c

c c c

cc


c  ccc
 c
c
cccccc
c
c
c
x

 
 c !c"!c

 
#$c$%$c
$c!c
c
"
&cc
''(c)*c
c
c
 ›
cc

'!
 ccc
c" +c
,c
c"

c
c
Energetic galactic-cosmic-ray particles penetrate several tens of centimeters into the
surface of planets with little or no atmosphere: Spallation reactions produce secondary
neutrons. Inelastic scattering and neutron capture  rays induced by these particles have
energies that are characteristic of atomic nuclei in the surface. Measurements of Ȗ ray spectra
provide information on abundances of most major elements and naturally radioactive Ȗ
emitters. Planetary Ȗ ray spectrometry was initially performed for the Moon and is being
planned for Mars.

Gamma rays are high-energy photons produced when an excited level of an atomic nucleus
decays into a lower level. The levels in each nucleus occur at specific quantized energies, and
the energy of a Ȗ ray made by a transition between nuclear levels can usually identify which
nucleus produced it. For example, the nucleus 28Si has its first excited level at an energy of
1.779 MeV (a mega-electron-volt, MeV, is 1.602 X 106 ergs), and when this level decays to
the ground state of 28Si, a Ȗ ray with energy 1.779 MeV is produced. Thus, when a Ȗ ray
spectrometer detects a Ȗ ray with this energy, it is very likely that that Ȗ ray was made by the
decay of the first excited level of 28Si. The intensity of the I. 779-MeV Ȗ rays detected from a
planet will be proportional to the abundance of silicon in the part of the planet below the
spectrometer. This picture is simplified, as will be discussed below, but represents the basis
for the use of Ȗ ray spectroscopy to map elemental abundances in a planet's surface.

As an introduction to planetary Ȗ ray spectroscopy, we will discuss the sources of Ȗ

rays, their transport through matter, how they can be detected, and how the observations are
converted into elemental abundances. In brief, Ȗ rays made by natural radioactivity, non
elastic-scattering reactions, and neutron-capture reactions can be used for planetary Ȗ ray
spectroscopy. Elements that can be mapped using Ȗray spectroscopy include all major ones,
many minor ones (including K and H), and some trace ones (especially Th and U). The rays
typically have energies from 0.2 to 10 MeV and are hard to stop; so many tens of centimeters
at the top of a planet¶s surface contribute to the observed flux, not just a very thin surface
layer. Most Ȗrays can pass through a fairly thin atmosphere, such as that on Mars, and can be
used to study that atmosphere. The basic concepts given below have been tested by the  Ȗray
spectrometers that flew on the Apollo 15 and 16 missions and by simulation experiments at
various accelerators.

The instrumentation used in planetary Ȗ ray spectrometers is presented below. The

interactions of rays with scintillator and semiconductor detectors and the resulting signals are
reviewed. Future planetary Ȗ ray spectrometers will probably be made with high- purity
germanium, and cooling and radiation damage are problems that need to be considered in the
design of such instruments. An important aspect of any planetary Ȗ-ray-spectrometer mission
is the extraction of information from the data returned by the instrument. Electronic drifts
need to be corrected so data from different times over a given region can be summed to get
better counting statistics. There are many backgrounds for which corrections are needed prior
to getting the signals of the planetary Ȗ rays. The counts from a spectrometer then need to be
converted to intensities of Ȗ rays from the planet. Finally, elemental and other results from
these Ȗ ray fluxes need to be determined and displayed.

The applications of the results from planetary remote-sensing Ȗ ray spectrometers will
not be discussed in detail here. The main product will be elemental maps of the regions of the
planet covered by the spectrometer. Such elemental abundances help to characterize the
present state of a planet and can be used to infer its origin and evolution. Some of the
scientific results for the Moon from the Luna- 10 and Apollo y ray spectrometers are given in
x  (1984),    (1976),

   (1977b),     (1973, 1974,
1977,     (1978), and in Chapter 15. Expected results from cometary and
martian Ȗ ray spectrometers are discussed by et (1986) and by x!  
(1987), respectively, and in Chapters 17 and 18. A discussion of the use of multielemental
maps as determined by remote sensing planetary spectroscopy for various planetary bodies is
given by (1976). Planetary neutron spectroscopy is closely related to planetary y
ray spectroscopy and is discussed in Chapters 10 and 19, including many of the topics
presented here (such as neutron production and transport). Planetary Ȗ ray spectroscopy
complements well elemental analyses by planetary X-ray spectroscopy, as can be seen in
Chapter 9 and in the joint X-ray/ Ȗ ray papers of "    (1973) and

  
(1977b).

Planetary Ȗ ray spectroscopy has been, and will generally be, applied to solid
planetary surfaces, especially those on the Moon and Mars. Other solid objects in the solar
system that could be studied from orbit by Ȗ ray spectroscopy include the planet Mercury,
comets, asteroids, and most planetary satellites (Cf.     1976). Gamma ray
spectroscopy of the naturally radioactive elements in the surface of Venus was done on the
Soviet Venera 8, 9, and 10 missions #x   1984), but other sources of Ȗ rays are not
present at the surface of Venus because it¶s very thick atmosphere removes essentially all
cosmic ray particles. Although it is possible, there are no plans to use Ȗ ray spectroscopy to
study from orbit the planets with very thick atmospheres: Venus and the four ³gas giants´
(Jupiter, Saturn, Uranus, and Neptune).

c
› c
c cc cc

There are two principal ways that excited levels of nuclei are formed in planetary
surfaces: the decay of radioactive isotopes and excitation by cosmic ray particles. Several
naturally occurring radioactive isotopes are present in planets, and their decay is the main
source of natural Ȗ rays seen at the Earth¶s surface. The latter process needs cosmic ray
particles, but nuclear reactions can occur with any nucleus. Gamma rays made by the decay
of natural radioelements are relatively easy to model, as usually only basic nuclear
parameters, such as isotopic abundances, half-lives, Ȗ ray yields per decay, and Ȗ ray transport
properties, need to be known. The production of y rays by cosmic rays is a complicated
process as one needs to start with the high-energy (GeV) particles in the galactic cosmic rays
(GCR) and follow the cascade of particles made in the planet until one of these particles
produces a y ray. For most elements, the rays used for planetary spectroscopy are made by
inelastic-scattering reactions of neutrons with a few MeV of energy or by capture of neutrons
near thermal (§0.02 eV) energies. Thus, the range of energies of the exciting particles
involved is huge from GeV particles in the GCR to thermal neutrons. Photons with energies
in the range of interest here (§0.2- 10 MeV) can be made by a number of nonnuclear
processes, but such rays are of little interest to planetary Ȗ ray spectroscopy. For example,
when high-energy electrons are accelerated in the electric field of a nucleus, photons with a
continuum of energies are emitted (a process called bremsstrahlung). [Electrons from Jupiter
produced bremsstrahlung that was a background in the Apollo y ray spectra, # 
1976).) Another source of nonnuclear photons is the decay of the neutral pion (ʌ°).cThe ʌ°cis
a meson made by cosmic ray interactions and almost immediately decays into two high-
energy photons. These photons, in turn, initiate an electromagnetic shower (photons,
electrons, and positrons) that results in many photons escaping from the planet¶s surface.
These photons resulting from ʌ° decay are an important part of the continuum #$ 
$  1970). "  (1972) estimates that most of these escaping photons from ʌ°
decay will have energies below 1 MeV. Most of the rest of the continuum of Ȗ rays from a
planet are from scattering of gamma rays or many weak nuclear Ȗrays. Below, %Ȗray´ will
usually refer to a nuclear Ȗray.

Yield/ Gamma Ray Flux

Element Nuclide Energy (MeV)
Decay (photons/cm2-min)

K 40K 1.4608 0.1067 2.39

208
T1
2.6146 0.36 2.19
228
AC 0.9689 0.175 0.66
228
Th AC 0.9111 0.29 1.05
208
T1 0.5831 0.307 0.92
212
Pb
0.2386 0.47 0.96

214
Bic 1.7645 0.159 0.64
214
Bi 1.1203 0.15 0.48
U 214
Bi 0.6093 0.46 1.12
214
Pb 0.3519 0.37 0.71

c

c

cc

One process that generates Ȗ rays is the decay of naturally radioactive elements.
Several elements have isotopes with half-lives long enough that many atoms have not
decayed since they were formed by nucleosynthetic processes over 4.6 Ga (4.6 X 1 years)
ago. For example, natural potassium includes 0.01 17% of the isotope 40K, which has a half-
life of 1.25 Ga. When 40K decays, 10.67% of the time the first excited level
of 40Ar at energy of 1.4608 MeV is produced, and in turn almost immediately decays to the
40Ar ground state. Other naturally radioactive isotopes used for planetary spectroscopy are
several in the decay chains of 14.05-Ga 232Th and 4.47-Ga 238 U. (Both of these long-lived
naturally-occurring radioisotopes emit a series of Į particles or ȕ particles and eventually
decay to isotopes of lead.) Several other elements have naturally occurring  ray-emitting
isotopes, e.g., 106Ga, 138La, but because of extremely long half-lives or very low abundances
their rays are exceedingly hard to detect.
c› !"
"c
!!
c
 cc

Particles in the solar system that are capable of producing nuclear reactions and Ȗ rays
include the galactic cosmic rays (GCR) and energetic particles associated with large flares on
the Sun (cf. ?"  1972; ? 1983). Solar energetic particles will not be
considered below as they are emitted sporadically, often do not produce many Ȗ rays, and
usually penetrate a detector and saturate a Ȗray spectrometer with high count rates. A few Ȗ ±
ray emitting radionuclides made by these solar particles have long enough half-lives that a
planetary Ȗ ray spectrometer will detect their radiations after the solar particle event, which
typically lasts several days. For example, 2.6yr 22Na or 0.7-Ma 26 Al made by solar particles
can be detected by a remote-sensing ray spectrometer. However, the fluxes of solar particles
for several half-lives prior to the mission need to be known. The ray fluxes from these solar-
produced radionuclides are usually fairly low and often these radionuclides are made from
several target elements, making data interpretation hard. Additional details on solar-particle-
produced Ȗ -ray-emitting radionuclides can be found in "  (1972), ?(1973),
and ?#&

'&(' ) * *** + * +,

, ,+ $-?+ *.*,
, *
,+  
 / 0*1 *   +2  *  + *      " ,,
 
, *  2, +,
 ,  )  2  *  *  2,
+,3/0*14 *+ , 2,2*1,
   *5  $-? )* ?#&6,  ?" #1
#
c 
#$c
#

There are several ways that an energetic particle can react with a nucleus. The
Simplest is where the particle elastically scatters from the nucleus, leaving the nucleus in its
ground state, but giving some of its energy to the nucleus in the form of kinetic energy. For
example, a neutron can scatter from a carbon nucleus and lose some of its energy. No Ȗ rays
are created by an elastic-scattering reaction. To create a Ȗ ray, an excited level needs to be
produced. Such a excited-level producing reaction means that the product nucleus is Changed
from the initial one, either by being excited or by creating a different nucleus. Reactions other
than elastic-scattering ones are called nonelastic scattering reactions. Note, in this chapter on
planetary Ȗ ray spectroscopy, we will not consider neutron capture to be a nonelastic-
scattering reaction.

›

# c%c
 c%c##
" 
#$c
# c

The complex cascade of GCR particles in a planet makes calculating the rates for
nonelastic scattering reactions fairly difficult several approaches have been used to calculate
the production rates of such Ȗ rays. All first determine the energy and depth distribution of
GCR particles, then produce and transport the y rays. "   (1972) used Monte Carlo
methods to determine the particles made by the cascade of GCR particles and then used a
transport code to get the neutron flux distribution below 15 MeV. $   $ 
(1970) used a simple model for the distribution of GCR-produced neutrons. ? et 
(1973) and ? (1978) used the lunar GCR particle flux model of ?  " 
(1972), which works we)) for predicting the production rates of cosmogenic(cosmic-ray-
produced) nuclides in the Moon. +(1981) and x!  (1987) used a source
term for GCR-produced neutrons and calculate the fluxes of neutrons below 14 MeV with a
neutrontransport code. Experimental cross sections for the production of 4 rays were used
with these particles fluxes to get their production rates as a function of depth. All considered
(n,nȖ) and similar low-energy reactions. Only ? (1978) considered, in addition, 4 ray
production by higher-energy particles, such as the 28Si (n,X Ȗ)20Ne reaction.

#c
 c
# c

cThe relative production  for the particles that make nonelastic. Scattering
reactions, such as neutrons, are not affected strongly by the bulk composition of a planet.
[However, the absolute flux of the MeV neutrons can be decreased if there is much hydrogen
to rapidly slow them (d +  1981;   x!    1987).] The bulk composition
does strongly affect the moderation of neutrons from MeV energies (the energies at which
almost all neutrons are produced) to thermal (eV) energies. For example, hydrogen rapidly
moderates neutrons (which is why water is often used in reactors). The fluxes of Ȗ rays made
by neutron-capture reactions can vary considerably with changes in the bulk composition of
the planetary surface. For this reason, Ȗ rays made by neutron-capture reactions are often
considered separately from the nonelastic-scattering ones.

  
c
c cc c

Once one knows the rate at which Ȗ rays are produced, then the transport of the Ȗ rays
from their sources to the detector must be considered. The source term, the rate at which a Ȗ
ray is produced, will be expressed as S(x), where x is the depth below the planet's surface.
This source term is a function of depth because usually a source is not uniform with depth.
For the GCR, the particle fluxes vary with depth (cf. Figs. 8.1 and 8.2). For the natural
radioactive isotopes, these elements, like any element, could be present in concentrations that
vary with depth. For the Ȗ-ray-emitting isotopes after radon in the decay chains of thorium
and especially 2J8U, diffusion of radon could distort the depth-vs.-concentration profiles of
these isotopes.

In going from the source to the detector, a Ȗ ray can vanish entirely (e.g., by giving its
energy to an electron, the photoelectric effect), scatter such that it loses some of its energy
(such as by Compton scattering), or arrive with its initial energy unchanged. In almost all
applications of Ȗ ray spectroscopy, only Ȗ rays that arrive at the detector without changing
energy are used to map elements. The continuum of energetic photons made by Compton
scattering, bremsstrahlung, ʌ decay, and other processes contain little information about a
planet's elemental composition, but often need to be considered in unfolding the observations
made by a Ȗ ray spectrometer.

In most cases, the Ȗ ray detector will be above the planet's surface, such as in an
orbiting satellite or on a roving vehicle. The basic expression given below for calculating the
flux ofȖ rays at a detector will assume this geometry of being above the planet's surface.
Note, however, that the detector could be in the medium of interest, such as on a penetrator
(e.g.,  1986; Chapter 17), in which case the basic expression below would have
to be modified. In the calculations given below, we also assume no absorbing medium, such
as an atmosphere or layer of ice, between the planet's surface and the detector. An absorbing
medium, such as the martian atmosphere or an ice cap, attenuates Ȗ rays but, as discussed in
Chapter 17, can be studied using the observed fluxes of Ȗ rays that get through the medium.

c›

#c%c c !c
!!
c
c&c

In a homogeneous semi-infinite medium (a slab geometry) with no absorbing medium

above it, the flux, F, of Ȗ rays escaping without undergoing an interaction is calculated (cf.
? 1973) by integrating over all upward angles and over depth the expression

Where S(x) is the source strength at depth x, șis the angle with respect to the normal,
and 7is the exponential mass attenuation coefficient for that Ȗ ray (Le., over a path length of
1/ 7  the intensity of the Ȗ ray decreases by a factor of 1/ e). The second term outside the
exponential is the area of the surface that a unit area of detector projects for an angle /with
respect to the normal. The integration over depth needs only to extend to depths below which
very few Ȗ rays escape with their initial energies, which is roughly a meter in a regolith (the
broken layer at the top consisting of soil and rock) with a density of ~2 g/cmJ (~200 g/cm2).

