Nuclear Instruments and Methods in Physics Research B 227 (2005) 591–596

www.elsevier.com/locate/nimb

Elastic recoil detection analysis of hydrogen with 7Li ionsusing a polyimide foil as a thick hydrogen reference

Primoz Pelicon, Alenka Razpet *, Sabina Markelj, Iztok Cadez, Milos Budnar

Department for Low and Medium Energy Physics, Jozef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia

Received 27 July 2004; received in revised form 17 September 2004

Abstract

Elastic recoil detection analysis (ERDA) with an absorber foil using a 4.2MeV 7Li2+ beam was utilized for evalu-

ation of hydrogen depth profiles. Since recoil cross-sections when using Li ions as projectiles are not well known,

the energy dependent ratio between the experimental yield and the yield calculated using the Rutherford recoil

cross-section was obtained from an ERDA spectrum of a thick polyimide (Kapton) sample. It was estimated that this

ratio does not significantly depend on sample composition. Therefore it was used for correction of measured spectra

analyzed by existing simulation and evaluation programs in which the Rutherford recoil cross-sections were applied.

The correction procedure has been verified in round-robin measurements of well-characterized Si:H thin layers. Appli-

cation of the method for determination of a hydrogen depth concentration profile in hydrogen-containing graphite sam-

ples is presented.

� 2004 Elsevier B.V. All rights reserved.

PACS: 82.80.Yc; 68.55.Nq; 06.90.+v; 85.75.DdKeywords: Thin films; Ion beam analysis; Elastic recoil detection analysis; Recoil cross-section

1. Introduction

Elastic recoil detection analysis (ERDA) with an

absorber foil is a well established technique formeasuring hydrogen depth profiles in samples [1].

Conventionally, helium ions are used as projectiles.

0168-583X/$ - see front matter � 2004 Elsevier B.V. All rights reserv

doi:10.1016/j.nimb.2004.10.005

* Corresponding author. Tel./fax: +386 1 5612 335.

E-mail addresses: primoz.pelicon@ijs.si (P. Pelicon), alenka.

razpet@ijs.si (A. Razpet).

The possibility of employing lithium ions has

recently been explored to achieve better depth

resolution and enhance separation of hydrogen iso-

topes [2]. Besides, lithium ions can easily be ob-tained using sputtering negative ion sources

available at most tandem accelerator facilities.

The major drawbacks of hydrogen-ERDA with a

Li beam, as pointed out by Mayer et al. [2], is the

lack of recoil cross-section data and increased

background in the spectra due to nuclear reactions.

ed.

592 P. Pelicon et al. / Nucl. Instr. and Meth. in Phys. Res. B 227 (2005) 591–596

In this work, we present a quantitative hydro-

gen depth analysis using polyimide (Kapton) as a

reference material for evaluation of hydrogen-

ERDA spectra measured with a 4.2MeV 7Li

beam. The accuracy of the procedure has beenconfirmed in ‘‘Hydrogen in Silicon’’ round-robin

measurements, organized by Bundesanstalt fur

Materialforschung und -prufung (BAM), Berlin,

where Si:H thin films were distributed for hydro-

gen analysis [3].

The applicability of the method was demon-

strated for a graphite sample from the test limiter

used in TEXTOR, a research tokamak at IPP-FZ Julich [4]. This study was initiated in order to

determine precisely the depth profiles of hydrogen

in typical tokamak material, graphite, that is fre-

quently used as a plasma facing material.

Fig. 1. RBS-ERDA set-up used in this work.

2. Experimental

Measurements were performed at the 2MV

Tandem accelerator at the Jozef Stefan Institute

(JSI) in Ljubljana. Lithium ions were obtained

from the sputtering source using a mixture of

LiOH and Ag powder as a target. The energy of

the incident beam was 4224keV for all the mea-

surements presented in this paper. The beam was

collimated by a 1 · 4mm rectangular shaping slitplaced in front of the entrance of the experimental

chamber which was equipped with two conven-

tional silicon detectors; the RBS detector at the

scattering angle h = 150� and the ERDA one at

the recoil angle u = 30� (Fig. 1). The incident beamangle a and the exit angle b as measured from the

normal to the sample surface were both 75�. An

11lm thick aluminum absorber foil was insertedin front of the ERDA detector to separate hydro-

gen recoils hitting the ERDA detector from scat-

tered Li ions.

