Hydrogen microdetermination in geological materials using elastic recoil detection analysis (ERDA)

340

Nuclear

Instruments

and Methods

HYDROGEN MICRODETERMINATION IN GEOLOGICAL RECOIL DETECTION ANALYSlS (ERDA) M. MOSBAH “, J. TIRIRA”,

R. CLOCCHIATTI

in Physics

MATERIALS

Research

B49 (1990) 340-344 North-Holland

USING ELASTIC

I), J. GOSSET ‘) and P. MASSIOT

”

Ii IRDI/DCAEA/SEAIN. .” CNRS/Groupe

Groupe de Microunu!rse Nuclcawe da Sciences de la Terre. Lrthorrutorrr Pmre Sue. CEA. CEN Sucfqv. !JI 191’ G!f Sur Yvette Cede.

France

For the first time. hydrogen has been determined (by means of nuclear microanalysis) within glassy mclusions trapped m volcanic minerals. Elastic recoil detection of protons has been used by means of the transmission and reflection geometry. The two major originalities of this work consist in the quantitative absolute method and the application of this technique to natural materials. Some of our results are close to those obtained with other methods. We reached a detection limit < 50 pg/g within the San Carlos &vine.

1. Introduction

2. Basic principles L I. Theoretical

Volatile species (HzO. COZ, F, Cl, N. etc.) play an important role in governing the nature and size of explosive volcanic eruptions; they contribute notably to the atmosphere of our planet. Particularly. the knowledge of the volatile element contents dissolved in magma is fundamental for the establishment of the formation conditions of volcanic melts and crystallization of minerals (eruptive products). The lavas lose their gases when they overflow at the surface, but fortunately dissolved gases can be preserved within glasses (or silicate melt inclusions) trapped during crystal growth [1,2] in magma chamber conditions (HT - 800-1200 o C, HP 0.5-500 MPa). This is why the study of melt inclusions is interesting. Nuclear microdeterminations of C and N were performed in natural glasses and volcanic minerals, leading to reliable data which agree with previously published works [3,4]. Hydrogen takes a preponderant place, but up to now, it has been difficult to characterize its presence and its distribution in these materials. Recent works [5,6] exhibit ion microprobe results on this type of materials. Because of the size of melt inclusions (between 50 and 300 pm width), a microanalytical tool is absolutely necessary. The nuclear microprobe consists in the interaction between a charged particle microbeam (of energy < 4 MeV) and the target to analyze. The hydrogen depth profiling by elastic recoil detection has been developed in recent papers [7,8]. In this paper, the quantitative absolute method to determine the hydrogen density based on elastic recoil spectrometry in transmission and reflection geometries is schematically discussed. Then our objective is to apply this method to natural materials like geological samples. 016~-5~3X/90/$03.~0 (North-Holland)

8~ Eisevier Science Publishers

B.V.

comiderotions

Generally. the binary collision is a long studied problem. In fact, the energy-angle dependence of scattered particles in elastic collision is fully described by the following equation [9] (fig. 1): E, = E,4M,

M2 co&/(

M, + M2 )‘,

(1)

with E,, the incident energy, M, and M, the respective masses of the projectile ( 4He) and the recoil nucleus ( 'H), B the recoil angle and E, the recoil nucleus energy. The collision geometry is fully characterized by the relation between the scattering angle Q and the recoil angle B according to the conservation principle of energy and impulsion [lo]:

RECOIL NUCLEUS

=f

Mz

MI

=

Ec

----

6’

,.

‘-7

0

)-_--______-

-._,

-“\‘i #

INCIDENT 4th

\

\ Fig. 1. Schematic

description

!

of the 4 He/’

SCATTERED PARTICLE

H elastic colliston.

M. Mosbah et 01. / Hydrqen microdetermination

b

Proton detector

kinematic factor, S and .S, the respective stopping power for incident ions and recoil protons, calculated from Ziegler [ll], a( 19. E(x)) is the proton elastic cross section [lo]. and G is the geometric factor. The recoil protons issued from a depth x are not monoenergetic but emerge with an energy distribution F(E). We assume that this function can be described as a Gaussian, where the energy spreading (FWHM) results from the contribution of several effects: straggling, multiple scattering in the target and in the absorber foil, and resolution of the detector system. 2.3. Simulation

Fig. 2. Elastic recoil detection in transmission mode. (b) Elastic recoil detection in reflection geometry.

