WorksPhysics and Astronomy
Broadband Diffractive Solar Sail
Abstract
The transverse radiation pressure force and acceleration is compared for two parametrically optimized designs: prismatic and two-pillar metasurface gratings. The numerical results were cross-verified with both Maxwell stress tensor and modal analysis. Solar blackbody irradiance was assumed for wavelengths ranging from 0.33 [um] to the grating cutoff at 1.5 [um], encompassing 83% of the solar constant. This multi-objective optimizer study found that neither design comprised Si3N4 performed as well as those corresponding to a low refractive index, low mass density material. The predicted transverse acceleration of the optimized low-index metasurface grating is compared to that of a state-of-the-art reflective solar sail.
Introduction
The in-space propulsion of sailcraft via solar radiation pressure was originally pioneered by in the 1920s by Tsander and Tsiolkovsky [1, 2]. In contrast to rockets which both transport significant amounts of fuel mass and make discrete orbit-changing burns, solar sails can attain extraordinarily high velocities given a low mass and continuous acceleration. Space organizations such as NASA, JAXA, and the Planetary Society, have improved the technical readiness level of solar sails in recent years, culminating in an assortment of proposed space science missions [3]. The advent of solar sailing has stimulated advanced concepts that consider the mission objectives as part of the sail design. For example, missions having a spiral trajectory toward or away from the sun benefit from a sail having an optimal “lift" force perpendicular to the sun line. To achieve lift a traditional reflective sail must be tilted away from the sun; consequently the maximum lift cannot be achieved owing to the reduced illumination projected area. In contrast, optical scattering mechanisms like diffraction provide alternative means of transferring photon momentum to the sail in a preferred sun-facing orientation [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]. The maximum transverse force on the sail occurs when sunlight is uniformly scattered at 90° with respect to the surface normal of a sun-facing sail.
Theory
To advance the understanding of diffractive sails we explore two designs: a triangular prismatic grating and a metasurface grating comprised of two pillars. Two material strategies are analyzed for each design. First we consider an arbitrary non-dispersive dielectric material having a refractive index placed on a thin substrate of index . Finite difference time domain (FDTD) methods are used to account for internal and external reflections of both polarization component of light, and moreover, the angular scattering distribution across a broad band of optical frequencies. Likewise, we determine the angular scattering distribution when the grating and thin substrate are made with . The schematic illustration shown in Figure 1 depicts a portion of a flat rigid infinitely periodic grating with period in the -plane of incidence for a sun-facing configuration, comprised of either (A) prism elements or (B) pillars on a thin substrate. Structural flexing and non-normal incidence angle are beyond the scope of this baseline study.

Figure 1. Schematic diagram of a solar sail with constituent (A) prism and (B) subwavelength pillar elements of period and . The sail diffracts incident light by into owing to , resulting in net radiation pressure force .
The grating period [], or equivalently the grating frequency [THz] was selected from a consideration of the spectral cut-off condition, the prism mass, and diffraction effects. The fraction of blackbody irradiance cut off from diffraction decreases with increasing value of , whereas the mass of a prism varies as . A large value of the transverse acceleration generally requires negligible spectral cut off and low mass, which combined with a diffraction analysis, provides a value of roughly [].
Light is transmitted or reflected light into discreet diffraction angles measured with respect to the back surface normal as depicted in Figure 1. In the reference frame of the sail, the incident and scattered wavelengths are equal, and thus, the respective wave vectors may be expressed and , where . The diffraction angles are governed by the grating equation: assuming normal incidence. We note that , where corresponds to transmitted (reflected) light. The order photon momentum transfer efficiency imparted to the sail at the optical frequency may be expressed , where is the speed of light, and normal incidence is assumed. For a light source having a spectral irradiance distribution the net momentum transfer efficiency may be found by integrating over all frequencies and summing over all allowed diffraction orders for both polarization modes [12]. For an unpolarized source like the sun, we assume the spectral irradiance is equally divided into and polarization states.
