Nuclear Fusion from First Principles — Vol.9: Fusion Propulsion

Series: Nuclear Fusion from First Principles (9 of 10)
Previous: Vol.8: Alternative Confinement — Beyond the Tokamak
License: MIT

---

Executive Summary

This volume examines fusion as a propulsion technology for spacecraft. The central argument is simple: chemical rockets cannot reach the outer solar system in human-relevant timescales, fission-thermal rockets roughly double the capability, but only fusion offers the specific impulse ($I_{\mathrm{sp}}$ > 10,000 s) needed to make Mars a weeks-long trip rather than a months-long ordeal — and to make Jupiter, Saturn, and interstellar precursor missions feasible at all.

The physics is governed by two equations. Tsiolkovsky's rocket equation sets the mass penalty for any given velocity change. The specific power constraint $\alpha = P_{\mathrm{thrust}} / m_{\mathrm{engine}}$ (kW/kg) sets the timescale. Together they define a design space in which fusion occupies a unique position: high enough $I_{\mathrm{sp}}$ to carry meaningful payload, high enough thrust to avoid decade-long spirals, and — crucially — a specific power that might be achievable with compact reactors.

The word "might" is load-bearing. No fusion propulsion system has ever operated. The closest real hardware is fission-thermal (NERVA, tested 1960s, $I_{\mathrm{sp}}$ ≈ 850 s), and even its modern revival (DRACO) was cancelled in 2025 when SpaceX's falling launch costs eroded the economic case. Every fusion propulsion concept discussed here is at TRL 1–3. The gap between paper designs and flight hardware is measured in decades and tens of billions of dollars.

This volume derives the physics, surveys the concepts, and — in keeping with this series' tradition — tells you exactly where the honest uncertainties lie.

Concept	Type	$I_{\mathrm{sp}}$ (s)	Thrust (N)	$\alpha$ (kW/kg)	Fuel	TRL	Key Player
Chemical (LOX/LH₂)	Combustion	450	10⁶	10³	H₂/O₂	9	Aerojet, SpaceX
NTP (NERVA-class)	Fission thermal	850–900	10⁵	~1	H₂ + U-235	5	BWXT (DRACO†)
NEP (ion/Hall)	Fission electric	1,600–5,000	0.01–1	0.01–0.1	Xe, Kr	7	NASA, Ad Astra
DFD / Starfire	Fusion direct	10,000–50,000	5–250	0.2–1	D-³He	2	Princeton/PSS
Helicity Drive	Fusion MIF	1,500–30,000	Pulsed	~1 (goal)	D-D/D-³He	1	Helicity Space
Pulsar Sunbird	Fusion DDFD	10,000–15,000	10–100	~1 (goal)	D-³He	2	Pulsar Fusion
ICF Pulse	Fusion inertial	10,000–100,000	Pulsed	1–10	D-T/D-³He	1	Conceptual
Orion / Mini-Mag	Nuclear pulse	5,000–100,000	10⁶+	10+	Fission/fusion	2	Historical

†DRACO cancelled May 2025.

---

Table of Contents

• §1. The Tyranny of the Rocket Equation
• §2. The Three Figures of Merit
• §3. From Fission to Fusion: A Propulsion Ladder
• §4. Direct Fusion Drive — Turning Plasma into Thrust
• §5. The Magnetic Nozzle Problem
• §6. Concepts in Development
• §7. Mission Analysis — Where Fusion Changes the Game
• §8. The Fuel Question in Space
• §9. DRACO's Death and the Economics of Propulsion
• §10. Uncertainties and Honest Unknowns
• §11. Decision Matrix
• §12. Conclusion
• References

---

§1. The Tyranny of the Rocket Equation

Every spacecraft obeys the same equation. Konstantin Tsiolkovsky derived it in 1903:

$$\Delta v = v_e \ln\left(\frac{m_0}{m_f}\right)$$

where:

• $\Delta v$ is the total velocity change required (m/s)
• $v_e$ is the effective exhaust velocity (m/s), related to specific impulse by $v_e = I_{\mathrm{sp}} \cdot g_0$
• $m_0$ is the initial mass (structure + payload + propellant)
• $m_f$ is the final mass (structure + payload, propellant spent)

The ratio $m_0 / m_f$ is the mass ratio. The logarithm is merciless: doubling the mass ratio only adds one more $v_e$ of $\Delta v$. To reach Mars from low Earth orbit requires approximately 4–6 km/s (Hohmann transfer with aerocapture) to 15–30 km/s (fast transfer, 90-day class). Jupiter requires 6–10 km/s minimum. Interstellar precursor missions (1,000 AU) need 100+ km/s.

For a chemical rocket with $v_e$ = 4.4 km/s (LOX/LH₂), achieving $\Delta v$ = 15 km/s requires a mass ratio of $e^{15/4.4}$ ≈ 30. That means 97% of the vehicle is propellant. There is essentially no payload capacity for a fast Mars transfer on chemical propulsion alone.

This is not an engineering problem that better materials will solve. It is an exponential function. The only escape is to increase $v_e$ — which means changing the energy source.

Propulsion Type	$v_e$ (km/s)	$I_{\mathrm{sp}}$ (s)	Mass Ratio for $\Delta v$ = 30 km/s
Chemical (LOX/LH₂)	4.4	450	891 (impossible)
Nuclear thermal (NERVA)	8.5	850	34.4 (marginal)
Nuclear electric (ion)	30	3,000	2.7 (but years of transit)
Fusion direct (DFD)	100	10,000	1.35 (weeks to months)
Fusion exhaust (advanced)	1,000	100,000	1.03 (interstellar class)

The table reveals the fundamental asymmetry. Chemical rockets live in a regime where the exponential is crushing. Fusion rockets live in a regime where the exponential is nearly flat — the mass ratio approaches 1, meaning almost all the vehicle mass is payload and structure. This is the physics that makes fusion propulsion not merely an improvement over chemical rockets, but a qualitative change in what missions are possible.

---

§2. The Three Figures of Merit

A propulsion system is characterised by three quantities. All three matter, and optimising one often degrades the others.

2.1 Specific Impulse ($I_{\mathrm{sp}}$)

Defined as thrust per unit weight flow rate of propellant:

$$I_{\mathrm{sp}} = \frac{F}{\dot{m} \cdot g_0}$$

Equivalently, $I_{\mathrm{sp}} = v_e / g_0$. Higher $I_{\mathrm{sp}}$ means less propellant mass for a given $\Delta v$. Chemical rockets peak at ~450 s. Fusion concepts promise 10,000–100,000 s.

2.2 Thrust ($F$)

The force applied to the spacecraft. Chemical rockets produce meganewtons. Ion drives produce millinewtons. The thrust determines how quickly the vehicle accelerates and therefore whether it can execute efficient impulsive burns (Oberth effect) or must spiral slowly outward.

