- Email: [email protected]
Elemental fractionation in ultraviolet laser ablation sampling of igneous silicate minerals relevant to Mars
Applied Surface Science 165 Ž2000. 166–177 www.elsevier.nlrlocaterapsusc
Elemental fractionation in ultraviolet laser ablation sampling of igneous silicate minerals relevant to Mars M.E. Taylor ) , D.L. Blaney, G. Cardell Jet Propulsion Laboratory, MS 302-231, 4800 Oak GroÕe Dr., Pasadena, CA 91109, USA Received 23 March 2000; accepted 11 May 2000
Abstract Laser ablation sampling ŽLAS. has significant potential for remote terrestrial and extraterrestrial applications if fractionation can be understood and controlled. This study focuses on acmite, albite, augite, diopside, forsterite, and labradorite, silicate minerals that are relevant to Mars, Earth, the moon, and asteroids. The minerals were sampled using a frequency-quadrupled Nd:YAG laser and the samples were deposited as films on graphite substrates. Rutherford backscattering spectrometry and X-ray photoelectron spectroscopy measurements of the minerals and their films indicate sodium, magnesium, silicon, calcium, and iron fractionation. Oxygen fractionation was not indicated. A cation replacement mechanism is a potential source of fractionation. These minerals can be easily distinguished by their film compositions despite fractionation, indicating that ultraviolet laser ablation sampling is feasible for silicate minerals. q 2000 Elsevier Science B.V. All rights reserved. PACS: 91.65.Vj; 83.80.Nb; 42.62.Eh Keywords: Ablation; Fractionation; Stoichiometry; Silicate minerals; ICP-AES; ICP-MS
1. Introduction Laser ablation sampling ŽLAS. is a technique that is often used to obtain vapor or droplet samples of parent materials for subsequent chemical analysis w1–4x. Chemical information is obtained through either direct or indirect analysis of samples. Direct approaches include optical spectroscopy and timeof-flight mass spectrometry w5,6x. In some indirect approaches, i.e. inductively coupled plasma atomic
) Corresponding author. Tel.: q1-818-354-2662; fax: q1-818393-4663. E-mail address: [email protected] ŽM.E. Taylor..
emission spectrometry ŽICP-AES. and inductively coupled plasma mass spectrometry ŽICP-MS., samples are introduced into an additional excitation source w7,8x. In other indirect approaches, samples are deposited on a substrate, a process called pulsed laser deposition ŽPLD. w1,2x. The resulting films are analyzed by standard characterization techniques such as Rutherford backscattering spectrometry ŽRBS. and X-ray photoelectron spectroscopy ŽXPS. w9–11x. LAS is particularly feasible for geological materials. Sampling of a variety of geological materials with different compositions and microstructures has been demonstrated w12–14x. Extensive preparation of parent materials is not required. Sampling areas of a few tens of micrometers and sampling rates of a few
0169-4332r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 Ž 0 0 . 0 0 4 4 2 - 6
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
nanometers per pulse are not atypical, facilitating both spatial profiling and depth profiling. Fractionation in LAS has been studied for a variety of geological materials w12–14x. The majority of these studies used ICP-AES and ICP-MS techniques and focused on trace element fractionation. Fractionation was found to correlate with absorbed power density and to be more significant for infrared lasers than for ultraviolet lasers. Fractionation that is often attributed to LAS may be due to other sources. In ICP-AES and ICP-MS, for example, fractionation may occur during transport to the ICP or in the ICP. Recent ICP-MS measurements suggest that large particulates are often either lost during transport to the ICP or incompletely digested in the ICP and that this can result in fractionation w15–17x. LAS has significant potential for remote terrestrial and extraterrestrial applications if fractionation can be understood and controlled. This study was undertaken with that objective. A PLD approach rather than an ICP-AES or ICP-MS approach was selected to avoid potential fractionation related to the ICP.
2. Mineral selection and preparation The Shergottite–Nakhlite–Chassignite ŽSNC. meteorites are believed to have originated on Mars and are reported to consist largely of feldspars, pyroxenes, and olivines w18x. The Earth’s continental crust is reported to consist of 58% feldspars, 13% pyroxenes and amphiboles, and 3% olivines by volume w19x. These types of minerals are also relevant to the moon and asteroids. They are therefore of general interest for remote terrestrial and extraterrestrial applications. The plagioclase feldspars albite and labradorite, the clinopyroxenes acmite, augite, and diopside, and the olivine forsterite were selected for this study. Table 1 shows the nominal compositions of these minerals w20x. All minerals used in this study are research-grade natural specimens obtained commercially.1 Multiple specimens of each mineral were polished using sili-
1
Wards, 800-962-2660, P.O. Box 92912, Rochester, NY 14692-9012, USA.
167
Table 1 Silicate minerals selected for this study w20x Mineral
Classification
Nominal composition
Acmite Albite Augite Diopside Forsterite Labradorite
Clinopyroxene Plagioclase feldspar Clinopyroxene Clinopyroxene Olivine Plagioclase feldspar
NaFeSi 2 O6 NaAlSi 3 O 8 ŽCa,Na.ŽMg,Fe.Si 2 O6 CaMgSi 2 O6 Mg 2 SiO4 30% NaAlSi 3 O 8 , 70% CaAl 2 Si 2 O 8
con carbide films with average particulate diameters of 68, 22 and 14 mm and diamond films with average particulate diameters of 6, 3 and 1 mm.
3. Experiments Fig. 1 shows the sampling system. The laser is a pulsed frequency-quadrupled Nd:YAG laser with an internal energy attenuator. The wavelength is 266 nm, the pulse repetition rate is 15 Hz, the maximum pulse energy is 5 mJ, the pulse width is 3–5 ns, and the sampling area is approximately 0.4 mm2 . The mineral target and the graphite substrate are separated by approximately 2.5 cm and are housed in a vacuum chamber with a base pressure of 5 = 10y8 Torr. Prior to sampling, the average powers incident on the target and reflected from the target are measured using a thermopile meter. During sampling, the position of the target is adjusted at intervals of 4500 pulses to provide a fresh target area. Sampling is stopped when a film is visible on the substrate. Films were deposited from the mineral targets at nominal power densities of 100 and 320 MW cmy2 . These power densities were calculated assuming a pulse width of 4 ns and a sampling area of 0.4 mm2 . They varied slightly due to variations in target reflectivity potentially due to composition and surface roughness. The lower power density was selected because it is the minimum power density required to produce visible plumes in all of the minerals. No film was deposited for labradorite at this power density because the deposition rate was very low. The higher power density was selected because it is
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
168
Fig. 1. Sampling system.
the maximum power density easily achievable with the current laser, lens, and geometry.
4. Characterization In RBS, the sample is irradiated with monochromatic ions w9–11x. The ions undergo Coulomb repulsion collisions with the sample atoms and are backscattered. The energy loss of each backscattered ion is a function of the penetration depth of the ion and the mass of the sample atoms and is not a function of the chemical environment of the sample atoms. The sampling area is typically a few millimeters in diameter and the sampling depth is typically a micrometer. In XPS, the sample is irradiated with monochromatic X-rays w9,10x. The sample electrons absorb the X-rays and are ejected. The energy of each ejected electron is determined by the electron binding energy and is a function of the sample atoms and their chemical environment. The sampling area is typically a few tenths of a millimeter in diameter. The sampling depth is limited by the escape depth of the ejected electrons and is typically several nanometers. Because XPS has a small effective sampling depth, surface contamination may affect results. Samples are often exposed to an ion beam prior to characterization to remove surface contamination. RBS was selected as the primary characterization technique because large sampling area, large sampling depth, and insensitivity to chemical environment make RBS preferable to XPS for mean composition measurements. The RBS system used in this study is a National Electrostatics Corporation pel-
letron system. It uses a beam of 2-MeV He 2q ions. The sampling area is approximately 25 mm2 . Sampled targets were repolished prior to characterization. The RBS spectra were analyzed using the software package RUMP 2 Fig. 2Ža. and Žb. show examples of spectra and fits. Because of non-negligible backgrounds potentially due to embedded particulates, linear background subtractions were performed for the O and Na peaks in the albite and labradorite films and for the O, Mg, and Si peaks in the lower power density augite film. Approximate fits were obtained using the simulation feature. Best fits were obtained using the Poisson fitting feature in conjunction with the approximate fits. The XPS system used in this study is a Physical Electronics Quantum 2000 system. It uses a focused beam of 1.4866-keV Al K a X-rays for characterization and a beam of 2-keV Arq ions for removal of surface contamination. The spherical section analyzer was used in a constant pass energy mode with a pass energy of 46.95 eV and a step size of 0.4 eV. The sampling area is approximately 0.3 mm2 . Sampled targets were repolished prior to characterization. To remove surface contamination, 5 nm of material was sputtered from targets and films prior to characterization. The XPS spectra were analyzed using the software package MultiPak.3 The calibration feature was used to correct the energy and pass energy dependence of the analyzer. Shirley background subtractions were performed. Fits were obtained using the peak areas in conjunction with the O1s, Na1s, Mg2s, Al2p, Si2p, Ca2s, and Fe2p3 sensitivity factors. Both RBS and XPS have limitations. RBS is unable to distinguish between Al and Si in the film spectra because of their similar masses. XPS is unable to reproducibly measure O in the targets, possibly due to incomplete removal of surface contamination. The compositions indicated by RBS and XPS differ slightly. These deviations are not atypical and most likely occur because sensitivity varies with factors such as chemical environment. In this study, 2
Computer Graphic Service, 607-277-4913, 52 Genung Circle, Ithaca, NY 14850, USA. 3 Physical Electronics, 952-828-6100, 6509 Flying Cloud Drive, Eden Prairie, MN 55344, USA.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
Fig. 2. RBS spectra and fits for a forsterite Ža. target and Žb. film deposited at the high power density.
the relative compositions of the targets and films are of interest and therefore the differences in the compositions indicated by the two techniques are not addressed.
