M
Collector Materials

By Section:
     1. Semiconductor Silicon
         1.1 General Techniques
         1.2 Si by INAA; Literature values
         1.3 Our own INAA analyses
         1.4 Si feasibility by Measurement Objective
         1.5 SOS
     2. Germanium
         2.1 General Characteristics
         2.2 Ge SIMS analysis
     3. Noble Metal Foils and Films
     4. Concentrator Target Materials
         4.1 Single Crystal Noble Metals
         4.2 Diamonds
                  4.2.1Diamond SIMS Study
                  4.2.2Diamond SIMS Results: Impurities
                  4.2.3Diamond SIMS Results: Diffusion
                  4.2.4Diamond SIMS Study Conclusions
         4.3 Comparisons Between Noble Metals and CVD Diamond
     5. Materials for Noble Gas Collection
         5.1 Solar Wind
         5.2 Energetic Particles
Appendix: Summary on oxygen in diamond, by K.M. Rutledge

Because of the rarity of most elements in the solar wind the primary requirement for the collector materials is that they have extremely low impurity levels. Since it is unlikely that any one material will have optimum characteristics for all elements and all analysis techniques, the collectors will combine several different materials. Generally speaking, analytical techniques, such as secondary ion mass spectrometry (SIMS) or resonance ion mass spectrometry (RIMS), that only analyze the surface layers containing the implanted solar wind have fewer materials purity difficulties than those that require bulk analysis of the entire collector thickness. For nominally bulk analytical methods such as radiochemical neutron activation analysis (RNAA), "surface stripping" techniques are required which preferentially extract the surface layer holding the implanted solar wind.

1. Semiconductor Silicon. Silicon is an excellent substrate for RIMS and RNAA, but not as good for SIMS because of molecular ion interferences.

From the aspect of purity, by far the most promising materials are semiconductors, with the most known about silicon. For many electronics products, the presence of 5-20 ppm oxygen is desirable because it getters lithophile impurities, causing them to lose their conductivity; consequently lithophilic impurities can be tolerated to some degree. This is the case for the Czochralski (CZ) method of crystal growth, in which a crystal is pulled from a SiO₂ crucible. For float-zone (FZ) growth the crystal is grown from a poly-crystalline rod by passing through an RF coil in a non-oxidizing atmosphere. The FZ method is necessary to obtain wafers of higher purity, particularly in the CNO content, but also presumably for any other elements which might be present in trace quantities in the SiO₂ crucible. Recent efforts to achieve the highest purity crystals have focused on using higher purity raw materials. This was motivated by a desire to reduce concentrations of B, which is not excluded during crystallization and is one of the principal acceptor impurities. The results are quite favorable for other conductive impurities as well, with reductions of up to a factor of ten. The resistivity of the semiconductor is a general measure of its conductive impurities, and by extension, should be an indicator of overall purity. FZ Si wafers grown from normal stock typically have impurity levels of ~2000 ohm cm; those produced from ultra-high purity poly-silicon average ~18,000 ohm cm, and we have obtained material with over 50,000 ohm cm.

Table M1. Impurities measured or attempted in float-zone silicon wafers by methods other than INAA.

Element	Source	Approx. Resistivity (Ohm-cm)	Analysis Technique	Detection Limit (cm-3)	Measured (cm-3)	Solar Wind4
B	1	18,000	PL5	-----	5 x 1011	6 x 1011
B	2	2,000	PL	5 x 1011	1 x 1012	"
C	2	2,000	IR	5 x 1015	-----6	6 x 1017
N	2	2,000	IR	1 x 1014	5 x 1014	2 x 1017
O	2	2,000	IR	1 x 1014	1 x 1015	1 x 1018
Al	1	18,000	PL	5 x 1010	-----	5 x 1015
P	2	2,000	PL	5 x 1011	1 x 1012	6 x 1014
P	1	18,000	PL	-----	5 x 1011	"
As	1	18,000	PL	5 x 1010	-----	5 x 1011
Mg	3	-----	SIMS	1 x 1014	-----	1 x 1016
Ca	3	-----	SIMS	1 x 1015	-----	1 x 1016

¹Maurits et al. (1990) abstr. #394, Fall Meeting Electrochem Soc., Pennington, NJ.
²Dreier (1990) Nucl. Meth. A288, 272.
³Our own unpublished SIMS analyses.
⁴Solar wind measurement is averaged over the topmost 100 nm for a 2 year normal flux. Multiply by 10^-5 to convert from cm^-3/(top 100 nm) to cm^-2.
⁵PL = Photoluminescence; IR = Infrared Spectroscopy; SIMS = Secondary Ionization Mass Spectrometry
⁶Where measured values are not given, the impurity is below the detection limit.

1.1 General Techniques. Table M1 gives impurity measurements made on Si wafers by techniques other than neutron activation analysis (NAA). For comparison, estimated solar wind concentrations are based on an H flux of 3 x 10⁸ cm^-2s^-1 and a solar wind Si/H ratio of 1 x 10^-4. Estimates of all the solar wind abundances are calculated by assuming that the relative abundance pattern is similar to CI chondrites for most nonvolatile elements (Document B), e.g., Anders and Grevesse (1989), and that the solar wind atoms are spread over the top 100 nm for light element targets (C, Si, Ge) and over 50 nm for heavy element targets (Pt, Au). This results in conservative solar wind abundance estimates.