For human missions, thrust matters enormously. A 100-tonne crewed Mars vehicle needs at least ~10 N of continuous thrust to reach Mars in under 90 days, and ~1,000 N for a ~30-day transfer. Ion drives at 0.1 N would take years.

2.3 Specific Power ($\alpha$)

The ratio of thrust power to engine mass:

$$\alpha = \frac{P_{\mathrm{thrust}}}{m_{\mathrm{engine}}} \quad [\mathrm{kW/kg}]$$

where $P_{\mathrm{thrust}} = \frac{1}{2} F \cdot v_e$.

This is the most important and least discussed figure of merit. A fusion engine with $I_{\mathrm{sp}}$ = 10,000 s but $\alpha$ = 0.01 kW/kg is useless for fast missions — the engine is too heavy relative to its power output, and the acceleration is too low to matter within a human lifetime.

The threshold for useful interplanetary propulsion is approximately:

• $\alpha$ > 0.1 kW/kg: Outer planet missions in years
• $\alpha$ > 1 kW/kg: Mars in weeks, Jupiter in months
• $\alpha$ > 10 kW/kg: Interstellar precursor missions in decades

Chemical rockets achieve ~1,000 kW/kg but only for seconds. Fission-electric systems achieve ~0.01–0.1 kW/kg. The fusion propulsion challenge is to reach 1 kW/kg sustained — a goal that no concept has demonstrated.

The relationship between these three quantities creates a fundamental constraint known as the Stuhlinger optimum: for a given mission ($\Delta v$, trip time), there exists an optimal exhaust velocity that minimises total vehicle mass. Go too low and propellant dominates. Go too high and the engine mass dominates (because extracting higher $v_e$ from a given power source requires a heavier engine). The optimum $v_e$ is approximately:

$$v_{e,\mathrm{opt}} \approx \frac{2}{3} \frac{\Delta v}{\sqrt{1 + (2\alpha \tau / \Delta v^2)}}$$

where $\tau$ is the trip time. This equation shows that $\alpha$ is the lever arm that determines whether a mission is physically achievable.

---

§3. From Fission to Fusion: A Propulsion Ladder

3.1 Nuclear Thermal Propulsion (NTP)

NTP heats hydrogen propellant by passing it through a fission reactor core. The propellant exits at 2,500–3,000 K, limited by the melting point of the fuel elements (typically UC-ZrC composites). This yields $I_{\mathrm{sp}}$ ≈ 850–900 s — roughly twice chemical.

The technology was demonstrated in the 1960s–70s under Project NERVA (Nuclear Engine for Rocket Vehicle Application). Twenty reactors were ground-tested, accumulating over 17 hours of run time. The Phoebus-2A reactor reached 4,000 MW thermal and 930 s $I_{\mathrm{sp}}$. NERVA was flight-qualified but never flew — cancelled in 1972 when the Mars mission it was designed for was shelved.

The modern revival was DARPA/NASA's DRACO program ($499M, Lockheed Martin/BWXT). It aimed to flight-test an NTP system by 2027 using high-assay low-enriched uranium (HALEU). DRACO was cancelled in May 2025 — a decision we analyse in §9.

NTP's fundamental limit: The exhaust temperature is bounded by material constraints. No matter how powerful the reactor, hydrogen cannot exit hotter than ~3,000 K before the fuel elements disintegrate. This caps $I_{\mathrm{sp}}$ at roughly 1,000 s. To go higher, you must abandon material walls and confine the working fluid with fields — which leads to fusion.

3.2 Nuclear Electric Propulsion (NEP)

NEP uses a fission reactor to generate electricity, which then powers an electric thruster (ion, Hall-effect, or magnetoplasmadynamic). The $I_{\mathrm{sp}}$ is determined by the thruster (1,600–5,000 s for current technology), but the thrust is limited by the electrical power available.

NEP's specific power is poor (~0.01–0.1 kW/kg including reactor, power conversion, radiators, and thruster) because the Rankine or Brayton cycle that converts thermal to electrical energy requires massive radiators to reject waste heat in vacuum. In space, the only way to reject heat is radiation, which scales as $T^4$ but requires large surface areas at practical temperatures.

The JIMO (Jupiter Icy Moons Orbiter) mission concept, cancelled in 2005, was an NEP design requiring a 200 kW reactor with a total power system mass of ~36,000 kg — a specific power of ~0.006 kW/kg. Transit time to Jupiter: 6 years.

3.3 Why Fusion is Different

Fusion offers an escape from the NTP temperature limit and the NEP power conversion bottleneck through direct drive: the fusion products themselves, at energies of 3.5–14.7 MeV, are the exhaust. No intermediate conversion step is needed.

Consider the D-³He reaction:

$$\mathrm{D} + {}^3\mathrm{He} \rightarrow {}^4\mathrm{He}\ (3.6\ \mathrm{MeV}) + \mathrm{p}\ (14.7\ \mathrm{MeV})$$

The 14.7 MeV proton has a velocity of ~53,000 km/s. Even after thermalisation and dilution with added propellant, exhaust velocities of 100–500 km/s ($I_{\mathrm{sp}}$ = 10,000–50,000 s) are in principle achievable.

Equally important, D-³He fusion produces <1% of its energy as neutrons (from parasitic D-D side reactions), compared to 80% for D-T. This dramatically reduces shielding mass — a critical advantage in space where every kilogram of shielding is a kilogram not available for payload.

The energy density comparison is stark:

Energy Source	Energy Density (J/kg)	Ratio to Chemical
LOX/LH₂ combustion	1.35 × 10⁷	1
Uranium fission	8.2 × 10¹³	6 × 10⁶
D-T fusion	3.4 × 10¹⁴	2.5 × 10⁷
D-³He fusion	3.5 × 10¹⁴	2.6 × 10⁷
p-¹¹B fusion	2.9 × 10¹⁴	2.1 × 10⁷

Fusion fuel contains 25 million times more energy per kilogram than chemical propellant. The engineering challenge is converting even a tiny fraction of this into directed thrust.

---

§4. Direct Fusion Drive — Turning Plasma into Thrust

4.1 The Concept

A direct fusion drive (DFD) generates thrust by exhausting fusion products — and optionally additional propellant heated by those products — through a magnetic nozzle. No turbines, no generators, no radiators for power conversion. The fusion plasma is the exhaust.