Equation (8.1) can only be completely integrated analytically when S(x) is, a constant
over depth, which is usually only the case for natural radioactivity in a homogeneous
medium. In this case

F=S/27 where S is the production rate of the Ȗ ray

 c

c 

c #c

For the case of a source that is constant with depth, Ȗ ray detector above the surface will see
equal fluxes from all parts of the surface, even near the horizon where only Ȗ rays produced
very near the surface reach the detector. Thus, fairly large areas of the planet are sampled by
an orbiting Ȗ ray spectrometer if the detector is isotropic (that is, it detects Ȗ rays from any
direction with the same efficiency). As Ȗ rays are very penetrating, it is hard to collimate a
detector without using a lot of material. As planetary missions are usually not able to carry
the large amount of material needed to collimate a Ȗ ray spectrometer, most planetary Ȗ ray
spectrometers are fairly isotropic in their detection efficiencies.

'&8
, 4  ) +**  **,
   *#   8& , //,,


'c ( › )c ccc( c

The fluxes of Ȗ rays calculated as escaping from the Moon by ?(1978) are shown in Fig.
8.4 as a function of energy. The most intense of these Ȗ rays are also given in Tables 8.1-8.3.
Note that the abundances of the major elements in Tables 8.1-8.3 are those of ? (1978)
and were used to get the fluxes shown in Fig. 8.4. However, several minor elements in Tables
8.2 and 8.3 are higher in abundance than the abundances used by ? (1978), so do not
appear in Fig. 8.4. Note that the higher-energy Ȗ rays tend to be ones made by neutron-
capture reactions (the major nonelastic-scattering Ȗ rays at higher energies being from 160,
which have no levels below 6 MeV, and silicon). Most of the Ȗ rays below -1 MeV are from
the decay of the natural radioelements. Few y rays of interest to planetary Ȗ ray spectroscopy
occur below 0.2 MeV. A more complete listing of Ȗ rays and their calculated fluxes from the
Moon is in ?(1978). These Ȗ ray fluxes would also apply to other planetary objects with
very low contents of hydrogen and carbon and no atmosphere, although a chemical correction
factor may need to be applied to the neutron-capture Ȗ rays.

*ccc+)  ›
c+ ›+
 c

Gamma ray spectroscopy uses nondispersive techniques for the detection of' Ȗ rays.
Nondispersive refers to the ability of the detector to measure the Ȗ ray energy directly as a
means of identification. Dispersive techniques rely on the separation of Ȗ ray photons by
wavelength for identification. The Ȗ ray loses some or all of its energy in the detector, and
this energy loss must be collected for the measurement. For geochemical analysis,
spectroscopy information is needed so that the energy collected in the detector will be
uniquely related to the energy of the incident Ȗ ray. For efficient detection, the detector
should have a large volume and be made of a dense material with elements having fairly high
atomic numbers ("high 2") to increase the probability of interactions and the total energy loss
of the Ȗ ray within the detector volume. For effective identification of a particular element,
the detector should have sufficient energy resolution to distinguish particular Ȗ rays from that
element.

There are several different types of interactions that can occur between a Ȗ ray and the
detector. The three principal interaction mechanisms are photoelectric absorption, Compton
scattering, and pair production. These processes can involve total and/or partial energy
absorption in the detector. The total interaction cross section is equal to the sum of each of
the individual cross sections. The relative strength of any of these mechanisms depends on
the incident photon energy and the detector material. See '*, 

 (1981) for
graphs of the principal interaction properties of common Ȗ ray detector materials.

Photoelectric absorption is the result of the interaction of a Ȗ ray with a bound electron of the
detector material. All the energy of the Ȗ ray is lost in photoelectric absorption. For Ȗ rays in
the 0.1-10 MeV range interacting in common detector materials, the probability of
photoelectric absorption decreases with increasing photon energy.

,ccc)  ›
 c

Two kinds of detectors are or will be used for geochemical remote sensing Ȗ ray spectroscopy
in the 0.1-10 MeV energy region: semiconductors and scintillators. [n semiconductor
detectors the interaction of a Ȗ ray leads to the creation of electron-hole pairs that can be
collected and provide a signal proportional to the energy of the Ȗ ray. [n scintillation
detectors, a light signal is produced by the interaction of a Ȗ ray. The collection of this light
by an optically coupled photomultiplier tube produces an electric signal proportional to the
energy of the Ȗ ray. For either kind of detector, the number of Charge carriers or photons
produced by an incident Ȗ ray needs 'to be maximized in order to increase the signal-to-noise
ratio of the detected signal and the resulting energy resolution of the detector. For a general
review of Ȗ ray detectors see Ô(1989).

,c #
#c) c

A number of different scintillation materials are used for Ȗ ray detection: NaI(TI), CsI(TI),
CsI(Na), and plastics. The first three of these materials are alkali halide crystals that are
activated with trace amounts of the element indicated in parentheses. By one or more of the
interaction processes discussed above, an incident Ȗ ray loses all or part of its energy in the
detector crystal. The energy that is lost is converted to energy of electrons in the crystal, and
this energy is in turn lost to ionization from energy losses of the charged particles as they
slow down.

In scintillators the energy band gap, the difference between the valence and conduction band,
is about 300 eV. Electrons excited to the conduction band canreturn to the valence band
through luminescencecenter levels with the emission of light. It is this light, "scintillations:'
that is collected and used to characterize the photon energy of the incident Ȗ ray.Scintillation
crystals that are used for geochemical spectroscopy analysis must have a linear relationship
between the energy deposited by the Ȗ ray and the light output due to the scintillations.
NaI(TI) crystals are often used because of the transparency of the crystal to its own light
output, its high light output per photon, and its short light-decay time. A drawback is that the
crystal is very hygroscopic and therefore needs special care in handling and storage. A
comparison of the properties of various scintillation crystals can be found in Ô(1989).

,c c 
c) c

In the past, scintillation detectors, both NaI and CsI, had been used on both US and Soviet
spacecraft. Plans for future missions include the use of solid-state detectors, particularly high-
purity germanium (HPGe), because of the much better energy resolution of these detectors. In
solid-state detectors, -y rays lose energy to the detector material in the same ways as in a
scintillation detector. This energy is used to move electrons into higher energy bands. The
result is that an electron-hole pair is produced. The high-energy electrons produced rapidly
lose energy to other electrons. Single -y ray interactions are detected by applying an electric
field to the detector and sweeping the electron-hole pair¶s through the material and measuring
an electrical pulse at the detector output.

As mentioned above, the energy resolution of a detector will depend on the number of charge
carriers or number of photons produced per unit energy absorbed in the detector. About 3 eV
is required to produce an electron-hole pair that is finally detected at the output of a
germanium semiconductor as compared with about 300 eV needed to create one photon at the
photocathode in the case of a NaI(Tl) scintillator. The energy resolution improves as about
the square root of the number of electron-hole pairs or photons produced; thus, the energy
resolution should be improved by about an order of magnitude for Ge semiconductor
detectors compared with NaI(Tl) scintillation detectors. The observed improvement in energy
resolution is even larger than this simple model would predict. Actual resolution
improvement by a factor of 40 is usual and can be accounted for by including the Fano factor
in the calculation. Details about solid-state detectors operations can be found in Ô#&

-c+
+
c (›+
c

The extraction of geochemical information from measured Ȗ ray spectra can be a difficult and
involved process. The uncertainties associated with the excitation source, the limited
accumulation times, the conversion of the measured pulse-height spectra to photon emission
spectra, and the conversion of photon spectra to elemental composition all contribute to this
process. Each of the last three processes will be discussed more fully below, along with some
discussion of methods used to compensate for these uncertainties in the analysis.

-c #c)%c

On a geochemical mapping mission it is necessary that spectra from one orbit be added to
spectra taken over the same region on subsequent orbits. This is to ensure that there are
enough counts in a Ȗ ray peak to allow the analysis of that peak with acceptable uncertainty.
Gamma ray spectroscopy is governed by Poisson statistics where the uncertainty in a count is
equal to the square root of that count. The greater the number of counts in a Ȗ ray peak, the
smaller the uncertainty in that peak #Ô  1989). The Mars Observer mission will be in a
polar orbit around Mars. The accumulation times per unit surface area during the complete
687 -day nominal mission vary considerably on Mars as a function of latitude (see Chapter
17). For most elements, even 6 hr of data accumulated over one spatial resolution unit (280 X
280 km) during the whole mission will not be enough time to allow its elemental
concentrations to be determined with reasonable uncertainty.

-c›#c%c.
/$# c

Many processes produce signals in a Ȗ ray spectrometer that can interfere with measurement
of Ȗ rays from the planetary surface. In order to extract the discrete-line signal that contains
geochemical information, the intensity and energy of this background signals must be
understood and accounted for in the data analysis scheme. A study of the background
components during the Apollo Ȗ ray experiment showed that there were eight identifiable
components to the background #  9  1976; see Fig. 8.10). Some of these
backgrounds produce discrete-linerays that interfere directly with the Ȗ rays from the surface,
while others produce only a continuum. On Apollo, where a low-resolution scintillation
detector was used, the characterization of the continuum was important in order to separate
the continuum from the discrete lines. On missions where high- resolution solid-state
detectors are used, like Mars Observer, the quantitative characterization of the continuum
background should be less important.
-'c c0$ c›# cc # c

Flux Conversion to extract the chemical information contained in the measured pulse height
(energy-loss) spectrum, the flux and energy spectrum of photons emitted from the planet
must first be determined. This must be obtained from the measurements of the energy lost in
the detector by these photons. Absorption by any material between the source region and the
detector must also be accounted for in the conversion. This includes the atmosphere (on a
Mars mission) and material surrounding the detector, such as a charged-particle shield or
passive cooler. In an orbital mission, changes in orbital altitude will also cause changes in the
intensity nd spectrum of measured Ȗ rays.

-1c #c&cc›! #c›# # c

The last and most important step in the conversion process is to relate the derived Ȗ ray
emission spectrum to the elemental composition of the planetary surface material. This
conversion can be quite complex because of the many mechanisms involved in generating Ȗ
rays. Secondary neutrons produce most of the Ȗ rays that contain compositional information.
Protons and other secondary charged particles do not contribute much to the discrete-line Ȗ
ray flux as they are usually stopped by ionization energy losses before they undergo nuclear
reactions. Neutrons can only lose energy through nuclear scattering. They can interact over
the full range of energies from GeV energies down to thermal energies. Neutrons can lose
much more energy in an elastic collision with a light nucleus like hydrogen than in a collision
with abundant nuclei such as oxygen or silicon. Variations in concentration of elements with
large neutron capture cross sections can change significantly the thermal neutron flux
distribution. Thus, changes in composition affect neutron reactions in all energy ranges and
thereby modify the resulting Ȗ ray emission spectrum #?  1978; +  1981; 
x!   &

-*c
 
c #
# c

Orbital data that illustrate general chemical trends can often be used to select or substantiate
different theories of planetary origin and/or evolution. The presentation of elemental results
from the analysis of Ȗ ray mapping data is often most effective when presented graphically.
Among the data products most widely used for interpretation are maps of the geochemical
information superimposed on geological features of the planetary surface. Examples of these
types of maps for the Apollo lunar data can be found in the frontispiece of the 5 *
,' ,  x**- *(1973),  #8 1977, 1979),


(1977b), :(1980), and Chapter 15. There are a number of ways to generate maps that
use the Ȗ ray measurements. The earliest maps used the total concentration of the radioactive
elements (Th, U, and K) around the Moon. This was due to the high count rate in the 0.55-
2.75-MeV region (dominated by the natural radioactivity) and the resulting statistical
precision and high spatial resolution of the results. Later work included concentrations of a
single element (e.g., Th or Fe), composite maps of the concentration of two or more elements
(e.g., Th and Fe), and comparisons of Ȗ ray results with other kinds of measurements over the
same regions (e.g., elemental concentrations and laser altimeter results). See the frontispiece
of the 5 *  , ,,    x** - * (1977) for some examples of
composite maps of the Moon from the Apollo program. These include both correlations
between Ȗ ray results for composition of different elements and between Ȗ ray results and
results from other instruments.

c
 ›
c2cc
Î c
c" +c(
 -!cc,

c
 ccc"!
cc
c
Solar X-rays and low-energy charged cosmic ray particles trapped in radiation belts can
excite X-ray radiation characteristic for elements in the outermost surface layers of
atmosphere-free planetary bodies. Abundances of elements with medium to high atomic
numbers can be determined from X-ray spectroscopy measurements. This technique was
successfully applied in initial lunar surface exploration.

X-ray fluorescence, or XRF, is a well-established technique in the laboratory for performing

both qualitative and quantitative elemental analysis of samples. Usually an X-ray source is
used to eject inner-shell electrons from the atoms in the sample by way of photoelectric
absorption. The subsequent rearrangement of the atomic electrons gives rise to the emission
of monoenergetic X-rays²the fluorescent X-rays² whose energies, in the 0.1- to 100-keV
range, are characteristic of the specific chemical elements in the sample. Spectral analysis of
these fluorescent X-rays can therefore provide us with information concerning the elemental
composition of the sample. A laboratory XRF instrument contains three essential
components: the X-ray source, the sample, and the X-ray detector/spectrometer. The
objective in remote-sensing planetary X-ray spectrometry is to scale up this well-known
laboratory technique to planetary dimensions in order to determine the elemental
compositions of planetary surfaces. Obviously such a task imposes unusual demands and
constraints upon each component of the XRF system. This chapter will examine these
requirements and discuss the type of instrumentation that can be designed to fulfill them.
Up to now the most successful remote sensing planetary XRF experiments are those carried
out during the Apollo 15 and 16 lunar missions. Consequently, the XRF instrument and
results from Apollo 15 and 16 will be discussed in some detail to serve both as lessons from
the past and as guideposts for the future.