The net radiation pressure force on the sail may be expressed , where where is the sail area and is the irradiance. For example the solar blackbody irradiance between and of a band-limited blackbody source a distance from the sun may be expressed
where [m] is the solar radius, [Js] is the Planck constant, [J/K] is the Boltzmann constant, and we assign as the effective absolute temperature of the sun. Below we assume corresponds to 1 [AU]. The case corresponds to the so-called solar-constant, [W/m]. Values of are plotted in Figure 2 as a function of the grating period for and two different values of : (blue line) and 900 [THz] (red line). The case used for our FDTD model, [] and [] ( [THz], and [THz]) includes up to four diffraction orders and spans 83% of the solar spectrum. Although wider bandwidths are of interest, FDTD run times become prohibitively long.

Figure 2. Fraction of integrated solar black body spectral irradiance for the range to , where [W/m]. Insert: Black body spectral irradiance with range [THz] ( [m]) and [THz] (). Arrows: Range of maximum mode number .
Following Ref [12] the net radiation pressure force on the sail owing to a band-limited source may be expressed
where and respectively correspond to the value of the spectral irradiance scattered into the th diffraction order for the and polarization states, and where depends on frequency owing to the grating equation which may be expressed, . The frequency-dependent cut-off mode numbers at the normal incident are given by (or equivalently ) where INT represents the integer value of the argument rounded toward zero. In a lossless system having no guided surface waves that extend to infinity, we expect
In general owing to polarization-dependent scattering.
The Maxwell stress tensor may be evaluated at each frequency as an alternative method to evaluate the net force :
where is an arbitrary surface enclosing the sail, is the elemental area vector and
where and are respectively the vacuum permittivity and permeability, and are respectively electric and magnetic field amplitudes, and is the Kronecker delta function. For a structure that is periodic in the plane of incidence as depicted in Figure 3 and extended over a distance out of the plane, the only elemental areas that contribute to (4) are . The force exerted across the area of an infinitely period grating may therefore be expressed

Figure 3. FDTD Schematic: Unit cell of period of (A) prism and (B) meta gratings with plane wave source (red line), field monitors (blue lines), and perfectly absorbing boundary layers (green areas).
where is an arbitrary distance from the grating, and the final integral includes the frequency-integrated stress tensor components and .
Numerical Methods
We used the open source FDTD numerical solver MEEP [17] to solve Eqs. (4) - (6), making use of fast built-in “methods” like ForceSpectra to calculate forces in a specified ForceRegion. To cross-validate the force values we randomly compared them to values obtained using Eq. (2), this time using diffraction mode options in MEEP. In both cases, Bloch periodic boundary conditions were employed. The power spectrum of a broadband source in MEEP is defined as the distribution function GaussianSource(fcen, fwidth) where fcen and fwidth are respectively the center and width of the Gaussian distribution. Force calculations are made in the frequency domain and we scaled them to correspond to the solar blackbody spectral irradiance. The red line in Figure 3 depicts a planar light source propagating in the direction. The blue lines represent so-called monitors where the electromagnetic fields and are evaluated for the determination of the Maxwell stress tensor, and where alternatively the spectral irradiance may be determined to evaluate Eq. (2). The green lines in Figure 3 represent perfectly matched layers. The square numerical grid elements were set to [nm]. The simulation ran until either or decayed to of the peak value.
The focus of this study was to determine optimized parameters of the two structures depicted in Figure 3, both having the same period [m]: (A) a prismatic grating and substrate having four optimization parameters , , , and ; and (B) a metasurface comprised of two pillars and a substrate having nine optimization parameters , , , , , , , , and . We employed a multi-objective optimizer NSGA-II (with 40 agents, 40 offspring, 150 generations [18] with the range of parameter values listed in Table 1. The objectives are to achieve the largest values of transverse force for both polarizations and to minimize the mass. A representative set of 40 solutions (called Pareto-optimal) were obtained. The same procedure was followed for silicon nitride () structures, but in this case and . Silicon nitride is relatively stable in a space environment, its optical properties are well characterized, and its lithographic fabrication techniques are mature. The refractive index varies from at 200 [THz] to at 900 [THz] [19]:
| max : | |
| min : | |
| such that : | |
| such that : | |
| such that : | |
| such that : |
Table 1. Multi-Objective Optimization Scheme: Nine variables, three objectives, and four constraints.