The concept was developed at Princeton Plasma Physics Laboratory by Samuel Cohen, based on the Princeton Field-Reversed Configuration (PFRC) reactor. Now commercialised as Starfire by Princeton Satellite Systems (PSS), the design parameters for a 1 MW engine are:

• Fuel: D-³He (aneutronic)
• Thrust: 5–10 N per MW of fusion power
• $I_{\mathrm{sp}}$: ~10,000 s (variable by propellant injection)
• Specific power: ~0.2 kW/kg (current estimate), targeting 1 kW/kg
• Electrical output: ~200 kW (simultaneous with thrust)
• Neutron fraction: <1.1% of total power
• Mass: ~5,500 kg for 1 MW engine
• Dimensions: ~2 m diameter × 8 m length

The PFRC uses an odd-parity rotating magnetic field (RMF) to heat the plasma and maintain the FRC equilibrium. This is a steady-state, driven system — it does not require ignition. The RMF is generated by external antennae, avoiding the need for material elements inside the plasma.

4.2 Thrust Augmentation

A key feature of DFD is variable $I_{\mathrm{sp}}$ through propellant injection. The fusion products (3.6 MeV ⁴He, 14.7 MeV protons) have extremely high velocities but low mass flow rates. By injecting additional propellant (deuterium or hydrogen) into the scrape-off layer (SOL), the fusion products transfer energy to the propellant through collisions. This increases the mass flow rate (and thrust) while decreasing the exhaust velocity.

The trade-off follows directly from energy conservation. If the fusion power is $P_f$ and the exhaust carries kinetic energy $\frac{1}{2}\dot{m}v_e^2$, then:

$$F = \dot{m} v_e = \frac{2 \eta P_f}{v_e}$$

where $\eta$ is the conversion efficiency. Higher $v_e$ means lower $F$ for fixed power. The pilot can adjust this ratio during a mission — high $I_{\mathrm{sp}}$ for cruise, higher thrust for planetary capture.

4.3 Power Budget

The Starfire/DFD power flow (1 MW fusion):

• 35% → thrust (via magnetic nozzle)
• 30% → electrical power (via direct conversion + Brayton cycle)
• 25% → waste heat (radiated)
• 10% → recirculated for RF heating

The 30% electrical output is significant. It provides ~200 kW of power to the spacecraft — enough for instruments, communications (including laser comm), and active thermal control. For comparison, the International Space Station's solar arrays generate ~120 kW.

4.4 Current Status

The PFRC-2 experiment at PPPL has demonstrated:

• Electron heating to >500 eV by RMF
• Plasma pulse durations up to 300 ms (>10⁴ × predicted tilt instability growth time)
• FRC formation and sustainment in superconducting flux conservers

What has NOT been demonstrated:

• Ion heating to fusion temperatures (>5 keV for D-³He)
• Any fusion reactions in the PFRC
• Magnetic nozzle thrust generation
• Sustained operation at power

The planned PFRC-3 (twice the size of PFRC-2) aims to reach 10-second FRC lifetimes at ~60 million K. PFRC-4 would use live D-³He fuel at ~600 million K. No timeline or funding for either has been confirmed as of February 2026.

---

§5. The Magnetic Nozzle Problem

5.1 Principle

A magnetic nozzle uses a diverging magnetic field to convert the random thermal energy of a plasma into directed kinetic energy, analogous to a de Laval nozzle for gas. The plasma is confined radially by the field and accelerated axially as the field diverges.

The advantage over a physical nozzle is obvious: no material wall needs to survive contact with a multi-keV plasma. The magnetic field is generated by superconducting coils that never touch the exhaust.

5.2 The Detachment Problem

The fundamental challenge is plasma detachment. In a magnetic nozzle, the field lines eventually curve back toward the spacecraft. If the plasma remains frozen to the field lines (as ideal MHD predicts for a perfectly conducting plasma), the exhaust would follow the field lines back and produce zero net thrust.

For the nozzle to work, the plasma must detach from the magnetic field at some downstream distance. Several mechanisms have been proposed:

1. Resistive detachment: Collisions break the frozen-in condition. Requires the plasma to cool sufficiently downstream, which may not happen fast enough.

2. Inertial detachment: When the plasma's kinetic energy density exceeds the magnetic energy density ($\beta > 1$), the plasma can stretch and break the field lines. This is the most promising mechanism for high-power fusion nozzles.

3. Recombination: Ions recombine with electrons to form neutrals, which are unaffected by the magnetic field. Requires very high collision frequencies, unlikely in typical fusion nozzle conditions.

4. Loss of adiabaticity: When the ion gyroradius becomes comparable to the scale length of the magnetic field variation, the particle trajectories decouple from the field lines.

A 2025 review in Physics of Plasmas (Wu et al.) concludes that inertial detachment is the dominant mechanism at high power, but experimental verification at fusion-relevant parameters (multi-MW, multi-keV) does not exist. All laboratory experiments have been conducted at <1 kW with <100 eV plasmas.

This is an unsolved physics problem for fusion propulsion. VASIMR (Variable Specific Impulse Magnetoplasma Rocket) has conducted the most relevant experiments, operating at up to 200 kW, but VASIMR is an electric thruster — not a fusion device.

5.3 Nozzle Efficiency

Even with successful detachment, the nozzle has efficiency losses:

• Divergence losses: The exhaust plume is not perfectly collimated. A half-angle of 15° reduces thrust efficiency by cos(15°) ≈ 0.97, but 30° gives cos(30°) ≈ 0.87.
• Electron cooling: Electrons trapped on closed field lines drain energy from the exhaust ions.
• Radiation losses: Bremsstrahlung and synchrotron radiation from the nozzle region is lost in all directions.

Estimates for overall nozzle efficiency range from 50% to 90%, but no measurement has been made at fusion-relevant conditions.

---

§6. Concepts in Development

6.1 Princeton Starfire / DFD

Described in §4. The most mature concept with the strongest physics basis (PFRC experimental heritage). NASA NIAC has funded mission studies (Pluto orbiter, Titan explorer, interstellar precursor). Five patents issued 2017–2022, including an in-space startup method. Current cost estimate: $8.9M for 1 MWe reactor, $24.5M for 10 MWe, assuming 20-unit production.

Key risk: D-³He fuel requires temperatures 3–5× higher than D-T. The PFRC has never achieved fusion. The gap from PFRC-2 (500 eV electrons) to a working engine (>5 keV ions in D-³He) is at least three experimental generations.

6.2 Pulsar Fusion Sunbird

UK-based Pulsar Fusion (founded 2011, Bletchley) is building what it claims will be the largest practical fusion propulsion chamber. The Sunbird concept uses a Dual Direct Fusion Drive (DDFD) based on the PFRC architecture, targeting:

• 2 MW power output (thrust + electrical)
• $I_{\mathrm{sp}}$: 10,000–15,000 s
• Static fire testing: 2025 (chamber firing, not fusion)
• In-orbit demonstration of core components: 2027
• Full-scale test flights: early 2030s (projected)

Pulsar has demonstrated Hall-effect thrusters (20 km/s exhaust) and hybrid chemical rockets. In January 2025, they tested what they describe as the largest plasma engine fired in the UK, at the University of Southampton. They hold a partnership with Princeton Satellite Systems for plasma modelling using AI.