5. Results Table 2 shows the compositions indicated by XPS for multiple areas on a forsterite target sampled at the higher power density. The first column shows the
169
composition obtained from a fit to a single spectrum collected from a sampled area that was not sputtered prior to characterization. The second column shows the compositions obtained from fits to three spectra collected from unsampled areas adjacent to the sampled area. Since XPS is unable to reproducibly measure O in the targets, the O target values were set equal to the O target values indicated by RBS. The Mg, Si, and Fe values corresponding to the sampled area are within the ranges defined by the Mg, Si, and Fe values corresponding to the unsampled areas. This suggests that ablation was not accompanied by formation of a melt layer in which processes such as surface segregation played a significant role. Table 3Ža. and Fig. 3Ža. and Žb. show the compositions indicated by RBS for the targets and films. Each reported film composition was obtained from a fit to a spectrum and the uncertainty shown is the two-sigma uncertainty in the fit. For each of the minerals, variations in composition between targets were not significant in comparison to variations in composition within a target. Each reported target composition is the weighted average of three compositions obtained from fits to spectra collected from multiple targets and the uncertainty shown is the two-sigma uncertainty in the weighted average. Weighting factors were calculated using the twosigma uncertainties in the fits. Since RBS is unable to distinguish between Al and Si in the plagioclase feldspar films, total Al q Si values are shown for these films. Since the O film values agree with the O target values to within the sum of the uncertainties in the values, the O film values were set equal to the O target values in Fig. 3Ža. and Žb.. Table 3Žb. shows the compositions indicated by XPS for the targets and films. Each reported film composition was obtained from a fit to a spectrum
Table 2 Compositions in atomic percents as indicated by XPS for sampled and unsampled areas on a forsterite target sampled at the high power density Element
Sampled area
Unsampled areas
Mg Si Fe
26.57 14.20 2.03
23.97–27.00 13.79–15.85 2.01–2.99
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
170 Table 3
Ža. Compositions in atomic percents as indicated by RBS for the targets Župper. and the films deposited at the low Žcenter. and high Žlower. power densities Element
Acmite
Albite
Augite
Diopside
Forsterite
Labradorite
O
58.88 " 1.35 58.2 " 1.4 59.1 " 0.9 4.86 " 0.45 2.6 " 0.3 2.7 " 0.2 4.80 " 0.40 4.6 " 0.3 4.4 " 0.3 – – – 19.12 " 0.32 21.4 " 0.5 20.5 " 0.3 – – – 6.65 " 0.15 7.0 " 0.2 7.8 " 0.2 5.69 " 0.06 6.2 " 0.1 5.5 " 0.1
62.16 " 1.21 62.0 " 2.8 60.9 " 1.7 7.77 " 0.49 9.2 " 0.6 9.4 " 0.5 – – – 6.08 " 0.55 – – 23.59 " 0.51 – – – 28.7 " 1.0 29.6 " 0.8 0.40 " 0.07 0.1 " 0.1 0.1 " 0.1 – – –
59.61 " 1.30 60.2 " 2.3 57.6 " 1.0 – – – 7.61 " 0.40 7.6 " 0.5 7.6 " 0.2 – – – 19.15 " 0.30 19.5 " 0.7 20.3 " 0.3 – – – 9.13 " 0.17 9.4 " 0.3 10.8 " 0.2 4.50 " 0.06 3.3 " 0.1 3.7 " 0.1
59.84 " 1.30 60.4 " 2.3 58.2 " 1.4 – – – 10.05 " 0.41 8.9 " 0.5 9.6 " 0.4 – – – 19.87 " 0.35 19.1 " 0.7 19.6 " 0.5 – – – 9.11 " 0.13 10.8 " 0.4 11.8 " 0.3 1.13 " 0.06 0.8 " 0.1 0.8 " 0.1
57.12 " 1.19 57.4 " 1.6 57.3 " 1.6 – – – 25.73 " 0.42 24.3 " 0.8 24.5 " 0.8 – – – 13.95 " 0.29 15.0 " 0.5 14.8 " 0.5 – – – – – – 3.20 " 0.06 3.3 " 0.1 3.4 " 0.1
61.94 " 1.16 – 61.2 " 1.9 2.77 " 0.43 – 3.5 " 0.3 – – – 11.86 " 0.38 – – 18.39 " 0.38 – – – – 30.4 " 0.7 4.84 " 0.06 – 4.8 " 0.1 0.20 " 0.06 – 0.1 " 0.1
Na
Mg
Al
Si
Al q Si
Ca
Fe
Žb.: Compositions in atomic percents as indicated by XPS for the targets Župper. and the films deposited at the high power density Žlower. Element
Acmite
Albite
Augite
O
61.31 61.31 " 0.35 2.08 " 0.04 1.58 " 0.04 3.70 " 0.13 3.93 " 0.15 – – 20.22 " 0.23 20.41 " 0.27 7.49 " 0.12 8.72 " 0.15 5.20 " 0.05 4.05 " 0.05
61.53 61.53 " 0.31 3.71 " 0.05 8.88 " 0.09 – – 8.15 " 0.14 8.26 " 0.19 25.64 " 0.20 21.20 " 0.25 0.97 " 0.04 0.13 " 0.02 – –
62.16 62.16 " 0.32 – – 5.21 " 0.12 5.62 " 0.16 – – 20.77 " 0.20 19.45 " 0.24 8.07 " 0.11 10.24 " 0.15 3.79 " 0.03 2.53 " 0.04
Na Mg Al Si Ca Fe
and the uncertainty shown is the two-sigma uncertainty in the fit. Each reported target composition is the weighted average of five compositions obtained from fits to spectra collected from multiple areas on a target and the uncertainty shown is the two-sigma
uncertainty in the weighted average. Weighting factors were calculated using the two-sigma uncertainties in the fits. Since XPS is unable to reproducibly measure O in the targets, the O target values were set equal to the O film values.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
Fig. 3. Compositions in atomic percents as indicated by RBS for the films deposited at the Ža. low and Žb. high power densities. The values are displaced along the horizontal axes for display. Where no fractionation is observed the values lie on the indicator lines.
The RBS values in Table 3Ža. were used to calculate the ratios R 1 and R 2 R 1 s Ž F y T . rT
171
In a first approach, significant fractionation effects were defined as those effects for which < R 1 < ) 0.1. Both acmite films are depleted in Na and the film deposited at the higher power density is enriched in Ca. Both albite films are depleted in Ca and enriched in Na. Both augite films are depleted in Fe and the film deposited at the higher power density is enriched in Ca. Both diopside films are depleted in Fe and enriched in Ca. The forsterite films are neither depleted nor enriched in any of the elements. The labradorite film is depleted in Fe and enriched in Na. XPS confirms these fractionation effects in the films for which data is available. In a second approach, significant fractionation effects were defined as those effects for which < R 2 < ) 1.0. Both acmite films are depleted in Na and enriched in Si, the film deposited at the lower power density is enriched in Fe, and the film deposited at the higher power density is enriched in Ca. Both albite films are depleted in Ca and enriched in Na. Both augite films are depleted in Fe and the film deposited at the higher power density is enriched in Ca. Both diopside films are depleted in Fe and Mg and enriched in Ca and the film deposited at the higher power density is depleted in Si. Both forsterite films are enriched in Si, the film deposited at the lower power density is depleted in Mg, and the film deposited at the higher power density is enriched in Fe. The labradorite film is neither depleted nor enriched in any of the elements. XPS confirms these fractionation effects in the films for which data is available. The selected approaches suggest general fractionation trends. The clinopyroxene films are depleted in Na and Fe and enriched in Ca. The olivine films may be enriched in Si. The plagioclase feldspar films are depleted in Ca and enriched in Na and may be depleted in Fe.