Additional measurements: We are planning to use thermal ionization mass spectrometry (TIMS) to make additional bulk purity measurements. Blank measurements to date at the Caltech TIMS lab indicate that , with < 10 g of material, the following elements can be assayed to blank levels comparable to INAA: Mg, Ca, Ti, Sr, Sm, Nd, and Ba. Inductively coupled plasma mass spectrometry will also be used to check detection limits of the following elements to below the expected solar wind concentration: Na, Mg, Al, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ge, Ru, Y, Ba, and Pt (R. Brown, personal communication).

1.2 Si by INAA: Literature values. Instrumental neutron activation analysis (INAA) is the primary technique used by industry to determine individual elemental impurities in bulk Si. INAA is well-suited for bulk analysis, and gives results for a large number of elements at once. Its sensitivity for the electrically active metals of interest to industry is quite good. Fig. M1 shows literature INAA results from several analyses for a large number of elements. They are plotted logarithmically, normalized to elemental solar wind concentrations from a 2-year fluence averaged over the top 100 nm, as above. The dashed line shows our materials purity goal of (impurity/s.w.) £ 0.01. It is important to note that only the filled symbols are actual measurements; open symbols (and X's) are detection limits. An upper limit below the dashed line is a significant result. An upper limit above the dashed line tells us nothing. Detection limits vary according to the neutron fluence, the size of the sample, and the counting times. For the most comprehensive study, by Takeuchi et al. (1993), a symbol is given at both upper and lower ends of the range of measurements, although nearly all of Takeuchi's lower end values are simply detection limits. In general, relatively few measurements were made above the detection limits, and most measurements of bulk impurities are below the solar wind concentration. Among these 48 elements, except for Au and As, there are no cases where the INAA results are consistently above the design goal.

Overall, Figure M1 demonstrates why we have emphasized Si as a collector material. It should be noted that the study of Takeuchi et al., which is most useful for our purposes, did not use FZ silicon which may be higher purity. Given the requirement to measure the solar wind element fluence to an accuracy of ±5% (1s), a very conservative purity criterion is that the background level of a given element be less than 1% of the solar wind concentration, e.g., as shown by the dotted horizontal line in Fig. M1. Using this criterion, Table M2 summarizes the situation, using data from Table M1 and Fig. M1, in terms of the specific science objectives given in proposal Table 1-2. Elements for which there are only upper limits which are too high are omitted from Table M2.

1.3 Si by INAA: Our own analyses. We are attempting to confirm the literature conclusions with our own analyses of Si from known vendors. During Phase A co-investigator M. Ebihara carried out several neutron activation analyses on Si crystals from two candidate vendors. In choosing materials, we singled out FZ crystals as likely to be more desirable than CZ crystals due to the absence of contaminants leached from the CZ crucible. Of the four known foreign FZ Si producers, Topsil was used because they had the highest resistivity material and were the most helpful. Within the US, only one producer, Unisil, has recently (1994) come on line, though another producer may be in the works.

A 19 g Topsil sample (resistivity > 50 kOhm) was irradiated in 1994 at the JRR-3 reactor in site HR-2 for approximately 2 days. Estimates of the flux from co-irradiated standards give 7.1 and 7.0 x 10¹³ n/cm²/sec, respectively for 1173 keV and 1332 keV. The sample was cooled for 1 month before measurement at the Japan Atomic Energy Research Institute (JAERI). Before measurement, 1.2% of the material was etched away to remove surface contaminants. The counting time was 2 x 10⁵ sec. each at distances of 0 and 5 cm from the Ge detector.

Irradiation of two additional Si crystals, one each from Topsil and Unisil, took place in 1995. This time the neutron fluence was three times the 1994 irradiation with other conditions essentially the same. Because of relatively high brehmstrahlung background (mentioned below) detection limits were only comparable to the 1994 data.

The Ebihara data are also plotted as detection limits on Fig. M1. Only one element (Au) was actually measured above detection limits. Comparison with the literature data in Fig. M1 shows that with some exceptions Ebihara's detection limits for the elements shown are generally in the same range as those in the literature. Fewer elements are shown in part because only one count was done, with no attempt to measure elements with short half lives.

Dr. Ebihara's analyses revealed a fundamental limit to bulk analysis of Si. The neutron-rich radioactive isotopes ^31,32Si have half-lives of 2.6 hours and 280 years, respectively. With a moderately high neutron flux, a significant amount of double neutron capture occurs on ³⁰Si to produce the relatively long-lived ³²Si. Consequently, radiochemistry (RNAA) will be required to produce lower detection limits than those shown in Fig. M1.

Fig. M2		31P 100%	32P 14.3d	33P 25.3d
28Si 92.2%	29Si 4.7%	30Si 3.1%	31Si 2.62h	32Si ~280a

 

1.4 Si feasibility by Measurement Objective. Based on the above results, proposal Table 2-4 shows that Si will be a suitable collector material for most of the 18 specific measurement objectives. As shown in Table 2-5, some of the specific measurement objectives encompass a number of elements. The following paragraph documents these objectives for Si on an element-by-element basis based on Table M2 and Figure M1.