Key risk: The gap between firing a plasma chamber and achieving fusion temperatures is enormous. Pulsar has not published peer-reviewed fusion results. The claimed timeline (fusion temperatures by 2027) would require an unprecedented acceleration of PFRC physics. In December 2025, Pulsar pivoted emphasis toward commercial plasma thrusters for satellites — a revenue-generating product while the fusion work continues.

6.3 Helicity Space — Helicity Drive

Pasadena-based Helicity Space (founded 2018) takes a fundamentally different approach: magneto-inertial fusion for propulsion. The Helicity Drive uses plectonemic plasma jets — double-helical, self-organized plasma structures — merged and compressed through a magnetic nozzle.

Physics basis:

• Plectonemes are tilted, twisted spheromaks observed in the MOCHI LabJet experiment to have "unexpectedly good" confinement times
• Magnetic reconnection during jet merging heats ions to fusion temperatures
• Peristaltic magnetic compression raises the triple product
• The exhaust plasma provides both thrust and (via direct conversion) electricity

Funding: $5M seed (December 2023, led by Airbus Ventures), Lockheed Martin Ventures investment (April 2024), NASA NIAC funding for heliosphere constellation mission study (2025).

Projected capabilities:

• Mars in 2 months, Jupiter in 1 year, interstellar space in 10 years
• Scalable from 100 kW to GW power levels
• Full prototype flying in space: ~10 years from 2024

Key risk: The concept is at TRL 1. No fusion-relevant experiments have been conducted. The physics of plectoneme merging at fusion temperatures is entirely theoretical. The claimed timelines are aspirational. However, the propulsion-first approach (space vacuum is a better environment for plasma confinement) has conceptual merit, and the advisory board (including former NASA Ames director Pete Worden) lends credibility.

6.4 ICF-Based Pulsed Propulsion

Inertial confinement fusion, demonstrated at NIF (Vol.8: target gain Q = 4.13 in April 2025), can in principle be adapted for propulsion. A spacecraft detonates fusion micro-targets behind a magnetic or material pusher plate at 5–15 Hz, generating thrust through repeated impulses.

Historical concepts:

• Project Orion (1958–1965): Nuclear pulse propulsion using fission/fusion bombs. Calculated $I_{\mathrm{sp}}$ ~6,000 s, thrust ~meganewtons. Killed by Partial Nuclear Test Ban Treaty (1963). The physics works — the engineering and politics do not.
• VISTA (Vehicle for Interplanetary Space Transport Applications, LLNL): ICF target-based propulsion using laser drivers. $I_{\mathrm{sp}}$ ~17,000 s, specific power ~5 kW/kg.
• Mini-Mag Orion: Scaled-down pulsed concept using z-pinch compression of magnetised targets.

Modern relevance: Pacific Fusion's IMG technology (Vol.8) — 90% electrical efficiency pulsed power — could in principle drive a repetitive ICF propulsion system. But the unsolved engineering challenges are staggering: target fabrication at 5–15 Hz, target injection and tracking, chamber clearing between shots, and the massive driver hardware.

Key risk: NIF's driver (192 laser beams, ~300 MJ wall-plug energy per shot) weighs thousands of tonnes. Even at Pacific Fusion's improved efficiency, the driver mass problem remains severe. No ICF propulsion system has ever been designed at engineering detail.

6.5 Other Concepts

• Gasdynamic Mirror: An elongated mirror machine with exhaust through one end. NASA MSFC funded studies. $I_{\mathrm{sp}}$ ~20,000 s, 1 kW/kg projected. Not actively pursued.
• Spheromak propulsion: Repetitive formation and ejection of compact torus plasmoids. Low hardware complexity. Very early stage.
• Fusion-driven rocket (FDR): University of Washington/MSNW concept using magnetically compressed metal liners around an FRC target. NASA NIAC Phase II funded. The company (MSNW) has dissolved.
• Focus Fusion / Dense Plasma Focus: LPPFusion's p-¹¹B concept. Extremely compact. No peer-reviewed demonstration of net energy.

---

Python: Figure 1 — Fusion Propulsion Physics Landscape (click to expand)

"""
Vol.9 Figure 1: Fusion Propulsion — Physics Landscape
Seed-fixed, fully reproducible.
"""
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches

np.random.seed(42)

fig, axes = plt.subplots(2, 2, figsize=(16, 12))
fig.suptitle('Figure 1: Fusion Propulsion — Physics Landscape',
             fontsize=16, fontweight='bold', y=0.98)

# Panel A: Isp vs Thrust
ax1 = axes[0, 0]
systems = {
    'Chemical (LOX/LH2)': {'isp': 450, 'thrust': 2e6, 'color': '#95A5A6', 'size': 300},
    'NTP (NERVA)':        {'isp': 850, 'thrust': 3.3e5, 'color': '#E67E22', 'size': 250},
    'Ion (NSTAR)':        {'isp': 3100, 'thrust': 0.092, 'color': '#3498DB', 'size': 200},
    'Hall (SPT-140)':     {'isp': 1770, 'thrust': 0.3, 'color': '#2980B9', 'size': 200},
    'VASIMR (200 kW)':    {'isp': 5000, 'thrust': 6, 'color': '#1ABC9C', 'size': 200},
    'DFD (1 MW)':         {'isp': 10000, 'thrust': 8, 'color': '#E74C3C', 'size': 350},
    'DFD (10 MW)':        {'isp': 10000, 'thrust': 80, 'color': '#C0392B', 'size': 350},
    'Helicity Drive':     {'isp': 15000, 'thrust': 50, 'color': '#9B59B6', 'size': 250},
    'ICF Pulse (VISTA)':  {'isp': 17000, 'thrust': 8e4, 'color': '#F39C12', 'size': 300},
    'Orion':              {'isp': 6000, 'thrust': 3.5e7, 'color': '#2C3E50', 'size': 300},
}
for name, d in systems.items():
    ax1.scatter(d['isp'], d['thrust'], s=d['size'], color=d['color'],
                alpha=0.8, edgecolors='black', linewidth=1, zorder=5)
    ax1.annotate(name, xy=(d['isp'], d['thrust']),
                 xytext=(d['isp']*1.08, d['thrust']*1.5),
                 fontsize=6.5, fontweight='bold', va='center',
                 arrowprops=dict(arrowstyle='-', color='gray', lw=0.5, alpha=0.5))
ax1.set_xlabel('Specific Impulse (s)'); ax1.set_ylabel('Thrust (N)')
ax1.set_title('(A) Propulsion Map: Isp vs Thrust', fontweight='bold')
ax1.set_xscale('log'); ax1.set_yscale('log')
ax1.set_xlim(100, 100000); ax1.set_ylim(0.01, 1e8)
ax1.grid(alpha=0.3)

