K
Solar Wind Concentrator

R.C. Wiens, L. Adamic, and D. Burnett
California Institute of Technology, Pasadena, CA

M. Neugebauer
Jet Propulsion Laboratory, Pasadena, CA


Submitted to AGU Monograph on Measurement Techniques for Space Plasmas

In the proposed Suess-Urey Discovery mission, solar matter is obtained during a two year exposure of ultra-high purity collector materials to the solar wind. For some elements, especially O, obtaining the required bulk purity levels for collector materials is very challenging. Further, most metals have a native oxide layer, and although it is possible in principle to resolve the oxide layer from implanted solar wind oxygen ions at roughly 100 Å depth, this is difficult in practice.

To get around these difficulties we have designed a solar wind concentrator, based on the principle of an electrostatic reflecting telescope, but with the additions of ion pre-acceleration and solar wind proton rejection. This concentrator increases the fluence of ions on the high purity target material by a factor of >20, providing a commensurate increase in the signal to noise for laboratory analysis of the solar wind ions. In addition to oxygen the concentrator covers the mass range of ions from ~4-20 amu which contains other scientifically important and highly contamination-sensitive elements such as Li, B ,C, N, and F. In addition to enhancing fluence the concentrator accelerates the solar wind ions by 10 keV/q, implanting them deep in the target material, beneath any surface contamination.

In this paper we describe the design and state of development of the solar wind concentrator. In Section 2 we provide an overview of the electro-optical design and describe the present mechanical design that is well under way to implement a flight-type solar wind concentrator. In Section 3 we discuss the simulations and calculations that have been performed to demonstrate the concentratorís capabilities, and describe the prototype development and testing carried out to verify this functionality. Finally, in Section 4 we summarize the state of the concentrator development.

Figure 1. Schematic diagram of a meridional cut through the solar wind concentrator. Heavy ions enter through the grounded grid, pass through the proton suppression grid, and are accelerated into the interior region by -10 kV on the accelerator grid. They are then reflected between the mirror grid and the solid electrode, and embedded deep into the target. The inset shows micro-machined steps in the electrode which serve to reflect light and heat back out into space rather than concentrating it also onto the target.

2. CONCENTRATOR DESIGN
The solar wind concentratorís basic design is schematically displayed in Figure 1. This sketch shows a meridional cut through the concentrator. Solar wind ions enter from the top through a series of three high transmission grids. The topmost grid is maintained at ground potential so that potentials inside the concentrator do not extend outside of it or affect the trajectories of particles prior to their entry. The next grid is biased at a variable positive high voltage. This voltage is set such that the bulk of the solar wind protons are rejected while essentially all of the heavier ions are transmitted. This is possible because solar wind ions all travel at nearly the same speed and their energies per charge (E/q) are selected by the suppression grid. For example, for a typical 440 km s^-1 solar wind, an ionís energy per charge is ~1 kV/q/amu such that the protons, alphas, and ¹⁶O⁺⁶ have E/q values of 1, 2, and 2.7 keV/q, respectively. For this example the suppression grid might be set at ~1.3 kV. The suppression voltage is variable and tracks the solar wind speed so as to provide the appropriate proton rejection as the solar wind speed varies.

The third grid is mounted to a metal structure which completely encapsulates the inner region of the concentrator and is maintained at a constant negative high voltage of -10 kV. This voltage accelerates the ions so that they all have at least 10 keV/q, and straightens out the ion trajectories such that they reflect more efficiently in the mirror section below. Ion reflection is accomplished in an electrostatic mirror, the front grid of which is part of the -10 kV inner capsule, and the back of which is biased at a variable positive high voltage. The mirror grid and its back electrode are equipotential parabolic surfaces separated by 2 cm on the axis of the concentrator. The mirror electrode potential is varied up to +10 kV and is again selected on the basis of the solar wind speed such that the primary ions of interest (e.g., oxygen) always reflect in the region of the mirror where the effects of the finite grid and mirror electrode structure are minimized.