A mesh charge collector set behind the shaping

slit was used to measure the number of ions hitting

the target [5]. It consisted of a tungsten mesh of

80% transmission placed between two cylindrical

electrodes at a negative voltage (�800V) to sup-press the secondary electron current from the

mesh. The normalization parameter of the mea-

sured spectra, i.e. the number N of ions hitting

the sample multiplied by the detector solid angle

XRBS, is proportional to the charge q collected by

the mesh charge collector:

NXERDA ¼ Rq: ð1ÞThe coefficient R was calibrated by measuring the

RBS yield from thin standard targets.

The ratio of RBS and ERDA detector solid an-

gles XRBS/XERDA, which in our case was equal to

0.82, was determined by an 241Am source posi-

tioned on the target wheel. The spectra of alphaparticles were accumulated simultaneously until

the statistical uncertainty in the solid angle ratio

XRBS/XERDA was lower than 1%. The normaliza-

tion parameter NXERDA needed for quantitative

analysis of spectra was calculated from the accu-

mulated charge on the mesh collector using rela-

tion (1).

3. Results and discussion

The energy spectrum of hydrogen recoils from a

polyimide film measured by the ERDA detector is

displayed and compared to a simulated spectrum

in Fig. 2. The calculation was done using the SIM-

0 10 20 30 40 50 60 70 80 900.0

0.5

1.0

1.5

2.0

Y/YRuth

Channels

Fig. 3. Correction factor F = Y/YRuth as a function of the

channel number in the ERDA spectrum.

0 20 40 60 80 100 120 1400

200

400

600

800

1000

1200

Yiel

d

Channels

experimentalcalculation ( Ruth)σ

Fig. 2. Experimental (thick solid line) and calculated (thin solid

line) ERDA spectra of hydrogen recoils from polyimide

(Kapton) sample (energy per channels is 9.15keV).

P. Pelicon et al. / Nucl. Instr. and Meth. in Phys. Res. B 227 (2005) 591–596 593

NRA code [6] with an experimentally determined

value of NXERDA, Rutherford recoil cross-sectionsrRuth(E) and correction factors for the stopping

powers in polyimide [7]. The computed spectrum

differs from the measured scattering yield since

the actual differential recoil cross-section r(E) is

non-Rutherford. A discrepancy between the exper-

imental and Rutherford values was also observed

when using He projectiles. For energies of He

above 2MeV the measured cross-sections wereabout twice as high as Rutherford predictions [1].

There was a hump present in the measured spec-

trum of polyimide between channels 10 and 40 (en-

ergy �300keV). To investigate the origin of this

hump, several thick homogenous samples contain-

ing hydrogen, such as SiN:H, were analyzed. The

shape of the recoil spectra matched that of the

hydrogen spectrum of polyimide, therefore it canbe concluded that the origin of the hump is a res-

onance in the recoil cross-section.

In order to use available simulation programs

based on the Rutherford recoil cross-sections, the

yield Y(EH) at the energy EH of detected recoils

was divided by the corresponding correction factor

F(EH) to compensate for the non-Rutherford val-

ues. The factor F(EH) was determined as the ratiobetween the measured yield in the spectrum of

polyimide and the calculated yield assuming Ruth-

erford cross-sections (Fig. 3).

Since the correction factor F was calculated at

the recoil energy EH, which depends on the stop-

ping power, rather than at the energy of Li ions

at the recoil event, the question arises if the estima-

tion of F for polyimide can be applied to other

materials. The energy of a hydrogen recoil hitting

the detector is

EH ¼ E1KERDA �Z 0

x1

SHm ðEÞ

dxcos b

�Z d

0

SHAlðEÞdx;

ð2Þ

where SHm ðEÞ is the stopping power in the material

m for hydrogen and SHAlðEÞ the stopping power for

recoils in the absorber foil of thickness d. The

energy E1 of the lithium when it knocks out ahydrogen atom at depth x1 is determined by the

equation

E1 ¼ E0 �Z x1

0

SLim ðEÞ

dxcos a

; ð3Þ

where E0 is the initial energy of the incident ionand SLi

m ðEÞ the stopping power for Li in m.

The dependence of the recoil energy before

entering the absorber foil Eout in front of the detec-

tor has been computed for the geometry used in

our set-up for polyimide, carbon and molybdenum

taking into account Ziegler–Biersack stopping

data [7]. These data for Li ions with initial energy

of 4MeV are shown in Fig. 4.

1000 2000 3000 4000

200

400

600

800

1000

1200

E out

[keV

]

E1

[ keV ]

in carbonin molibdenin kapton

Fig. 4. Dependence of the energy of recoils before entering the

absorber foil Eout on the energy of Li ions in the recoil event E1

for different target materials.