R2 s Ž F y T . rŽ f q t . for Na, Mg, Si, Al q Si, Ca, and Fe. F is the film value, T is the target value, f is the uncertainty in the film value, and t is the uncertainty in the target value. Table 4 and Fig. 4Ža. and Žb. show the R 1 and R 2 values. As in Fig. 3Ža. and Žb., the O film values were set equal to the O target values. Negative R 1 and R 2 values indicate depletion effects and positive R 1 and R 2 values indicate enrichment effects.
6. Discussion Potential sources of fractionation include laser– target interactions, laser–vapor interactions, vapor– vapor interactions, and vapor–substrate interactions. A comprehensive investigation of these interactions
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
172
Table 4 R1 and R 2 values calculated from the RBS values in Table 3Ža. for films deposited at the low Župper. and high Žlower. power densities Element
Function
Acmite
Albite
Augite
Diopside
Forsterite
Labradorite
Na
R1
y0.47 y0.44 y3.07 y3.31 y0.06 y0.08 y0.39 y0.54 0.10 0.08 2.37 2.42 – – – – 0.04 0.18 0.69 3.40 0.07 y0.03 2.56 y1.00
0.18 0.17 1.28 1.37 – – – – – – – – y0.04 y0.03 y0.53 y0.56 y0.75 y0.75 y1.76 y1.76 – – – –
– – – – 0.01 y0.05 0.11 y0.63 0.03 0.01 0.63 0.32 – – – – 0.04 0.13 0.87 3.22 y0.26 y0.22 y7.19 y6.13
– – – – y0.10 y0.09 y1.11 y1.05 y0.03 y0.05 y0.47 y1.25 – – – – 0.20 0.24 3.41 5.31 y0.28 y0.32 y2.00 y2.25
– – – – y0.05 y0.04 y1.03 y0.92 0.08 0.07 1.46 1.15 – – – – – – – – 0.04 0.07 0.75 1.31
– 0.24 – 0.92 – – – – – – – – – y0.01 – y0.30 – y0.03 – y0.81 – y0.50 – y0.63
R2 Mg
R1 R2
Si
R1 R2
Al q Si
R1 R2
Ca
R1 R2
Fe
R1 R2
and their potential fractionation effects are beyond the scope of this study. A simple cation replacement mechanism was developed to try to predict significant fractionation effects in geological materials. Ablation occurs when the incident power density is greater than a threshold power density that is a function of several parameters, including laser wavelength and material composition. At smaller power densities, thermal and laser desorption can occur. Since the laser produces pulses with non-rectangular spatial and temporal profiles and the estimated sampling rates were on the order of a nanometer per pulse, a significant fraction of material was exposed to an incident power density smaller than the threshold prior to exposure to an incident power density greater than the threshold. Consequently, there was an opportunity for desorption to occur. Fractionation effects observed in previous studies have been attributed to thermal and laser conditioning of the target surface w7,12,21x. Similarly, the fractionation effects observed in this study may be due to a cation replacement mechanism triggered by
thermal desorption, laser desorption, or other processes. In the first step, species that are characterized by weak bonds are preferentially released from lattice sites. In the second step, species that are characterized by strong bonds are preferentially captured by empty lattice sites. Since desorbed species typically lack the energy and directionality of ablated species and are less likely to arrive at the substrate, the second step can potentially occur at the target surface, at the film surface, or at both surfaces. Silicate minerals are characterized by a tetrahedral structural unit comprised of four anion sites occupied by O 2y and a cation site w19,22x. The tetrahedral cation sites ŽT. are occupied by cations with 3 q or 4 q charge states w19,22x. The RBS values in Table 3Ža. suggest that the selected silicate targets contain two cations with 3 q charge states, Al 3q and Fe 3q, and one cation with a 4 q charge state, Si 4q. The cation-site set is the set of cations compatible with a cation site, ordered by decreasing compatibility. The accepted cation-site set for the T sites is Si 4q, Al 3q, Fe 3q 4 w22x.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
173
Fig. 4. R1 and R 2 values calculated from the RBS values in Table 3Ža. for films deposited at the Ža. low and Žb. high power densities. The symbols represent Na Žup triangle., Mg Žsquare., Si or Al q Si Žcircle., Ca Ždown triangle., and Fe Ždiamond. in acmite Žsolid., albite Žopen., augite Žhorizontal bar., diopside Žvertical bar., forsterite Žplus., and labradorite Žcross..
There are two additional cation sites in clinopyroxenes, a small site ŽP1 . with sixfold coordination and a large site ŽP2 . with eightfold coordination w19,22x. The T, P1 , and P2 sites occur in a 2:1:1 ratio. If the T sites are occupied by cations with 4 q charge states, the P1 and P2 sites are occupied by cations with 2 q charge states w19,22x. The RBS values in Table 3Ža. suggest that the selected clinopyroxene targets contain three cations with 2 q charge states, Mg 2q, Ca2q, and Fe 2q. The accepted cation-site set for the P1 and P2 sites is Mg 2q, Fe 2q, Ca2q 4 w22x. A slightly different cation-site set for the P2 sites, Ca2q, Mg 2q, Fe 2q 4 , was used in this
study. This change reflects a possible preference of the P2 sites for cations with large radii, i.e. Ca2q and Naq. The P2 sites in many natural clinopyroxenes are typically occupied by Ca2q and Naq w19,22x. There are two additional cation sites in olivines, sites ŽO 1 and O 2 . with sixfold coordination w19,22x. The T, O 1 , and O 2 sites occur in a 1:1:1 ratio. The O 1 and O 2 sites are occupied by cations with 2 q charge states w19,22x. The RBS values in Table 3Ža. suggest that the selected olivine target contains two cations with 2 q charge states, Mg 2q and Fe 2q. The accepted cation-site set for the O 1 and O 2 sites is Mg 2q, Fe 2q 4 w22x.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
174 Table 5
Ža. Estimated distributions of cations in the clinopyroxenes before Župper. and after Žlower. the cation replacement mechanism has taken effect Site type
Element
Acmite
Augite
Diopside
T
Si
46.50 50.00 3.50 0.00 11.67 13.26 2.99 0.00 10.34 11.74 11.82 0.00 0.00 0.00 13.18 25.00 0.00 0.00
47.41 50.00 2.59 0.00 17.20 17.20 0.00 0.00 7.80 7.80 0.00 0.00 1.64 0.00 22.61 25.00 0.75 0.00
49.48 50.00 0.52 0.00 22.90 22.90 0.00 0.00 2.10 2.10 0.00 0.00 2.12 0.00 22.69 25.00 0.19 0.00
Fe P1
Mg Ca Fe
P2
Na Mg Ca Fe
Žb. Estimated distributions of cations in the olivine before Župper. and after Žlower. the cation replacement mechanism has taken effect Site type
Element
Forsterite
T
Si
32.53 33.33 0.80 0.00 60.01 60.01 6.66 6.66
Fe O1 and O 2
Mg Fe
Žc. Estimated distributions of cations in the plagioclase feldspars before Župper. and after Žlower. the cation replacement mechanism has taken effect Site type
Element
Albite
Labradorite
T
Na
0.53 0.00 16.07 20.00 62.34 60.00 1.06 0.00 0.00 0.00 20.00 20.00 0.00 0.00
0.00 0.00 31.16 20.00 48.32 60.00 0.00 0.00 0.52 0.00 7.28 20.00 12.72 0.00
Al Si Ca Fe F
Na Ca
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
There is one additional cation site in plagioclase feldspars, a site ŽF. with six or sevenfold coordination w19,22x. The T and F sites occur in a 4:1 ratio. If the T sites are occupied by cations with 4 q charge states, the F sites are occupied by cations with 1 q charge states w19,22x. If the T sites are occupied by cations with 3 q charge states, the F sites are occupied by cations with 2 q charge states w19,22x. The RBS values in Table 3Ža. suggests that the selected plagioclase feldspar targets contain one cation with a 1 q charge state, Naq, and two cations with 2 q charge states, Ca2q and Fe 2q. The accepted cationsite set for the F sites is either Naq4 or Ca2q, Fe 2q 4 depending on the cations that occupy the T sites w22x. The placement of a cation in the cation-site set is a reflection of the relative bond strength between the cation and the site and can be used to predict fractionation effects due to a cation replacement mechanism triggered by thermal desorption, laser desorption, or other processes. In the selected silicates, the expected effects are replacement of Al 3q and Fe 3q by Si 4q in T sites, replacement of Ca2q by Mg 2q and Fe 2q in P1 sites, replacement of Mg 2q and Fe 2q by Ca2q in P2 sites, and replacement of Ca2q and Fe 2q by Naq in F sites. Since Mg 2q and Fe 2q have the same charge state and similar radii, significant replacement of Fe 2q by Mg 2q in P1 , P2 , O 1 , and O 2 sites is not expected to occur. If sites are occupied by cations not present in the corresponding cation-site sets, i.e. Naq in P2 sites in acmite, replacement of these cations by cations present in the corresponding cation-site sets is expected to occur. Table 5Ža., Žb., and Žc. show the estimated distributions of cations in sites for the selected silicates before and after the cation replacement has taken effect. The initial distributions were obtained using the RBS values for the targets in Table 3Ža.. Charge states of 1 q for Na, 2 q for Mg and Ca, 3 q for Al, and 4 q for Si were assumed. Since published analyses indicate that Fe 2q is much more common than Fe 3q in pyroxenes and olivines, a charge state of 3 q was assumed for the Fe required for complete occupancy of T sites and a charge state of 2 q was assumed for all other Fe w22x. Since Mg 2q and Fe 2q have the same charge state and similar radii, they were assumed to be indistinguishable and were assigned to sites according to their relative quantities.