A solar sail is typically used to achieve a spiral trajectory toward or away from the sun. In this case, the flight time may be minimized when the transverse (lift) component of acceleration is a maximum, where is the total mass of the sailcraft, is the mass of the diffractive sail material, is the mass of the payload and structural support mechanisms, and . The transverse acceleration is optimized when both and are maximized and is minimized. The sail mass of our two designs may be expressed
where is the number of grating periods across the sail, and is the area of a square sail. Ignoring the payload mass () and writing the transverse component of force we obtain the transverse acceleration for our unladen structures:
where is the fill factor, and for convenience we associate the refractive index and mass density with a proportionality factor : . Using the space qualified polyimide material CP1 [20] as an example, with a specific gravity s.g. = 1.54 and a mean refractive index of 1.57 we obtain [kg/m]. For our silicon nitride structures we instead combine its specific gravity, s.g. = 3.17 [21] with the mean index, 2.02, to obtain [kg/m]. As seen in Eq. 9 the transverse acceleration is independent of the sail area and is implicitly dependent on the grating period via the efficiency factor (which is found by numerically determining the transverse force value).
Results & Analysis
Forty representative Pareto-optimal solutions are plotted in Figure 4 for the two gratings having nine arbitrary parameters (A) and (B), and for the two gratings comprised of (C) and (D). The net transverse radiation pressure force is plotted against the total mass of the sail, . In all cases a trend in the data appears: Higher mass sails provide higher forces. To select the most optimal design for each structure we use the greatest value of the transverse acceleration as the deciding factor (see straight line in Figure 4). The parameters for the Pareto-optimal solution that intersects this line are tabulated in Table 2 for the four different cases.
| Parameters | A | B | C | D |
| [] | 0.76 | 1.12 | 1.02 | 0.62 |
| [] | - | 1.26 | - | |
| [] | - | 0.32 | - | 0.16 |
| [] | - | 0.16 | - | 0.24 |
| [] | - | 0.06 | - | 0.38 |
| [] | - | 0.44 | - | 0.1 |
| Prism Angle | 26.9° | - | 34.2° | - |
| 2.43 | 1.55 | |||
| 1.5 | 1.5 | |||
| [] | 0.1 | 0.1 | 0.1 | 0.11 |
| Force [nN] | 785 | 787 | 722 | 416 |
| mass [ kg] | 1.07 | 0.73 | 1.93 | 0.84 |
| [] | 731 | 1080 | 373 | 494 |
Table 2. Optimized parameters and cost function values for (A) prism and (B) meta gratings of arbitrary dispersionless materials, and (C) prism and (D) meta gratings for , each with period [], [m], [m], .

Figure 4. Pareto optimal solutions for (A) prismatic and (B) metasurface gratings having arbitrary refractive indexes, and for (C) prismatic and (D) metasurface gratings comprise of silicon nitride. A sun-facing square sail of area illuminated with a band-limited solar black body is assumed. The optimal transverse acceleration for each case is determined from the slope of the straight line, and the corresponding design parameter values for the intersecting points are given in Table 1.
We find that both the prismatic and metasurface structures having arbitrary refractive indexes are able to produce large values of , as is evident in Figure 4 for Case A and Case B. However, owing to the lower mass of the metasurface structure, its optimal acceleration [] is 48% greater than that of the prism grating. The structures, Case C and Case D, depict significantly less values of optimized acceleration. These values may be compared with a conventional aluminized polyimide sail [22] which is roughly 3 [] thick and achieves a momentum transfer efficiency of roughly 90% of the ideal value of 0.77: []. This comparison suggests that an optimized metasurface sail is a competitive alternative to a conventional reflective sail. However, amongst the many unknown fabrication, packaging, unfurling, and space weathering issues is whether a large robust metasurface grating can be fabricated on a thin (< 1 []) substrate [14].