# Panel B: Mass Ratio vs Delta-v
ax2 = axes[0, 1]
dv = np.linspace(1, 100, 200)
for name, isp, color in [('Chemical (450 s)', 450, '#95A5A6'),
    ('NTP (850 s)', 850, '#E67E22'), ('NEP (3000 s)', 3000, '#3498DB'),
    ('DFD (10000 s)', 10000, '#E74C3C'), ('Advanced (50000 s)', 50000, '#9B59B6')]:
    ve = isp * 9.81 / 1000
    mr = np.clip(np.exp(dv / ve), 1, 1000)
    ax2.plot(dv, mr, label=name, color=color, linewidth=2)
ax2.set_xlabel('Delta-v (km/s)'); ax2.set_ylabel('Mass Ratio')
ax2.set_title('(B) Tsiolkovsky: Mass Ratio vs Delta-v', fontweight='bold')
ax2.set_yscale('log'); ax2.set_xlim(0, 100); ax2.set_ylim(1, 1000)
ax2.axhline(y=10, color='red', linestyle='--', alpha=0.3)
ax2.legend(fontsize=8, loc='upper left'); ax2.grid(alpha=0.3)

# Panel C: Energy Density
ax3 = axes[1, 0]
sources = ['LOX/LH2', 'Kerosene/LOX', 'U-235 Fission', 'D-T Fusion', 'D-3He Fusion', 'p-11B Fusion']
energies = [1.35e7, 1.0e7, 8.2e13, 3.4e14, 3.5e14, 2.9e14]
colors_e = ['#95A5A6', '#BDC3C7', '#E67E22', '#3498DB', '#E74C3C', '#9B59B6']
ax3.barh(sources, energies, color=colors_e, alpha=0.85, height=0.6, edgecolor='black', linewidth=0.5)
for bar, val in zip(ax3.patches, energies):
    ax3.text(bar.get_width()*1.1, bar.get_y()+bar.get_height()/2,
             f'{val:.1e} J/kg', va='center', fontsize=8, fontweight='bold')
ax3.set_xlabel('Specific Energy (J/kg)'); ax3.set_xscale('log')
ax3.set_title('(C) Energy Density by Source', fontweight='bold')
ax3.set_xlim(1e6, 1e16); ax3.grid(axis='x', alpha=0.3)

# Panel D: Specific Power vs Isp
ax4 = axes[1, 1]
concepts = {'Chemical': (450, 1000, '#95A5A6'), 'NTP': (850, 1, '#E67E22'),
    'NEP (JIMO)': (3000, 0.006, '#3498DB'), 'VASIMR': (5000, 0.05, '#1ABC9C'),
    'DFD (est.)': (10000, 0.2, '#E74C3C'), 'DFD (target)': (10000, 1.0, '#C0392B'),
    'ICF (VISTA)': (17000, 5.0, '#F39C12')}
for name, (isp, alpha, color) in concepts.items():
    ax4.scatter(isp, alpha, s=200, color=color, alpha=0.8, edgecolors='black', linewidth=1, zorder=5)
    ax4.annotate(name, xy=(isp, alpha), xytext=(isp*1.15, alpha*1.3), fontsize=7, fontweight='bold')
ax4.axhline(y=0.1, color='orange', linestyle='--', alpha=0.4)
ax4.axhline(y=1.0, color='green', linestyle='--', alpha=0.4)
ax4.axhline(y=10, color='blue', linestyle='--', alpha=0.4)
ax4.set_xlabel('Specific Impulse (s)'); ax4.set_ylabel('Specific Power (kW/kg)')
ax4.set_title('(D) The Alpha Frontier', fontweight='bold')
ax4.set_xscale('log'); ax4.set_yscale('log')
ax4.set_xlim(100, 100000); ax4.set_ylim(0.001, 10000); ax4.grid(alpha=0.3)

plt.tight_layout(rect=[0, 0, 1, 0.95])
plt.savefig('vol9_fig1_propulsion.png', dpi=200, bbox_inches='tight',
            facecolor='white', edgecolor='none')
plt.close()
print("Figure 1 saved.")

IMAGE_URL_PLACEHOLDER

§7. Mission Analysis — Where Fusion Changes the Game

7.1 Delta-v Requirements

Mission	$\Delta v$ (km/s)	Chemical Mass Ratio	Fusion ($I_{\mathrm{sp}}$=10,000 s) Mass Ratio
LEO → Mars (Hohmann)	4–6	2.5–3.9	1.04–1.06
LEO → Mars (90-day)	15–30	30–891	1.16–1.34
LEO → Jupiter (Hohmann)	6–10	3.9–9.7	1.06–1.10
LEO → Saturn orbit	8–13	6.2–19.5	1.08–1.14
LEO → Pluto flyby	12–16	15.2–37.5	1.13–1.17
LEO → 550 AU (gravitational lens)	~50	>10⁵ (impossible)	1.66
LEO → Alpha Centauri (0.1c)	30,000	Absurd	~20 ($I_{\mathrm{sp}}$=100,000 s)

The contrast is dramatic. For inner solar system missions, fusion propulsion reduces propellant to a few percent of vehicle mass, freeing capacity for payload, radiation shielding, and life support. For outer solar system missions, fusion makes them possible at all. For interstellar missions, only the highest-performance fusion concepts ($I_{\mathrm{sp}}$ > 100,000 s) are even conceivable.

7.2 DFD Mission Studies

Princeton Satellite Systems has published detailed mission analyses:

Pluto orbiter and lander (NASA NIAC Phase I/II):

• 1 MW DFD, ~1,000 kg payload
• Transit time: ~4 years (vs. 9.5 years for New Horizons flyby)
• Arrives with sufficient $\Delta v$ to orbit (not just fly by)
• 2 MW of power for instruments and laser comm upon arrival
• Return HD-equivalent video from Pluto

Titan explorer:

• DFD orbital transfer vehicle + fusion-powered electric aircraft
• Transit to Saturn: <2 years
• Aircraft powered by a second PFRC configured as a closed-loop generator
• Extended surface exploration capability

Interstellar precursor (550 AU):

• Reach the solar gravitational lens focus in ~20 years
• Conventional propulsion: >100 years or impossible

Mars crew transport:

• 2× DFD (2 MW total), ~100 tonne vehicle
• Transit time: 90 days (one-way)
• Continuous acceleration/deceleration profile
• Crew radiation exposure reduced by factor of ~3 vs. chemical (shorter transit)

7.3 The Speed Hierarchy

For a 100-tonne spacecraft to Mars (one-way):

Propulsion	Transit Time	Propellant Mass	Payload Fraction
Chemical (LOX/LH₂)	7–9 months	~300 tonnes	~15%
NTP (NERVA-class)	4–6 months	~150 tonnes	~25%
NEP (100 kW ion)	2–4 years	~10 tonnes	~60% (but slow)
DFD (1 MW fusion)	90 days	~5 tonnes	~85%
DFD (10 MW fusion)	30 days	~3 tonnes	~90%

The transition from chemical to fusion is not incremental. It is a phase transition in mission capability.