175
All other cations were assigned to sites according to their placements in the cation-site sets. The final distributions for the clinopyroxenes and the olivine are assumed to be those which would occur if the cation replacement mechanism proceeded to completion. In the plagioclase feldspars, charge balance requires that complete replacement of Al 3q and Fe 3q by Si 4q in T sites be coupled to complete elimination of cations from F sites. Therefore, the final distributions for the plagioclase feldspars are assumed to be those which would occur if the cation replacement mechanism proceeded until the Si 4q to Al 3q ratio in T sites is at the maximum value consistent with complete occupancy of F sites. Tables 4 and 6 show R 1 values for Na, Mg, Al, Si, Al q Si, Ca, and Fe in the selected silicates. The experimental values, shown in Table 4, were calculated from the RBS values in Table 3Ža.. The theoretical values, shown in Table 6, were calculated from the cation distributions in Table 5Ža., Žb., and Žc.. In Table 7, the elements are ordered by increasing R 1 value for each mineral. The cation replacement mechanism correctly predicts Fe, Mg, Si, Ca4 for augite and diopside and Fe, Ca, Al q Si, Na4 for labradorite. For albite it correctly predicts Ca, Al q Si4 but incorrectly predicts the placement of Na4 . For forsterite it correctly predicts Mg, Si4 but incorrectly predicts the placement of Fe4 . For acmite it correctly predicts Na, Mg, Ca4 and Na, Fe, Si4 but incorrectly predicts the relative placements of Fe, Si4 and Mg, Ca4 . It should be noted that the placement of Fe4 is not consistent for the two forsterite films and that the placement of Ca4 is not consistent for the two acmite films. The incorrect predictions for forsterite and acmite should be considered in that context.
Table 6 R1 values calculated from the estimated cation distributions in Table 5Ža., Žb., and Žc. Element Acmite Albite Augite Diopside Forsterite Labradorite Na y1.00 y0.03 – – Mg 0.14 – y0.09 y0.08 Si 0.08 – 0.05 0.01 AlqSi – 0.02 – – Ca 0.55 y1.00 0.11 0.10 Fe y0.15 – y0.30 y0.25
– 0.00 0.02 – – y0.11
1.75 – – 0.01 y1.00 y1.00
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
176
Table 7 The elements in order of increasing R1 value. The experimental orders, calculated from the R1 values in Table 4, are for films deposited at the low Župper. and high Žlower. power densities. The theoretical orders were calculated from the R1 values in Table 6 Mineral
Theoretical order
Experimental order
Acmite
Na Fe Si Mg Ca
Albite
Ca Na AlqSi
Augite
Fe Mg Si Ca
Diopside
Fe Mg Si Ca
Forsterite
Fe Mg Si
Labradorite
Fe and Ca AlqSi Na
Na Mg Ca Fe Si Na Mg Fe Si Ca Ca AlqSi Na Ca AlqSi Na Fe Mg Si Ca Fe Mg Si Ca Fe Mg Si Ca Fe Mg Si Ca Mg Fe Si Mg Si Fe – Fe Ca AlqSi Na
The cation replacement mechanism proposed here is too simple to predict fractionation effects with complete accuracy. Replacement of Mg 2q and Fe 2q by Fe 3q and replacement of Fe 2q by Mg 2q in P1 , P2 , O1 , and O 2 sites may occur w22x. Fe 2q shows a slight preference for O 1 sites in some olivines, suggesting that replacement of Mg 2q by Fe 2q in O 1 sites may occur w22x. These effects and others were neglected for the sake of simplicity and for the lack of sufficient physical data.
7. Conclusions Acmite, albite, augite, diopside, forsterite, and labradorite are silicate minerals comprised of O, Na, Mg, Al, Si, Ca, and Fe. No O fractionation effects were observed. The clinopyroxene films are depleted in Na and Fe and enriched in Ca. The olivine films may be enriched in Si. The plagioclase feldspar films are depleted in Ca and enriched in Na and may be depleted in Fe. These effects may be due in part to a cation replacement mechanism. These minerals can be easily distinguished by their film compositions despite the observed fractionation effects. These results are of interest because they indicate that ultraviolet LAS is feasible for silicate minerals.
Acknowledgements This research was performed at the Center for Space Microelectronics Technology at JPL, a facility sponsored by NASA and located at the Jet Propulsion Laboratory ŽJPL., and at the Environmental Molecular Sciences Laboratory, a facility sponsored by DOE and located at the Pacific Northwest National Laboratory ŽPNNL.. It was supported by the JPL Director’s Research and Development Fund and by the NASA Cross Enterprise Technology Development Program. The authors gratefully acknowledge M.H. Engelhard ŽPNNL. for XPS characterization of the mineral samples and M.L. Alexander ŽPNNL., R.S. Kowalczyk ŽJPL., R.A. Beach ŽCalifornia Institute of Technology., J.O. McCaldin ŽCalifornia Institute of Technology., S. Nikzad ŽJPL. and K.S. Min ŽIntel. for their scientific and technical contributions.
References w1x J.C. Miller, R.F. Haglund, Laser Ablation and Desorption, Academic Press, New York, 1998. w2x D.B. Chrisey, G.K. Hubler, Pulsed Laser Deposition of Thin Films, Wiley, New York, 1994. w3x A. Vertes, R. Gijbels, F. Adams, Laser Ionization Mass Analysis, Wiley, New York, 1993. w4x L. Moenke-Blankenburg, Laser Microanalysis, Wiley, New York, 1989. w5x R. Kelly, A. Miotello, A. Mele, A. Giardini Guidoni, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 5. w6x D.B. Geohegan, Pulsed Laser Deposition of Thin Films, Wiley, New York, 1994, Chap. 5. w7x R.E. Russo, X.L. Mao, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 8. w8x I. Jarvis, K.E. Jarvis, Chem. Geol. 95 Ž1992. 1–33. w9x M. Schleberger, S. Speller, W. Heiland, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 6. w10x L.C. Feldman, J.W. Mayer, Fundamentals of Surface and Thin Film Analysis, Prentice-Hall, Englewood Cliffs, NJ, 1986. w11x W.K. Chu, J.W. Mayer, M.A. Nicolet, Backscattering Spectrometry, Academic Press, New York, 1978. w12x T.E. Jeffries, N.J.G. Pearce, W.T. Perkins, A. Raith, Anal. Commun. 33 Ž1996. 35–39. w13x K.E. Jarvis, J.G. Williams, Chem. Geol. 106 Ž1993. 251–262. w14x L. Moenke-Blankenburg, D. Gunther, Chem. Geol. 95 Ž1992. ¨ 85–92.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177 w15x M.L. Alexander, M.R. Smith, J.S. Hartman, A. Mendoza, D.W. Koppenaal, Appl. Surf. Sci. 127–129 Ž1998. 255–261. w16x D.J. Figg, J.B. Cross, C. Brink, Appl. Surf. Sci. 127–129 Ž1998. 287–291. w17x P. Goodall, S.G. Johnson, E. Wood, Spectrochim. Acta 50B Ž1995. 1823–1835. w18x H.Y. McSween Jr., Meteoritics 29 Ž1994. 757–779. w19x W.G. Ernst, Earth Materials, Prentice-Hall, Englewood Cliffs, NJ, 1969.
177
w20x C.W. Chesterman, National Audubon Society Field Guide to North American Rocks and Minerals, Chanticleer Press, New York, 1995. w21x Z. Ball, R. Sauerbrey, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 7. w22x W.A. Deer, R.A. Howie, J. Zussman, An Introduction to the Rock-Forming Minerals, Addison Wesley Longman, Essex, 1992.