Objective 8: The A=80-90 group of elements consists of Se,Br,Rb,Sr, and Y. Literature data show adequate purity for Se. In Phase B we shall obtain RNAA measurements of Se. Br is more difficult. Literature INAA data for Rb give an upper limit of about 0.1 times the SW value, but the geochemically similar element, Cs, is present at about 0.001 SW levels. Rb and Sr will be analyzed in Phase B by thermal ionization mass spectrometry (TIMS). Y analysis by ICPMS should have detection limits below the solar wind value (R. Brown, personal communication) and we can also attempt to get a measurement by synchrotron radiation X-ray fluorescence using the high energy photons from the Argonne facility. Y is geochemically similar to the mid-heavy REE elements. Literature Si analyses for Tb and Yb are distinctly below the 0.01 limit. The A=120-140 elements are Sn, Sb, Te, I, Cs and Ba. The literature data indicate that impurity levels of Cs and Sb are sufficiently low. Takeuchi et al report some samples with Sn concentrations at about 0.05x SW levels. As this is the least important of the A=120-140 element group, somewhat larger errors on the Sn abundance might be acceptable. RNAA measurements of Se and Te will be carried out in Phase B. Given the relatively short Phase B time scale measurement of I might not be possible, but we will explore I and Br analysis by measurement of neutron capture Xe and Kr respectively. Ba will be analyzed by ICPMS, and/or thermal ionization mass spectrometry in Phase B.

Objective 16: Literature data are available for 8 REE . Some samples analyzed by Takeuchi show light REE (La,Ce,Nd) concentrations around 0.03x SW values, but other samples with upper limits around 0.01x SW. In our analyses one Topsil sample had high La and Ce levels of about 0.3-0.4x SW. The other samples have upper limits of about 0.1x SW. In general our REE upper limits are about a factor of 10-50 higher than those of Takeuchi et al; the reasons for this are being investigated. If the INAA limits cannot be improved, we can use simple RNAA techniques for La, Ce, Tb, Yb, and Lu along with TIMS measurements of Nd and Sm to characterize the light REE.

Objective 13: The Pt-metal group of elements (Mo, Ru, Rh, Pd, W, Re, Ir, Os, Pt, Au) could be more difficult as these elements may tend to alloy with Si given the reducing conditions required to produce Si metal, and exposure to at least some of these elements occurs in the bulk Si reduction and crystal-producing processes. Although Mo, W, and Au appear too high, Ir is quite low and the Os upper limit is £5%. We can afford to lose a few of these elements and still test for systematic Pt-nugget fractionation in the solar nebula; in principle the Ir abundance alone can accomplish this objective.

1.5 SOS. Epitaxial silicon on sapphire (SOS), which is commercially available for applications where radiation hardening is desirable or where an insulator is desired for the structural backbone, was initially considered. The intrinsic electrical properties of the Si in SOS are nearly as good as FZ Si, but very little is known about individual element impurities. At present SOS remains a possible back-up material for Si wafers in the unlikely event that etching of Si wafers to catch only the top 100 nm proves too difficult. Al is an acceptable material for high flux reactor irradiation, and the sapphire is not easily etched away. The Si can be easily stripped from the sapphire substrate after irradiation, prior to gamma counting.

2. Germanium. While we have baselined Si for collection of solar wind, a small amount of Ge will be used for collection as well, due to its advantages for analysis by SIMS. However, testing the purity of Ge is much more difficult than for Si because the high neutron capture cross sections of the Ge isotopes renders NAA useless.

2.1 General Characteristics. Hyper-pure zone-refined Ge is produced up to 7 cm diameter for nuclear detectors. This material is about as low as Si in electrical impurities, with net impurity concentrations (the difference between p-type and n-type carriers) of ~10⁹ cm^-3 available. Like Si, some data are available by techniques other than NAA. The best crystals have 10¹³-10¹⁴ cm^-3 each of O and Si if grown in a silica crucible, and ~10¹⁴ cm^-3 C if grown in a graphite crucible (because of its viscosity and density, it is not possible to produce Ge by the crucible-less FZ method on Earth). Of the electrically active species, typically the most abundant impurities are Al, P, B, and In, all generally below 10¹⁰ cm^-3 (e.g., Haller, 1991). By comparison with Table M1, Ge is as good as or better than Si in these elements, well below the solar wind concentrations for these elements.

2.2 Ge SIMS Analyses. To check the purity of Ge, and also demonstrate sufficiently low detection limits, a SIMS analysis was carried out on two Ge samples using the PANURGE ion probe at Caltech. Elements targeted were Mg, Al, Si, Ca, Ti, Sc, Cr, Mn, and Fe, as these are common elements that have favorable sensitivity factors (RSFs) using positive secondary ions. The RSFs were taken from Wilson et al. (1989), using O₂⁺primaries, but should be applicable for O^-within a factor of 2. The dark current backgrounds were 0.04-0.05 cps. The Ge matrix signal was monitored at mass 73, and ranged from 2-5 x 10⁴ cps. Charging apparently limited us from using a more intense primary beam. Mass resolution was set at M/DM ~ 1930, so that molecular interferences can generally be ignored.