To better understand the spectral force characteristics of the four sails examined in this study, we plot the transverse spectral force distribution in Figure 5. The blue line represents the FDTD-obtained values corresponding to the Maxwell stress tensor calculations, whereas the circles represent the values corresponding to our FDTD modal analysis. The excellent agreement between these two approaches provides a level of cross-validation of the methods. Fluctuations of the value of are indicative of pronounced diffractive variations of the transmitted and reflected light at different optical frequencies, as expected for a small period grating [4]. Also plotted in Figure 5 are theoretical values of force for the ideal limit (black line) and the ideal reflective sail (red line): . These results suggest that the diffractive sails explored in this study may equal or exceed the acceleration of a

Figure 5. Spectral transverse force distribution determined from Maxwell stress tensor (blue line) and modal analysis (open circles) for the four cases described in the text, and for an idealized reflective sail (red line) and the upper theoretical bound (black line). An area of 1 m is assumed.
reflective sail only if there is a small-mass advantage of the former. The prism and pillar designs suffer from the effects of external and internal reflections which can scatter light that opposes the desired transverse scattering direction. For example front surface reflections from the prism in Figure 3 (A) have positive values of which oppose the transmitted (refracted) rays. Those reflected rays carry 17% of incident beam power owing to Fresnel reflections. Less than two thirds of the incident radiation is refracted out the back surface owing to internal reflections and shadowing effects from the steep side facets. It is yet unknown whether the added mass of anti-reflection coatings would provide increased the transverse acceleration. Other unknowns that are beyond the scope of this paper include the practical limits of assumptions about the rigidity of the sail, the coherence properties of the incident sunlight, and whether the sail can be packaged and unfurled without changing its optical properties.
Conclusions
We performed FDTD simulations coupled with a NSGA-II multi-objective optimizer to determine design parameters for four different grating structures, each having a period of and a sail area of . The small grating period was selected to satisfy a small desired mass and a marginal cutoff wavelength of the solar blackbody spectrum. Our optimization study included 3 objectives and up to 9 variables, as well as both s and p polarization. The transverse component of radiation pressure force was determined for a truncated solar black body radiator (200-900 [THz] or equivalently, 0.33 to ) at for the purpose of two-orbit changing maneuvers in space. An optimized metasurface grating comprised of two pillars per period was found to provide 48% more transverse acceleration than an optimized prism grating owing to the small mass of the former grating. We found that Silicon Nitride did not perform well for either the prism or two-pillar metasurface design. Although none of the structures provided radiation pressure force values exceeding those of an ideal flat reflective sail, the diffractive sail may nevertheless provide an acceleration advantage if the proposed sun-facing diffractive sail spacecraft has a total lower mass than a reflective sailcraft. The design of alternatives to flat reflective sails is an emerging area of research and we therefore believe continued exploration of diffractive designs such as hybrid reflective/transmissive structures will provide more efficient solar sails in the future.
Funding. National Aeronautics and Space Administration (NASA) (80NSSC19K0975, 80MSFC22F0165), Johns Hopkins University Applied Physics Lab (177864).
Acknowledgment. We are grateful to Charles (Les) Johnson and Andrew F. Heaton (NASA George C. Marshall Space Flight Center, Huntsville, AL), and to Amber L. Dubill (Johns Hopkins Applied Physics Laboratory) for discussions related to solar sailing. We also thank Rajesh Menon and Apratim Majumder (U. Utah, Salt Lake City, UT) for meta-material and FDTD modeling discussions.
Disclosures. The authors declare no conflicts of interest.
Data availability. Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
- [1]K. Tsander, “From a scientific heritage, nasa technical translation ttf-541, 1967, a translation of iz nauchnogo naslediya,” (1924).