Elemental fractionation in ultraviolet laser ablation sampling of igneous silicate minerals relevant to Mars M.E. Taylor ) , D.L. Blaney, G. Cardell Jet Propulsion Laboratory, MS 302-231, 4800 Oak GroÕe Dr., Pasadena, CA 91109, USA Received 23 March 2000; accepted 11 May 2000
Abstract Laser ablation sampling ŽLAS. has significant potential for remote terrestrial and extraterrestrial applications if fractionation can be understood and controlled. This study focuses on acmite, albite, augite, diopside, forsterite, and labradorite, silicate minerals that are relevant to Mars, Earth, the moon, and asteroids. The minerals were sampled using a frequency-quadrupled Nd:YAG laser and the samples were deposited as films on graphite substrates. Rutherford backscattering spectrometry and X-ray photoelectron spectroscopy measurements of the minerals and their films indicate sodium, magnesium, silicon, calcium, and iron fractionation. Oxygen fractionation was not indicated. A cation replacement mechanism is a potential source of fractionation. These minerals can be easily distinguished by their film compositions despite fractionation, indicating that ultraviolet laser ablation sampling is feasible for silicate minerals. q 2000 Elsevier Science B.V. All rights reserved. PACS: 91.65.Vj; 83.80.Nb; 42.62.Eh Keywords: Ablation; Fractionation; Stoichiometry; Silicate minerals; ICP-AES; ICP-MS
1. Introduction Laser ablation sampling ŽLAS. is a technique that is often used to obtain vapor or droplet samples of parent materials for subsequent chemical analysis w1–4x. Chemical information is obtained through either direct or indirect analysis of samples. Direct approaches include optical spectroscopy and timeof-flight mass spectrometry w5,6x. In some indirect approaches, i.e. inductively coupled plasma atomic
) Corresponding author. Tel.: q1-818-354-2662; fax: q1-818393-4663. E-mail address: [email protected] ŽM.E. Taylor..
emission spectrometry ŽICP-AES. and inductively coupled plasma mass spectrometry ŽICP-MS., samples are introduced into an additional excitation source w7,8x. In other indirect approaches, samples are deposited on a substrate, a process called pulsed laser deposition ŽPLD. w1,2x. The resulting films are analyzed by standard characterization techniques such as Rutherford backscattering spectrometry ŽRBS. and X-ray photoelectron spectroscopy ŽXPS. w9–11x. LAS is particularly feasible for geological materials. Sampling of a variety of geological materials with different compositions and microstructures has been demonstrated w12–14x. Extensive preparation of parent materials is not required. Sampling areas of a few tens of micrometers and sampling rates of a few
0169-4332r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 Ž 0 0 . 0 0 4 4 2 - 6
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
nanometers per pulse are not atypical, facilitating both spatial profiling and depth profiling. Fractionation in LAS has been studied for a variety of geological materials w12–14x. The majority of these studies used ICP-AES and ICP-MS techniques and focused on trace element fractionation. Fractionation was found to correlate with absorbed power density and to be more significant for infrared lasers than for ultraviolet lasers. Fractionation that is often attributed to LAS may be due to other sources. In ICP-AES and ICP-MS, for example, fractionation may occur during transport to the ICP or in the ICP. Recent ICP-MS measurements suggest that large particulates are often either lost during transport to the ICP or incompletely digested in the ICP and that this can result in fractionation w15–17x. LAS has significant potential for remote terrestrial and extraterrestrial applications if fractionation can be understood and controlled. This study was undertaken with that objective. A PLD approach rather than an ICP-AES or ICP-MS approach was selected to avoid potential fractionation related to the ICP.
2. Mineral selection and preparation The Shergottite–Nakhlite–Chassignite ŽSNC. meteorites are believed to have originated on Mars and are reported to consist largely of feldspars, pyroxenes, and olivines w18x. The Earth’s continental crust is reported to consist of 58% feldspars, 13% pyroxenes and amphiboles, and 3% olivines by volume w19x. These types of minerals are also relevant to the moon and asteroids. They are therefore of general interest for remote terrestrial and extraterrestrial applications. The plagioclase feldspars albite and labradorite, the clinopyroxenes acmite, augite, and diopside, and the olivine forsterite were selected for this study. Table 1 shows the nominal compositions of these minerals w20x. All minerals used in this study are research-grade natural specimens obtained commercially.1 Multiple specimens of each mineral were polished using sili-
1
Wards, 800-962-2660, P.O. Box 92912, Rochester, NY 14692-9012, USA.
167
Table 1 Silicate minerals selected for this study w20x Mineral
Classification
Nominal composition
Acmite Albite Augite Diopside Forsterite Labradorite
Clinopyroxene Plagioclase feldspar Clinopyroxene Clinopyroxene Olivine Plagioclase feldspar
NaFeSi 2 O6 NaAlSi 3 O 8 ŽCa,Na.ŽMg,Fe.Si 2 O6 CaMgSi 2 O6 Mg 2 SiO4 30% NaAlSi 3 O 8 , 70% CaAl 2 Si 2 O 8
con carbide films with average particulate diameters of 68, 22 and 14 mm and diamond films with average particulate diameters of 6, 3 and 1 mm.
3. Experiments Fig. 1 shows the sampling system. The laser is a pulsed frequency-quadrupled Nd:YAG laser with an internal energy attenuator. The wavelength is 266 nm, the pulse repetition rate is 15 Hz, the maximum pulse energy is 5 mJ, the pulse width is 3–5 ns, and the sampling area is approximately 0.4 mm2 . The mineral target and the graphite substrate are separated by approximately 2.5 cm and are housed in a vacuum chamber with a base pressure of 5 = 10y8 Torr. Prior to sampling, the average powers incident on the target and reflected from the target are measured using a thermopile meter. During sampling, the position of the target is adjusted at intervals of 4500 pulses to provide a fresh target area. Sampling is stopped when a film is visible on the substrate. Films were deposited from the mineral targets at nominal power densities of 100 and 320 MW cmy2 . These power densities were calculated assuming a pulse width of 4 ns and a sampling area of 0.4 mm2 . They varied slightly due to variations in target reflectivity potentially due to composition and surface roughness. The lower power density was selected because it is the minimum power density required to produce visible plumes in all of the minerals. No film was deposited for labradorite at this power density because the deposition rate was very low. The higher power density was selected because it is
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
168
Fig. 1. Sampling system.
the maximum power density easily achievable with the current laser, lens, and geometry.
4. Characterization In RBS, the sample is irradiated with monochromatic ions w9–11x. The ions undergo Coulomb repulsion collisions with the sample atoms and are backscattered. The energy loss of each backscattered ion is a function of the penetration depth of the ion and the mass of the sample atoms and is not a function of the chemical environment of the sample atoms. The sampling area is typically a few millimeters in diameter and the sampling depth is typically a micrometer. In XPS, the sample is irradiated with monochromatic X-rays w9,10x. The sample electrons absorb the X-rays and are ejected. The energy of each ejected electron is determined by the electron binding energy and is a function of the sample atoms and their chemical environment. The sampling area is typically a few tenths of a millimeter in diameter. The sampling depth is limited by the escape depth of the ejected electrons and is typically several nanometers. Because XPS has a small effective sampling depth, surface contamination may affect results. Samples are often exposed to an ion beam prior to characterization to remove surface contamination. RBS was selected as the primary characterization technique because large sampling area, large sampling depth, and insensitivity to chemical environment make RBS preferable to XPS for mean composition measurements. The RBS system used in this study is a National Electrostatics Corporation pel-
letron system. It uses a beam of 2-MeV He 2q ions. The sampling area is approximately 25 mm2 . Sampled targets were repolished prior to characterization. The RBS spectra were analyzed using the software package RUMP 2 Fig. 2Ža. and Žb. show examples of spectra and fits. Because of non-negligible backgrounds potentially due to embedded particulates, linear background subtractions were performed for the O and Na peaks in the albite and labradorite films and for the O, Mg, and Si peaks in the lower power density augite film. Approximate fits were obtained using the simulation feature. Best fits were obtained using the Poisson fitting feature in conjunction with the approximate fits. The XPS system used in this study is a Physical Electronics Quantum 2000 system. It uses a focused beam of 1.4866-keV Al K a X-rays for characterization and a beam of 2-keV Arq ions for removal of surface contamination. The spherical section analyzer was used in a constant pass energy mode with a pass energy of 46.95 eV and a step size of 0.4 eV. The sampling area is approximately 0.3 mm2 . Sampled targets were repolished prior to characterization. To remove surface contamination, 5 nm of material was sputtered from targets and films prior to characterization. The XPS spectra were analyzed using the software package MultiPak.3 The calibration feature was used to correct the energy and pass energy dependence of the analyzer. Shirley background subtractions were performed. Fits were obtained using the peak areas in conjunction with the O1s, Na1s, Mg2s, Al2p, Si2p, Ca2s, and Fe2p3 sensitivity factors. Both RBS and XPS have limitations. RBS is unable to distinguish between Al and Si in the film spectra because of their similar masses. XPS is unable to reproducibly measure O in the targets, possibly due to incomplete removal of surface contamination. The compositions indicated by RBS and XPS differ slightly. These deviations are not atypical and most likely occur because sensitivity varies with factors such as chemical environment. In this study, 2
Computer Graphic Service, 607-277-4913, 52 Genung Circle, Ithaca, NY 14850, USA. 3 Physical Electronics, 952-828-6100, 6509 Flying Cloud Drive, Eden Prairie, MN 55344, USA.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
Fig. 2. RBS spectra and fits for a forsterite Ža. target and Žb. film deposited at the high power density.
the relative compositions of the targets and films are of interest and therefore the differences in the compositions indicated by the two techniques are not addressed.