Table M3: SIMS analyses of Ge crystals

	Sample #1		Sample #2		Expected Solar Wind
Element	Measured	Det. Limit	Measured	Det. Limit
Mg	3.1 ppba	1. ppba	----	1.0 ppba	4000. ppba
Al	1.5	0.4	1.2	0.3	320.
Si	28.	3.	240	3.	3800.
Ca	0.2	0.03	0.3	0.05	220.
Sc	0.2	0.07	5.6	0.11	0.13
Ti	----	0.3	----	0.5	9.1
Cr	----	0.3	----	0.5	50.
Mn	----	0.4	----	0.6	36.
Fe	----	1.7	----	2.6	3300.
Ni	----	----	----	----	190
Rb	----	----	----	----	0.08
Ba	----	----	----	----	0.02

Results are given in Table M3. Doubly charged GeO and GeO₂ did not seem to contribute any counts. Initially all of masses 24, 25, 26, 27, 28, 29, 30, 40, 44, 45, 48, 50, 52, 53, 55, and 56 were monitored, but after no unusual ratios turned up, only the major isotopes were monitored. Si was clearly measured in both samples. The mass 40 signal seemed to be Ca, as the 40/44 ratio was approximately correct for Ca. Likewise, the 25/24 and 26/24 ratios were approximately correct for Mg in the Tennelec sample. Signals at 27 and 45 were slightly above the dark current backgrounds, but it is less certain that these were actually Al and Sc.

From talking to the suppliers, it appears that most of the actual measurements except for Si are probably due to low levels of instrumental contamination.

This analysis, while rather cursory, shows that Ge purity is clearly sufficient for Mg, Al, Si (1 sample < 1% of solar wind), Ca, Cr, Mn, and Fe. By comparison with Table M3, a SIMS analysis of Si had detection limits for ~50 ppba for Ca and Ti due to interfering isobars (Document J). The best SIMS instruments have a factor of ~5 lower dark currents. Even with the present instrument, other elements to be tested in Ge could include F, S, Cl, Se, Br, Te, and I for negative secondaries. For positive secondaries, Ni, Rb, and Ba will be tried.

3. Noble Metal Foils and Films. The highest purity metal foil commercially available is 99.9999% (6-9s) pure gold foil. This turns out to be not as pure as Si for most elements, but it may have some advantages. For example, because gold has only one isotope at relatively high mass, SIMS analyses will have almost no interferences (see Document J). Like Ge, gold is also not analyzable by NAA, so less information is available on the impurities in gold. Fig. M3 plots the impurities as assayed by the manufacturer (ESPI) ratioed to the expected solar wind concentrations. Several elements are above the background. Notable is Si, which will not be measurable in Si wafers. However, Fig. M3 does not tell the whole story. SIMS analyses of foil samples revealed that many of the impurities are concentrated near the surfaces, a product of the rolling process. Therefore, it is important to use a batch of foil coming off freshly electropolished rollers; however, even this may not adequately remove particulate contamination.

Figure M3

Electron beam deposited Au on sputter-cleaned 6-9s Au foil could be an adequate collector material for selected elements, such as Si, light alkalis and Fe (which diffuse rapidly in Si), or S (which has high surface concentrations on polished Si wafers). The Au single crystals considered as backup material for the concentrator target (section 4.1) are too small for the large areas needed for the collector arrays. Epitaxial Au or Pt films might be ideal; however, there are problems finding an acceptable substrate. Chang (IBM research labs) has produced multilayer epitaxial films beginning with Si crystals: Cu, then Pd, then Au or Pt. This is necessary to have a suitable lattice spacing match for Au/Pt crystal growth. However, the chemical complexity of the layers is a major disadvantage. We have explored possible substrates which have a sufficiently similar lattice spacing to grow Au/Pt crystals directly Only LiF looks practical, but Li and F are both high priority elements, ruling out LiF because of the cross- contamination risk.

Because at least one other material besides Si should be used for general-purpose solar wind collection, an effort will be made in Phase B to characterize both Ge and Au foils/films to see which of these materials, if not both, should be used.

4. Concentrator Target Materials. As a primary requirement the concentrator target must not have a native surface oxide layer, and must have bulk oxygen concentrations lower than 1% of the expected solar wind averaged over the top 100 nm. Additionally it is desirable to have low concentrations of other light elements, particularly N, C, and F. We are considering two materials: noble metals and diamond, with diamond presently favored.

4.1 Single Crystal Noble Metals. Grain boundaries contain the majority of all impurities in any polycrystalline material. Single crystals therefore have significantly lower oxygen impurity than foils, and also do not have the impurities (usually oxide grains) associated with rolling foils. They still can have surface impurities from the polishing process, though these are usually less than from rolling. Single crystal metals are in a cost and size range that make them suitable for the concentrator target but not for bulk collectors. We have used SIMS to analyze gold crystals in several different orientations and two different companies, and a Pt crystal. All samples were from Monocrystals except for one <100> Au crystal from ESPI; the Pt crystal had a <111> orientation. The Monocrystal Au samples were first mechanically polished and then electropolished (gold is so soft as to be difficult to polish well mechanically). The Pt crystal was mechanically polished. It had a much better looking surface, but turned out to have contamination grains buried just below the surface. The gold had some similar contamination, but not as many, and they tended to be more on the surface. Our analyses focused on finding the matrix purity and we were able to look between any such contaminants. Hence we looked for the lowest elemental concentrations within each SIMS profile.