The target is a six centimeter disk mounted onto a structural spider that is in the plane of, and supports, the third (-10 kV) grid. The baseline target material is pure single crystal gold, however, other materials such as chemical vapor deposition diamond and silicon are also presently under consideration and the ultimate concentrator target may be some combination of several materials. All concentrator ions are deeply imbedded in the target material which maintains the same -10 kV bias as the rest of the interior of the device.

Figure 2. Isometric drawing of the actual concentrator mechanical design with a 90° section removed. For simplicity of construction, the mirror grid is mounted onto thirty, 12° sectors which are curved on separate parabolic surfaces rather than trying to form a single, large paraboloid of revolution.

The mechanical design of a flight solar wind concentrator is presently in an advanced state. Figure 2 displays an isometric, cutaway view of the actual concentrator design. The structural exterior hatbox has an open aperture of 40 cm diameter and an outside envelope of 46 cm wide by 21 cm tall. It is designed in a modular fashion in three cylindrical sections for ease of assembly: these sections carry the grids and the mirror electrode either directly or via ceramic insulators. The solid electrode that acts as the back of the ion mirror is aluminum. As this surface will also act as an excellent photon concentrator, much like an optical telescope, it is necessary to microstep the concave surface of the electrode in order to prevent heating the concentrator target to unacceptable temperatures with reflected solar radiation (see blowup in Figure 1). The microsteps are machined into the surface such that incident photons are rejected directly back into space and not concentrated onto the target. The step size is sufficiently small (<100 µm) that the electric field near the surface of the mirror remains smooth throughout the ion bounce region.

The mirror grid is the most difficult surface in the concentrator to form. Ideally, this surface would be a very high transmission parabolic grid with no or few support structures. However, our search for a partner capable of electroforming or otherwise fabricating such a structure met with little success. We therefore decided to approximate the ideal by forming the shape from a number of pie-shaped frames. These frames are parabolically curved in the radial dimension but flat in the azimuthal dimension; they are supported both at the outer edge and in the mirror center. High-transmission, electroformed grids with 70 lines-per-inch (lpi) and >90% transmission are stretched across and attached to these frames. The design currently uses thirty frames in order to bring the mirrored ions to an adequate focus, but recent simulations show that in the future this number of sections might be reduced by perhaps a factor of two.

The three entrance grids are all planar and each is attached by an embroidery hoop capture mechanism around the edges and further attached along six radial supports. These grids are the same as the high-transmission ones used in the mirror grid section. Analysis shows that six radial supports per plane are sufficient to prevent drumheading during launch vibration. In addition, the centers of the three grid planes are structurally tied together with an insulator to further enhance the structural stability. The central insulator also carries the target. All high voltage power supplies (HVPSs) are located external to the concentrator for cleanliness and feedthroughs from the HVPSs are arranged along one edge of the hatbox. The concentrator has an estimated mass of 4.3 kg. About half of this mass is in the support hatbox and the remainder in the electro-optical components.

As contamination of the target is of great concern, a number of measures are employed to insure that there can be no sputtering of undesirable elements with subsequent implantation into the target. Materials used to fabricate the concentrator are limited to aluminum, stainless steel and ceramic. The grids are all electroformed from pure gold. All interior surfaces (except for the insulators) are gold-plated just prior to final assembly. In addition, all insulators are shielded from direct ion bombardment in order to preclude their sputtering onto the target. These measures insure that the only internally-generated contaminant that can be sputtered and perhaps implanted in the target is pure gold, which is not of concern. Contamination by organic volatiles from the rest of the spacecraft is minimized as 1) there is no line-of-sight path to the target surface, 2) the concentrator will operate at fairly high temperatures which will prevent most volatiles from condensing, and 3) only deeply implanted contaminants will cause subsequent analytical problems.

Figure 3. 3-D simulation of the concentrator with 2.5 keV/q ions incident at an aperture angle of 6° to the axis. The inset shows the detail of the ion trajectories hitting the target for this configuration of incident ions.