As M, > M, ( 4He/‘H collision), + + 0 5 90 O. In this way, two experimental configurations are possible for the detection of the recoil protons: _ For t9 = 0 transmission geometry (fig. 2a). the energy of the recoil proton is maximum and a slow variation of B around this position has a very slight influence on the recoil proton energy values. _ Another experimental configuration can be chosen for the glancing angle 0 - with respect to the incident beam - (fig. 2b) where 0t can be varied between 0 o and 90”. In the case of reflection geometry. a Mylar absorber foil of sufficient thickness, to stop the scattered helium is placed in front of the recoil detector.

2.2. Interpretation

of recoil spectrum

The interpretation of the energy proton spectrum can be made from H( E,)6E, recoil yield, or the number of recoil protons detected around the 13direction in the solid angle s2 and in the energy range E, over the energy interval SE,. H(E,) can thus be written as the following convolution product (eq. (3)) [7.8]:

H(K) = ,:f

AfJQN(x)o(@,

G+

E(x))

S(E(x))

K’sr(K,E(x))

x F( E’, E,) dE’ d-u, with N(x) the hydrogen atomic density, of incident ions, t the analyzed depth,

1 (3)

Q the number K’ the recoil

code

Energy loss and energy spreading determinations and the final integration required accurate algorithms and complex numerical treatment [7,8]. The computer program “GABY” had to be made in order to automate the different processes. Thus the whole thickness of the sample is divided into homogeneous layers of equal width (typically 100 nm for transmission configuration and 5 nm for reflection configuration) where the average particle energy is assumed to be constant. The recoil cross section and stopping power are calculated. All the processes which contribute to the total energy spreading are independently estimated and then quadratically combined to obtain the total energy resolution of the system. Finally, the simulation code allows the experimental spectrum to be reconstituted and the experimental data to be interpreted.

3. However, a suitable compromise between the incident energy and the sample thickness has to be found. For example. in the transmission mode. thin targets (for Z < 30, E, = 3 Mev, thickness = 30 pm) are required. whereas for thick samples the reflection geometry is more appropriate.

341

Experimental

3. I.

The nuclear microprobe

The measurements are conducted in the analysis chamber of the French AEC nuclear microprobe. The 4He beam is delivered by a 4 MV Van de Graaff. The beam line is equipped with an adjustable object aperture and a magnetic quadruplet (Harwell system) [12]. The vacuum is maintained around 2 X 10e6 Torr using a turbomolecular pump so that surface contamination of the samples can be reduced. The typical beam area used is 100 urn* and the respective current is 1 nA. The number of incident helium ions is monitored using a current chopper placed just before the quadrupoles. In the case of transmission geometry, the recoil protons are recorded at 0 o with a surface barrier detector (200 mm*. 100 pm) covered by a 2 mm collimator. In reflection geometry, the recoil angle is fixed at 25 O after a prealignment of the target surface and the detector position in the scattering plane with respect to the incident beam axis using an optical system of alignment. V. GEOLOGICAL/ARCHAEOLOGICAL

APPLICATIONS

For both geometries. the distance between the focussing plane and the detector is around 90 mm and the solid angle AS2 depends on the beam spot size (for 10 pm. A0 is around 1.48 + 0.02 x 10-s sr). The incident 4He energy varies between 1.8 and 3 MeV.