---

Python: Figure 2 — Mission Analysis & Transit Comparison (click to expand)

"""
Vol.9 Figure 2: Fusion Propulsion — Mission Analysis & Transit Comparison
Seed-fixed, fully reproducible.
"""
import numpy as np
import matplotlib.pyplot as plt

np.random.seed(42)
fig, axes = plt.subplots(2, 2, figsize=(16, 12))
fig.suptitle('Figure 2: Fusion Propulsion — Mission Analysis',
             fontsize=16, fontweight='bold', y=0.98)

# Panel A: Transit Time to Mars
ax1 = axes[0, 0]
prop_types = ['Chemical (Hohmann)', 'Chemical (fast)', 'NTP (NERVA)',
              'NEP (100kW ion)', 'DFD (1 MW)', 'DFD (10 MW)']
transit_days = [259, 150, 120, 900, 90, 30]
colors_t = ['#95A5A6', '#BDC3C7', '#E67E22', '#3498DB', '#E74C3C', '#C0392B']
bars = ax1.barh(prop_types, transit_days, color=colors_t, alpha=0.85,
                height=0.6, edgecolor='black', linewidth=0.5)
for bar, days in zip(bars, transit_days):
    label = f'{days} days ({days/365:.1f} yr)' if days > 365 else f'{days} days'
    ax1.text(bar.get_width()+10, bar.get_y()+bar.get_height()/2, label,
             va='center', fontsize=9, fontweight='bold')
ax1.set_xlabel('One-way Transit Time (days)')
ax1.set_title('(A) Mars Transit Time by Propulsion', fontweight='bold')
ax1.set_xlim(0, 1100); ax1.grid(axis='x', alpha=0.3)

# Panel B: Payload Fraction vs Destination
ax2 = axes[0, 1]
destinations = ['Mars (Hohmann)', 'Mars (90-day)', 'Jupiter', 'Saturn', 'Pluto', '550 AU']
dv_vals = [5, 20, 9, 12, 14, 50]
for ve, label, color in [(4.4, 'Chemical', '#95A5A6'), (8.5, 'NTP', '#E67E22'),
                          (98.1, 'DFD', '#E74C3C')]:
    pf = [max(0, (1.0/np.exp(dv/ve))-0.1)*100 for dv in dv_vals]
    x = np.arange(len(destinations))
    offset = {'Chemical': -0.25, 'NTP': 0, 'DFD': 0.25}[label]
    ax2.bar(x+offset, pf, 0.25, label=f'{label} ({int(ve/9.81*1000)} s)',
            color=color, alpha=0.85, edgecolor='black', linewidth=0.5)
ax2.set_xticks(range(len(destinations))); ax2.set_xticklabels(destinations, fontsize=8)
ax2.set_ylabel('Payload Fraction (%)'); ax2.set_ylim(0, 100)
ax2.set_title('(B) Payload Fraction by Destination', fontweight='bold')
ax2.legend(fontsize=9); ax2.grid(axis='y', alpha=0.3)

# Panel C: DFD vs Conventional Missions
ax3 = axes[1, 0]
missions_dfd = [('Pluto Orbiter', 4), ('Titan Explorer', 2), ('Mars Crew (RT)', 0.5),
                ('Jupiter Orbiter', 1.5), ('550 AU', 20), ('1000 AU', 30)]
missions_conv = [('New Horizons (Pluto flyby)', 9.5), ('Cassini (Saturn)', 7),
                 ('Mars Crew Chemical (RT)', 2.5), ('Juno (Jupiter)', 5),
                 ('Voyager 1 (~150 AU)', 47)]
y = 0; labels = []
for name, years in missions_dfd:
    ax3.barh(y, years, 0.6, color='#E74C3C', alpha=0.85, edgecolor='black', linewidth=0.5)
    ax3.text(years+0.3, y, f'{years} yr', va='center', fontsize=9, fontweight='bold')
    labels.append(name); y += 1
y += 0.5
for name, years in missions_conv:
    ax3.barh(y, years, 0.6, color='#95A5A6', alpha=0.6, edgecolor='black', linewidth=0.5)
    ax3.text(years+0.3, y, f'{years} yr', va='center', fontsize=9, fontweight='bold')
    labels.append(name); y += 1
ax3.set_yticks(range(len(labels))); ax3.set_yticklabels(labels, fontsize=7)
ax3.set_xlabel('Mission Duration (years)')
ax3.set_title('(C) DFD vs Conventional', fontweight='bold')
ax3.set_xlim(0, 55); ax3.grid(axis='x', alpha=0.3)

# Panel D: Propulsion Ladder
ax4 = axes[1, 1]
ladder = [('Chemical', 4.4, '#95A5A6'), ('NTP (NERVA)', 8.5, '#E67E22'),
          ('NEP (Ion)', 30, '#3498DB'), ('DFD (D-3He)', 100, '#E74C3C'),
          ('ICF Pulse', 167, '#F39C12'), ('Advanced Fusion', 1000, '#9B59B6')]
names = [l[0] for l in ladder]; ve_vals = [l[1] for l in ladder]
colors_l = [l[2] for l in ladder]
ax4.barh(range(len(ladder)), ve_vals, color=colors_l, alpha=0.85,
         height=0.6, edgecolor='black', linewidth=0.5)
for i, (n, ve, c) in enumerate(ladder):
    ax4.text(ve*1.1, i, f'{ve} km/s', va='center', fontsize=9, fontweight='bold')
ax4.set_yticks(range(len(names))); ax4.set_yticklabels(names)
ax4.set_xlabel('Exhaust Velocity (km/s)'); ax4.set_xscale('log')
ax4.set_title('(D) The Propulsion Ladder', fontweight='bold')
ax4.set_xlim(1, 5000); ax4.grid(axis='x', alpha=0.3)

plt.tight_layout(rect=[0, 0, 1, 0.95])
plt.savefig('vol9_fig2_missions.png', dpi=200, bbox_inches='tight',
            facecolor='white', edgecolor='none')
plt.close()
print("Figure 2 saved.")