2c)++
c
› cc

2c 
c("
 cc

In the scaled-up version of the XRF instrument for remote sensing measurements, the sample
of interest is a planetary surface tens to thousands of square kilometers in extent. In order to
map the global chemical composition of this planetary surface, the detector spectrometer will
be placed on board a spacecraft either in orbit around the planetary body or in a flyby mode.
At orbiting altitudes or fly-by distances, the X-ray source would have to be extremely
powerful, for example an X-ray laser, in order to generate detectable amounts of fluorescent
X-rays from the planetary surface. Present technology is not mature enough for this type of
X-ray generator to be considered in realistic terms. Fortunately, for the inner planetary
bodies, the Sun proves to be an adequate source of X-rays for this kind of remote sensing
XRF experiment. In fact, solar X-rays have been used as the excitation source for the XRF
experiments onboard the Apollo 15 and 16 spacecrafts and have provided geochemical
mapping of about 10% of the lunar surface.
Solar X-rays should be more closely examined as a source of excitation for XRF. First of all,
unlike a laboratory X-ray generator, the solar X-ray output is neither constant nor
controllable. Not only is the solar X-ray intensity variable, but its X-ray spectral distribution
is also variable. Figure 9.1 shows a differential solar X-ray emission spectrum at 1 AU
(astronomical unit) during quiescent Sun periods of Apollo 15. This spectrum was arrived at
semiempirically by fitting the measured data from the solar observatory, SOLRAD, during
Apollo 15 with theoretical calculations based on a model developed by
*   Ô 
(1971). The calculations combined two temperatures, 1.58 X l06 K for the quiet corona and
1.36 X 106c K for the more active regions, in proportions that are consistent with SOLRAD
data. The energies of the K- shell absorption edge of Mg, Al, Si, P, and S are also indicated in
Fig. 9.1. The K edge of an atom is the binding energy of its innermost electron shell, the K
shell. In order to eject a K-shell electron from an atom through photoabsorption, thereby
causing the emission of a fluorescent K X-ray, the incident X-rays must have energies greater
than the K-shell binding energy. Therefore, as a minimum requirement for XRF, only those
solar X-rays in Fig. 9.1 with energies on the more energetic side of the K edge of an element
are capable of exciting that element. We see that the differential energy spectrum of solar X-
ray emission falls very sharply with increasing energy, almost following a power law.

Applying Fig. 9.1 to the lunar case, it can be shown that during quiet-Sun conditions
only the excitation of a few light elements, up to Si, can be expected. Although C and 0 can
be excited by the solar X-rays, their fluorescent X-rays will not be easily measured for two
reasons. First, the fluorescence yields of C and O are extremely low, about 0.001 arid 0.003
respectively. That is, for a thousand absorbed solar X-ray photons, only one C and three O
fluorescent X-ray photons are emitted on the average. Second, the very low energy of the K
X-rays of C and 0, 0.2 82 key and 0.523 key respectively, causes additional difficulties.c As
Fig. 9.1 clearly shows, the incident solar flux rises very steeply in this low-energy region.
Although the energies of these solar X-rays are too low for inner- shell photoelectric
absorption by most of the elements, they can be coherently scattered (i.e., scattered with no
change in energy) very efficiently from the lunar surface and be detected by the XRF
spectrometer. In fact, it was estimated that the coherently scattered solar X-ray flux would be
about six times more intense than the flurescent X-rays from the very light elements such as
C and 0. Under such unfavorable signal-to-background conditions, the detection of these very
light elements becomes highly questionable, even if one could fly an ultrathin window
detector such as a gas-flow proportional counter. Consequently, in quiet-Sun conditions, one
can expect to measure mostly the elements Mg, Al, and Si. In fact, by using detectors with
thin Be windows on Apollo 15 and 16, the Be effectively filtered out most of the low- energy
X-rays and thereby improved the signal-to- background ratio for the Mg, Al, and Si X-rays.
Although it may seem disappointing to be able to measure only three major elements by
remote-sensing XRF, these three happen to be extremely important elements geochemically.

During periods of high solar activities, the differential solar X-ray spectrum becomes
considerably ³hardened´ from that shown in Fig. 9.1. That is, in addition to an overall
increase in intensity during active Sun, there may also be proportionally many more X-rays
with higher energies. Under these conditions, heavier elements such as Ti and Fe may be
exited by the solar X-ray flux.
 Knowing the solar X-ray spectrum and the chemical elements that can be excited by it
places some requirements and limitations on remote sensing XRF experiments. The variable
solar X-ray flux demand that one must monitor its spectrum concurrently with fluorescent
X-ray measurements, in order to interpret the experimental results correctly later on. The fact
that both the incident solar X-rays and the fluorescent X-rays are overwhelmingly low in
energy means that remote sensing XRF experiments can only be carried out on planetary
bodies without atmospheres, such as the Moon, Mercury, comets, and asteroids, where these
low-energy X-rays can enter and escape from the planetary surface without absorption by the
atmosphere. Finally, because the energies of both the exciting solar X-rays and the
fluorescent X-rays are only a few key, the penetration and escape depths of these X-rays in a
planetary surface are on the order of tens of micrometers. Therefore, remote sensing XRF is
by necessity a surface measurement. As shall be seen from the Apollo results, this surface
sensitivity has its advantages.

2cCharged Particles

At the distances of the outer planetary bodies, solar X-ray flux is no longer sufficient
to excite measurable amounts of fluorescent X-rays from the planetary surface. However, in
the case of Jupiter, its strong magnetic field traps a substantial number of charged particles,
forming the jovian radiation belt. Then it may be possible to measure charged-particle-
induced fluorescent X-rays from the surface of the Galilean satellites of Jupiter.
ë
 

 
  
   

   
 

   
  
 


 
 


   
 

 




  



 



 


 

 
    








 



 


' $$c
,   ;  +*  **    !  x  *
, Ô

 +, *

The preponderance of useful fluorescent X-rays will be generated by the inner-shell

ionization due to electron or proton impact. The inner-shell ionization cross section (i.e.,
probability for ionization) from proton impact is comparable to that from electron impact if
the protons have energies about 1000 times that of electrons. For example, 5 key electrons
and 4 MeV protons have a cross section of about 2 X 10-20cm2 in the K-shell ionization of
an aluminum target #  1972;

  1972). Using the Einstein X-ray astronomy
satellite,  . (1983) have detected X-rays in the energy band 0.2-3.0 keV coming
from the polar regions of Jupiter itself. For the Galilean satellites of Jupiter, both charged
particles and X-rays could be a source of direct and scattered background. However, they
could also enhance the fluorescence excitation of Mg, Al, and Si from the satellite surfaces.
Solar-wind charged particles also produce fluorescent X-rays from the lunar surface, but the
absence of a strong lunar magnetic field made their contribution to the total lunar fluorescent
X-rays negligible in comparison to those excited by solar X-rays.

2c  
cc
2c cc(c  !c

The first approach to measure the lunar fluorescent X-rays was made on the Soviet
lunar orbiting spacecraft Luna 12 #9, 1968). Several small geiger counters
having a bandpass between 8 and 14 A (1.55-0.89 key) were flown. These alternately looked
at the lunar surface and into deep space for comparison. Although little compositional
information was obtained, the data showed the solar X-rays did produce measurable amounts
of fluorescent X-rays from the lunar surface.

'1
,- x * #-x  
 ,  
  # <   **    , x** 9    #x9  
* 


By far the most successful remote sensing XRF measurements on the chemical
compositions of the lunar surface were performed during the Apollo 15 and 16 missions
#"     l972a, b, c, 1973a, b, 1974). It is therefore worthwhile to discuss these XRF
instruments in some detail in terms of their design philosophy and their performance. Both
Apollo 15 and 16 carried a variety of remote sensing instruments in the Scientific Instrument
Module (SIM), which was a part of the Command Service Module (CSM). The CSM was the
portion of the spacecraft that orbited the Moon while the lunar module landed on the lunar
surface. Figure 9.2 shows the CSM in orbit as seen by the astronauts of Apollo 15 from the
lunar module. The instruments in the SIM bay are clearly visible. The XRF instrument is
located at the base of the SIM bay as shown in Fig. 9.3.
ë
 

 
  
   

   
 

   
  
 


 
 


   
 

 




   



 







 
   
















'8
,x**9  #x9
,;?' #
 ,
*,
,x9
ë
 

 
  
   

   
 

   
  
 


 
 


   
 

 





    



 



  


 
 
     

 





 
 






'=' ** ,"+;?'+*  

,, 
+ + *    

2,+ + *  , 
 , 22
,,  ,*
 *,>// 
2*  ,2" ,+ + *  #,  2
 ,++,+** 


Because the flux of lunar fluorescent X-rays was far too low to permit the use of high-
resolution wavelength dispersive devices such as crystal spectrometers, large-area (25 cm2)
proportional counters were chosen as the detectors. Detection area also indirectly affects the
achievable spatial resolution on the lunar surface. The active area of 25 cm2 was chosen so
that statistically significant measurement could be achieved during an approximately 16-s
time interval. In this way, a single pass over a lunar feature could provide elemental
composition information, which is correlated with excellent spatial resolution. Figure 9.4
shows the details of the XRF instrument aboard Apollo 15 and 16. There were three gas-filled
(90% Ar, 9.5%c C02, 0.5% He) proportional counters mounted side by side. All three
proportional counters had 0.025-mm- thick Be windows. Unlike Geiger counters such as
those on Luna 12, whose output pulses are uniform and independent of the X-ray energies,
the output pulses from proportional counters have amplitudes that are proportional to the
energies of the absorbed X-rays (hence the name). Thus, in principle, a proportional counter
is an energy-dispersive spectrometer that can be used to differentiate the chemical elements
by the energies of their characteristic fluorescent X-rays. However, the energy resolution of
proportional counters is quite poor. For example, the full width at half maximum (FWHM)
spread of the Mn K X-ray peak at 5.895 key was about 12 key for the Apollo 15 and 16
proportional counters. This resolution of about 20% is barely sufficient to resolve Mn from
its neighbor Fe (K X-ray at 6.40 key). The situation is further exacerbated if the two peaks
have widely different intensities. Because the energy resolution is inversely proportional to
the square root of the energy, the anticipated resolution at 1.487 key (Al K X-ray was about
40% or 0.6 key). Such a resolution would certainly be inadequate to resolve the three major
elements, Mg (1.254 key), Al (1.487 key), and Si (1.740 key), excitable by the solar X-rays.
Therefore, selected filters were chosen to enhance energy discrimination. For the three
proportional counters on Apollo 15 and 16, one was operated without a filter, one had an Mg
filter of 5.1 X I 0 cm thicknesses, and the third had a Al filter of 6.4 x l0 cm thickness. As the
following table shows, the Mg filter strongly absorbs Al and Si K X-rays, whereas the Al
filter selectively filters the SiK radiation. The transmissions of the Be windows, which is
common to all detectors, are also listed.

Percentage Transmission

Filter Mg K, Al K, Si K,
1.254 1.487 1.740
keV keV keV

0.005 I-mm Mg 64.1 2.7 8.6

0.0064-mm Al 32.5 49.1 0.5

0.0254-mm Be 23.7 42.2 58.6

In order to correctly interpret the lunar XRF data, one must also concurrently monitor
the incident solar X-ray flux. For this purpose another small proportional counter with
essentially a pinhole-sized Be window was mounted on the opposite side of the spacecraft
from the lunar-surface detectors. In addition to this onboard solar monitor, data from the
SOLRAD satellite, which was operating during the same time periods as the Apollo 15 and
16 missions, were also used to monitor the solar activities. For Apollo 16, an additional thin
Be filter was placed in front of the solar monitor detector window to permit the measurement
of high solar X-ray fluxes without the gain shift experienced on the Apollo 15 flight for that
monitor.

As shown in Fig. 9.4, there was also an inflight calibration device consisting of a
calibration rod with two radioactive sources that produced Mn and MgK X-rays. The sources
normally faced away from the proportional counters. The rod could be rotated 180° by
internal command from the X-ray processor assembly to expose the sources to the detectors
to monitor their gain, resolution, and efficiency. The calibration cycle was repeated every 16
mm and lasted for 64 s. During both Apollo 15 and 16, no gain shift was observed. All three
detectors shared a common mechanical collimator. The collimator consisted of multicellular
baffles that combined a large sensitive area with high spatial resolution, but were restricted in
the field of view (FOV). The total angular field of view presented by the collimator was 60°.
At an orbital altitude of 110km, the total instantaneous field of view is an area approximately
110 km on edge on the lunar surface.

Pulse shape discrimination was used on the proportional counter signals to reject the
slow-rise non X-ray events due to € rays and cosmic rays. The true X-ray pulses were fed
into eight energy bins covering the range from about 0.75 to 2.75 key. In order to see higher
energy X-rays should the Sun become more active than anticipated, the bare (unfiltered)
detector was programmed to operate in two modes. At the high-gain (normal) mode it
covered the energy range 0.75 to 2.75 key, identical to the other two detectors. In the mode
with attenuated gain it covered the range from about 1.5 to 5.50 key so as to see the high-
energy X-rays. During flight this detector alternately operated for six hours in the normal
mode and two hours in the attenuated mode. The filtered detectors, as well as the solar
monitor, operated continuously in the normal gain range of 0.75 to 2.75 key. Counts were
accumulated for 8-s intervals. However, many analyses of the data were performed using 24-
s integration times, which, for Apollo 15 from an orbital altitude of 110 km, corresponds to a
ground coverage area of approximately 110 X 150 km due to spacecraft motion. An example
of typical X-ray spectra obtained by the three detectors at the subsolar point is shown in Fig.
9.5. In this example the integration time was 56 s.
ë
 

 

  
 ë
 
  


  
 

  




 



   



 


 
 
 

'$$c
+*; +* 
 +
,, + + 
*  
,+*  />1?, 

*,
, ,)+2?>

During the Apollo 15 mission, the XRF instrument made more than 100 hr of
measurement in a circular orbit around the Moon at an altitude of approximately 110 km.
During Apollo 16, the XRF instrument operated for at least 60 hr in circular orbits at 110-km
altitudes. The orbital inclination was 26° for Apollo 15 and 9° for Apollo 16. Consequently
the Apollo 15 ground track covered a larger projected area than that of Apollo 16. The
overlap of orbital coverage between the two flights enabled us to show that the XRF
measurements from both flights were reproducible to better than 10% so that no
normalization of data was required. Because the XR.F spectra were solar excited, the data
covered only the sunlit portion of the lunar surface. Furthermore, the best measurements were
obtained at subsolar points. Data obtained near both sunrise and sunset terminators were less
reliable because of poor counting statistics.

During the return trip from the Moon, also known as the trans -Earth coast, the
spacecraft was put into a rotisserie mode where the X-ray detectors were alternately exposed
to the Sun and away from it. It is noteworthy that although the detectors went into saturation
each time they viewed the Sun directly, their operation always returned to normal when
pointing away from the Sun, without any shift in gain.

The XRF instrument was also used to make galactic X-ray observations during the
trans-Earth coast. In this regard, a so-called y ray burst was observed by the Apollo y ray
spectrometer as well as by µy ray detectors in Earth orbit during the trans-Earth portion of
Apollo 16. The XRF instrument made measurements of both temporal and spectral
distribution of X-rays in the 1-8 key regions. Because the X-ray detectors have directional
sensitivity, it was fortunate that when the burst occurred they were pointed in the direction of
the burst. From this information, it was determined that the origin of the burst was in the
vicinity of the Large Magellanic Clouds #

    1974;      1974).

2cc)

c#c

The outputs from the proportional counters are in the form of voltage pulses whose
amplitudes are proportional to the energies of the absorbed X-rays. As mentioned earlier,
these pulses are accumulated for 8 S and stored in eight pulse-height or energy bins. Ideally, a
variation in intensity in a given energy bin results directly from a variation in chemical
concentration of a given element on the lunar surface with fluorescent X-rays of that energy.
The fact that a change in solar activity may change both the intensity and the spectral
distribution of the incident solar X ray has already been discussed. Since the incident solar
spectrum was measured by the solar-monitor proportional counter, methods have been
developed to correct for such variations (e.g., -  1979). The effects due to other sources
such as solar illumination angle, lunar surface roughness, and matrix effects within the
surface are more elusive and difficult to account for quantitatively. However, these effects are
essentially eliminated if we use Mg/Si and Al/Si intensity ratios #"     1972a,b,c)
rather than absolute intensities. As a result, the detected Mg/Si and Al/ Si intensity ratios can
be directly related to their concentration ratios on the lunar surface. Furthermore, because the
concentration of Si is high and generally does not vary by more than 5% over the lunar
surface, variations in the Mg/Si and Al/Si ratios actually indicate variations in the Mg and Al
concentrations on the Moon.