5. Results Table 2 shows the compositions indicated by XPS for multiple areas on a forsterite target sampled at the higher power density. The first column shows the
169
composition obtained from a fit to a single spectrum collected from a sampled area that was not sputtered prior to characterization. The second column shows the compositions obtained from fits to three spectra collected from unsampled areas adjacent to the sampled area. Since XPS is unable to reproducibly measure O in the targets, the O target values were set equal to the O target values indicated by RBS. The Mg, Si, and Fe values corresponding to the sampled area are within the ranges defined by the Mg, Si, and Fe values corresponding to the unsampled areas. This suggests that ablation was not accompanied by formation of a melt layer in which processes such as surface segregation played a significant role. Table 3Ža. and Fig. 3Ža. and Žb. show the compositions indicated by RBS for the targets and films. Each reported film composition was obtained from a fit to a spectrum and the uncertainty shown is the two-sigma uncertainty in the fit. For each of the minerals, variations in composition between targets were not significant in comparison to variations in composition within a target. Each reported target composition is the weighted average of three compositions obtained from fits to spectra collected from multiple targets and the uncertainty shown is the two-sigma uncertainty in the weighted average. Weighting factors were calculated using the twosigma uncertainties in the fits. Since RBS is unable to distinguish between Al and Si in the plagioclase feldspar films, total Al q Si values are shown for these films. Since the O film values agree with the O target values to within the sum of the uncertainties in the values, the O film values were set equal to the O target values in Fig. 3Ža. and Žb.. Table 3Žb. shows the compositions indicated by XPS for the targets and films. Each reported film composition was obtained from a fit to a spectrum
Table 2 Compositions in atomic percents as indicated by XPS for sampled and unsampled areas on a forsterite target sampled at the high power density Element
Sampled area
Unsampled areas
Mg Si Fe
26.57 14.20 2.03
23.97–27.00 13.79–15.85 2.01–2.99
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
170 Table 3
Ža. Compositions in atomic percents as indicated by RBS for the targets Župper. and the films deposited at the low Žcenter. and high Žlower. power densities Element
Acmite
Albite
Augite
Diopside
Forsterite
Labradorite
O
58.88 " 1.35 58.2 " 1.4 59.1 " 0.9 4.86 " 0.45 2.6 " 0.3 2.7 " 0.2 4.80 " 0.40 4.6 " 0.3 4.4 " 0.3 – – – 19.12 " 0.32 21.4 " 0.5 20.5 " 0.3 – – – 6.65 " 0.15 7.0 " 0.2 7.8 " 0.2 5.69 " 0.06 6.2 " 0.1 5.5 " 0.1
62.16 " 1.21 62.0 " 2.8 60.9 " 1.7 7.77 " 0.49 9.2 " 0.6 9.4 " 0.5 – – – 6.08 " 0.55 – – 23.59 " 0.51 – – – 28.7 " 1.0 29.6 " 0.8 0.40 " 0.07 0.1 " 0.1 0.1 " 0.1 – – –
59.61 " 1.30 60.2 " 2.3 57.6 " 1.0 – – – 7.61 " 0.40 7.6 " 0.5 7.6 " 0.2 – – – 19.15 " 0.30 19.5 " 0.7 20.3 " 0.3 – – – 9.13 " 0.17 9.4 " 0.3 10.8 " 0.2 4.50 " 0.06 3.3 " 0.1 3.7 " 0.1
59.84 " 1.30 60.4 " 2.3 58.2 " 1.4 – – – 10.05 " 0.41 8.9 " 0.5 9.6 " 0.4 – – – 19.87 " 0.35 19.1 " 0.7 19.6 " 0.5 – – – 9.11 " 0.13 10.8 " 0.4 11.8 " 0.3 1.13 " 0.06 0.8 " 0.1 0.8 " 0.1
57.12 " 1.19 57.4 " 1.6 57.3 " 1.6 – – – 25.73 " 0.42 24.3 " 0.8 24.5 " 0.8 – – – 13.95 " 0.29 15.0 " 0.5 14.8 " 0.5 – – – – – – 3.20 " 0.06 3.3 " 0.1 3.4 " 0.1
61.94 " 1.16 – 61.2 " 1.9 2.77 " 0.43 – 3.5 " 0.3 – – – 11.86 " 0.38 – – 18.39 " 0.38 – – – – 30.4 " 0.7 4.84 " 0.06 – 4.8 " 0.1 0.20 " 0.06 – 0.1 " 0.1
Na
Mg
Al
Si
Al q Si
Ca
Fe
Žb.: Compositions in atomic percents as indicated by XPS for the targets Župper. and the films deposited at the high power density Žlower. Element
Acmite
Albite
Augite
O
61.31 61.31 " 0.35 2.08 " 0.04 1.58 " 0.04 3.70 " 0.13 3.93 " 0.15 – – 20.22 " 0.23 20.41 " 0.27 7.49 " 0.12 8.72 " 0.15 5.20 " 0.05 4.05 " 0.05
61.53 61.53 " 0.31 3.71 " 0.05 8.88 " 0.09 – – 8.15 " 0.14 8.26 " 0.19 25.64 " 0.20 21.20 " 0.25 0.97 " 0.04 0.13 " 0.02 – –
62.16 62.16 " 0.32 – – 5.21 " 0.12 5.62 " 0.16 – – 20.77 " 0.20 19.45 " 0.24 8.07 " 0.11 10.24 " 0.15 3.79 " 0.03 2.53 " 0.04
Na Mg Al Si Ca Fe
and the uncertainty shown is the two-sigma uncertainty in the fit. Each reported target composition is the weighted average of five compositions obtained from fits to spectra collected from multiple areas on a target and the uncertainty shown is the two-sigma
uncertainty in the weighted average. Weighting factors were calculated using the two-sigma uncertainties in the fits. Since XPS is unable to reproducibly measure O in the targets, the O target values were set equal to the O film values.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
Fig. 3. Compositions in atomic percents as indicated by RBS for the films deposited at the Ža. low and Žb. high power densities. The values are displaced along the horizontal axes for display. Where no fractionation is observed the values lie on the indicator lines.
The RBS values in Table 3Ža. were used to calculate the ratios R 1 and R 2 R 1 s Ž F y T . rT
171
In a first approach, significant fractionation effects were defined as those effects for which < R 1 < ) 0.1. Both acmite films are depleted in Na and the film deposited at the higher power density is enriched in Ca. Both albite films are depleted in Ca and enriched in Na. Both augite films are depleted in Fe and the film deposited at the higher power density is enriched in Ca. Both diopside films are depleted in Fe and enriched in Ca. The forsterite films are neither depleted nor enriched in any of the elements. The labradorite film is depleted in Fe and enriched in Na. XPS confirms these fractionation effects in the films for which data is available. In a second approach, significant fractionation effects were defined as those effects for which < R 2 < ) 1.0. Both acmite films are depleted in Na and enriched in Si, the film deposited at the lower power density is enriched in Fe, and the film deposited at the higher power density is enriched in Ca. Both albite films are depleted in Ca and enriched in Na. Both augite films are depleted in Fe and the film deposited at the higher power density is enriched in Ca. Both diopside films are depleted in Fe and Mg and enriched in Ca and the film deposited at the higher power density is depleted in Si. Both forsterite films are enriched in Si, the film deposited at the lower power density is depleted in Mg, and the film deposited at the higher power density is enriched in Fe. The labradorite film is neither depleted nor enriched in any of the elements. XPS confirms these fractionation effects in the films for which data is available. The selected approaches suggest general fractionation trends. The clinopyroxene films are depleted in Na and Fe and enriched in Ca. The olivine films may be enriched in Si. The plagioclase feldspar films are depleted in Ca and enriched in Na and may be depleted in Fe.