3. SIMULATIONS AND PROTOTYPE TESTING
Several different computational techniques were used to model the solar wind concentrator. In the ITRACK code, ion trajectories within the instrument were simulated using the numerical differential equation solver, Mathematica. Figure 3 illustrates the grids, electrode, and target of the concentrator and a set of ion trajectories passing through the instrument to the target. The proton rejection and acceleration grids are represented by three parallel equipotential planes at the entrance aperture of the concentrator. The mirror grid and backplate are paraboloids with a common geometric focus. The use of a common focus minimizes the shift of the focal plane as a function of ion energy. The vacuum electric fields used in this code are highly accurate representations of the anticipated concentrator. The effects of the discrete grid fields were considered and determined to be small; the fringing field effects will be small and localized to where the entry grids meet the outer wall of the concentrator. This ray tracing code demonstrated that the parabolic grid design focuses a beam parallel to the axis onto a sharp point. For the 6° offset rays shown in Figure 3, the focus is imperfect, but the vast majority of the rays still fall on the target (see inset). These results are consistent with geometric optics, and enable the use of simpler, analytic models for the bulk of the design optimization below.

The second modeling code, ANNION, is a 3D representation of the concentrator where the trajectories undergo a single refraction at the entry plane grids in accordance with the net axial acceleration of the ions. The mirror electrodes are represented by specular reflection on a paraboloid, consistent with the virtual mirror that would reproduce the reflected trajectories seen in the ray tracing code described above. This geometric representation enables one to write an analytic transformation which converts a point in the phase space of the solar wind as it enters the concentrator to points on the mirror and target where the ions will fall. The transformation can be reversed to project from points on the target and mirror to the corresponding source points in phase space. The distribution of solar wind ions reaching the target compared to the total flux of ions entering the concentrator determines the collection efficiency of the instrument.

Two different modeling codes were developed which were similar to ITRACK, but were operated as Monte-Carlo models. One code used analytical solutions while the other incorporated aspects of SIMION [Dahl et al., 1990]. Up to 50 million trajectories were launched in statistical distributions representing the solar wind. Then, the ratio of trajectories striking the target to those entering the concentrator was computed to determine collection efficiency. The forward and reversed trajectory computation techniques were developed independently by two groups (RM and RW/LA); their good agreement provides an excellent check on the accuracy of the calculations.

The efficiency of implanting ions into the target, and hence the concentration factor, is a function of the angle of incidence of the ions with respect to the symmetry axis of the device. The expected angular distribution of solar wind ions has been estimated from a year and a half of solar wind data collected by the ISEE-3 spacecraft [Bame et al., 1978] in 1978-80, which was roughly the same phase of the solar cycle as the planned Suess-Urey mission. Five minute averages of the proton vector velocities, densities, and temperatures were used to construct the angular distribution shown in Figure 4. For this calculation it was assumed that the concentrator would always point 4 degrees in advance of the center of the Sun in order to remove the average aberrational offset caused by the motion of the spacecraft around the Sun. Another assumption inherent in the applicability of the calculation is that the heavy ions in the solar wind have the same angular distributions as the protons; while this is not always exactly true, the difference between the assumption and reality corresponds to a slight over-estimate of the width of the heavy-ion distribution.

Figure 4. The integrated probability that the flux of solar wind ions will be within a given angle of the concentrator axis based on a one and one half year study of ISEE-3 solar wind data.

If the concentrator were operated without the entrance grids, it would behave very much like a simple reflecting telescope with the same optics for all ions. The entrance grids are needed to achieve proton rejection (beam deceleration) followed by pre-acceleration (increased beam energy) for enhanced implantation in the target. The introduction of these grids introduces another variable determining the distribution of ions on target. One might consider placing the accelerating grids directly below the target, but that would introduce a severe mass fractionation for ions approaching the target obliquely. By placing the accelerating grids at the entrance aperture, there is very little mass fractionation, while there is a significant gain in beam concentration. The grids add axial velocity to the ions without changing the transverse (horizontal) velocity. The added axial velocity reduces the angular distribution of the ions before they are reflected by the concentrator mirror.