Based on these discussions, the task in data reduction becomes that of first converting
the raw data into Mg/Si and Al/Si intensity ratios, and then converting these intensity ratios
into concentration ratios on the lunar surface. This was accomplished in the following way:
(1) All three proportional counters have identical characteristics, (2) the detectors are 100%
efficient for the X-rays transmitted through the Be detector window, and (3) all three
detectors cover the same energy range (in the normal mode) as determined by the calibration
sources.

Assumption (2) is justified because the absorption of the K X-rays of Mg, Al, and Si
(maximum energy of 1.74 key) by the Ar gas in the detector is for all practical purposes
100%, once they are transmitted by the Be window.

The X-rays detected by the proportional Counters consisted of not only the
characteristic X-rays of Mg, Al, and Si, but also the scattered solar X-rays. as well as
background radiation from various other sources. This background intensity was determined
by measurements made on the dark side of the Moon. The observed background turned out to
be quite constant, and consequently could simply be subtracted from the total spectra. The
problem of scattered radiation was more complex. We divided the scattered radiation into
three regions: 0.6-1.36 key, 1.36² 1.61 key, and 1.61- 2.6 key. The regions were chosen to
include the absorption edges of Mg, Al, and Si respectively. Because of the nature of the
solar spectrum (see Fig. 9.1) and the fact that Al and Si concentrations are generally high in
the lunar soil, the scattered radiation is much less intense than the fluorescence radiation in
the second and the third regions, i.e., 1.36²2.6 key. In the region of 0.6²1.36 key, the
combined effect of solar spectral distribution and the transmission of the 0.025- mm Be
window produces a transmitted spectral distribution with an average energy of about 1.2 key,
very close to the Mg K X-ray energy of 1.254 key. Therefore, the scattered radiation in this
region was treated as essentially monoenergetic and the total flux was considered a
combination of Mg K X-rays and scattered X-rays of the same energy. In this way the data
from the three detectors can be summed up

««««««« (9.1)

Where Gj is the total counts in detector i after background subtraction: i = 1 is the

bare detector, i = 2 is the detector with the Mg filter, and i = 3 is the detector with the Al
filter. Ij t is the total X-ray intensity at energy j, with j = I, 2, 3 representing the energy of the
Mg, AI, and Si K X-rays respectively.

The approach to convert the intensity ratios 11/13 (Mg/si) and l2/l3 (Al/Si) into Mg/si
and Al/Si concentration ratios is partly theoretical and partly empirical. The theoretical
calculations are based on the assumption of a quiet Sun and a coronal temperature of 4 X 106
K. These conditions give a solar X-ray spectral distribution composed of both a continuum
and characteristic line emissions similar to Fig. 9.1, which are consistent with the solar
monitor observations. With this distribution and various compositions of lunar-type soils, a
nearly linear relationship is established between the Mg/Si and Al/Si intensity ratios and their
respective chemical concentration ratios. This relationship was further confirmed by ground-
truth concentration values of soil samples from the Apollo Il site at Tranquility Base and
from the Luna 16 Site in Mare Fecunditatis, which were along the Apollo 15 flight path.

2'cc*c
#c,c(c  cc

The primary concern of this chapter is the application of remote sensing XRF techniques to
the geochemical mapping of planetary surfaces. With this in mind, the results of the Apollo
15 and 16 missions will be briefly discussed as examples of the end product in the successful
application of the technique. The voluminous geological information obtained from the
careful analyses and interpretation of these results is clearly beyond the scope of this chapter
and can be found in the open literature (for example,
  1975, 1982).
The combined XRF measurements of Apollo 15 and 16 covered about 10% of the lunar
surface. Figure 9.6 shows the variation of Al/Si and Mg/Si weight concentration ratios along
the projected ground tracks for Apollo 15 and 16 #"  1 973a,b). The upper value of
each pair corresponds to Al/Si, the lower value to Mg/Si. Because the orbits of Apollo 15
(upper envelope) had a higher inclination, they covered a larger ground track than Apollo 16
(lower envelope). In Fig. 9.7, the Al/Si ratio is color coded such that blue corresponds to low
ratios and red corresponds to high ratios #"   1977). It is immediately evident that
there is an east-west trend in the chemical composition. The Sun proved to be relatively
stable during both missions except for a few brief periods of increased activity. Since the
XRF data in the overlap areas of the two missions are reproducible to within about 10%, no
normalization of the data between the two missions was necessary.
 Figure 9.8 gives detailed Al/Si values vs. longitude for the Apollo 15 ground track
#"     1972b). Again, one notices the trend that the Al/Si ratio generally increases
from west to east. Closer examination also shows a systematic difference between highland
and mare regions. The Al/Si ratio is generally high in the highlands and low in the mare
regions. The extreme values differ by about a factor of 2. Figure 9.9 gives the complementary
picture for the Mg/Si ratio #"    1972b). Here, one sees the opposite trend; i.e., the
Mg/Si ratio decreases from west to east, and is low in the highlands and high in the mare
regions. Data from Apollo 16 confirmed both of these features: The variation of Al/Si and
Mg/Si ratio with topography and the inverse correlation between Al and Mg. Furthermore,
the correlation between the Al/Si ratio and topography is corroborated by similar correlations
observed between the Al/Si ratio and laser altimeter data.

The chemical difference between crater material and its surroundings, even on a very
fine scale, is illustrated in Fig. 9.10 #"  1978). This shows color-coded Mg/Al ratio
variations near the Mare Cnslum area, with dark brown indicating high values of Mg/Al. The
spatial resolution is about 20 km. We notice that Picard Crater has a clearly much higher
Mg/Al ratio than its surroundings. This difference in elemental composition is an indication
of the Mg-rich subsurface basalt material that was excavated during the cratering event.

It is important to emphasize that there is good agreement between the Al/Si and Mg/Si
ratios from the Apollo XRF experiments and those obtained from returned surface lunar
samples from the same location. This agreement gives proof to the reliability of the orbital
measurements as a guide to the surface chemical compositions of the Moon. In fact, because
the depth sensitivity of the XRF experiment is only tens of micrometers, any thin blanket of
excavated material from cratering or volcanic events can be unambiguously identified by the
XRF technique, and thereby can provide a vertical chemical profile of the disturbed portion
of the lunar landscape.

Another important correlation discovered from the Apollo XRF experiments is that
between the Al/Si ratio and optical albedo, shown in Fig. 9.11 #"     1973b). This
strong correlation suggests that high optical albedo is associated with high Al2 O3 content, and
that difference in optical albedo indicates a difference in chemical composition in addition to
topography.
 

'&"0x**     ,   
 *"+9x
,    *  *, *
)@*, *   2 , ,, 
*2,,,2  ,  

'0x**     ,    *
"+?
, ,  "0x0x  *  2
  ,,  2,,
 ë

 

 
 
 ë

  


  


  




 



  



  


  
 

'/- *0" +, -   ,2  

 **  1/  *#"  & *2
0"6+
2*,,A"5* -  , *,,, 0",
  
, ++ * ,+B00222* + 0
0
=/0/?1=/1&>, 
ë

 

 
 
 ë

  


  


  




 



  



  


  
 

'$$c"0x +*
  "+>
   *
, * 
2"0x +*
! 
 

'1-+ 
2, * ;  ) 

,  ,;  ) ,   * 
,;?'
*  "+?", ,,  ) *, 

>8,2* *

Although the solar activity remained fairly constant during the Apollo missions, there
were a few exceptions Figure 9.12 shows the integrated intensities registered by the solar
monitor and the XRF detectors facing the lunar surface on Apollo 15 during approximately
the same period of surface measurements (Adler eta)., I 973b). There is good agreement
between the solar monitor and the surface detectors and, for the most part, the solar flux
varied less than 30% of the mean value. However, orbits 67 and 73 showed a definite
increase in activity. The influence of high solar activity on the XRF data is demonstrated in
Fig. 9.13. During orbit 67, with the higher solar activity, there was a corresponding decrease
in the Al/Si and Mg/Si intensity ratios as compared with those of orbit 68 (quiet Sun) at the
same longitude, but slightly displaced latitude This decrease in the Al/Si and Mg/ Si intensity
ratios indicates that during the increased solar activity there was also a concurrent
preferential increase in the high-energy portion of the solar spec tru that excited Si more
efficiently than either Al or Mg.
'8 **  * +"0x90x C
>,,, 
 *,
>&#! x ,"0x0x +  ,, 
 *#
>
* ,, , ; +*  2,*,)*
x *,"

2'c 
 c  +c(c+c0 c cc
We have seen that remote sensing XRF provided large-scale chemical mapping of the
lunar surface and established the correlation of chemical composition and lunar topography.
These data suggest to the lunar scientist that global differentiation took place with the
formation of the lunar crust, which is rich in Al as seen in the highlands. Subsequent impact
and volcanic events excavated and exposed deeper crustal and mantle materials that are richer
in Mg as seen in the mare regions. They also provide quantitative numbers about the solar
and fluorescence ;  flux at 1 AU so that future XRF missions to other planetary bodies
can be contemplated with some confidence and realism. In contrast to the low-inclination
orbits of Apollo 15 and 16, future missions should employ polar orbits to allow global surface
coverage by the XRF detectors. For large bodies such as the Moon and Mercury, the
planetary surface completely fills the field of view of the spectrometer on an orbiting
spacecraft so there is no background X-ray contribution from galactic sources. The absence
of a magnetic field ensures negligible charged-particle background. Therefore, for orbiting or
rendezvous missions to these bodies one can take advantage of long measurement times with
excellent signal-to-background ratios. In the case of Mercury, the incident solar X-ray flux is
a factor of 4 to 6 higher than that on the Moon because of its closer proximity to the Sun.
Thus, one would expect a correspondingly higher fluorescence X-ray flux from the Mercury
surface. As the high temperature in the vicinity of Mercury precludes the use of cryogenically
cooled detectors, proportional counters will once again be the likely detectors of choice.
Many of the spacecraft systems are spin stabilized. For such spacecrafts the XRF system
must operate only when it is pointing toward the planetary surface and must
be shielded when it is pointing toward the Sun. For Mercury, additional thermal shields may
be needed to protect the detector windows and yet not attenuate significantly the Mg, Al, and
Si X-rays.

For small bodies such as asteroids, or the Galilean satellites of jupiter where the
intense charged-particle background trapped by the strong jovian magnetic field prevents the
XRF measurement to be carried out near the satellite surface, the object of interest will only
fill a portion of the field of view of the XRF spectrometer. Galactic X-ray sources will
interfere with the desired XRF measurements. One faces a similar situation in a fly-by
mission where the size of the body of interest, as viewed by the spectrometer, changes with
time as the spacecraft is approaching or receding from the body. In these cases, it may be
necessary to design special collimators to enhance the signal- to-background ratio.
Alternatively, it may be possible to use X-ray imaging systems. The sky background
could then be eliminated by considering only the X-rays emitted from the imaged planetary
body. For an asteroid fly-by mission, on account of the reduced solar X-ray flux, calculations
indicate that in order to obtain statistically significant data the closest approach to the asteroid
needs to be about one-half of the asteroid diameter, with a fly-by speed of less than 1 km/s.

Although measurements are possible on a comet rendezvous mission, the anticipated

results are believed to be marginal. There are several difficulties. The soil of a comet is most
likely encased in ice and thus it will be difficult to observe its X-ray fluorescence. Even if the
soil is not encased in ice, its concentration is low, and consequently the X-ray fluorescence
intensity will also be low. In addition, the surface material on a comet may not be
characteristic of the subsurface composition because they have been modified by the
evaporation and recondensation of materials caused by the periodic perihelion passes.
Furthermore, XRF measurements can only be made when the comet is greater than 2 AU
from the Sun. This is because at closer distances the spacecraft will have to move away from
the comet nucleus in order to avoid the materials that are evaporated from the nucleus
forming the comet tail.

Proportional counters were used in the Apollo missions because of their large detection area
and their stability. Their energy resolution is very poor. For Mg, Al, and Si, energy
discrimination was enhanced somewhat by the use of filters. In future missions, additional
enhancement of energy discrimination and improvement of signal-to-background ratio are
possible through better selection of filter thicknesses as well as improved proportional
counter designs. However, in order to detect higher Z elements with very low intensities,
which may be excited by solar flares or charged particles, it is better to have high-resolution
detectors to improve the signal-to-background ratio. Thermoelectrically cooled Si(Li)
detectors, ambient- temperature Hg12, and gas scintillation proportional counters (GSPC) are
possible candidates. The two solid-state detectors, Si(Li) and HgI2, presently have very small
sensitive areas, on the order of a few to tens of square millimeters. Since tens of square-
centimeter sensitive areas will be required, it will be necessary to form mosaics of these
solid-state detectors. An additional advantage of mosaic detectors is that the total ensemble
can also be used for imaging. The energy resolution of gas scintillation proportional counters
is not as good as the solid-state detectors, but it can have large detection areas as well as
position sensitivity. Of course for any long-duration space flight mission the long-term
stability of detectors under space environment is also a matter of concern.

21c›
› +
c
The Apollo 15 and 16 missions not only proved conclusively the feasibility of remote sensing
XRF analysis of planetary surfaces but, more importantly, they provided us with a rich
harvest of scientific knowledge. This kind of success in the past should give us the
confidence and the impetus for remote sensing XRF missions to other planetary bodies in the
future.

c
c

c
 ›
c3c

  +c
  ccc!
.c

 cc

c
c
3c+##cc

Earlier chapters have demonstrated that spectral features of most rock-forming minerals in
the visible to near-infrared region originate from the presence of transition elements in their
crystal structures. Iron and titanium have higher crustal abundances on terrestrial planets
relative to other transition elements and, consequently, are expected to contribute
significantly to the reflectance spectra of planetary surfaces. Spectral profiles of sunlight
reflected from planetary surfaces, when correlated with measured optical spectra of rock-
forming minerals, may be used to detect the presence of individual transition metal ions, to
identify constituent minerals, and to determine modal mineralogies of regoliths on terrestrial
planets. The origin and applications of such remote-sensed reflectance spectra measured
through Earth-based telescopes are described in this chapter.

