Major Compositional Units of the Moon: Lunar Prospector Thermal and Fast Neutrons

Science  04 Sep 1998:
Vol. 281, Issue 5382, pp. 1489-1493
DOI: 10.1126/science.281.5382.1489


Global maps of thermal and fast neutron fluxes from the moon suggest three end-member compositional units. A high thermal and low fast neutron flux unit correlates with the lunar highlands and is consistent with feldspathic rocks. The South Pole–Aitken basin and a strip that surrounds the nearside maria have intermediate thermal and fast neutron flux levels, consistent with more mafic rocks. There appears to be a smooth transition between the most mafic and feldspathic compositions, which correspond to low and high surface altitudes, respectively. The maria show low thermal and high fast neutron fluxes, consistent with basaltic rocks.

Neutrons are generated by interactions between galactic cosmic rays and surface material in all planetary bodies that have sufficiently thin atmospheres. Subsequent interactions of the neutrons with surrounding material produce a steady-state, equilibrium energy spectrum that spans from the fast neutron range, where neutrons are born (energies, E, greater than several hundred thousand electron volts), to the thermal energy range (E < 0.3 eV), where neutrons are absorbed. Neutron energy spectra are therefore expected to reflect the composition of near-surface planetary layers. Simulations of equilibrium spectra indicate that the fast neutrons provide information primarily about the Fe and Ti content of soils (1–3). Epithermal neutrons (energies between ∼0.3 eV and several hundred thousand electron volts) reflect primarily the abundance of hydrogen (4), and thermal neutrons reflect the abundance of neutron-absorbing nuclei, primarily Fe, Ti, K, Gd, and Sm (4–7).

The Lunar Prospector (LP) neutron spectrometer (NS) measures the flux of thermal, epithermal, and fast neutrons. Thermal and epithermal neutrons are measured using two 3He-filled gas proportional counters and associated electronics (8). One of the counters is covered with a 0.63-mm-thick sheet of Cd, which, because of its high (>10,000 barns) absorption cross section, shields the counter from neutrons with energies less than ∼0.3 eV. The second counter is covered with an identical thickness of Sn (to ensure a similar response to background fluxes) and responds to neutrons having energies up to ∼1000 eV. Because the two counters are matched, the difference in their counting rates yields a measure of the thermal neutron flux (E < 0.3 eV). Fast neutrons are measured using the anticoincidence shield (ACS) of the LP gamma-ray spectrometer (8).

Neutron flux backgrounds in space are low because free neutrons are unstable, with a mean life to beta decay of ∼900 s. To be detected by LP, neutrons must therefore be produced locally. All of the science instruments on the LP were placed at the ends of 2.5-m booms to minimize spacecraft neutron flux backgrounds (9). This separation and the spacecraft's low mass were sufficient to reduce backgrounds to acceptable levels (measured during transit between Earth and moon; Fig. 1). The first two orbits of LP around the moon had a 12-hour period, followed by seven 3.5-hour-period orbits, then by many intermediate mapping orbits (∼90 km periselene and 150 km aposelene), until 16 January 1998, when LP was placed into its final mapping orbit at an altitude of 100 ± 20 km (Fig. 1). Residual variations in detector counting rates after midday on 13 January reflect, for the most part, compositional variations of the lunar surface. Here, we report on neutron fluxes measured between 16 January and 27 June 1998.

Figure 1

Counts registered by the tin-wrapped (HeSn) and cadmium-wrapped (HeCd) NS sensors during successive 32-s integration periods from instrument turn-on (8 January 1998) to 16 January 1998. LP was launched on 6 January 1998, the NS was operational by 8 January, and lunar orbit insertion occurred on 11 January. Backgrounds were recorded by each counter operated at its low energy threshold during the first 3.5 days of operation when the spacecraft was far from both Earth and moon. The reality of this background was verified by lowering the high voltage on the HeSn detector for a short time on 11 January. The time sequence of HeSn counting rates responded with a step function decrease that was not recorded by the HeCd detector. Counter thresholds were commanded to their operational values late on 14 January, as seen by the sharp decrease in counting rates in both counters.

Thermal and fast neutron maps. The difference in counts registered by the Sn- and Cd-covered counters in successive 32-s counting intervals was binned into equal area pixels, equivalent to 2° by 2° latitude-longitude bins at the equator. These data were segregated into individual 2-week map cycles that each correspond to complete coverage of the moon. Adjacent odd and even cycles were first registered for overlapping data and then added together to compensate for day-night and hot-cold differences that result from the combined effects of lunar gravity and Doppler effects. Lunar gravity reduces the energy of escaping thermal neutrons as they travel from the lunar surface to the spacecraft, and it bends neutrons away from the zenith because their orbits are elliptical or hyperbolic (10). Accounting for the Doppler effect is important because LP orbits the moon at a speed (1.64 km s–1) that is comparable to the speed of a thermal neutron (2.2 km s–1 at a temperature of 293 K) (11). Accepted counts were then corrected for the nonspherical response function of the NS and for variations in the flux of galactic cosmic rays.