IMAGE_URL_PLACEHOLDER

§8. The Fuel Question in Space

Volumes 2, 5, and 6 established the physics of fusion fuels. In space, the calculus shifts:

8.1 D-T: The Wrong Fuel for Space

D-T fusion is the easiest to achieve (lowest temperature, highest cross-section) but the worst choice for propulsion:

• 80% of energy in 14.1 MeV neutrons → cannot be magnetically directed → wasted for thrust
• Neutrons require massive shielding (~1 m of lithium hydride or borated polyethylene) near crew
• Shielding mass destroys the specific power advantage
• Tritium is radioactive (t₁/₂ = 12.3 yr), creating handling and safety issues
• Tritium breeding in space is impractical (no lithium blanket infrastructure)

For terrestrial power, D-T's ease of ignition outweighs these disadvantages. For space, they are disqualifying.

8.2 D-³He: The Goldilocks Fuel

D-³He produces 18.3 MeV per reaction, almost entirely as charged particles:

• 14.7 MeV proton: magnetically directable for thrust
• 3.6 MeV alpha: magnetically directable
• Neutron fraction <1.1% (from parasitic D-D side reactions)
• No radioactive fuel components

The physics barriers (Vol.2, Vol.6): D-³He requires ~5× higher temperature than D-T, and the triple product requirement is ~17× higher. No device has achieved D-³He burning conditions.

The ³He supply problem: Terrestrial ³He is scarce (~15,000 litres/year from tritium decay). Lunar regolith contains ~1.1 million tonnes of ³He implanted by the solar wind. Mining it requires processing ~100 tonnes of regolith per gram of ³He — an enormous industrial operation, but one that becomes feasible if lunar infrastructure exists for other reasons.

Cohen's "Goldilocks" argument: D-³He is harder than D-T but far easier than p-¹¹B. For a small, low-power space reactor (1–10 MW), the physics requirements may be achievable with compact FRC or mirror devices. You do not need to reach ignition — driven, sub-ignition operation is sufficient if the engineering gain $Q_{\mathrm{eng}}$ exceeds 1.

8.3 p-¹¹B: Still Impossible in Space Too

Vol.2 demonstrated that p-¹¹B thermal ignition is impossible because bremsstrahlung losses exceed fusion power at all temperatures. This is a thermodynamic fact independent of confinement geometry.

Some propulsion concepts invoke non-thermal mechanisms (beam-target, aneutronic resonances) that might partially circumvent this barrier. None have been demonstrated experimentally. TAE Technologies pursues p-¹¹B for terrestrial power (Vol.8) but has not published any propulsion-specific designs.

For Vol.10 (Valkyrie), we will examine whether advanced physics — alpha channelling, non-Maxwellian distributions, or exotic confinement — could eventually make p-¹¹B propulsion viable. For now, D-³He is the only realistic advanced fuel for fusion propulsion.

---

§9. DRACO's Death and the Economics of Propulsion

The cancellation of DRACO in May 2025 is the most significant propulsion policy event in a decade. The lessons apply directly to fusion propulsion.

9.1 What Happened

DARPA launched DRACO in 2021 ($499M with NASA) to flight-test a nuclear thermal rocket by 2027. Lockheed Martin would build the spacecraft; BWXT would build the HALEU reactor. A preliminary design review was completed in August 2024.

On May 30, 2025, the FY2026 budget zeroed all NTP and NEP funding. DARPA Deputy Director Rob McHenry explained: "When DRACO was originally conceived, that was pre the precipitous decrease in launch costs driven largely by SpaceX. As launch costs came down, the efficiency gain from nuclear thermal propulsion relative to the massive R&D costs started to look like less and less of a positive ROI."

In other words: the $\Delta v$ advantage of NTP ($I_{\mathrm{sp}}$ = 850 s vs. 450 s for chemical) can be offset by simply launching more mass on cheap Falcon 9 / Starship flights. If launch costs drop from ~$10,000/kg to ~$100/kg, it is cheaper to launch 10× more propellant than to develop a new propulsion technology.

9.2 Implications for Fusion Propulsion

The DRACO logic applies to fusion at inner-solar-system destinations. If Starship reduces Earth-to-orbit costs to ~$100/kg, a chemical Mars mission with multiple refuelling depots becomes economically competitive with any advanced propulsion system that costs >$10B to develop.

But the logic breaks for three classes of missions:

1. Fast crewed Mars: Radiation exposure scales linearly with transit time. A 30-day transfer vs. a 7-month transfer reduces cosmic ray exposure by ~7×. This has quantifiable health benefits that chemical propulsion cannot match regardless of launch costs.

2. Outer solar system: No amount of cheap launches helps you reach Jupiter in less than 2 years on chemical propulsion. The $\Delta v$ is simply too large. Gravity assists help but constrain launch windows and trajectories.

3. Interstellar: Chemical propulsion cannot reach even the nearest star in less than ~80,000 years. No launch cost reduction changes this physics.

DRACO's death was rational for NTP, where the $I_{\mathrm{sp}}$ improvement (2×) is modest. Fusion propulsion, with its ~20× improvement, occupies a different regime entirely. But it also faces a much longer development timeline — and the lesson of DRACO is that political patience for long-development propulsion programs is thin.

9.3 The Nuclear Electric Alternative

DARPA's McHenry noted that "nuclear electric is probably a more optimal long-term solution." The Air Force Research Laboratory's JETSON program (Joint Emergent Technology Supplying On-orbit Nuclear Power) is developing fission reactors for space electrical power (Lockheed Martin, Westinghouse, Intuitive Machines contracts, 2023).

Nuclear electric provides power (not propulsion) but can drive high-$I_{\mathrm{sp}}$ thrusters. It may reach flight before any fusion concept. However, NEP's low thrust means trip times of years for outer planet missions — acceptable for robotic probes, not for crew.

---

§10. Uncertainties and Honest Unknowns

This series requires explicit acknowledgement of what we do not know.

1. No fusion propulsion system has ever produced thrust. All performance numbers in this volume are projections from theoretical models and subscale experiments. The gap between PFRC-2 (500 eV electron temperature, no fusion) and a flight-ready DFD (>5 keV ion temperature, sustained fusion, integrated nozzle) is at least 3–4 experimental generations spanning 15–25 years.

2. The magnetic nozzle at fusion parameters is untested. Laboratory nozzle experiments operate at <1 kW with <100 eV plasmas. Fusion nozzles must handle multi-MW power at multi-keV temperatures. Whether efficient detachment occurs at these parameters is an open physics question.

3. Specific power projections are optimistic. The DFD's 0.2 kW/kg estimate assumes mature HTS magnets, efficient RF systems, and compact shielding. Real systems typically weigh 2–5× more than paper designs predict. Achieving 1 kW/kg would be remarkable.