3c› !
c! #c%c c 
c
# cc

Properties of the terrestrial planets that are central to this chapter are summarized in table
10.1 and information about element abundances is contained in Appendix 1. The crustal
abundance data for the Earth indicate the presence of relatively high concentrations of Fe,
and to a lesser extent Ti, compared to other first-series transition elements. However, the
terrestrial abundance of Fe is smaller than its cosmic abundance due to chemical fractionation
during major Earth-forming processes involving Core formation, Mantle evolution and
chemical weathering of igneous minerals exposed to the atmosphere. Other first-series
transition elements, notably Cu in porphyry calc-alkaline rocks ( 8.6.2), Ni in ultramafic
rocks ( 8.6.3), and Mn in sub-aqueous fissures ( 8.7.3), may be concentrated in local areas
where deeply eroded igneous rocks, ore deposits, or hydrothermal veins outcrop at the Earth's
surface. On Earth, the most common and stable oxidation states of transition metal-bearing
minerals occurring in near-surface environments are Fe(II), Fe(III), Mn(IV), Mn(II), Ti(IV),
Ni(II), Cu(II), Cr(III) and Cr(VI). Electronic transitions involving these cations dominate the
spectra of terrestrial minerals described in earlier chapters. Earth's atmosphere contains N2
and O2, which are spectrally inactive in the visible to near-infrared region. However, the
small amounts of CO2, water vapour and ozone have important telluric effects on light
reaching spectrometers attached to Earth-based telescopes observing planetary bodies.cc

Different evolutionary histories of other terrestrial planets have influenced the relative
concentrations of the transition elements compared to their cosmic abundances, as suggested
by geochemical data for surface rocks on the Moon, Mars and Venus (Appendix 1). Chemical
analyses of lunar samples returned from the Apollo and Luna missions show that minerals
and glasses occurring on the Moon contain high concentrations of Fe and Ti existing as
oxidation states Fe(II), Ti(III) and Ti(IV). Some lunar minerals, notably olivine and opaque
oxides, also contain significant amounts of Cr(II), Cr(III) and Mn(II). The lack of an
atmosphere on the Moon simplifies interpretation of remote-sensed reflectance spectra of its
surface.

On Mars, the   X-ray fluorescence analyses of its surface during the Viking Lander
experiments indicate an iron-rich regolith in which Fe(III) appears to predominate. Although
the atmospheric pressure of Mars is much lower than that on Earth, its constituent CO2
influences reflectance spectral profiles of the Martian surface. Chemical analyses of the
surface of Venus performed during the Venera and Vega missions indicate that the
composition of the Venusian surface resembles that of Earth by having similar crustal abun-
dances of Fe and Ti. However, the high atmospheric pressure on Venus with its constituent
CO2, SO2 and other spectrally-active gases prevents light reflected from the Venusian
surface reaching Earth. Asteroids such as Vesta and Ceres which may be sources of basaltic
achondrites such as eucrites and diogenites, also appear to be characterized by Fe(II), Ti(IV)
and perhaps Ti(III)-bearing phases. Although no geochemical data exist for Mercury, this
planet may contain relatively low concentrations of transition metal ions, perhaps existing in
low oxidation states.

Plate tectonic activity, which is responsible on Earth for subduction zones, spreading centres
and obducted ophiolites, as well as associated ore deposits of Cu, Cr and Ni described in
8.6, appears to have been less significant on other terrestrial planets. As a result, local
enrichments of these and other transition elements (apart from Fe and Ti) are probably absent
on the Moon, Mercury, Venus, Mars and the asteroids. Since Fe and Ti minerals are
predominant on terrestrial planets, electronic spectra of Fe2+ and Fe3+ in silicates and oxides
influenced by Ti4+ and Ti3+ are expected to dominate remote-sensed spectra of their surfaces.

The primary silicate minerals containing Fe and Ti cations on terrestrial planets are olivine,
pyroxenes and, perhaps, amphiboles and micas, together with glass and opaque oxide phases.
During weathering processes on planetary surfaces, a variety of secondary minerals
dominated by Fe3+ ions, including oxides, oxyhydroxides, clay silicates and sulphates, are
formed by interactions with volatiles in the atmosphere or by hydrothermal activity
associated with impact events. How well the electronic spectra of primary ferromagnesian
silicate phases can be distinguished from one another and from those of secondary ferric-
bearing assemblages is discussed in this chapter.

Each terrestrial planet has spawned its own unique set of problems that have had to be solved
in order to better understand the mineral assemblages contributing to remote-sensed
reflectance spectral profiles of light scattered from its surface. Selected examples are
described throughout the chapter. First, origins of reflectance spectra of minerals and their
generation by sunlight penetrating the surface of a planet are discussed. Reflectance spectra
from the Moon are then described together with laboratory investigations of mineral
mixtures. Effects of temperature on reflectance spectra are considered next, because they
refine interpretations of lunar spectra and set constraints on mineral constituents of the
surface of Mercury. The need to understand spectral features arising from oxidative
weathering products of mafic igneous rocks leads to a description of reflectance spectra of
ferric oxide minerals because their spectral characteristics underlie interpretations of remote-
sensed spectra of Mars, as well as Venus, which are also complicated by atmospheric
absorption at critical wavelengths. Finally, the information that spectral measurements of
meteorites brings to bear on the mineralogy of asteroids is discussed. Emerging from such
examples is the central theme that information about the surface mineralogy of terrestrial
planets, complemented by measurements of samples from the Earth, Moon, Mars and
meteorites, has resulted from reflectance spectroscopy used as a remote-sensing tool with
Earth-based telescopes. The focus is on optical spectra of pyroxenes, olivine and iron oxide
phases, because these minerals appear to dominate the surface mineralogy and visible to near-
infrared spectrum profiles, and are particularly diagnostic of the evolution of terrestrial
planets.

3'c
$#c%c%
#c 
c

Regoliths or soils on surfaces of planets consist of an intimate mixture of different

minerals, including primary igneous rocks and alteration products formed by chemical and
mechanical weathering and impact events. Such assemblages are present in a range of particle
sizes which scatter and absorb sunlight incident on them. The total reflectance,?
from such
a multicomponent polycrystalline assemblage is the sum of the surface or specular
reflectance,? and volume or diffuse reflectance,?  The surface reflectance is the portion
of the reflected radiation that has not penetrated inside any particle, whereas the volume
reflectance is that portion which has been transmitted through some part of one or more
particles. There are limiting conditions under which either? or? is predominant. When the
absorption coefficient and mean size of the scattering particles are large, then ?
§ ?
However, when absorption coefficients and particle sizes are small, which applies to most
regolith materials over wide wavelength ranges, then?
§? and the Kebulka-Munk theory
of diffuse reflectance becomes valid (Wendlandt and Hecht, 1966; Kortum, 1969; Morris
 1982).

The Kebulka-Munk remission function f(?
), is defined as

Where?
is the reflectance from an infinitely thick sample which, for most powdered silicate
and oxide minerals, occurs within depths of just a few millimetres. The parametersc 4 and x
are the absorption coefficient (table 3.2) and scattering coefficient, respectively. Equation
10.1 assumes that the incident radiation is monochromatic and that the scattering processes
show no wavelength dependence. Wavelength dependence may occur when dimensions of
scattering particles are comparable to or less than the wavelength,; of the incident radiation.
In visible to near-infrared spectra, this corresponds to dimensions of 0.4 to 2.5 microns (table
3.1). Equation (10.2) is then expressed as
Where C is a constant and exponent lies within the limits 1 < < 4, the value = 4 being
characteristic of Rayleigh scattering.
ë
 

 
  
   

   
 

   
  
 


 
 


   
 

 

 


   

 

 


 
 

 
 
 
      

 






 







ë
 

 
 
    

  
 

   
  
 



 

 
   
 

 

 


   

 

 



  

 

 
     

 





 


 


 

' /-+ 
 =  +* ***+ ) 
 *,5 
 ++* *  * 2,
, **+* +2 +, 
 ë
 

 
 
    

  
 

   
  
 



 

 
   
 

 

 


   

 

 



  

 

 
     

 





 


 


 

' /1"
 ++* '#999),  
 #

    # ,# x, DE &?#$,6#

+* *6#*,6#,  **+* 2 * 

 ++* 
++*,Ô
  *
, *
 
*
+#
/1


Reflectance spectra are usually measured using a diffuse reflectance acces sory with an
integrating sphere attached to a spectrophotometer. Spectra are referenced against a
reflectance standard, such as smoked MgO, barite or Halon powder. The latter is a
commercial fluorocarbon that does not absorb water or suffer radiation damage as does MgO.
Each of these standards is vir tually free of spectral features in the wavelength range 03 to 1?
m.

In reflectance spectral profiles, regions of minimum reflectance correspond to peak

maxima in absorption spectra measured in transmitted light. Comparisons for a pyroxene are
shown in fig. 10,1, in which polarized spectra of single crystals are correlated with the diffuse
reflectance spectrum of the powdered mineral. Much resolution is lost in reflectance spectral
profiles of mineral powders. Thus, bands are broadened and contrasts between band
intensities are diminished in reflectance spectra compared to singlecrystal polarized spectra.
Nevertheless, significant differences of spectrum profiles and peak positions do exist between
different minerals, and this forms the basis for identifying mineral asemblages by
remotesensing measurements of planetary surfaces. By using the Kebulka²Munk functions,
eqs (10.1) or 10.2), diffuse reflectance spectra m ay be transformed into profiles for direct
comparisons with transmission spectra. Exam ples are shown for Fe3 oxide minerals in fig.
10,2. Here, reflectance spectra measured for goethite, lepidocrocite, maghemite and hem atite
have been converted to absorption spectra by application of the KebulkaMunk function
(Sherman and Waite, &?
 ë
 

 
  
   

   
 

   
  
 


 
 


   
 

 




   



 



 




    

















' /8? *  , !     

,  *,# *- D" 

31c
!#c%c c%
#c 
c

In remote-sensed reflectance spectra of planetary surfaces measured through Earth-based

telescopes, the Sun is the illuminating source. Light reaching Earth from a reflecting
planetary object such as the Moon or Mars has travelled a complex path from the Sun. If a
planet has an atmosphere, the sunlight is first scattered and absorbed by constituents in the
atmosphere before undergoing surface reflection and diffuse scattering by heated particles in
the regolith. The Sun as a quasi-black-body radiator at 6,000 °C is an efficient radiator of
energy in the visible to near-infrared region. However, the planetary body itself becomes a
radiator somewhere in the spectral region between about 1.6 m and 5 m, depending on the
distance of a terrestrial planet from the Sun, and emits back into space a greater proportion of
thermal energy from its surface than the energy of scattered light received from the Sun. For
example, the spectral flux received on Earth from a square kilometre of sunlit mare basalt on
the Moon's surface is plotted in fig. 10.3. Here, the solar radiation flux consists of two parts: a
specular reflectance component that has been scattered by particles on the Moon' surface; and
a diffuse reflectance component that has been absorbed by the regolith, converted to heat and
emitted as thermal radiation. Other shining planetary objects have similar energy flux profiles
to that of the Moon. However, the relative contribution of the emitted thermal radiation flux
increases from hotter planetary surfaces such as Mercury which has a higher daytime surface
temperature (-650 K) than that on the Moon (-400 K). When the emitted thermal radiation
becomes the major fraction of the reflected energy at a given wavelength, interpretations of
measured reflectance spectra become very difficult. Figure 10.3 also shows that in the
ultraviolet region (wavelengths shorter than about 0.4 m), the reflected solar flux decreases
rapidly. Therefore, the region of most interest in remote-sensed spectroscopic studies of
planetary objects is confined to the wavelength range

0.35 to 5 m where reflected incident solar light is usually the primary source flux from
surfaces and atmospheres of sufficient energy to be measured accurately.

Upon entering the Earth's atmosphere, some absorption by atmospheric gases occurs. Below
0.3 m, absorption by ozone results in the atmosphere being effectively opaque to ultraviolet
radiation. Absorption by CO2 occurs at 1.33, 1.44, 1.60, 1.95, 2.02 and 2.08 m, while traces
of H2O in the atmosphere absorb very weakly at 0.82 and 0.95 m and strongly at 1.4, 1.9
and 2.85 to 3.1 m. These telluric effects may be eliminated, however, by ratioing the energy
flux from the reflecting planet to that from a nearby standard star measured in the heavens at
the same time. The spectral reflectance is usually scaled to unity at a selected wavelength
(e.g., 0.56 or 1.02 m) to standardize albedo effects (i.e. overall reflectance) so that attention
can be focussed on absorption spectral features rather than background.

' /=:  **+* +2    *'1#
5*+ 6#
 ,+ )6#*****+ )6##* 
- 5  

3*c%
#c 
c%c% c 
 c#c#c/ c

Ferrous iron in silicate minerals typically gives rise to crystal field bands in the 1 m
region, as demonstrated by the spectra illustrated throughout chapter ? The positions and
intensities of these Fe2 CF bands serve as calibration standards for identifying individual
minerals in diffuse reflectance spectra of powdered rocks and surfaces of planets (e.g., Burns
   1972a,b; Adams, 1974, 1975; Burns and Vaughan, ?6 Bell    ?6 Vaughan
and Burns, 1977; Burns, l989a). Examples are illustrated in fig. 10.4. Lunar calcic plagioclase
feldspars containing small amounts of Fe2+ replacing Ca2+ ions in large coordination sites,
produce a weak crystal field band centered around 1.25 Fe2 + #?  The 1.05 Fe2 + band in
olivine spectra originates from Fe2+ ions in M2 sites, while Fe2 ions in Ml sites produce the
features at about 0.85 and 1.25 m (5,4,3,2). Pyroxenes, alone, produce crystal field bands in
the 2 m region originating from Fe2+/M2-site cations (5,5), the positions of which are more
diagnostic of the pyroxene structure-type than the band near 1 m. Note, however, that
tetrahedrally coordinated Fe2 ions in oxide structures and silicate glasses also absorb in the
1,8 to 1,9 m region (Bell    1975; Nolet    1979), which can complicate
interpretations of pyroxene crystal chemistry (White and Keester, 1966; Bancroft and Burns,
1967a) and remote-sensed reflectance spectra of basaltic glass-bearing assemblages (Farr 
 1980; Dyar and Burns, 1981).

' />?+*  +  ,F *

# $ &?*+*+*  *** /17
 G*+* 6 ,B  
 +  ,
   ,* ) /87>7,

#, c+,"+> #
,,
 

,"+#*2
 
 x #+ **
+
  2

The spectra illustrated in figs 10.1 and 10Ab,c indicate that there are structural and
compositional dependencies of the pyroxene absorption bands, which are demonstrated in
plots of the 'one micron' band versus the 'two micron' band such as that illustrated in fig. 10.5
(Adams, 1974; Singer and Roush, 1985). The occurrence of Fe2+ ions in the very distorted,
asymmetric pyroxene M2 sites is responsible for these two bands. As noted in chapter 5, the
positions of the intense pyroxene Fe2+/M2-site CF bands in the 0.9 to 1.03 m and 1.8 to
 2.3 m regions serve to distinguish orthorhombic and monoclinic pyroxenes, to
indicate pyroxene structure-type, and to estimate Fe2+ compositions ( 5.5). These spectral
characteristics are embodied in a variety of pyroxene determinative curves (e.g., Burns et ai,
1972a; Adams, 1974; Hazen et al, 1978; Singer and Roush, 1985; Cloutis and Gaffey, 1991).
The one illustrated in fig.

10.5 is particularly applicable to powdered samples and has been widely used to
interpret remote-sensed reflectance spectra of planetary surfaces. Note, however, that fig.
10.5 is applicable only to iron-bearing pyroxenes in which Fe2+ ions are present in M2 sites.
Pure enstatite (MgSiO3), stoichiometric calcium pyroxenes along the diopside-hedenbergite
series [Ca(Mg,Fe2+)Si206], and Ti-bearing pyroxenes would remain unidentified on the basis
of' 1 micron' band versus '2 micron' band correlations. Note also that the intensities of the ' 1
m band are affected by overlapping bands from Ml-site Fe2+ ions, particularly in
orthopyroxenes, contributions from which might be indicated, though, by the other pyroxene
Ml-site Fe2+ CF transitions around 1.15 m (cf. figs 5.14 and 5.15). This relatively weak
pyroxene band, in turn, overlaps olivine and plagioclase Fe2+ CF spectral features (fig. 10.4).
Also, as noted earlier, tetrahedrally coordinated Fe2+ ions in basaltic glasses and spinels
which also absorb near 1.8 to 2.0 m may interfere with the pyroxene 2 m bands. Despite
these problems and effects of temperature described later, the '1 m versus 2 m pyroxene
determinative curve shown in fig. 10.5 has been widely used to map basaltic rocks on
terrestrial planets by remote-sensing reflectance spectroscopy.