0 10 20 30 40 50 60

0.5

1.0

1.5

2.0

18

20

22

24

26

Si:H on Si (BAM 99) [3]alumooxynitride film, ITN S5 [8]polyimide (kapton)

tota

l H c

onc

[1017

at./c

m2 ]

H a

tom

ic c

once

ntra

tion

in k

apto

n [%

]

Cumulative dose [1013 ions/cm2]

Fig. 5. Measured hydrogen concentration by ERDA as a

function of cumulative incident dose of 7Li ions for polyimide

(right axis) and two other targets (left axis).

594 P. Pelicon et al. / Nucl. Instr. and Meth. in Phys. Res. B 227 (2005) 591–596

For simplicity, it was assumed that the hydro-

gen content in carbon and molybdenum sample

does not influence the stopping power. From the

results shown in Fig. 4 it can be seen that the dif-ferences between Eout for different materials at a

constant E1 are smaller than 20keV, which is a

value comparable to the detector resolution. The

three materials taken into consideration in Fig. 4

actually cover a wide range of stopping powers.

Therefore, it can be concluded that the correction

factor F(EH) calculated for polyimide can be used

for other samples as well. The correction factorF(EH) is a fair approximation of the ratio of the re-

coil cross-sections r/rRuth.

In order to assure that polyimide can be used as

a reference for ERDA analysis, the stability of this

material under irradiation with the 7Li beam was

investigated. ERDA spectra of polyimide were col-

lected in several consecutive steps at the same spot

on the sample. The yield in a selected energy inter-val, corresponding to the hydrogen located at

depths from zero to 500nm in the polyimide, was

measured as a function of cumulative ion dose.

The results representing the hydrogen loss in the

upper 500nm layer during irradiation of the sam-

ples by 4.2MeV 7Li ions, is given in Fig. 5 together

with the loss from two other thin film samples

[3,8]. From Fig. 5 it can be seen that the hydrogencontent in the uppermost 500nm of the polyimide

foil decreases with cumulative dose. However, dur-

ing a typical spectrum acquisition time, when

the sample is exposed to a dose of less than

1 · 1014 ions/cm2, the change in hydrogen concen-

tration is within the experimental error. To accu-

mulate hydrogen spectra of polyimide for

determination of the correction function F(EH),

we therefore suggest measuring a series of spectrawith doses preferably below 5 · 1013 ions/cm2 at

several initially unexposed spots on the polyimide

foil. As also shown in Fig. 5, hydrogen loss from

the two thin film samples was not observed for

typical doses required to acquire hydrogen spectra

with good statistics.

The evaluation procedure applying Kapton

correction to measured recoil yields has beentested in ‘‘Hydrogen in Silicon’’ round-robin

measurements, organized by Bundesanstalt fur

Materialforschung und -prufung (BAM), Berlin.

Amorphous Si:H thin films on silicon were distrib-

uted for hydrogen analysis [3]. Fig. 6(a) shows the

measured spectrum of a Si:H sample labelled 99.

The full line represents the best simulation by

SIMNRA where the target was modelled as a filmconsisting of three layers and the dotted line repre-

sents single layer simulation. It can be seen that the

three-layer model simulation matches the experi-

mental values better. This suggests that the hydro-

gen concentration in the film is not homogeneous.

Four concentration profiles are plotted in Fig.

6(b): (i) the result of the three-layer simulation,

20 40 60 80 100 120

0

50

100

150

200

Cou

nts

Channels

MeasurementThree-layer simulationSingle layer simulation

0 50 100 150 200 250

0.00

0.02

0.04

0.06

0.08

0.10

0.12

Hco

ncen

tratio

n[a

t.%]

Depth [nm]

Li-ERDA. three-layer modelLi-ERDA. one-layer modelBAM reference valueRound-robin average value

(b)(a)

Fig. 6. (a) Measured hydrogen-ERDA spectrum of Si:H thin film, labelled 99 and selected as a potential thin film hydrogen standard

within the round-robin ‘‘Hydrogen in Silicon’’. Measured energy spectrum of a standard sample compared to a model spectrum

calculated by SIMNRA [6] and ‘‘Kapton correction’’. (b) Hydrogen concentration profiles obtained from the single and three layer

models compared to the BAM reference value and the average value of the round-robin.