The global map of thermal neutron counting rates (Fig. 2) reveals several areas of high and low intensity. The lowest intensity overlaps the maria that fill the large nearside basins, as seen in visible reflectance maps of the moon. These low intensities reflect, for the most part, the combined large absorption cross sections of Fe, Ti, K, Gd, and Sm and their relatively large concentration in mare basalt (5–7).

Figure 2

Global map of the thermal neutron counting rate (given by counts per 32-s spectral integration period). Data acquired during mapping cycles 1 through 12 (16 January to 27 June 1998) are combined and have been partially corrected for instrument response function and variations in the flux of galactic cosmic rays. A basemap constructed using Clementine albedo data (19), showing various lunar features, overlays the thermal neutron counting rates.

A well-defined intermediate intensity generally fills the South Pole–Aitken (SPA) basin (56°S, 180°W). It is similar to that registered from Crisium (18°N, 59°E), Smythii (2°S, 87°E), Marginis (20°N, 84°E), Australe (52°S, 95°E), and a rim that surrounds all of the nearside basins. Although some of these rates may reflect the relatively poor spatial resolution of the LP thermal neutron sensor (footprint diameter ∼450 km), which cannot fully resolve small-scale features such as Orientale and Moscoviense, this reason does not necessarily apply to SPA or Australe. Instead, the intermediate counting rates there reflect the presence of rocks that are distinct from those in surrounding areas and, perhaps, the mixing of basaltic and highlands material in soils at a scale size (<100 km) that is not resolved by the NS.

A roughly annular region of high thermal neutron intensities appears to ring the SPA basin. This region corresponds to the highlands, as shown by the topography measured using the Clementine laser altimeter (12) and modeled from gravity maps (13). This annulus is broken by local regions of low intensity that mark individual impact craters or basins. Specific examples are Orientale (20°S, 95°W), Moscoviense (25°N, 150°E), Australe (52°S, 95°E), Humorum (24°S, 40°W), and Nubium (21°S, 15°W).

The three general regions are delineated in peaks in the histogram of thermal neutron intensities at about 290, 400, and 520 counts per 32-s spectrum (14). They probably correspond to three broadly different surface composition units. We infer that these units correspond to a wide range of mare basalts, mafic highland rocks, and feldspathic highland rocks, respectively.

A map of integrated fast neutron counts is shown in Fig. 3. These counts weight the low-energy portion of the fast neutron flux spectrum most heavily because the efficiency of the ACS to neutrons decreases asE –1 (15). Enhanced counting rates mark the locations of the various basin-filled maria on the nearside of the moon, and weaker enhancements mark the SPA, Australe, Orientale, Marginis, Smythii, Humboldtianum, and Moscoviense formations. This pattern matches the regions of high Fe and Ti abundances inferred from spectral reflectance measurements made by Clementine (16), and those of low thermal neutron flux (Fig. 2) [see also (7)]. Enhanced fast neutron emission from Fe and Ti probably reflects the higher number of neutrons relative to protons within Fe and Ti nuclei (each containing an excess of four neutrons) than found in lower mass nuclei (which typically have an equal number of neutrons and protons). It is therefore reasonable to expect that more neutrons will escape from Fe and Ti after being hit by a galactic cosmic ray than would emerge from O or Si, a fact confirmed by simulations (1, 2).

Figure 3

Global map of the fast neutron counting rate (given by counts per 32-s spectral integration period). Data acquired during mapping cycles 1 through 12 (16 January to 27 June 1998) are combined and have been partially corrected for instrument response function and variations in the flux of galactic cosmic rays. A basemap constructed using Clementine albedo data (19), showing various lunar features, overlays the fast neutron counting rates.

Correlation with composition of returned samples. Translation of thermal counting rates to surface composition requires intercalibration with surface samples of known composition. This information is available from measurements of the composition of soils and regolith breccias returned to Earth by the Apollo and Luna missions (17). The correlation between the measured ratio of fast-to-thermal neutron intensities and the macroscopic absorption cross sections from the samples (Fig. 4) is fair, with a correlation coefficient r = 0.71. The measured correlation can be improved slightly by repositioning the Apollo and Luna subsatellite footpoints within the spatial resolution elements of the neutron sensor, yielding r = 0.78. In contrast, the correlation is better between the fast-to-thermal neutron flux ratios simulated using ONEDANT (18) and the macroscopic absorption cross section, also shown in Fig. 4 (r = 0.99).