3,c c#5c6! c%c!#
c!& c

The close proximity of the Moon to Earth and its lack of an atmosphere have resulted
in the compilation of a large number of telescopic reflectance spectra from diverse areas of
the lunar surface (McCord   1981), including mare basalts (Pieters, 1978), pyroclastic
mantling deposits (Gaddis   1985) and lunar highlands (Pieters, 1986). Representative
spectra are illustrated in fig. 10.6. A prominent feature of fig. 10.6 (left), and, indeed, all
reflectance spectra of the Moon, is the positive continuum slope (i.e. an overall increase of
reflectivity towards longer wavelengths). Superimposed on this positive continuums are
absorption features in the 1 and 2 m regions, suggesting that pyroxenes are the dominant
minerals at each site. These pyroxene features are accentuated after a straight line continuum
extending from 0.73 to 1.6 m has been removed, as in fig. 10.6 (right). Correlations with the
'1 m ' versus '2 m ' determinative curve in fig. 10.5 indicate the presence of orthopyroxene
in highland soil sampled at the Apollo 16 mission landing site (fig. 10.6a), pigeonite and sub
calcicaugite in high-Ti mare basalt at the Apollo 17 site (fig. 10.6&), and calcic augite in
low-Ti basalt at Mare Serenitatis (fig. 10.6c) (Gaddis    1985). However, since mono-
minerallic assemblages are extremely unlikely on the Moon, the pyroxene-dominated spectra
shown in fig. 10.6 must have contributions from Fe2+ in other phases in the regolith,
particularly in the low-albedo reflectance spectrum from pyroclastic mantled mare at the
Littrow region (fig. /> Mineralogical and optical spectral studies of samples returned
from the Moon indicate that pyroxenes coexisting with olivine, plagioclase feldspar, volcanic
and impact glasses, and opaque ilmenite and spinels are the most likely phases to modify the
pyroxene 1 and 2 m bands. The problem of how one mineral interferes with the spectral
features of another has been addressed by laboratory reflectance spectral measurements of
various mineral mixtures (e.g., Nash and Conel, 1974; Singer, 1981; Cloutis   1986;
Mustard and Pieters, 1987; Sunshines 1990).

' /?**+* ) 
# x 

&   ,+ ) #  &> '= *,+ ) #E = ? ' )   ,
 ,+ )#'&? '?)  + +  )+ 2+ *
@,2+ ), ) +* C* 
 ,
+ )

,  +  1?7

Examples of experimental measurements of mixed-mineral assemblages are

illustrated by the spectral data for suites of orthopyroxene-clinopyroxene and orthopyroxene-
olivine mixtures in fig. 10.7. The gradual shifts of the 1 and 2 m bands are clearly seen for
orthopyroxene-clinopyroxene mixtures (fig. 10.7; left). While deleterious broadening of the 1
m band occurs, the 2 m band appears to be resolvable into component bands in pyroxene
mixtures. The orthopyroxene-olivine mixtures (fig. 10.7; right) demonstrate that the diagnos-
tic pyroxene 2 m band is conspicuous even when pyroxene is a minor constituent. Thus, in
the 75 per cent olivine/25 per cent orthopyroxene mixture, the pyroxene 2 m band is already
prominent, while the pyroxene 1 m band has almost obscured the characteristic features for
olivine at 0.9,1.05 and 1.25 m. Figure 10.8 showing reflectance spectra of mixtures of
orthopyroxene with plagioclase feldspar and magnetite vividly demonstrates how the
presence of an opaque oxide phase drastically lowers the albedo (i.e. overall reflectance)and
intensities of the diagnostic pyroxene absorption features. Metallic iron in chondritic
meteorites is also detrimental to pyroxene spectra (Gaffey, 1976).

' /&:  ** ,+ )#5 ?C;2,+*+ 

#5"$   #"$ #  "  =   
  -  5 
  + +    )+   2 +  * @ ,2 , +!  ) +, 
2+,*+ )
 17

Such results for mixed-mineral assemblages enable more elaborate interpretations to

be made of the remote reflectance spectra of the Moon shown in fig. 10.6 (Gaddis  
1985). Olivine and mixed-pyroxene assemblages are responsible for the overall profiles
obtained from mare basalts and lunar highlands (figs 10.6a-c). However, band-broadening
and low-albedo spectra of dark mantling pyroclastic deposits (fig.106d) indicate the presence
of volcanic glass which has been partially devitrified to opaque ilmenite-bearing
assemblages, such as that those found in Fe-Ti orange glass spherules from Shorty Crater at
the Apollo 17 site (Vaughan and Burns, 1973).
 The ultimate objective of remote reflectance spectral measurements is to obtain
quantitative estimates of the modal mineralogy of unexplored surfaces of the Moon and,
indeed, other terrestrial planets. This necessitates difficult and elaborate spectrum-curve
fitting procedures, which has been the focus of detailed research (e.g., Roush and Singer,
1986; Cloutis   1986; Mustard and Pieters, 1987; Sunshine   1990). Nevertheless,
valuable petrological information may be deduced from the reflectance spectral profiles
alone. For example, telescopic reflectance spectra for §5 km diameter areas within the 95 km
diameter lunar crater Copernicus shown in fig. 10.9 indicate that wall (fig. 10.9a) and floor
(fig./
 areas with the weak band centred near 0.92 m diagnostic of orthopyroxenes are
typical highland soils of noritic composition and predominate in Copernicus (Pieters  
1985). Other floor areas (fig. 10.9c) contain high proportions of glass-bearing impact melt.
However, central peaks of Copernicus (fig. 10.9d) are quite different from the walls and
floors. They exhibit a broad multiple band centred near 1.05 m indicating that olivine is the
principal ferromagnesian silicate mineral (Pieters, 1982).

Troctolite is thus believed to be the major rock-type forming the central peaks of
Copernicus. Reflectance spectra of rays emanating from Copernicus contain more calcic-rich
pyroxenes indicative of pigeonite-augite assemblages (Pieters 1985).

' /?**+* 
   ,
*+ #H?

     2, , -+ *  *    ,  #  5     &?
B ** *    /1 76 ,B   
 +   * 
x+*   * #E#
  * ,+ )
  *
 **+6#* *+ ) 
)+*6#* +* *
 *

3-cc
#c c#5c6! c%c $ c! 
 c

Since reflectance spectra measured through telescopes are produced by sunlight

impinging upon the surface of a planet, the effects of temperature on spectral reflectivities of
minerals need to be carefully assessed, particularly when large differences exist between
planetary surfaces and ambient conditions on Earth under which calibration measurements
have been made. Such situations arise for Mercury (§650 K), Venus (§740 K), and the Moon
(§400 K), day-time high temperatures of which at their equators are higher than those on
Earth (table 10.1). Corresponding sunlit surfaces of Mars (§280 K) and asteroids (§175 K),
on the other hand, have much lower temperatures. Numerous laboratory investigations of
temperature-induced changes of mineral spectra have been undertaken, therefore, in order to
interpret modal miner

' //
+   *  **+* '1I
 
* 
*
#+ x D? , &? #C 
'& '6#
 ,+ )  &> '=6#**+ ) E =1 ?'6#
*

 *+  ,+ ) +   2, +*
+ @,,+   ,J7JJ17J
 ,2
+ ) +/?

Two important consequences of elevated temperatures on visible to near-infrared

spectra discussed in chapter 9 are: first, positions and intensities of crystal field bands within
a transition metal-bearing mineral are affected; and, second, thermal emission by the host
mineral, itself, occurs. The effects of temperature on reflectance spectral profiles are
demonstrated by the data for Fe2+ in olivine, pyroxenes, and basaltic assemblages shown in
fig. 10.10. The three components of the olivine spectra centred around 1 m become better
resolved at lower temperatures but show insignificant shifts of band minima (fig. 10.10a).
However, reflectance spectra of pyroxenes (figs 10.10& and * show dramatic changes of
band shape with rising temperature, particularly at the longer wavelength edges, which may
result from thermal population of vibrational levels of the crystal field states (cf. §3.9.5; fig.
3.14). In common with olivines, the pyroxene 1 m bands also show only minor wavelength
shifts. However, in marked contrast to the 1 m bands, the pyroxene 2 m bands show major
differences of temperature-induced shifts; for orthopyroxenes, this band increases from 1.80
to §1.90 m between 80 K and 448 K (fig. 10.10&), whereas for clinopyroxenes it decreases
from § 2.35 to §2.25 m over the same temperature range (fig. 10.10c). As a result, two
pyroxenes are clearly resolvable in basaltic assemblages at low temperatures (fig. 10.10d).
However, such resolution of individual pyroxene bands is not achievable at the higher
temperatures applicable to sunlit surfaces of the Moon (§400 K). Thus, broadband features in
the 2 m region observed in lunar remote-sensed reflectance spectra (figs 10.6 and 10.9) are
suggestive of two-pyroxene assemblages and, perhaps, contributions from tetrahedral Fe2+ in
basaltic glass (Farr 1980; Dyar and Burns, 1981).

The contrasting temperature-induced shifts of the pyroxene 1 and 2 m bands could

lead to erroneous estimates of the composition and, to a lesser extent, structure-type of a
pyroxene-bearing mineral assemblage deduced from the remote-sensed reflectance spectrum
of a hot or cold planetary surface if room-temperature determinative curves, such as that
shown in fig. 10.5, are used uncritically. For example, remote-sensed spectra of planets with
hot surfaces, such as Mercury and the Moon, would lead to overestimates of Fe2+ contents of
the orthopyroxenes and underestimated Fe2+ contents of the clinopyroxenes (Singer and
Roush, 1985). Planets with cold surfaces, such as Mars and the asteroids, could produce
opposite results. On the other hand, the room-temperature data underlying the pyroxene
determinative curve shown in fig.

10.5 may impose constraints on the compositions of pyroxenes deduced from

telescopic spectra of a planet with very high surface temperatures, such as Mercury.

The hot surface of Mercury, as well as reflecting sunlight, also behaves as a quasi-
black-body radiator and emits thermal energy back into space. As a result, the spectral
reflectance of Mercury rises sharply above §1.5 m due to its thermal emissivity obscuring
any possible contribution from a Fe2+-pyroxene 2 m band. The close proximity of Mercury
to the Sun also makes telescopic reflectance spectral measurements of its surface very
difficult, so that attempts to identify pyroxenes on Mercury and to estimate their
compositions from the 1 m band alone have produced ambiguous results (McCord and
Clark, 1979; Vilas, 1985). For example, a weak broad band resolved at 0.89 m (McCord and
Clark, 1979) and corresponding to the lowest limit for an orthopyroxene in the room-
temperature pyroxene determinative curve shown in fig. 10.5 might imply the presence of
enstatite in low-iron basalts on the surface of Mercury. However, temperature-induced
variations of the orthopyroxene 1 m band, such as those portrayed in fig. 10.10& showing
that Fe2+ contents of orthopyroxenes are overestimated from high-temperature spectra,
suggest that the wavelegth of the 0.89 m features is impossibly low to be assigned to Fe2+ in
orthopyroxene on Mercury. Pure end-member enstatite or diopside could occur on Mercury,
but these pyroxenes would not be identified by spectral reflectance. The 0.89 m band, if it
exists, could be indicative of ferric-bearing augites (Burns 1976; Straub 1991) in
the regolith of Mercury since Fe3+ crystal field transitions intensify considerably but do not
shift much at elevated temperatures (table 9.2; Parkin and Burns, 1980).
3c7# c
#c
 5c6! c%c
!   c
#c %
c8
 #$c c

The surface of the Moon and Mercury are conducive to telescopic spectral
measurements, albeit complicated by effects of high temperatures, because they lack
atmospheres. The presence of atmospheric gases on Venus and Mars, however, impose
severe problems on the measurement and interpretation of Earth-based remote-sensed
reflectance spectra obtained from these planets. The problem is most acute for Venus because
its hot surface is masked by the dense atmosphere which strongly absorbs and scatters visible
to near-infrared radiation. However, Soviet Venera missions to the Venusian surface have
yielded spectrophotometric data in the form of multispectral images at three wavelengths in
the visible region (0.44,0.54 and 0.63 m) from which surface mineralogy has been deduced
(Pieters 1986). After correcting for effects of orange colouration due to the atmosphere,
the surface of Venus appears to be dark without significant colour. Correlations with high-
temperature laboratory reflectance spectra of oxidized basaltic materials suggested that the
basaltic surface of Venus contains ferric-bearing minerals possibly formed from oxidation of
pyroxenes (Straub 1991) and olivine (Pieters 1986).

The surface of Mars, on the other hand, is visible and has been accessible to several
Earth-based telescopic reflectance spectral measurements in the visible and infrared regions
(McCord 1982; Singer, 1982; Bell et al., 1990a,b; Pinet and Chevrel, 1990), as well as 
  multispectral images of the surface taken during the 1977 Viking orbiter and lander
experiments (Singer, 1985) and the 1989 Phobus II spacecraft mission to Mars (e.g., Erard
 1991). However, while thermal emissivity of the relatively cold surface of Mars becomes
dominant only beyond the 5 m wavelength region, the presence of atmospheric CO2 and
traces of H2O mask critical regions in the near-infrared spectra needed for positive
identification of ferromagnesian silicates in the Martian regolith. To a close approximation,
Mars' surface is composed of bimodal high- and low-albedo regions (Singer, 1985; Bell
1990a,b) which give rise to the typical 'bright-region' and 'dark-region' spectra illustrated in
fig. 10.11. 'Bright-region' spectra are dominated by spectral features at §0.87 m and §0.62
m, attributed to crystal field transitions in Fe3+ ions (table 5.15); atmospheric CO2 is
responsible for the sharper features at 1.45,

1.62 and 1.9 to 2.1 m. The Fe3+ spectral features lack specificity and a variety of
ferric-bearing phases have been suggested as oxidative weathering products on the Martian
surface, including a variety of poorly crystalline oxides, clay silicate and sulphate phases
(Sherman 1982; Singer, 1985; Burns, 1986, 1988; Morris 1985, 1989, 1990). In
the 'dark-region' spectra, although features due to Fe3+ ions may be present, they do not
obscure contributions from Fe2+ ions in the 1 m region (Singer, 1985; Pinet and Chevrel,
1990). These spectral features indicate the presence of subcalcic pyroxenes and, perhaps,
olivine in iron-rich basalts believed to occur on Mars' surface. Unfortunately, interference by
atmospheric CO2 at 1.9 to 2.1 m makes it extremely difficult to identify the pyroxene
structure-type and composition from its diagnostic 2 m band. However, correlations of
telescopic spectra of dark regions of Mars with reflectance spectra of shergottites have been
made (McFadden, 1989; Singer and McSween, 1992). These meteorites, which are believed
to have originated from the surface of Mars (§10.9), contain pigeonite-augite ( olivine)
assemblages giving broad spectral features at 1 and 2.1 m, indicative of basaltic rocks on the
Martian surface.