P. Pelicon et al. / Nucl. Instr. and Meth. in Phys. Res. B 227 (2005) 591–596 595

(ii) the result of the single-layer simulation, (iii)

BAM reference value and (iv) the round-robin

average value.One of the motivations for this work was to

study the interaction of a hydrogen plasma with

solid surfaces. The hydrogen depth profile mea-

sured for the graphite sample from the test limiter

of the TEXTOR tokamak [4] that was exposed to

a hydrogen plasma for many hours is shown in

Fig. 7. The energy spectrum was also converted

into a depth concentration profile by an iterative

0 1000 2000 3000 4000 5000 6000 70000

1

2

3

Hco

ncen

tratio

n[a

t.%

]

depth [1015at./cm2]

iter. algorithm [9]SIMNRA

Fig. 7. Hydrogen depth profiles in a graphite sample exposed to

a hydrogen plasma as calculated by SIMNRA and a depth

conversion algorithm from [9].

algorithm originally written for conversion of

ERDA multielemental spectra [9]. In this calcula-

tion the sample is divided into several layers ofthickness Dxi. An energy interval DEi correspond-

ing to Dxi is calculated in the first order approxi-

mation using stopping power values of the initial

element composition. New concentration values

are then obtained by comparison of the calculated

yield from the layer and the measured yield in the

corresponding energy interval. The results of the

iterative method were compared to SIMNRA val-ues obtained from the four-layer simulation. It can

be concluded that the iterative algorithm gives

comparable results and can be used when fast

analysis of data is needed.

4. Conclusions

Experimental hydrogen elastic recoil spectra

obtained using a 4.2MeV Li beam cannot be eval-

uated directly by existing simulation codes since

hydrogen recoil cross-sections for ERDA with7Li ions at the relevant energies and angles are

not available in the existing data bases [6,10]. As

shown in the case of polyimide they deviate from

the Rutherford values. We estimated that the cor-rection factor which follows from the measured

ERDA spectrum on polyimide (Kapton) is not

596 P. Pelicon et al. / Nucl. Instr. and Meth. in Phys. Res. B 227 (2005) 591–596

strongly dependent on the sample composition.

Therefore, the values obtained for a thick bulk ref-

erence sample such as polyimide can serve for con-

version of energy spectra into hydrogen depth

profiles. The accuracy of the depth scale is limitedto the reliability of the stopping power data, simi-

larly as for other heavy ion ERDA techniques.

Our results confirm the predictions of Mayer

et al. [2] that a Li beam is a suitable alternative

to a He beam for hydrogen-ERDA. The method,

with a carefully performed normalization proce-

dure, yielded very accurate results in the recent

round-robin measurements on thin hydrogenlayers in silicon, initiated by BAM Berlin [3].

Further experiments with samples also containing

deuterium are planed in order to be able to inves-

tigate the adsorption of hydrogen isotopes in

fusion relevant materials.

Acknowledgements

The authors would like to thank U. Reinholtz

and P. Weise from Bundesanstalt fur Materialfors-

chung und -prufung (BAM), Berlin. We are in-

debted to I. Bogdanovic Radovic for discussions

about recoil cross-sections. Part of this research

was supported by EURATOM, contract no.

FU06-CT-2003-00010.

References

[1] J. Tirira, Y. Serruys, P. Trocellier, Forward Recoil

Spectrometry, Plenum Press, New York, 1996.

[2] M. Mayer, J. Roth, K. Ertl, Nucl. Instr. and Meth. B 190

(2002) 405.

[3] U. Reinholtz, H.P. Weise, Bundesanstalt fur Materialfors-

chung und -prufung (BAM), Berlin, round robin test

‘‘Hydrogen in Silicon’’, results sent to the participants of

the round-robin, unpublished.

[4] http://www.fz-juelich.de/ipp/textor_en.

[5] M. El Bouanani, University of North Texas, in press.

[6] M. Mayer, Tech. Rep. IPP 9/113, Max-Planck-Institut fur

Plasmaphysik, Garching, 1997.

[7] J.F. Ziegler, J.R. Biersack, U. Littmark, The Stopping and

Ranges of Ions in Matter, Vol. 1, Plenum Press, New York,

1985, see also http://www.srim.org/.

[8] N.P. Barradas et al., Nucl. Instr. and Meth. B, submitted

for publication.

[9] C. Spaeth, F. Richter, S. Grigull, U. Kreissig, Nucl. Instr.

and Meth. B 140 (1998) 243.

[10] Sigmabase, Available at http://www.physics.isu.edu/sigma-

base/data/.