigation, and control.

The fact that EP has infiltrated all aspects of space endeavors does not mean that its best days are behind it. Quite the contrary, the best days for electric propulsion are most certainly in the future. This is driven by the inescapable fact that the solar system is big and that there will always be a desire to get around it faster. This will drive missions to ever higher ΔV’s driving the need for significantly better EP systems. Accompanying this drive to better EP systems will be a corresponding development of ever larger, lighter, less expensive solar arrays to provide the required power.