3c 
c%c%c& c

The vulnerability of igneous ferromagnesian silicate and iron sulfide minerals to

atmospheric oxidation on terrestrial planets results in the formation of numerous Fe3+-bearing
phases, including a variety of oxide, oxyhydroxide, clay silicate and sulphate minerals found
in soils on Earth, gossans capping sulphide ore deposits, and oceanic sediments. Because
some of these minerals could have formed in Martian regolith and on the surface of Venus,
numerous visible to near-infrared spectral measurements have been made on candidate
Fe(III) phases (Sherman  ., 1982; Sherman and Waite, 1985; Morris    1985, 1989,
1990; Morris and Lauer, 1990; Townsend, 1987; Straub    1991). Representative
absorption spectral profiles of four such minerals, goethite, lepidocrocite, maghemite and
hematite, are illustrated in fig. 10.2. Table 10.2 summarizes structural data and peak
assignments of these and other Fe3+-bearing minerals that have figured in discussions of Mars
surface mineralogy.

The visible and near-infrared spectra of goethite, lepidocrocite, nontronite and jarosite
resemble one another and reflect similarities between the coordination environments in each
mineral. Thus, Fe3+ ions occur in edge-shared octahedral sites formed by oxygen and
hydroxyl ligands. The broad band¶s centred around 0.90 m ( 11,100 cm"1 ) and 0.64 m
(~15,100 cm-1) are at typical energies for the>" Y =
(=$) and> " Y =
1(=$) spin-forbidden
CF transitions within Fe3+ ions octahedrally coordinated to oxygen ligands (table 5.15). The
spectrum profile of hematite, however, differs from those of the other Fe (III) minerals,
particularly in the visible region, where bands centred near 0.55 and

0.43 m are conspicuous. Their intensification results from the unique structural
environment of Fe3+ ions in hematite.

In the hematite crystal structure which is isostructural with corundum (fig. 5.4),
planes of hexagonal close-packed O2- ions are stacked along the* axis and accommodate Fe3+
ions in two-thirds of the available octahedral sites. Pairs of Fe3+ ions are located along the*
axis above and below vacant octahedral sites and the adjacent [FeO6] octahedra share three
oxygens across a common triangular face linking the two octahedra. The Fe3+-0 distances to
the three face-shared oxygens are shorter (194.5 m) than those to the other three oxygens
(211.5 pm), so that the Fe3+ ions are not centrally located in their coordination sites. The Fe3+-
Fe3+ distances between pairs of face-shared octahedra are 289 pm parallel to the* axis. Each
[FeO6 ] octahedron also shares edges with three neighbouring [FeO6 ] octahedra perpendicular
to the* axis, and Fe3+-Fe3+ distances across edge-shared octahedra are 297 m. Although all
Fe3+ sites in the hematite structure are crystallographically equivalent, there are two
magnetically inequivalent sites. All cations in a given plane of edge-shared octahedra
perpendicular to the* axis are equivalent but magnetically inequivalent to cations in adjacent
planes separating the face-shared octahedra. Below the Curie temperature of crystalline
hematite (955 K), Fe3+ ions are strongly antiferromagnetically coupled.
 The presence of Fe3+ ions in asymmetric trigonally distorted octahedral sites
contributes to the overall intensification of spin-forbidden bands in crystal field spectra of
hematite. However, magnetic coupling between the antiferromagnetically ordered Fe3+ ions
in crystalline hematite also intensifies these bands and results in an additional band at 0.53
m (18,700 cm-1 ). This band represents a paired transition between cations in the face-shared
[FeO6] octahedra, by analogy with that seen in spectra of isostructural yellow sapphire (fig.
3.16). Small particle sizes of Fe2O3 and atomic substitution of Al3+ for Fe3+ ions diminish
magnetic coupling interactions, thereby lowering intensities of visible-region spectra of
nanocrystalline and aluminous hematites (Morris 1985,1989; Morris and Lauer, 1990).

Although magnetic coupling between Fe3+ ions in edge-shared [FeO6 ] octahedra also
occurs in goethite, lepidocrocite, nontronite and jarosite, leading to intensification of ferric
spin-forbidden Fe3+ CF bands relative to magnetically-dilute minerals such as garnets and
coquimbite, intensities of paired transitions in the 0.485 to 0.550 m region are significantly
reduced relative to hematite (fig. 10.2). Molecular orbital calculations (Sherman, 1985b)
support experimental evidence (Rossman, 1975, 1976) showing that intensification of Fe3+
CF transitions is smaller when Fe3+ are bridged by OH- rather thanO2anions.

3c"6
#$c
!6
$ c#c
 c

The overall colour of Mars, together with telescopic reflectance spectral resolution of
the 0.87 and 0.62 m bands attributable to the> " Y  =
(=$) and>" Y  =
1(=$) crystal field
transitions in octahedral Fe3+ ions, suggests that ferric-bearing phases are present in the rusty
'bright' regions of Mars. 'Dark-region' reflectance spectra, on the other hand, show broad
band¶s near 1.0 and 2.1 m, attributable to Fe2+ CF bands in relatively unoxidized pyroxenes
and olivine. The high iron contents of Martian regolith were demonstrated during the 1977
Viking Lander XRF experiments (Toulmin, 1977), in which high sulphur concentrations
(assumed to be sulphate) were also analysed (see Appendix 1). Although a magnetic phase
inferred to be maghemite (Ȗ-Fe2 O3) (Hargraves 1979), feroxyhyte (į-FeOOH) (Burns,
1980) or nanophase hematite (Morris   1985, 1989) was demonstrated by the Viking
magnetic properties experiment, the overall mineralogy of Martian regolith has not been posi-
tively identified.

The more obvious crystalline Fe(III) minerals most commonly found on oxidized
surfaces of the Earth, such as hematite and goethite, appear to be eliminated on Mars on
account of mismatches of relative intensities of the crystal field bands in their visible-region
spectra and the rapid decrease of reflectance below 0.7 m. Magnetic ordering in crystalline
hematite adversely enhances the intensities of the >" Y  =
 (~0.60 m) and > " Y    =E, =

(~0.44 m) transitions relative to remote-sensed reflectance spectral profiles of Mars' surface
in the 0.40 to 0.70 m region (cf. figs. 10.2 and 10.11). Alternative materials whose visible to
near-infrared spectra compare more favourably with Martian 'bright-region' spectra, include
nanocrystalline Fe2O3 phases (Sherman   1982; Morris   1985,1989), Fe 3+-
exchanged clays (Banin  1985) containing ferrihydrite (Bishop 1992), palagonite
(Gooding, 1978; Morris   1990), and gossaniferrous ferrihydrite-jarosite-opal-clay
silicate assemblages (Hunt and Ashley, 1979; Burns, 1987b, 1988). Each of these candidates
comprises poorly crystalline or X-ray amorphous Fe(III) phases, an essential property for
reducing magnetic ordering of constituent Fe3+ ions and decreasing the intensity of crystal
field bands in the visible region. Inferences can be drawn from each of these terrestrial
analogues about weathering processes on Mars. Thus, nanophase Fe(III) oxides could be pro-
duced by ablation of surface basalt flows during global dust-storms and photochemical
oxidation of ferromagnesian silicates exposed to the Martian atmosphere. Palagonites suggest
the intrusion of basaltic magma into permafrost on the frozen Martian surface. Clay silicates
indicate chemical weathering of igneous minerals in an aqueous environment. And, gossans
imply the presence of sulphide mineralization and perhaps ore deposits associated with the
iron-rich basalts on Mars. Oxidative weathering of such sulphide mineralization produces
acid groundwater which facilitates chemical weathering of the silicate minerals (Burns,
1987b, 1988; Burns and Fisher, 1990). Thus, near-surface ablation, dissolution, oxidation and
hydrolysis of basaltic materials would produce poorly crystalline and colourful ferric sulphate
(jarosite, botryogen, copiapite, coquimbite), oxide (ferrihydrite, hematite, goethite), and clay
silicate (montmorillonite, nontronite, hisingerite) minerals all of which could contribute to the
red colour and ferric-like spectra of the Martian surface.

3'c"6
#$c
!6
$ c#c7# cc

Evidence for ferric-bearing minerals on the surface of Venus, which was derived from
spectral reflectance data measured during Soviet Venera missions (Pieters 1986), poses
a dilemma. Hematite is thermodynamically unstable (Fegley   1991) and is rapidly
converted to magnetite (Straub and Burns, 1992) under temperature and oxygen fugacity
conditions believed to exist on the surface of Venus. Ferrifayalites and laihunite ( 4.8.3.4),
too, are rapidly reduced to Fe2+-olivines under Venusian
/1 surface conditions. However,
other mixed-valence Fe2+-Fe3+ silicates could exist on the surface of Venus, including ilvaite
( 4.8.2) and dehydroxylated silicates such as oxyhornblendes and oxybiotites (Burns and
Straub, 1992). The high electrical conductivities of Fe2+-Fe3+ silicates such as ilvaite at
Venusian surface temperatures render these mixed-valence minerals viable candidates, in
addition to magnetite (Fe3O4), pyrite (FeS2) and perovskite (CaTiO3) (Fegley et al., 1992;
Klose 1992), that could be responsible for the high radar-reflectivity surfaces observed
on most mountainous terranes on Venus (Pettengill 1992).

32c%
#c 
c%c! c
#c
 c

Before the advent of the Apollo and Luna missions which retrieved samples from the
Moon's surface, meteorites provided the only source of extraterrestrial materials and raised
questions about their sources from parent bodies such as asteroids. Visible to near-infrared
reflectance spectroscopy, therefore, has been applied extensively to laboratory investigations
of meteorites and to remote-sensed measurements of many asteroids (Gaffey, 1976; Gaffey
and McCord, 1978; McFadden et.al., 1982, 1984; Bell and Keil, 1988).

Spectral reflectance curves for the range of meteorite types are illustrated in fig.
10.12. These spectra demonstrate the diagnostic features of the various
 ' /1@ +*  ***  ,  +
# $ >

meteorite types, including the presence or absence of absorption bands, their positions
and relative intensities, their symmetries and widths, and other properties such as continuum
slope, curvature, and inflection points (Gaffey, 1976). Again, noteworthy features of the
meteorite spectra are the prominent pyroxene 1 and 2 m bands, the relative intensities and
asymmetries of which have been used to characterize the pyroxenes and to estimate
olivine/pyroxene ratios in meteorites. Relating these meteorite spectra to those of asteroids
has been difficult due to the faintness of these objects in space. Nevertheless, telescopic
spectral measurements have led to discoveries of several large asteroids containing oli vine-
rich dunite-like rocks (Cruikshank and Hartmann, 1984), as well as other asteroids with wide
ranges of olivine/pyroxene ratios, which are inconsistent with an ordinary chondritic
composition but suggestive of affinities with stony-iron meteorites (Bell and Keil, 1988). The
spectra of some asteroids, including 4 Vesta and 1915 Quetzalcoatl, resemble basaltic achon-
drites (McFadden   1982) suggesting that they underwent internal heating and
differentiation. Thus, asteroids like Vesta and Quetzalcoatl are very likely sources of the
large and varied meteorites of the eucrite, howardite, and diogenite groups.
 Meteorites collected from Antarctica in the past decade have yielded some unique
specimens which appear to have originated from the terrestrial planets themselves. They
include several meteorites with lunar affinities and others belonging to the SNC group of
meteorites, comprising the shergottites, nahklites and chassignites, believed to have
originated from Mars. One of the lunar meteorites (ALHA 81005) is a regolith breccia
analogous to rock-types found in the lunar highlands. A composite diffuse reflectance
spectrum of this meteorite shows a band centred near 0.98 |iim indicative of subcalcic augite,
together with features attributed to olivine and Fe2+-bearing plagioclase feldspars (Pieters
 1983). However, this spectrum profile does not match telescopic reflectance spectra
obtained from §150 small areas three to twenty km in diameter on the Moon's surface,
including young highland craters. The spectral data suggest that this lunar meteorite is
derived from a surface unit on the Moon not previously sampled, and that the most probable
source area is the near-side limb or the far-side of the Moon. The Antarctic shergottite EETA
79001, with impact-glass pockets containing trapped gases resembling the composition of the
Martian atmosphere, contains calcite, gypsum and sulphur-rich aluminosilicate phases which
were suggested to represent chemical weathering products from Mars (Gooding and
Muenow, 1986; Gooding    1988). However, negligible ferric iron was resolved in
Mossbauer spectral measurements of the impact glasses, while the pyroxene-dominated
matrix of EETA 79001 contains less than 2 wt per cent Fe3+ (Solberg and Burns, 1989).

33cc!
!# c
#c! # c

Reflectance spectroscopy has proven to be the most powerful and versatile remote-
sensing technique for determining surface mineralogy, chemical compositions and lithologies
of planetary objects, as well as constituents of their atmospheres. Table 10.1 summarizes
information that has been deduced for the terrestrial planets based on spectral properties of
light in the visible and near-infrared regions reflected from their surfaces.

The planetary reflectance spectral data constituting Table 10.1 traditionally have been
obtained with Earth-based telescopes. Such remote-sensed measurements are limited by
telescope availability, favourable observational conditions, and optimum viewing alignments
of the planetary objects. As a result, comparatively few high quality telescopic spectra
(perhaps numbering several hundred) are available for the solar system planets and their
satellites. However, this situation could change dramatically.

Future spacecraft missions to solar system objects are primarily being oriented
towards remote-sensing experiments, in contrast to the soft-landed   experiments and
sample-return initiatives during the 1970's and 1980's. Because reflectance spectroscopy has
become one of the most important investigative techniques in the planetary sciences, current
and planned space missions for the 1990's and 21st century should include visible and near-
infrared spectrometers in their instrument pay loads. Reflectance spectral measurements from
space would provide more favourable viewing geometries, eliminate problems due to telluric
water and CO2, and improve the resolution of areas scanned on a nearby planetary surface.
 Studies of a qualitative nature described earlier in the chapter have demonstrated that
spectral reflectance profiles of planetary surface materials are influenced primarily by the
chemical compositions and abundances of constituent minerals. There is now an increasing
awareness of second-order effects, such as temperature, viewing geometry, grain size and
particle packing, on positions and intensities of diagnostic mineral spectral features. Research
is needed to quantify these effects and to more closely simulate physical properties of
planetary regoliths when they reflect sunlight. Such projects include spectral measurements
for mineral assemblages having different temperatures, particle sizes, grain-size packings,
and modal mineral proportions, over a range of angles of incident and reflected light and in
confining atmospheres having a variety of pressures and compositions. Other effects such as
atmospheric weathering and radiation sputtering processes under different intensities and
exposure times also need to be assessed. The ultimate goal of such laboratory investigations
is to develop a theoretical model for deducing the modal mineralogy of an area on a planetary
surface from its reflectance spectrum measured under known lighting conditions.

c
 cc
 
c"
ccc
Alpha-particle spectroscopy is possible on planetary surfaces with atmospheric
abundances of less than a few milligrams per cm2. Alpha particles emitted from radon, a
uranium daughter trapped by gravity, and from radioactive radon daughters provide
information on uranium distributions. Seismic activity and impact events promote radon
diffusion to the surface and lead to areas of enhanced a-particle activity. Alpha-particle
spectroscopy was first applied to the lunar surface.