R2 s Ž F y T . rŽ f q t . for Na, Mg, Si, Al q Si, Ca, and Fe. F is the film value, T is the target value, f is the uncertainty in the film value, and t is the uncertainty in the target value. Table 4 and Fig. 4Ža. and Žb. show the R 1 and R 2 values. As in Fig. 3Ža. and Žb., the O film values were set equal to the O target values. Negative R 1 and R 2 values indicate depletion effects and positive R 1 and R 2 values indicate enrichment effects.
6. Discussion Potential sources of fractionation include laser– target interactions, laser–vapor interactions, vapor– vapor interactions, and vapor–substrate interactions. A comprehensive investigation of these interactions
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
172
Table 4 R1 and R 2 values calculated from the RBS values in Table 3Ža. for films deposited at the low Župper. and high Žlower. power densities Element
Function
Acmite
Albite
Augite
Diopside
Forsterite
Labradorite
Na
R1
y0.47 y0.44 y3.07 y3.31 y0.06 y0.08 y0.39 y0.54 0.10 0.08 2.37 2.42 – – – – 0.04 0.18 0.69 3.40 0.07 y0.03 2.56 y1.00
0.18 0.17 1.28 1.37 – – – – – – – – y0.04 y0.03 y0.53 y0.56 y0.75 y0.75 y1.76 y1.76 – – – –
– – – – 0.01 y0.05 0.11 y0.63 0.03 0.01 0.63 0.32 – – – – 0.04 0.13 0.87 3.22 y0.26 y0.22 y7.19 y6.13
– – – – y0.10 y0.09 y1.11 y1.05 y0.03 y0.05 y0.47 y1.25 – – – – 0.20 0.24 3.41 5.31 y0.28 y0.32 y2.00 y2.25
– – – – y0.05 y0.04 y1.03 y0.92 0.08 0.07 1.46 1.15 – – – – – – – – 0.04 0.07 0.75 1.31
– 0.24 – 0.92 – – – – – – – – – y0.01 – y0.30 – y0.03 – y0.81 – y0.50 – y0.63
R2 Mg
R1 R2
Si
R1 R2
Al q Si
R1 R2
Ca
R1 R2
Fe
R1 R2
and their potential fractionation effects are beyond the scope of this study. A simple cation replacement mechanism was developed to try to predict significant fractionation effects in geological materials. Ablation occurs when the incident power density is greater than a threshold power density that is a function of several parameters, including laser wavelength and material composition. At smaller power densities, thermal and laser desorption can occur. Since the laser produces pulses with non-rectangular spatial and temporal profiles and the estimated sampling rates were on the order of a nanometer per pulse, a significant fraction of material was exposed to an incident power density smaller than the threshold prior to exposure to an incident power density greater than the threshold. Consequently, there was an opportunity for desorption to occur. Fractionation effects observed in previous studies have been attributed to thermal and laser conditioning of the target surface w7,12,21x. Similarly, the fractionation effects observed in this study may be due to a cation replacement mechanism triggered by
thermal desorption, laser desorption, or other processes. In the first step, species that are characterized by weak bonds are preferentially released from lattice sites. In the second step, species that are characterized by strong bonds are preferentially captured by empty lattice sites. Since desorbed species typically lack the energy and directionality of ablated species and are less likely to arrive at the substrate, the second step can potentially occur at the target surface, at the film surface, or at both surfaces. Silicate minerals are characterized by a tetrahedral structural unit comprised of four anion sites occupied by O 2y and a cation site w19,22x. The tetrahedral cation sites ŽT. are occupied by cations with 3 q or 4 q charge states w19,22x. The RBS values in Table 3Ža. suggest that the selected silicate targets contain two cations with 3 q charge states, Al 3q and Fe 3q, and one cation with a 4 q charge state, Si 4q. The cation-site set is the set of cations compatible with a cation site, ordered by decreasing compatibility. The accepted cation-site set for the T sites is Si 4q, Al 3q, Fe 3q 4 w22x.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
173
Fig. 4. R1 and R 2 values calculated from the RBS values in Table 3Ža. for films deposited at the Ža. low and Žb. high power densities. The symbols represent Na Žup triangle., Mg Žsquare., Si or Al q Si Žcircle., Ca Ždown triangle., and Fe Ždiamond. in acmite Žsolid., albite Žopen., augite Žhorizontal bar., diopside Žvertical bar., forsterite Žplus., and labradorite Žcross..
There are two additional cation sites in clinopyroxenes, a small site ŽP1 . with sixfold coordination and a large site ŽP2 . with eightfold coordination w19,22x. The T, P1 , and P2 sites occur in a 2:1:1 ratio. If the T sites are occupied by cations with 4 q charge states, the P1 and P2 sites are occupied by cations with 2 q charge states w19,22x. The RBS values in Table 3Ža. suggest that the selected clinopyroxene targets contain three cations with 2 q charge states, Mg 2q, Ca2q, and Fe 2q. The accepted cation-site set for the P1 and P2 sites is Mg 2q, Fe 2q, Ca2q 4 w22x. A slightly different cation-site set for the P2 sites, Ca2q, Mg 2q, Fe 2q 4 , was used in this
study. This change reflects a possible preference of the P2 sites for cations with large radii, i.e. Ca2q and Naq. The P2 sites in many natural clinopyroxenes are typically occupied by Ca2q and Naq w19,22x. There are two additional cation sites in olivines, sites ŽO 1 and O 2 . with sixfold coordination w19,22x. The T, O 1 , and O 2 sites occur in a 1:1:1 ratio. The O 1 and O 2 sites are occupied by cations with 2 q charge states w19,22x. The RBS values in Table 3Ža. suggest that the selected olivine target contains two cations with 2 q charge states, Mg 2q and Fe 2q. The accepted cation-site set for the O 1 and O 2 sites is Mg 2q, Fe 2q 4 w22x.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
174 Table 5
Ža. Estimated distributions of cations in the clinopyroxenes before Župper. and after Žlower. the cation replacement mechanism has taken effect Site type
Element
Acmite
Augite
Diopside
T
Si
46.50 50.00 3.50 0.00 11.67 13.26 2.99 0.00 10.34 11.74 11.82 0.00 0.00 0.00 13.18 25.00 0.00 0.00
47.41 50.00 2.59 0.00 17.20 17.20 0.00 0.00 7.80 7.80 0.00 0.00 1.64 0.00 22.61 25.00 0.75 0.00
49.48 50.00 0.52 0.00 22.90 22.90 0.00 0.00 2.10 2.10 0.00 0.00 2.12 0.00 22.69 25.00 0.19 0.00
Fe P1
Mg Ca Fe
P2
Na Mg Ca Fe
Žb. Estimated distributions of cations in the olivine before Župper. and after Žlower. the cation replacement mechanism has taken effect Site type
Element
Forsterite
T
Si
32.53 33.33 0.80 0.00 60.01 60.01 6.66 6.66
Fe O1 and O 2
Mg Fe
Žc. Estimated distributions of cations in the plagioclase feldspars before Župper. and after Žlower. the cation replacement mechanism has taken effect Site type
Element
Albite
Labradorite
T
Na
0.53 0.00 16.07 20.00 62.34 60.00 1.06 0.00 0.00 0.00 20.00 20.00 0.00 0.00
0.00 0.00 31.16 20.00 48.32 60.00 0.00 0.00 0.52 0.00 7.28 20.00 12.72 0.00
Al Si Ca Fe F
Na Ca
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
There is one additional cation site in plagioclase feldspars, a site ŽF. with six or sevenfold coordination w19,22x. The T and F sites occur in a 4:1 ratio. If the T sites are occupied by cations with 4 q charge states, the F sites are occupied by cations with 1 q charge states w19,22x. If the T sites are occupied by cations with 3 q charge states, the F sites are occupied by cations with 2 q charge states w19,22x. The RBS values in Table 3Ža. suggests that the selected plagioclase feldspar targets contain one cation with a 1 q charge state, Naq, and two cations with 2 q charge states, Ca2q and Fe 2q. The accepted cationsite set for the F sites is either Naq4 or Ca2q, Fe 2q 4 depending on the cations that occupy the T sites w22x. The placement of a cation in the cation-site set is a reflection of the relative bond strength between the cation and the site and can be used to predict fractionation effects due to a cation replacement mechanism triggered by thermal desorption, laser desorption, or other processes. In the selected silicates, the expected effects are replacement of Al 3q and Fe 3q by Si 4q in T sites, replacement of Ca2q by Mg 2q and Fe 2q in P1 sites, replacement of Mg 2q and Fe 2q by Ca2q in P2 sites, and replacement of Ca2q and Fe 2q by Naq in F sites. Since Mg 2q and Fe 2q have the same charge state and similar radii, significant replacement of Fe 2q by Mg 2q in P1 , P2 , O 1 , and O 2 sites is not expected to occur. If sites are occupied by cations not present in the corresponding cation-site sets, i.e. Naq in P2 sites in acmite, replacement of these cations by cations present in the corresponding cation-site sets is expected to occur. Table 5Ža., Žb., and Žc. show the estimated distributions of cations in sites for the selected silicates before and after the cation replacement has taken effect. The initial distributions were obtained using the RBS values for the targets in Table 3Ža.. Charge states of 1 q for Na, 2 q for Mg and Ca, 3 q for Al, and 4 q for Si were assumed. Since published analyses indicate that Fe 2q is much more common than Fe 3q in pyroxenes and olivines, a charge state of 3 q was assumed for the Fe required for complete occupancy of T sites and a charge state of 2 q was assumed for all other Fe w22x. Since Mg 2q and Fe 2q have the same charge state and similar radii, they were assumed to be indistinguishable and were assigned to sites according to their relative quantities.