Figure 5. Calculation of the percent hitting a 6 cm diameter target as a function of the radius of curvature of the electrostatic mirror. The solar wind angular distribution shown in Figure 4 was used in this calculation.

To arrive at a specific concentrator design, the radius of curvature of the parabolic grids and the size of the target plate were selected from a number of runs of the ANNION code described above. Figure 5 is an example of such a run where the ISEE-3 solar wind was modeled entering a concentrator with a 20 cm radius parabolic mirror having a 3 cm radius target plate. The mirror was the virtual mirror equivalent of the simulations ITRACK code, and the radius of curvature at the center of the paraboloid was varied from 20 cm to 40 cm. The target plate was placed at the focal point of the mirror, half of the radius of curvature. Using the ISEE-3 angular distributions, the percentage hitting the target (PHT) was computed and compared relative to the total beam entering the aperture; see Figure 5. This example indicates that a radius of curvature of 30 cm results in ~96% of the 2.5 keV/q ions hitting the target, while only 2% more would be gained by going to a much larger 38 cm radius of curvature.

To calculate the overall concentrator efficiency, the ratio of ions entering the concentrator must be reduced by a number of factors. This starting point is simply the ratios of areas of aperture to target or (40/6)2 = 44.4. Ions pass through the three entrance grids once each and through the single mirror grid twice. For worst case grid transmissions of 0.9, this gives a reduction in transmission of (0.9)5 = 0.59. Additional support structure will obscure about another 0.1 of the incident ions giving another factor of 0.9. Finally, the PHT given in Figure 5 for our baseline design was slightly greater than 0.96. Multiplying all of these factors together gives a total concentration factor of 22.6. Similar calculations for worst case, higher energy ions, yield a concentration factor of 20.

Figure 6. Spin- and flow-direction-averaged radial distribution of ¹⁶O⁺⁶ (dashed line) and ¹⁸O⁺⁶ (solid line) on the concentrator target as a function of distance from target center.

The one final concern about the functionality of the solar wind concentrator is whether slight differences in ion E/q will lead to electro-optical mass fractionation that could mimic exactly the fine elemental and isotopic differences the concentrator is designed to sample. Figure 6 shows the calculated distribution of ¹⁶O⁺⁶ and ¹⁸O⁺⁶ ions as a function of distance from the center of the target. The calculation was based on the angular distribution shown in Figure 4 for an ion energy of 1 keV/amu (corresponding to a solar wind speed of 440 km/s), an acceleration voltage of 10 kV at the entry plane, and a voltage of -10 kV on the back electrode of a parabolic mirror. Because the ¹⁸O⁺⁶ ions undergo less acceleration, and hence less refraction, in the entry grids than do the ¹⁶O⁺⁶ ions, their distribution on the target falls slightly more toward the outside. The simulations treat the spacecraft as spinning, so irregularities in solar wind direction are averaged over the azimuthal angle. Averaged over the entire 3 cm radius target, 95.4% of the ¹⁸O⁺⁶ and 96.1% of the ¹⁶O⁺⁶ are collected; at the target radius of ~1.2 cm the fluence of the two isotopes is equal. In practice, the distributions shown in Figure 6 will be smeared out, and hence differences between the distributions of the two isotopes lessened, by several effects not taken into account in this idealized calculation.

The science goal of <0.1% fractionation of oxygen isotopes can be readily achieved because of several effects. First, the intrinsic fractionation will be even smaller than the idealized calculation shown in Figure 6. Second, the solar wind distributions measured by the ion monitor on Suess-Urey will be used to model the concentrator; the ion fluence across the target is a predictable function of the solar wind properties and ion charge-to-mass ratios (e.g., Figure 6). Finally, and perhaps most importantly, the actual radial variation of isotope ratios can be measured after the target is returned by sampling several spots. If the target is subdivided into annuli, one can find rings that have virtually no difference in ion deposition on that ring. As an example, in Figure 6, the percent of ¹⁶O⁺⁶ and ¹⁸O⁺⁶ beams hitting the target in an annulus between radii of 0.75 cm and 1.75 cm are equal.