4. D-³He burning has never been demonstrated. The reaction requires ion temperatures of ~60 keV — conditions not yet achieved in any FRC device. The cross-section peaks at 200 keV. Whether compact FRCs can reach and sustain these conditions is the central physics question.

5. ³He supply is insufficient for large-scale use. Current terrestrial ³He production (~15,000 litres/year) could fuel approximately one 1 MW engine for one mission. Lunar mining is theoretically possible but requires infrastructure that does not exist.

6. Private company timelines are unreliable. Pulsar Fusion's claim of "fusion temperatures by 2027" has no published physics basis. Helicity Space's "10 years to flight prototype" is aspirational. Historical fusion timelines have consistently underestimated development time by factors of 3–10×.

7. The economics are uncertain. DRACO's cancellation demonstrates that advanced propulsion competes not only against physics but against economic alternatives (cheaper launch, more propellant). Fusion propulsion must demonstrate value that cannot be replicated by brute-force chemical approaches.

8. Crew safety in proximity to a fusion reactor is unstudied. Even aneutronic concepts produce some neutrons (from D-D side reactions and activation). The shielding requirements for a crewed vehicle with an operating fusion reactor have not been engineered.

9. This author has no propulsion engineering background. The analysis is based on published literature and first-principles physics. Errors in engineering mass estimates, thermal management, and systems integration are likely.

---

§11. Decision Matrix

Weighted comparison of fusion propulsion concepts for a crewed Mars mission (90-day class):

Criterion (Weight)	DFD/Starfire	Helicity Drive	Pulsar Sunbird	ICF Pulse	NEP (Baseline)
$I_{\mathrm{sp}}$ achievable (15%)	4	3	4	5	4
Thrust for crewed mission (15%)	3	3	3	5	1
Specific power (20%)	3	2	2	3	1
Physics maturity (20%)	3	1	2	3	5
Shielding / crew safety (10%)	4	3	4	2	4
Scalability (10%)	4	4	3	3	3
Development cost (10%)	3	4	3	1	4
Weighted Total	3.25	2.55	2.85	3.05	2.90

Score: 1 (poor) to 5 (excellent).

No concept scores above 3.5. This reflects the fundamental reality: fusion propulsion is not ready. The DFD concept leads because it has the strongest experimental basis (PFRC) and the most detailed mission analyses. ICF pulse concepts score well on raw performance but are penalised by driver mass and development cost.

NEP — the only technology close to flight readiness — scores poorly on thrust and specific power but is included as the realistic baseline against which fusion concepts must compete.

---

§12. Conclusion

The physics of fusion propulsion is sound. The Tsiolkovsky equation demands either enormous mass ratios (chemical, NTP) or high exhaust velocities (fusion). For missions beyond Mars — to the outer planets, the interstellar medium, or anywhere that requires $\Delta v$ > 30 km/s — fusion is the only known physics that produces sufficient $I_{\mathrm{sp}}$ at sufficient thrust to be useful.

The engineering gap between "sound physics" and "flight hardware" is the widest in this series. Vol.3 showed that tritium breeding has never achieved TBR > 1.0 in experiment. Vol.4 showed that no structural material has been tested beyond 20 dpa of fusion neutron damage. Those gaps are serious. The gap in fusion propulsion is worse: no fusion propulsion device has ever produced a single newton of thrust.

The nearest-term concept, the Princeton DFD/Starfire, is a decade away from a ground demonstration and perhaps two decades from flight. Private ventures (Pulsar, Helicity) inject energy and ambition but have not yet produced peer-reviewed results at fusion-relevant parameters. DRACO's cancellation shows that even fission propulsion — a demonstrated technology — struggles to survive the political economy of space.

And yet.

The missions that fusion propulsion enables — Pluto orbiters, Titan aircraft, 90-day Mars transits, interstellar precursors — are precisely the missions that justify having a space program at all. Chemical rockets can reach Mars. Only fusion (or something equally exotic) can make the solar system accessible.

DRACO died because its $I_{\mathrm{sp}}$ improvement (2×) could be offset by cheaper launches. Fusion's improvement (20×) cannot. That asymmetry is the case for continued investment.

The rocket equation does not negotiate. It does not care about budgets, politics, or timelines. It only cares about exhaust velocity. And the only energy source that offers the exhaust velocity to reach the stars is the same one that powers them.

---

References

1. Tsiolkovsky, K. E. (1903). "Exploration of Outer Space by Means of Rocket Devices." Nauchnoe Obozrenie.
2. Cohen, S. A. et al. (2019). "Direct fusion drive for interstellar exploration." JBIS, 72, 37–50.
3. Razin, Y. S. et al. (2014). "A direct fusion drive for rocket propulsion." Acta Astronautica, 105, 145–155.
4. Thomas, S. et al. (2017). "Fusion-Enabled Pluto Orbiter and Lander." NASA NIAC Phase I/II Final Report.
5. Galea, P. et al. (2023). "The Princeton Field-Reversed Configuration for Compact Nuclear Fusion." Journal of Fusion Energy, 42, 4.
6. Wu, K. et al. (2025). "A review of plasma acceleration and detachment mechanisms in propulsive magnetic nozzles." Physics of Plasmas, 32, 040501.
7. Stuhlinger, E. (1964). Ion Propulsion for Space Flight. McGraw-Hill.
8. Wikipedia contributors. "Demonstration Rocket for Agile Cislunar Operations." Retrieved February 2026.
9. McHenry, R. (2025). Mitchell Institute remarks on DRACO cancellation. Breaking Defense, June 2025.
10. Paluszek, M. et al. (2020). In-space startup method for fusion rocket engines. US Patent 10,811,143.
11. Helicity Space (2023). "$5M Seed Round" press release. BusinessWire, December 2023.
12. Pulsar Fusion (2025). Sunbird DDFD specifications. pulsarfusion.com.
13. Adams, R. et al. (2003). "Conceptual Design of In-Space Vehicles for Human Exploration of the Outer Planets." NASA/TP-2003-212691.
14. Williams, C. H. et al. (2001). "A spherical torus nuclear fusion reactor space propulsion vehicle concept for fast interplanetary travel." NASA/TM-2001-210289.
15. Kammash, T. (1995). Fusion Energy in Space Propulsion. AIAA Progress in Astronautics and Aeronautics, Vol. 167.
16. Wurden, G. et al. (2019). "A New Vision for Fusion Energy Research: Fusion Rocket Engines for Planetary Defense." Journal of Fusion Energy, 35, 123–133.
17. Princeton Satellite Systems. "Starfire: Compact Fusion Power and Propulsion." psatellite.com.
18. Space Launch Schedule (2025). "UK plasma thruster test positions Pulsar Fusion." December 2025.