In recent years radon has literally become a household word. It is apparent that the diffusion
of radon gas from below the ground can substantially increase radioactivity levels of the air
in some homes; this effect is characterized by significant local variations and can be time
dependent. Geologists were familiar with the process of radon diffusion long before the
public became aware of it, as radon had been observed in uranium mines and in the emission
of gases from volcanoes. Furthermore, radon diffusion can operate anywhere in the solar
system where natural concentrations of U and Th exist. It is expected that radon diffusion will
be more pronounced at sites of high soil porosity or where radon is swept along the path of
more common gases diffusing to the surface.

On the lunar surface, in the absence of an atmosphere, the diffusion of radon (Rn) has
some interesting consequences. For example, heavy Rn atoms are trapped by the Moon¶s
gravity. Radon atoms will follow ballistic trajectories until they decay. Upon decay,
characteristic particles are emitted and heavy recoil nuclides are deposited on the lunar
surface. The surface deposit is itself unstable against radioactive decay and will be a further
source of characteristic 4c particles. These nuclides also emit  rays and will enhance the
* of the surface beyond the values expected from U and Th concentrations. There are
two distinct signals of interest. The first is from 5.486 MeV a particles emitted by the U
daughter product 222Rn (half-life 3.8 days) and her daughter products. The ratio of the
intensity in the 2 Rn line to the intensities in the daughter product lines (except that of 210Po)
is expected to have an equilibrium value of two. Approximately one-half the heavy recoils
from the gravitationally bound 222Rn atoms are directed downward and are deposited on the
surface; the rest have sufficient energy to escape the Moon¶s gravity. Polonium- 210, which
emits 4 particles with energy of 5.305 MeV, gives a second type of signal because its
production is held up by the 21-year half-life of 210Pb. Thus the instantaneous activity ratio
of 222Rn to 210Po will depend on the history of Rn emanation over a few 210Pb half-lives.
These two processes are illustrated in F.ig. 11:1. A third type of signal is expected from the
short-lived Th daughter 220Rn (half-life 55 s) and her daughters. However, this signal is not
expected to be significant because 220Rn has considerably less time to diffuse through the
regolith than 222Rn.

Prior to any measurements the a-particle activity of the lunar surface was estimated
from a model based upon terrestrial concentrations of U and terrestrial rates of diffusion
#Ô   1966). The predicted rate was 4 disintegrations per second. This proved to be
too large by nearly 2 orders of magnitude.

The first observations of 4 -particle activity was made at Mare Tranquillitatis by the
Surveyor 5 4 - particle scattering experiment #
*,   1970). The observed rate of
210
Po was 0.09 ö0.03 a disintegrations per second. At the Surveyor 6 and 7 landing sites, the
210
Po activity was at least a factor of 2 smaller. Hence, assuming the landing sites were not
disturbed by the spacecraft, the Surveyor experiments gave indications of a spatially varying
deposit of210Po on the lunar surface.

c

c ( + cc

The lunar orbiting command modules of the Apollo 15 and Apollo 16 spacecrafts
contained an instrument known as the Alpha Particle Spectrometer. It consisted of an array of
10 totally depleted silicon surface barrier detector units. These devices are simple diodes with
a 50-volts bias. They provide an output pulse that is directly proportional to the amount of
energy deposited by ionization. Each detector was approximately 100 m thick, had an active
area of 3 cm2 (the area of a U.S. five-cent coin), and was collimated to a 90° field of view.
Except for custom packaging, they were essentially identical to detectors commonly used in
the laboratory. The main source of background is solar and galactic cosmic rays. Most cosmic
rays are protons with energy above 50 MeV. The relationships between energy, range, and
pecific ionization for singly charged protons is quite different than that of doubly charged a
particles. The thickness of the detectors was selected such that the mean energy deposit of
protons (and deuterons) of all energies by ionization is always less than 5 MeV while the
output signal for a particles in the energy range 4.7 to 9 MeV is directly proportional to
energy. This range includes the Įparticles of Rn and her daughter products. The background
is due to higher-energy deposits from proton scattering and ionization of heavier nuclides
found in cosmic rays. It should be a continuum that varies very little over the band. Five of
the ten detectors contained a 208Pb calibration source (5.11 MeV). If an energy deposit is
between 4.7 and 9 MeV, it is digitized and transmitted along with a signal that identifies the
detector unit. Monoenergetic a particle sources such as the decaying nuclides 222Rn and
210
Po are seen as peaks in a pulse height spectrum above a background continuum.
Measurements consisted of simply pointing the detectors downward as the Apollo spacecraft
orbited the Moon and recording all signals in the range of 4.7 to 9 MeV along with the time.
By accumulating energy spectra for arbitrary selected time intervals ones constructs a map of
the distribution of 222Rn and 210Po over the Apollo ground track.

c
c
6 
# c

The data, 210Po and 222Rn, are discussed by considering first general trends and then
specific correlations with lunar features. The important point to bear in mind is that the decay
of 222Rn represents activity that was occurring at the time of observation, whereas 210Po
activity is caused by 222Rn reaching the lunar surface any time up to about 60 years (three
times the 210Pb half-life) prior to the observation. Therefore the observed 222Rn to 210Po ratio
is an indicator of variability in radon emanation.

c6
c# cc
 Figure 11.2 shows the observed count rate for 222Rn (plus her two prompt daughters)
as a function of longitude for the Apollo 15 ground track. The data are displayed in four
panels, each containing 90° of lunar longitude, the top panel starting at 0° and proceeding to
the east (toward Mare Crisium).

The background level shown in Fig. 11.2 consists of two components, a diffuse Rn
contribution and cosmic rays. Two methods were used to estimate the diffuse Rn
contribution. The first method is based upon the difference in counting rates in the 222Rn
energy channels between the lunar orientation and a relatively brief deep-space pointing of

the detectors. The second method consists of assuming the cosmic ray background is uniform
in the 5- to 9-MeV range. The count rates in energy channels not associated with Į particles
from Rn and daughter products is taken as the background level for all energy channels.
Results from both methods are consistent within counting statistics. The mean level of 111?
activity over the Apollo 15 ground track at the time of the Apollo 15 mission (224.. hour
period commencing 14:45 GMT 29 July 1971) was (1.00 ö0.22)* 10-3 dis/cm2 s. The cosmic
ray background counts are nearly three times as high as the mean radon counts. Count rates
detected on the far side of region 90°E²180°E are consistent with a cosmic ray background.

While the ability to observe detailed structure in these data is limited by the counting
statistics, there is an increase in the amount of 222Rn observed over the western quadrant
(Oceanus Procellarum and Mare Imbrium) and to a lesser extent over the eastern quadrant
(Maria Serenitatis, Tranquillitatis, Fecunditatis, and Crisium). The signal bin labeled ³Arist.´
(for Aristarchus) in Fig. 11.2 has been excluded from the average and will be discussed later
in this chapter. General trend of the count rate is in agreement with the observed U and Th
concentrations over the same region #  1972; Chapter 15).
' 
C+B   +*  
     '* 
,    

   ,   , ,    
   , *   ,  
B
,#   +* # * " 
 ,  ,,  *     ,   2 ++    + )* 
, +* 
, )*    1C+C *  *  

, 111? )*  
  * **
 ,   ,! 
    1C+C  111?
)*
,*   


c+  +
5c 7+) › c
c+ c›+7+c

c !!
c%c  c

On the Moon, 222Rn has been detected at a level of about 10-3 of terrestrial values. It is one
of perhaps only two gaseous species in the lunar atmosphere that is native to the Moon; the
other, as reported by GK(1973), is 40Ar. It is reasonable to expect that the rather
short half-life of 222Rn gas (3.8 days) could lead to the preservation offeatures that appear
initially in the spatial distribution. These features include emission that is associated with
certain well-known lunar craters and maria edges, as well as a diffuse component of222Rn
and 210pO that exists more or less over the entire Moon. Theoretical models must explain the
following observations:

1. The process of Rn emission takes place over a large portion of the Moon, as evidenced by
the widespread distribution of 222Rn and 210PO.

2. Spatial variations of 222Rn and 21Dpo activity exist across specific regions and across
maria edges; these variations are more rapid than the variation of the concentration of U, the
source of Rn.

3. The process of nonuniform Rn emission is operating at the present time and increases in
the rate of emanation have occurred during the last 21 years (effective half-life of the 21OpO
surface deposit).

4. Any )  process that causes a release of Rn (e.g., meteor bombardment, thermal
seismic activity, mass slumping) must churn a thickness of several tenths g/cm2 of material
per year, a very substantial mass, to explain the observed enhancement in 222Rn or 210Po.

In view of the persistence of lunar surface features on a timescale of 109 years, the last
constraint appears to rule out extern processes as a source of surface concentrations f Rn.
Hence, the mechanism responsible for the e ission of Rn is a time-varying, internal process.

c
9
#c

Several of the Rn and Po features observed from Apollo are localized to within 5° of
lunar longitude, i.e., within 150 km. The craters Aristarchus and Grimaldi are examples. The
21O
Po maria edge effect may also be a feature no greater than ?L The observed extent is
consistent with a narrow perimeter around the mare. The observed localization implies that
free Rn atoms stick on the lunar surface a large fraction of the time. If Rn atoms did not stick,
the spatial extent of the Į particle features would be much larger even on the farside of the
Moon. The mean diameter of a feature should be approximately equal to two times the square
root of the average number of trajectories a Rn atom undergoes during its half-life multiplied
by the average displacement per trajectory. On the farside, at a temperature of 100 oK, initially
localized Rn atoms undergoing successive free trajectories without sticking would have
spread to a diameter of about 103 at the end of their 3.8 day half life . This is severaltimes the
apparent size of the features. Thus at 100° K, Rn atoms emitted from the surface are in free
flight only a fraction of the time. This result has implications for the migration of Rn through
regolith. A sticking effect is consistent with the fact that the emission ofRn from the lunar
regolith is low compared to the Earth in spite of similar U concentrations. It also has possible
implications for the transport of other gases across the lunar surface, although other gases
which are much lighter than Rn, may not stick as much to the lunar surface.
'c›
#c8 c
# #cc #!#
c

Two of the brightest Q particle hot spots occur in the vicinity of the well-known
craters Aristarchus and Grimaldi. The former is seen in the 222Rn data while the latter is seen
in the 21OpO data. The significance of these two craters is that they have been reported
repeatedly by ground-based observers to be sites of transient optical events. Reports on
Aristarchus Crater go back several hundred years as noted by K,  (1967) and
- (1972), who have studied historical records of transient optical events on the lunar
surface. If the correlation between Į hot spots and lunar transient event is real we can predict
other specific correlations. For example, Alphonsus Crater is another frequently reported site
of transient optical events and should appear as a site of enhanced a activity. However, the
Apollo 15/16 Alpha Particle Spectrometer did not pass over Alphonsus Crater, so this
prediction cannot be tested. The second-largest 222Rn feature occurs in the Apollo 15 data at
a lunar longitude of -1300 E (Fig. I in G ,K 8Although not as statistically
significant with respect to the lunar average as Aristarchus Crater (4.3 standard deviations),
this feature is about 3 standard deviations above the lower local average over the farside
highlands. The feature is close to Tsiolkovsky Crater, which could not have been seen as a
site of transient optical events because it is on the farside of the Moon. However, there is
additional evidence for a correlation between Q activity and transient optical events. Along
with the discrete repeating sites such as Aristarchus, Grimaldi, and Alphonsus there are a
large number of reports of smaller optical events. K, (1967) has noted that the sites
of these events tend to group at the mare edges and near prominent young craters. x

(1968) have mapped the spatial distribution of dark-haloed craters on the visible lunar
surface and have found that they tend to occur at the edges of the maria. These craters are
generally small, are surrounded by low-albedo material, and may be of volcanic origin
#x,  >1 A factor that could be common to the Po effect, the transient optical
events, and the occurrence of darkhaloed lunar craters is transient venting of volatile
materials at the edges of the maria. In light of the fact that we see a rather striking 210pO
effect at the edges of the maria, we suggest that there is indeed a real spatial correlation
between Rn hot spots and sites of transient lunar phenomena. Radon is likely to be merely a
trace component of more common gases when transient emission occurs. On the Earth Rn
comprises only about one part in 1014 of soil gas. Certainly the diffusion of other more
abundant gases through lunar material will promote the diffusion of Rn, and it is indeed
difficult to imagine an increase in the rate of Rn diffusion without the involvement of other
gases. Perhaps optical events are detectable only at times when the gas venting is most
pronounced and localized fit is true that dark-haloed craters are of volcanic origin, the
venting might be expected. The Apollo results indicate that at least one gas, Rn, is being
emitted from the edges of the maria at the present time.

1c›
#c8 c
 c
!# c

Another correlation may be seen between the slower spatial variation of 222Rn and
the spatial variation of Ȗ ray activity across the Apollo ground track as reported by the Apollo
Gamma Ray Spectrometer #  9 1972, 1973; see also Chapter 15). For both the Ȗ
and Mdata, the highest levels are seen in the quadrant containing Mare Imbrium and Oceanus
Procellarum and the lowest levels are seen over the farside highlands. Both depend on the
variation of U, so a positive correlation is not surprising. However, the local rate of M
emission also depends on the local Rn conductance through the regolith. As we have noted,
this varies as a function of time. If there were sufficient counting statistics, a detailed
comparison of the Mand Ȗ results would identify regions of high conductance.

'c c+›+
 c
c0"c+› c  ›
›
cc

Alpha-particle spectroscopy can play a useful role as a probe of Rn emanation in

future planetary investigations. The technique of detecting characteristic Į particle energies
with spectroscopic detectors is applicable to future spacecraft that orbit the Moon. it is also
applicable to an orbital study of Mercury, certain asteroids, and the satellites of the giant
planets, which contain either no atmosphere or atmospheres too tenuous to completely stop 5-
MeV a particles.

The optimum platform for a study of the Moon¶s Rn emission would be a polar
orbiting satellite that operates for a year or more. In contrast to the Apollo 15 and 16
missions, which acquired only 13% and 3.5% coverage respectively, a polar-orbiting satellite
would have the advantage of obtaining complete coverage of the Moon. Another advantage is
more sensitive and precise results owing to longer collection times up to a factor of 10 over
the Apollo sites and up to a factor of fifty over the Polar Regions. There would be a much
better opportunity to observe transient events as they occur during the mission as regions of
the Moon are observed repeatedly at intervals of two weeks or less. Furthermore, substantial
improvements in electronics have taken place in the two decades since the construction of the
Apollo instruments. An instrument of equal weight could be built now with about two to
three times more collecting area. Hence, the total area time factor over the Moon would be
increased by at least an order of magnitude. Unlike photography, the a-particle investigations
are limited by counting statistics. Hence, higher area-time factors result in more sensitive
measurements and better spatial resolution.

An investigation of Mercury would proceed in essentially the same way as the Moon
except that the higher temperatures of the surface would require a more sophisticated thermal
design. Investigations of the moons of the giant planets that do not contain atmospheres with
total column densities exceeding I mg per cm2 could also be performed in the same manner.
If a moon had an atmosphere with a column density between I and S mg per cm2, studies of
Rn emanation and their spatial and temporal variations could be still performed from orbit,
though at somewhat reduced sensitivity. However, alterations in thea-particle spectra
resulting from slowing down in the thin atmosphere would provide a measurement of the
column density. In summary, a-particle spectroscopy is potentially quite useful in future lunar
and planetary investigations.