TABLE M4 SINGLE CRYSTAL Au and Pt
Everything is an upper limit, in parts per million atomic (ppma)

Element	Sample				Solar Wind Conc.
	Mono Au		ESPI Au	Mono Pt
	<111>	<100>	<100>	<111>
C	0.69	0.34	34	0.54	250
O	0.48	0.12	1.8	0.003	580
F	0.02	0.01	----	0.19	0.021
Al	----	0.3	----	----	6.2
Si	0.042	0.42	0.63	0.42	73
S	0.068	0.068	0.102	0.25	25
Cl	0.26	2.3	0.2	4.1	0.13

• Conditions were varied considerably over course of analysis. These are lowest measured for a given sample. UCLA Cameca 1270 analysis 8-17-95
• Considerable range in detection limits, so can't conclude that ESPI less pure than Mono
• Calibration factors (RSF) for Au matrix from Wilson et al., (1994)
• RSF for F and Cl assumed same as Br
• Elements in Pt assumed to have same RSFs as measured for Au
• Al assumed to have same RSF as Au.
• 9-29 implant calibration for O would give lower concentrations by a factor of 2.5
• Concentration factor of 20 assumed; this will be somewhat less for Al and heavier elements.

The SIMS results are shown in Table M4 for O, C, F, and several other elements. Comparison is given with the solar wind, concentrated by a factor of 20x and averaged over the top 100 nm. It is clear that the oxygen background is below the desired 1% of the solar wind, and in some cases it is much lower. Carbon is also below this goal, but F will require further study. The results are upper limits because of instrumental background (see Document J).

Nitrogen in single crystal Au was also analyzed by static mass spectrometry. The bulk sample had 0.63 ppma, just under the 1% design goal for blanks.

4.2 Diamond is the only other known material with essentially no native oxide layer and potentially low enough O (and N) concentrations. The purity levels and diffusion properties of natural diamonds have been studied for some time because of their potential for revealing information about the earth's mantle where they formed. For example, the aggregation state of nitrogen in natural diamonds has been used to study the thermal history of mantle diamonds. According to IR analyses, nitrogen diffuses very slowly in single crystal diamonds and requires high temperatures. For example, D was estimated ~ 10^-14 cm²/s at 1900^oC (Chrenko et al., 1977; see also Allen and Evans, 1981). Some other elements (e.g., helium, Wiens et al. , 1994; lithium, Cytermann et al. , 1994) have been demonstrated to have extremely low diffusivities in natural diamond.

CVD diamonds differ from natural diamonds in their formation process and in the fact that they are polycrystalline. CVD diamond producers have mentioned detection of a number of elements as impurities, in contrast to the relatively high purities of inclusionless type IIa natural diamonds. For species such as N and O, IR analyses can yield relative abundances of particular molecular bonds, but may not give quantitative results on the actual impurity concentration if some of the impurity has a different bond structure or is trapped in inclusions. This has motivated investigation of other methods for impurity analyses (e.g., McNamara et al. , 1994).

4.2.1 Diamond SIMS Study. We undertook a SIMS study of CVD diamonds, both for purity levels and diffusion properties, to determine the usefulness of CVD diamonds as substrates for collecting light elements (Li-Ne) in the solar wind. The study included H implantation to simulate conditions of our solar wind target. Our results include purity constraints for F and Cl as well as N and O, and diffusivity constraints for H, N, and O.

We obtained sample diamond films from a variety of sources, including GE, Norton, DeBeers, and several university and government laboratories. All diamonds were isotopically normal (¹²C/¹³C = 89) except one sample designated "B¹²C", which was essentially pure ¹²C. Initial IR analyses on some of the polished samples could not detect any nitrogen.

Initial attempts to analyze diamonds by SIMS indicated that surface charging is too severe in the absence of a conductive coating for at least some of the samples. Most vapor-deposit coaters introduce moderate levels of O and other impurities, which might compromise detection-level work. To avoid this, diamond samples to be analyzed for impurities were coated in a UHV Riber MBE coater designed and used exclusively for semiconductors. Previous analysis had shown that Pt coats from this instrument had < 60 parts per million (ppm), atomic ratio, of compared with at least several hundred ppm from typical Pt or gold coaters. The diamonds were first cleaned by ultrasonicating and heating in TCE, acetone, and isopropanol, and dried. After introduction to the MBE system, the samples were heated to 490^oC for 30 minutes. The initial pressure in the chamber was 5 x 10^-11 Torr, and did not rise above 2 x 10^-8T. during heating. A 100 nm Pt coat was transferred at 0.05 nm/sec. by electron beam evaporation.