175
All other cations were assigned to sites according to their placements in the cation-site sets. The final distributions for the clinopyroxenes and the olivine are assumed to be those which would occur if the cation replacement mechanism proceeded to completion. In the plagioclase feldspars, charge balance requires that complete replacement of Al 3q and Fe 3q by Si 4q in T sites be coupled to complete elimination of cations from F sites. Therefore, the final distributions for the plagioclase feldspars are assumed to be those which would occur if the cation replacement mechanism proceeded until the Si 4q to Al 3q ratio in T sites is at the maximum value consistent with complete occupancy of F sites. Tables 4 and 6 show R 1 values for Na, Mg, Al, Si, Al q Si, Ca, and Fe in the selected silicates. The experimental values, shown in Table 4, were calculated from the RBS values in Table 3Ža.. The theoretical values, shown in Table 6, were calculated from the cation distributions in Table 5Ža., Žb., and Žc.. In Table 7, the elements are ordered by increasing R 1 value for each mineral. The cation replacement mechanism correctly predicts Fe, Mg, Si, Ca4 for augite and diopside and Fe, Ca, Al q Si, Na4 for labradorite. For albite it correctly predicts Ca, Al q Si4 but incorrectly predicts the placement of Na4 . For forsterite it correctly predicts Mg, Si4 but incorrectly predicts the placement of Fe4 . For acmite it correctly predicts Na, Mg, Ca4 and Na, Fe, Si4 but incorrectly predicts the relative placements of Fe, Si4 and Mg, Ca4 . It should be noted that the placement of Fe4 is not consistent for the two forsterite films and that the placement of Ca4 is not consistent for the two acmite films. The incorrect predictions for forsterite and acmite should be considered in that context.
Table 6 R1 values calculated from the estimated cation distributions in Table 5Ža., Žb., and Žc. Element Acmite Albite Augite Diopside Forsterite Labradorite Na y1.00 y0.03 – – Mg 0.14 – y0.09 y0.08 Si 0.08 – 0.05 0.01 AlqSi – 0.02 – – Ca 0.55 y1.00 0.11 0.10 Fe y0.15 – y0.30 y0.25
– 0.00 0.02 – – y0.11
1.75 – – 0.01 y1.00 y1.00
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177
176
Table 7 The elements in order of increasing R1 value. The experimental orders, calculated from the R1 values in Table 4, are for films deposited at the low Župper. and high Žlower. power densities. The theoretical orders were calculated from the R1 values in Table 6 Mineral
Theoretical order
Experimental order
Acmite
Na Fe Si Mg Ca
Albite
Ca Na AlqSi
Augite
Fe Mg Si Ca
Diopside
Fe Mg Si Ca
Forsterite
Fe Mg Si
Labradorite
Fe and Ca AlqSi Na
Na Mg Ca Fe Si Na Mg Fe Si Ca Ca AlqSi Na Ca AlqSi Na Fe Mg Si Ca Fe Mg Si Ca Fe Mg Si Ca Fe Mg Si Ca Mg Fe Si Mg Si Fe – Fe Ca AlqSi Na
The cation replacement mechanism proposed here is too simple to predict fractionation effects with complete accuracy. Replacement of Mg 2q and Fe 2q by Fe 3q and replacement of Fe 2q by Mg 2q in P1 , P2 , O1 , and O 2 sites may occur w22x. Fe 2q shows a slight preference for O 1 sites in some olivines, suggesting that replacement of Mg 2q by Fe 2q in O 1 sites may occur w22x. These effects and others were neglected for the sake of simplicity and for the lack of sufficient physical data.
7. Conclusions Acmite, albite, augite, diopside, forsterite, and labradorite are silicate minerals comprised of O, Na, Mg, Al, Si, Ca, and Fe. No O fractionation effects were observed. The clinopyroxene films are depleted in Na and Fe and enriched in Ca. The olivine films may be enriched in Si. The plagioclase feldspar films are depleted in Ca and enriched in Na and may be depleted in Fe. These effects may be due in part to a cation replacement mechanism. These minerals can be easily distinguished by their film compositions despite the observed fractionation effects. These results are of interest because they indicate that ultraviolet LAS is feasible for silicate minerals.
Acknowledgements This research was performed at the Center for Space Microelectronics Technology at JPL, a facility sponsored by NASA and located at the Jet Propulsion Laboratory ŽJPL., and at the Environmental Molecular Sciences Laboratory, a facility sponsored by DOE and located at the Pacific Northwest National Laboratory ŽPNNL.. It was supported by the JPL Director’s Research and Development Fund and by the NASA Cross Enterprise Technology Development Program. The authors gratefully acknowledge M.H. Engelhard ŽPNNL. for XPS characterization of the mineral samples and M.L. Alexander ŽPNNL., R.S. Kowalczyk ŽJPL., R.A. Beach ŽCalifornia Institute of Technology., J.O. McCaldin ŽCalifornia Institute of Technology., S. Nikzad ŽJPL. and K.S. Min ŽIntel. for their scientific and technical contributions.
References w1x J.C. Miller, R.F. Haglund, Laser Ablation and Desorption, Academic Press, New York, 1998. w2x D.B. Chrisey, G.K. Hubler, Pulsed Laser Deposition of Thin Films, Wiley, New York, 1994. w3x A. Vertes, R. Gijbels, F. Adams, Laser Ionization Mass Analysis, Wiley, New York, 1993. w4x L. Moenke-Blankenburg, Laser Microanalysis, Wiley, New York, 1989. w5x R. Kelly, A. Miotello, A. Mele, A. Giardini Guidoni, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 5. w6x D.B. Geohegan, Pulsed Laser Deposition of Thin Films, Wiley, New York, 1994, Chap. 5. w7x R.E. Russo, X.L. Mao, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 8. w8x I. Jarvis, K.E. Jarvis, Chem. Geol. 95 Ž1992. 1–33. w9x M. Schleberger, S. Speller, W. Heiland, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 6. w10x L.C. Feldman, J.W. Mayer, Fundamentals of Surface and Thin Film Analysis, Prentice-Hall, Englewood Cliffs, NJ, 1986. w11x W.K. Chu, J.W. Mayer, M.A. Nicolet, Backscattering Spectrometry, Academic Press, New York, 1978. w12x T.E. Jeffries, N.J.G. Pearce, W.T. Perkins, A. Raith, Anal. Commun. 33 Ž1996. 35–39. w13x K.E. Jarvis, J.G. Williams, Chem. Geol. 106 Ž1993. 251–262. w14x L. Moenke-Blankenburg, D. Gunther, Chem. Geol. 95 Ž1992. ¨ 85–92.
M.E. Taylor et al.r Applied Surface Science 165 (2000) 166–177 w15x M.L. Alexander, M.R. Smith, J.S. Hartman, A. Mendoza, D.W. Koppenaal, Appl. Surf. Sci. 127–129 Ž1998. 255–261. w16x D.J. Figg, J.B. Cross, C. Brink, Appl. Surf. Sci. 127–129 Ž1998. 287–291. w17x P. Goodall, S.G. Johnson, E. Wood, Spectrochim. Acta 50B Ž1995. 1823–1835. w18x H.Y. McSween Jr., Meteoritics 29 Ž1994. 757–779. w19x W.G. Ernst, Earth Materials, Prentice-Hall, Englewood Cliffs, NJ, 1969.
177
w20x C.W. Chesterman, National Audubon Society Field Guide to North American Rocks and Minerals, Chanticleer Press, New York, 1995. w21x Z. Ball, R. Sauerbrey, Laser Ablation and Desorption, Academic Press, New York, 1998, Chap. 7. w22x W.A. Deer, R.A. Howie, J. Zussman, An Introduction to the Rock-Forming Minerals, Addison Wesley Longman, Essex, 1992.