I

Anaiysed req,on

(

3.2. Sample preparation

We select a thin Kapton foil (C22H,,0,N2) of 25 pm manufactured by DuPont de Nemours to test the spectrum reconstitution. It is a homogeneous sample of which hydrogen content is already known. The surface is covered with a (105 i: 5 nm) gold layer to enhance thermal and electric charge transport. The choice of the geometry depends on the sample mo~hology: for the trans~ssion geometry the samples were thinned down to about 25 pm by Al,O, polishing (0.3 km) and then a 80 + 5 nm gold layer was deposited on the surface; in the case of reflection geometry, the thickness of the samples is around 2 mm, and the surface was polished with AI,O, (without any gold layer). Two volcanic glasses have been selected according to their geodynamic environment: (1) the first glass from Pantelleria Island (distension zone) is relatively poor in water (4 to 5%); (2) the second glass from Guadeloupe (subduction zone) has an Hz0 content around 7 to 8% (table 1). The composition of the glasses was determined by electron microprobe analysis.

4. Resultsand discussion For the thin Kapton film, the analyzed depth is 6 pm. the hydrogen atomic density is 2.5 X 1O22 atoms/

Fig. 3. Elastic recoil detection on Kapton in transmission geometry (solid curve: simulated spectrum, dots: experimental spectrum. E,, = 3 MeV, i = 1 nA, beam diameter = 100 pm, f = 900 Sk

ems, which agrees well with the polymer stoichiometry (fig. 3). Figs. 4 and 5 illustrate the typical recoil spectrum, respectively, obtained from the microprobe examination of a 150 km melt inclusion trapped in a Guadeloupe quartz and in a Pantelleria quartz. in similar experimental conditions. From the experimental spectrum and the best reconstitution obtained using the simulation code, the average hydrogen atomic density for the different inclusions (Guadeloupe and Pantelleria) is compiled in table 2, both with the H,O content, if we consider that hydrogen is engaged in this molecule. The H,O contents obtained by nuclear microanalysis confirm the results deduced by the deficit in the chemical analysis conducted with the electron microprobe: - H,O content is 9.3 it 0.8 wt.% in the Guadeloupe

r

I

Table 1 Composition ent origins

of the glasses (wt.%) trapped

Pantelleria

in quartz

GUADELOUPE

of differ-

Guadeloupe 1131

SiO, TiO, A’ ~0, Fe0 MgG Na,O Rz.0 Cl CaO Total Deficit in the chemical analysis

68.9 ~0.8 0.34 & 0.06 7.4 kO.2 1.3 +0.3 0.05 + 0.02 6.4 +0.3 4.2 io.09 0.98 + 0.06 0.33 + 0.06 96.039

4%

71.9 10.6 0.05 * 0.09 11.5 50.2 1.6 kO.2 0.07 + 0.08 4.01+0.13 2.07+0.12 0.31 f 0.01 1.5 kO.13 93.02

7%

ENERGY

(KC+‘)

Fig. 4. Guadeloupe glass inclusion (Pumice fall): Transmission ERDA spectrum recorded from a glass inclusion, trapped in a Guadeloupe quartz bombarded with a 3 MeV ‘He microbeam (i = 1 nA. beam diameter = 100 pm, t = 1800 s).

M. Mosbah et al. / Hvdrogen

microdetermination

1.7 x lo22 atoms/cm3

343

with an analyzed

depth

of 0.4

pm.

ENERGY

An olivine mineral does not contain hydrogen engaged; the calculated low limit detection is around 50 ppm in this material. We note that H contents obtained in transmission and reflection geometries agree well with the expected value in the confidence space. In contrast, the total analyzed depth depends both on incident energy and sample thickness (transmission geometry - table 2) and is greater than what can be achieved in reflection geometry (0.4 urn).

(KeYI

Fig. 5. Pantelleria glass inclusion (Mountain Grande Pumice fall): Transmission ERDA spectrum recorded from a glass inclusion trapped in a Pantelleria quartz bombarded with a 3 MeV 4He microbeam (i = 1 n4, beam diameter = 100 pm. t = 1800 s).

glass

and

consider

5 + 0.5 wt.% in Pantelleria glass, that hydrogen is engaged in the Hz0

if we

molecule. _ For the Pantelleria and the Guadeloupe inclusions, the comparison between the experimental spectrum and the simulated spectrum permits one to deduce a homogeneous hydrogen distribution. The average hydrogen atomic density agrees with the ion microprobe results previously published on glasses of similar origin [5]. In the case of reflection geometry, the microbeam examination of a 150 urn Guadeloupe inclusion leads to an average atomic density of hydrogen evaluated to:

Table 2 Atomic hydrogen density, water contents and analyzed depth in Pantelleria and Guadeloupe inclusions by means of transmission geometry Origin

Guadeloupe

Average Pantelleria

Average

Atomic H content [10z2 atoms/ cm’]

Water content (wt.%)

1.41 rf-0.13 1.45+0.13 1.89kO.17 1.6 f0.14 1.48+0.13 1.28+0.12 1.31 f0.12 1.49kO.13

8.8*0.8 9.OkO.8 11.8+1.1 10.0+0.9 9.2*0.8 8.OkO.7 8.2kO.7 9.3 + 0.8

1.9 1.8 1.8 2.3 2.3 1.4 1.4

A A B C D E E

0.85 f 0.08 0.57 f 0.05 1.09*0.9 0.84 k 0.07

5.1+0.5 3.4kO.3 6.6 kO.6 5.0+0.5

1.7 1.5 2.2

A A C

Analyzed depth

Glass type a’

(pm)

a) A. B, C, D, E correspond to the different glass types.

5. Conclusion The formalism described here provides an absolute and quantitative method to determine the hydrogen distribution in the near-surface region of solids. In order to carry out absolute measurement, certain conditions must be met: (1) the relative quantities of the other components of the target must be known: (2) the size of the target surface has to be large enough to accommodate the beam spot broadening in glancing incidence, moreover in transmission geometry, it is possible to analyze an area no bigger than the incident beam spot size; (3) the surface roughness is not such a critical parameter in the transmission geometry. The absolute character of this method permits one to study complex samples for which a standard matrix is difficult to obtain. In contrast to reflection geometry, transmission geometry is supple and most of all, less caution is needed for the alignment of the sample surface with respect to the incident direction. Our tool does not give any information about the H speciation (H,O or OH-). Raman spectroscopy and IR spectrometry are more adapted for this type of investigation [14]. This method is fully adapted to the study of hydrogen distribution in geological materials (minerals, glasses) and must be developed in other fields (biology, medicine, etc.).

References [l] E. Roedder, Rev. Mineralogy 12 (1984) 1. [2] R. Clocchiatti, MCmoires de la SocietC Geologique de France 122 (1975) 1. [3] M. Mosbah, Rapport CEA-R-5456 (1988) 1. [4] N. Metrich and M. Mosbah, Bull. Mineral. 111 (1988) 511. [5] V.I. Kovalenko, R.L. Hervig and M.F. Sheridan, Am. Mineralogist 73 (1988) 1038. V. GEOLOGICAL/ARCHAEOLOGICAL

APPLICATIONS

[6] R.L. Hetvig. N. Dunbar. H.R. Westrich and P.R. Kyle, J. Volcanology and Geothermal Res. 36 (1989) 293. [7] J. Tirira, P. Trocellier, J.P. Frontier. and P Trouslard. Proc. 9th Ion Beam Analysis Conf., Kingston (Ontario) (June 26-30. 1989) Nucl. Instr. and Meth. B45 (1990) 203. [S] J. Tirira. P. Trocellier and J.P. Frontier, Proc. 9th Ion Beam Analysis Conf.. Kingston (Ontario) (June 26630. 1989) Nucl. Instr. and Meth. B45 (1990) 147. [9] A. Turos and 0. Meyer, Nucl. Instr. and Meth. B4 (1984) 92.

[lo] J. Tirira. P. Trocellier. J. Radioanal. and Nucl. Chem. Articles 130 (2) (1989) 311. [ll] J.F. Ziegler. C. Wu. P. Williams et al.. Nucl. Ins&. and Meth. 149 (1978) 19. [12] C. Engelmann and J. Bardy. Nucl. Instr. and Meth. 218 (1983) 209. [13] N. Metrich. C.R. Acad. Sci. Paris, vol. 307. ser. llp (1988) 1887. [14] E. Stolper. Contrih. Mineral. Petrol 81 (1982) 1.