For the diffusion experiment, a polished diamond designated TD939 was implanted with the fluences and energies given in Table M5. These are the expected solar wind energies for ions passing through the electrostatic concentrator. Fluxes were below 0.3 microAmp/cm² in every case. Prior to implanting, the Pt coat had been removed from this sample by soaking in warm aqua regia, which caused a slight amount of surface roughness in addition to removing the Pt. Profilometry indicated surface roughness was on a scale approaching 100 nm. The sample was subsequently broken into fragments a, b, and c. Pieces a and b were vacuum sealed (< 10^-6 Torr) in quartz glass and heated to 400 and 850 ±10^oC, respectively, for 2.60 x 10⁵ s. Piece c served as an unheated control. These pieces were then gold coated in a standard vapor deposition coater, as higher impurities in the conducting coat could be tolerated in the implanted samples.

Table M5. Implanted ions, listed in order of implantation

Isotope	Energy (keV)	Fluence (cm-2)
1H	11a	2.0 x 1016  a
13Cb	72	1.2 x 1014
18O	86	3.0 x 1014
15N	79	4.0 x 1013

^aImplanted as 1 x 10¹⁶ cm^-2 of 22 keV H₂
^bBy-product; not expected to have significant effect

SIMS analyses were done using the Cameca IMS 1270 instrument at UCLA and an IMS 4F at Charles Evans & Associates, both run in negative secondary ion mode. In all cases, samples were kept under vacuum at least 12 hours prior to analysis. For detection limit work, the primary Cs⁺ beam current was as high as practical, ~100 nA; lower currents were used for diffusion experiments. A normal incidence electron flood gun was used to help control charging except during H analyses. The analytical area was 30 microns diameter. The C matrix counts were monitored on a Faraday cup, with everything else measured on an electron multiplier. Faraday-multiplier calibrations were performed within a day of the analyses. Hydrogen and oxygen were typically analyzed at low mass resolution (m/Dm ~ 500) since there are no expected interferences. Nitrogen was analyzed as the CN^- molecule at mass 27. Mass 27 had two peaks, one from ¹³C¹⁴N and one from ¹²C¹⁵N. Detection limit analyses were typically done at low mass resolution, combining the contributions from both ions. However, some samples, including an isotopically pure ¹²C diamond, were analyzed at high mass resolution (m/Dm ~ 4,000). Diffusion analyses for N were done at high mass resolution as well, monitoring both the implanted ¹⁵N and contaminant ¹⁴N.

4.2.2 Diamond SIMS Results: Impurities. Upon initial analysis using lower sputtering rates, the diamonds appeared to have consistently high concentrations of N and O, and relatively small but measurable amounts of F and Cl. However, these turned out to be instrumental artifacts. For most samples the counting rates for N and O were essentially constant, independent of the carbon matrix counting rates. Figure M4 illustrates this point by showing a sample that did have measurable N but no measurable O or F. The analysis started with a raster size of 100 microns and a field aperture of 30 microns. After conditions equilibrated, the raster was reduced to a spot, significantly increasing the matrix counts. The N counts increased along with the C, indicating that N was from the sample. The O and F counts did not increase, indicating that these were not associated with carbon from the matrix. The source of the background is discussed below. Because this instrumental background was constant, the best detection limits were obtained with the highest possible sputtering rate, with the primary beam striking completely within the field aperture, as in the right side of Figure M4. The background, and hence the detection limits, varied somewhat from sample to sample. What was initially thought to be Cl was actually C₂B in at least some samples, confirmed by measurement of the correct proportions for the two isotopes of boron. In other samples, the mass 35 signal did not increase with the matrix, and was therefore all instrumental background.

The results for N, O, and F are shown in Table M6. Chlorine is not reported, but following the above discussion, it should be significantly below 0.05 ppm, measured at mass 35 in several samples. Note that only one sample ("G1") had oxygen detected above the instrumental background. The lowest O limit was 1.3 ppm in samples KMR1216 and "B¹²C". Nitrogen was detected above background in KMR1216 and 1146, and in both of the ANL samples. The lowest N limit was 0.12 ppm. The F measurements were verified by a high mass resolution scan over mass 19, which confirmed the existence of a single peak at the proper mass. However, like O and most N analyses, the fact that the F signal did not rise with C when the primary beam was changed to spot mode indicates that the F measured was instrumental contamination. The lowest F detection limit was 0.0025 ppm in both the "B Normal" and KMR 1216 samples.

TABLE M6 DIAMOND
Unless noted, all values are Upper Limits.

	O/C	O	F	N	S	Cl
Sample	cps ratio	ppma (1)	ppma (2)	ppma (3)	ppma (4)	ppma (2)
B Normal	1.07E-06	1.7	0.0025	0.12	0.012	0.062
B 12C	1.97E-06	1.3	0.0033	0.13	0.008	0.18
B IF-1	1.2E-06	2.0	0.05	0.10
GE 5 mm pol.	1.2E-06	2.0	0.02	1.5*
GE 7 mm	3.0E-07	0.5	0.03	0.6*
G1	8.10E-06	13	0.07	0.95	0.73	0.051
KMR 1216	8.31E-07	1.3	0.008	25*	0.016	0.048
KMR 1251	2.31E-06	3.7	0.06		0.014	0.15
KMR 1146*	9.78E-07	1.6	0.0025	20*	0.016	0.1
KMR 1175	1.4E-06	2.4	0.04	3.0*
KMR 1184	9.0E-07	1.5	0.36	21*
KMR 1192	9.0E-07	1.5	0.03	9.3*
KMR 1242	1.9E-06	3.3*	0.04	36*
KMR 1267	1.5E-06	2.6	0.04	7.7*
Norton TF068	2.56E-06	4.1	0.062	0.48	0.041	0.12
Norton TD 939	2.20E-06	3.6	0.057	0.33	0.029	0.12
ANL 941012		1.5	0.07	7.5*	0.01
ANL 940928		2.5	2.5*	125*	0.08
Solar wind, conc (5)		170	0.006	22	7.4 ppma	.04 ppma
	O/C cps ratio from UCLA Cameca 1270 instrument.