In order to test the most critical electro-optical component of the concentrator, we built a prototype concentrating mirror section. This section is full scale with the exception that it has only nine of the thirty sectors, covering 108°. In addition, each sector has circular rather than parabolic curvature, consistent with our initial, pre-optimized design. This prototype, however, is of sufficiently high fidelity to allow complete testing of the basic design features.

Testing was carried out in the CASYMS plasma instrument facility at the University of Bern, Switzerland [Steinacher et al., 1995], using singly-charged 1 kV beams and varying the mirror electrode voltage to simulate different solar wind E/q ratios. The tests were designed to check 1) optics calculations in general, 2) performance of ions near the outside edge, 3) scattering of ions by grid wires, 4) effective transparency of the grids, and 5) the effects of small wrinkles in the high-transparency grids. Most tests were carried out with a 1 cm diameter beam, directed at various points on the mirror, but we also carried out a wide (> 20 cm diameter) beam test. Two types of targets were used. A 250 x 250 pixel, 40 mm diameter imaging detector (Surface Science Laboratories/Quantar) was used in the target position for about two thirds of the tests. For the remainder of the tests aluminum foils were used at the target position, with control foils implanted before and after each mirror test. Use of beams consisting of different noble gases (He, Ne, Kr, & Xe) allowed several tests to be made between foil changes. Foils were analyzed by noble gas mass spectrometry at Washington University.

Data analysis is still in progress, nevertheless, the ion trajectory simulations described above have already been generally confirmed by prototype testing. We also learned that the effects of small wrinkles in the grid mesh were greater than anticipated; this result drove us to incorporate an improved design for stretching the grids. The prototype test results, backed up by calculations, also showed that the effects of grid scattering would be unacceptably large using an original plan to reflect unaccelerated ions approximately halfway between the mirror grid and back electrode. The present arrangement, with accelerating grids upstream of the mirror, and mirror tracking voltages slaved to the solar wind speed, causes all ions to reflect near the same point, closer to the electrode, and farther from the grid, resulting in a greatly reduced effect of scattering from grid wrinkles. Finally, extrapolation of the mass fractionation test results confirm our expectation that a fractionation of <0.1% is readily achievable.

4. SUMMARY
This paper summarizes the state of development of a solar wind concentrator to provide a new capability for measuring certain critical elemental and isotopic components of the solar wind and hence the nebula that formed the solar system. Here we have described the basic electro-optical design of the concentrator as well as many of the details of a real mechanical design. The design is based on the combination of three independent simulation approaches to optimize its various aspects. In addition, we have built and tested a prototype mirror section, confirming critical aspects of the design and identifying weak points which we have subsequently addressed by design improvements.

While a complete solar wind concentrator has never been built and flown, all of the components (i.e., large area grids, HVPSs, electro-optical mirrors) have substantial flight heritage. Considering the present, advanced state of design and development, there can be little doubt that the solar wind concentrator is fully ready to be built and flown on a near future Discovery mission. The scientific return of measuring the detailed isotopic composition of the solar oxygen in particular, and the mid-range (4-20 amu) atoms, in general, would be immeasurable.

Acknowledgements We gratefully acknowledge valuable contributions from Rudy Abeyta, Phil Barker, Peter Eberhardt, Charles Hohenberg, Charlie Kehm, Martin Steinacher, and Steve Storms. The work at Los Alamos was carried out under the auspices of the United States Department of Energy with support from NASA.

Bame, S.J., J.R. Asbridge, H.E. Felthauser, J.P. Glore, H.L. Hawk, and J. Chavez, ISEE-C solar wind plasma experiment, IEEE Trans. Geosci. Elect., GE-16, 160-162, 1978.

Dahl, D.A., J.E. Delmore, and A.D. Appelhans, SIMION PC/PS2 electrostatic lens design program, Rev. Sci. Inst., 61, 607-609, 1990.

Steinacher, M., Jost, F., and Schwab, U., A modern and fully automated calibration system for space ion mass spectrometers. Rev. Sci. Instrum., 66, 4180-4187, 1995.

Back to the Genesis Homepage