- [2]K. E. Tsiolkovsky, “Extension of man into outer space,” in Proceedings of the Symposium on Jet Propulsion, vol. 2 (1921).
- [3]L. Johnson, G. A. Swartzlander, and A. Artusio-Glimpse, “An overview of solar sail propulsion within nasa,” Adv. Sol. Sail. pp. 15–23 (2014).DOI
- [4]G. A. Swartzlander, “Radiation pressure on a diffractive sailcraft,” J. Opt. Soc. Am. B 34, C25 (2017).
- [5]Y.-J. L. Chu, E. M. Jansson, and G. A. Swartzlander, “Measurements of radiation pressure owing to the grating momentum,” Phys. Rev. Lett. 121, 063903 (2018).DOI
- [6]Y.-J. L. Chu, N. V. Tabiryan, and G. A. Swartzlander, “Experimental verification of a bigrating beam rider,” Phys. Rev. Lett. 123, 244302 (2019).DOI
- [7]P. R. Srivastava, Y.-J. L. Chu, and G. A. Swartzlander, “Stable diffractive beam rider,” Opt. Lett. 44, 3082 (2019).DOI
- [8]P. R. Srivastava and G. A. Swartzlander, “Optomechanics of a stable diffractive axicon light sail,” The Eur. Phys. J. Plus 135 (2020).DOI
- [9]Y.-J. L. Chu, M. Meem, P. R. Srivastava, R. Menon, and G. A. Swartzlander, “Parametric control of a diffractive axicon beam rider,” Opt. Lett. 46, 5141 (2021).DOI
- [10]S. Firuzi, Y. Song, and S. Gong, “Gradient-index solar sail and its optimal orbital control,” Aerosp. Sci. Technol. 119, 107103 (2021).DOI
- [11]Y. Chu, S. Firuzi, and S. Gong, “Controllable liquid crystal diffractive sail and its potential applications,” Acta Astronaut. 182, 37–45 (2021).DOI
- [12]G. A. Swartzlander, “Theory of radiation pressure on a diffractive solar sail,” J. Opt. Soc. Am. B 39, 2556 (2022).DOI
- [13]A. S. Swakshar, S. M. Kim, and G. A. Swartzlander, “Broadband radiation pressure on a small period diffractive film,” Opt. Express 30, 45279 (2022).DOI
- [14]M. I. Abdelrahman and F. Monticone, “How thin and efficient can a metasurface reflector be? universal bounds on reflection for any direction and polarization,” (2022).arxiv.org/abs/2208.05533
- [15]P. Zhang, S. Firuzi, C. Yuan, X. Gong, and S. Gong, “General passive stability criteria for a sun-pointing attitude using the metasurface sail,” Aerosp. Sci. Technol. 122, 107380 (2022).DOI
- [16]O. Borgue and A. M. Hein, “Transparent occulters: A nearly zero-radiation pressure sunshade to support climate change mitigation,” Acta Astronaut. 203, 308–318 (2023).DOI
- [17]A. F. Oskooi, D. Roundy, M. Ibanescu, P. Bermel, J. Joannopoulos, and S. G. Johnson, “Meep: A flexible free-software package for electromagnetic simulations by the FDTD method,” Comput. Phys. Commun. 181, 687–702 (2010).DOI
- [18]K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: Nsga-ii,” IEEE transactions on evolutionary computation 6, 182–197 (2002).
- [19]K. Luke, Y. Okawachi, M. R. E. Lamont, A. L. Gaeta, and M. Lipson, “Broadband mid-infrared frequency comb generation in a si_3n_4 microresonator,” Opt. Lett. 40, 4823 (2015).DOI
- [20]NeXolve LaRC CP1 Polyimide, https://www.matweb.com/search/datasheet_print.aspx?matguid=290379e273ae48139ac2628a39610e4. Accessed: Jan 2023.
- [21]W. M. Haynes, D. R. Lide, and T. J. Bruno, CRC handbook of chemistry and physics (CRC press, 2016).
- [22]C. R. McInnes, Solar Sailing (Springer London, 1999).