*Asterisk indicates measured value, rather than upper limit.
Sample analysis conditions and detection limits vary considerably, producing large range in upper limits. When data from two instruments available, lowest upper limit is tabulated.
CVD Diamond analyses: UCLA Cameca 1270 SIMS; 9-8-95 , Evans 7-6-95 Q-SIMS, Evans 4.5f 10-95, UCLA 6-97. Samples with no S, Cl analyses carried out 6/97.
(1) Based on implant standard, analyzed at time of analyses.
(2) Relative Sensitivity Factor from Wilson (1991)
(3) Based on implant standard, analyzed at time of analyses.
(4) Based on RSF for O primary ion from Wilson (1991)
(5) Two year fluence, averaged over outer 1000 Angstroms; concentrator concentration factor =20

4.2.3 Diamond SIMS Results: Diffusion. Implant profiles for N, O, and H in the unheated and heated pieces of TD939 are compared in Figs. M5, 6, and 7. The shoulders near the surface are predominantly due to the Au conductive coat. Concentrations are ratioed to C, which is low in the coat relative to the diamond matrix, while N and O are higher in the Au than the diamond. In Fig. M5 the geometric means of two to three profiles are plotted for each temperature. Reproducibility is discussed below. For N and O, the backgrounds were subtracted from the ¹⁵N and ¹⁸O profiles using the ¹⁴N and ¹⁶O profiles. This subtraction did not eliminate the near-surface shoulders completely because the signal was dropping so rapidly in this region that small differences in the analysis times of the individual isotopes became significant. The rising portion of the main peak matches the TRIM (IBM Theoretical Ranges of Ions in Matter) simulation well, while all of the distributions have longer tails than predicted by TRIM, due to channeling within individual crystals.

The important point to note is that the heated and unheated profiles are identical within the limits of expectation with a slightly rough surface. Diffusion would cause an increase in the dispersion around the mean, given by

Diffusion: Solar Wind Applications. For the purpose of solar wind collection, it is important to know whether the elements of interest (N, O, Ne, Li, etc.) diffuse in CVD diamond at temperatures up to 200^oC. Since we determined only upper limits on diffusivity at 400^oC and 850^oC, it is not possible to estimate D(200^o) from an Arrhenius plot. Instead, a most conservative approach would be to say that an upper limit is given by the upper limits determined at 400 and 850^oC. This is ~5 x 10^-18 cm²/s. Over an exposure period of approximately two years, this would lead to a characteristic diffusion length of sqrt(4Dt) = 23 nm. This is smaller than the concentrator target's 100 nm depth for the implantation peak, but it is possible that the atoms will only need to diffuse to a grain boundary before they can escape.

More realistically, Chrenko et al. estimated D ~ 10^-14 cm²/s for N in diamond at 1900^oC. Using this value and our upper limit at 850^oC, one can extrapolate a diffusivity upper limit of 2 x 10^-27 cm²/s, which translates to less than a fraction of an Angstrom in two years. One can also use estimates of the activation energy for nitrogen in diamond, based on the changes of IR spectra as a function of heating. One is from Chrenko et al., who estimates the activation energy to be 60 kcal/mol. Using this value, and his diffusivity estimate of 10^-14 cm²/s at 1900^oC, one gets D ~ 6 x 10^-36 cm²/s at 200^o, meaning that atoms are firmly held in place. It is thus expected that the actual D values for N, O in diamond are many orders of magnitude lower than we could measure.

4.2.4 Diamond SIMS Study Conclusions. The oxygen and nitrogen upper limits were in at least several cases below 1% of the expected concentrated solar wind values, meeting our purity goals. In addition, F concentrations are below that expected for the concentrated solar wind. Though not shown here, S concentrations were also below 1% of the expected solar wind, which would be applicable if CVD diamond is also used as a collector for unconcentrated solar wind.

Diffusion studies indicate that solar wind-implanted N and O will be retentively held at temperatures of up to 200^oC for the duration of the mission, though longer duration studies are presently underway to confirm this.

A final result of our study is that for diamond there is no detectable radiation damage from the fluence levels expected for the solar wind. This study substituted H for He, so the conclusions should be tested again, but from studies of similar He fluences and energies in metals, no damage is expected.

4.3 Comparisons between noble metals and CVD diamond. There are a number of differences between diamond and noble metal foils besides those mentioned above. Obviously solar wind carbon cannot be measured in a diamond matrix. This means that either diamond must be used in tandem with another material for the concentrator target, or solar wind carbon must be analyzable without concentration. Secondly, the chemistry during sample extraction differs between the two matrices. Solar wind oxygen extracted from diamond would come out predominantly as CO, greatly simplifying the procedure for GSMS of O as CO molecules (see Appendix J). On the other hand, GSMS of O as O2 may be simpler by extraction from a noble metal matrix. A disadvantage of diamond is that it is not a good electrical conductor. A diamond target would need to be very thin in order to avoid charging. Further tests will need to be done in this regard.