Figure 4

Correlation of the simulated (filled-square symbols) and measured (open-square symbols) fast/thermal neutron counting rates (both arbitrary units) as a function of the calculated macroscopic neutron absorption cross section. In order of increasing cross section, the points correspond to Apollo 16, Luna 20, Apollo 17, Luna 24, Apollo 15, Luna 16, Apollo 14, Apollo 12, and Apollo 11. Bulk compositions of the soils and regolith breccias from the six Apollo landing sites and the three Luna returned-sample sites (17) have been used to calculate the effective absorption cross sections for each site. The line gives the linear regression between simulated counts and the macroscopic absorption cross section (r= 0.99).

These results imply either that measured thermal neutron intensities are not uniquely related to composition, or that the composition of soils returned from discrete landing sites does not adequately represent the heterogeneous composition of the larger volumes (more than 450 km in diameter by ∼50 cm deep) sampled by neutron measurements made from orbit at an altitude of 100 km. We believe the last interpretation is more consistent with the data. Otherwise it would be difficult to explain the good correlation obtained between measured thermal-to-fast neutron flux intensities and the Fe and Ti abundances inferred from Clementine spectral reflectance data (7).

Correlation with albedo and topography. The thermal and fast neutron intensities (Figs. 2 and 3) resemble the visible albedo map of the moon measured by Clementine (19) (r = 0.80 and r = –0.76 for thermal and fast neutrons, respectively). The principal cause of lunar albedo variations is the presence or absence of Fe-rich mare basalts. The correlation of thermal counts with the albedo is improved by eliminating terrain covered by KREEP (potassium, rare-earth elements, and phosphorus) basalt (7). The correlation in the area bounded by ±30° latitude and 20° to 180°E is 0.91.

The overall correlation between thermal and fast neutron counting rates and surface topography is not as good. However, scatter plots of thermal and fast neutron counts as a function of smoothed height above 1738 km (Fig. 5) are revealing. Three major spurs are evident in both plots. The nearly vertical spurs between –4 and –5 km altitude correspond to the nearside maria. The downward (upward) sloping spur to the left in the thermal (fast) correlation corresponds to the SPA basin. It connects smoothly to the spur on the right that corresponds to the highlands. The last two spurs appear to be a single entity that extends from a mafic compositional unit that marks the SPA (intermediate abundance of Fe and Ti) to a feldspathic compositional unit that marks the highlands (relatively low Fe and Ti abundances). Also apparent in the fast neutron-to-topographic altitude correlation (at the bottom of Fig. 5) is a weak but distinct spur between the main SPA and mare basalt spurs at heights between about –4 and –5 km. These data correspond to the Crisium and Smythii basins. Their distinction from the main vertical spur must reflect a distinct basaltic composition.

Figure 5

Correlation between measured thermal neutron (top) and fast neutron (bottom) counting rates (given by counts per 32-s spectral integration period), and the height of the lunar surface above the mean datum measured by Clementine. Clementine-determined altitudes were smoothed to 14.5° spatial resolution to match the footprint size resolution of the thermal neutrons in the upper plot, and to 5.5° resolution to match that of the fast neutrons in the lower plot.

Summary and discussion. Thermal neutron flux intensities measured using the NS cover a substantial dynamic range (about a factor of 3.5 for thermal neutrons and a factor of 1.25 for fast neutrons) that correlates reasonably well with visible and topographic features on the moon. Three end-member compositional units are delineated. The first, consisting of generally low thermal and high fast intensities, corresponds to the maria deposits that fill the large nearside basins. The second, consisting of generally high thermal and low fast counting rates, delineates the highlands that form a rough annulus centered on SPA. The third unit, characterized by intermediate thermal and fast intensities, is associated with SPA, Humboldtianum, and a rim that surrounds the nearside basins. These last deposits may reflect a separate rock type that is mafic in composition, or may merely reflect numerous, unresolved small-area basalt deposits that dot these regions.

The overall correlation is sufficient to suggest that the measured thermal and fast neutron fluxes, on spatial scales on the order of 200 km (fast neutrons) to 450 km (thermal neutrons) diameter areas (for an LP altitude of 100 km), reveal a smooth transition from a predominantly mafic composition at low altitudes to a predominantly feldspathic composition at high altitudes. The more mafic deposits result from excavation of highlands material by the impacts that created all of the big basins such as SPA, thereby exposing material from the lower crust and perhaps the upper mantle (13), whereas the feldspathic composition reflects the top of the crust that is exposed in the highlands.

  • * To whom correspondence should be addressed. E-mail: wfeldman{at}


View Abstract


Navigate This Article