An important difference is that the ions are implanted quite differently in the two matrices due to the large difference in atomic number between C and noble metals. Actual ranges and straggles (standard deviations) for two charge states are compared in Table M8. Implantation in noble metal gives a relatively broad shallow peak. Approximately 20% of the ions are backscattered out of the target. This fraction is slightly dependent on atomic mass, with TRIM calculations showing approximately 5 permil/amu fractionation for oxygen of the same charge state and normal incidence. More calculations to cover the range of angles and charge states need to be done to determine the overall mass fractionation. By comparison, implantation in diamond gives a deeper peak (Fig. M5), which makes solar wind easier to separate from surface contamination and has negligible mass fractionation.

Fig. M5 Comparison of 16 keV ¹⁶O⁺⁶ ion profiles for normal incidence into Pt and diamond with and without -10 kV/q acceleration, as planned for the concentrator.

Both diamond and noble metals are matrices characterized by very low diffusion of trace elements, as was shown above for diamond.

The overall differences between diamond and noble metals are summarized in Table M9. Our present plan is to use diamond for the concentrator target. However, we have to deal with the potential charging problems. At present Au is a back-up material.

Table M9 Concentrator Target Materials Characteristics Summary
	Diamond		Noble Metals
O Purity	proven	OK	proven	OK
N Purity	proven	OK	proven	OK
C Analysis	no	X	proven	OK
O Analysis by GSMS	as CO or O2	?	as O2	?
Surface Cleanliness	relatively good	OK	relatively poor	X
Low Diffusion	proven	OK	in testing	?
Electrically Conductive	no	X	yes	OK
Implant Depth and Straggle	~1000 A, rel. tight	OK	~500 A, rel. broad	X
Mass Fractionation by Backscattering	no	OK	yes	X

5. Materials for Noble Gas Collection. Metal foils are preferred for noble gases because of their history dating back to the Apollo solar wind foils experiment, along with the fact that noble gas backgrounds are typically very low in most materials. For example, the Xe blank in an off-the-grocery-shelf Reynolds Wrap foil is 0.0018 parts per trillion atomic, or 3.9 x 10^-4 times the 2-year solar wind fluence averaged over the top 100 nm. The lighter noble gas blanks are even better relative to the solar wind. It turns out that the Bern foils used for the Apollo solar wind experiment have nearly the same blanks as Reynolds Wrap except for Ne, where the Bern foils are a factor of five better.

5.1 Solar wind. Noble gas mass spectrometry has traditionally been a bulk analysis technique. However, even with these low blank concentrations, consuming the whole foil results in blanks that are up to a few percent of the bulk solar wind fluence. Sampling noble gases from individual types of solar wind, such as coronal mass ejections (CMEs), is a high science priority. The CME fluence could be as low as 10% of the total solar wind fluence, in which case, a blank from the whole foil would be too high. Because of this, we are planning to use laser extraction to remove only the near-surface layer rather than melting the whole foil. This may be most easily accomplished if the noble gases are implanted into a thin metal layer on a refractory substrate. The thin layer would be volatilized by a pulsed laser, leaving the substrate untouched. There are two such materials we are currently investigating: silicon on sapphire (SOS) and aluminum on sapphire (AOS). The SOS (mentioned in 1.5 above) is made from ultra-high purity starting materials. AOS has been used for low-blank Kr work for a number of years (N. Thonnard, personal communication). Samples of both materials are at C. Hohenberg's lab awaiting analysis using the laser extraction system. Aluminum is preferred due to its heritage and because Si surfaces has been reported to adsorb and hold atmospheric Xe very tightly (e.g., Garrison et al., 1988).

5.2 Energetic Particles. Efforts are underway to collect high energy solar particles (SEP) for separate analysis. The existence of a SEP noble gas component for Ne, Kr, and Xe with a composition distinct from solar wind has been postulated for the last fifteen years, though it has never been confirmed by direct analysis in space. We have recently added this to our list of objectives (Table 1-2 in proposal). It may be possible to separate SEP noble gases by stepped heating, as has been done for lunar grains. However, an unambiguous way to measure SEP would be to collect it separately in a multilayer target. Preliminary plans are for a three-layer target consisting of a 0.2 micron Si film deposited on a 1.0 micron Au foil, physically supported by a high transparency (>90%) mesh. The bottom layer would be a ~4 micron Au foil to capture ions in the > 100 keV/amu range (Table M10). To separate solar wind (< 10 keV/amu) from the intermediate (10-100 keV/amu) range, the Si thin film is removed from the 1 micron foil by an acid etch.

Table M10. Implantation ranges of noble gases in Au foil
E (keV/amu)	Ne Range (microns)	Xe Range (microns)
10	0.11±0.09	0.12±0.06
100	0.83±0.29	1.15±0.30
1000	3.88±0.41	5.26±0.49

Range ± (1 std. dev.)
Source: TRIM