Hydrogenated tantalum: A convenient calibration substance for hydrogen profile analysis using nuclear resonance reactions

Nuclear Instruments and Methods North-Holland, Amsterdam

HYDROGENATED FOR HYDROGEN B. HJijRVARSSON,

in Physics

Research

B42 (1989) 257-263

257

TANTALUM: A CONVENIENT CALIBRATION SUBSTANCE PROFILE ANALYSIS USING NUCLEAR RESONANCE REACTIONS J. RYDkN,

T. ERICSSON

and E. KARLSSON

Department of Physics, Uppsala University, Box 530, S-751 21 Uppsala, Sweden Received

2 May 1988 and in revised form 30 January

1989

The preparation of tantalum-hydride samples for calibration purposes is described. By comparing the average atomic H/Ta-ratios from profiling, using the 6.385 MeV resonance of the reaction ‘H(“N, ay)‘*C, with those obtained by weighing, the long term stability of the composition of the hydride is verified. These calibration samples are shown to be UHV-compatible, i.e. resistant to heat treatment at 420 K in vacuum for an extended time. Furthermore, the y-yields of the resonances at 13.35 MeV and 6.385 MeV are compared.

1. Introduction Nuclear resonance reactions provide powerful nondestructive ways of studying the hydrogen depth distribution in a sample. Using the 6.385 MeV resonance of the reaction ‘H(15N, ay)“C [1,2], good spatial resolution can be obtained close to the surface due to the narrow resonance width of 1.8 keV. To calculate the hydrogen concentration in a sample the following formula is used: C=(h’-BAt)

dE/dxF(O,

[ atoms/cm3 ]

r,da(Eres,

0))/Q, (I)

where N, B At d E/dx

= the number of gammas detected during the time At, = the background contribution per time unit, normally 0.1-0.2 counts/s, = the time for measuring one point (typically of the order 60-180 s), = the specific energy loss at the resonance

energy, = the total number of particles hitting the sample during the time interval At, D = the solid angle of the y-detector, E = the overall efficiency of the y-detection system, 8) = the differential cross-section for the reacWErew tion. Thus, the function F includes the differential cross section for the reaction and the overall efficiency of the detection system, including the solid angle of the detector. To use calibration standards is preferable to the use of absolute cross-section data, since one avoids sys-

Q

0168-583X/89/$03.50 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

tematic errors arising from uncertainties in the crosssection data, experimental geometry, detector efficiency and angular dependence of the yield. F is replaced by a calibration constant, K, which is determined by making a measurement on a sample with known hydrogen concentration, C,, . K=&,Q/((N,-BAt)

dE/dx).

(2)

Using eq. (2) involves one crucial simplification, the off-resonance contribution is ignored. In section 3 we will discuss the error induced by this simplification. For the time being three categories of samples are proposed for calibration purposes: (A) chemical compounds containing hydrogen in stoichiometric proportions, e.g. plastics and other organic compounds [3,4]; (B) hydrogen implanted targets, e.g. silicon [5]; (C) metal hydrides [3]. In an earlier article [3], tantalum hydride was suggested as one such appropriate candidate. In this paper the previous work has been followed up and a simple method of preparing stable calibration samples is described. It is also shown that these samples are UHVcompatible by simulating a typical bake-out of a UHVsystem.

2. Experimental details 2. I. Sample preparation Polycrystalline tantalum foils (purity > 99.95%) of the size 10 x 10 x 0.5 mm, from Johnson Matthey Chemicals were annealed in a vacuum chamber at a temperature of 1100 K and a pressure of 5 x lo-’

B. Hjiirvarsson et al. / Hydrogen tantalum calibration standard

258

Table 1 The average concentration for the depth range 300-1000 nm for the samples Al-Al0 and Bl-B5. The concentrations measured for Al and A5 are also included after annealing at 420 K for 48 h, and for B3, B4 and BS after pohsbing. The uncertainty in the values obtained from weighing is f0.006 Sample

Wfa weighed

H/Ta profile

Al A2 A3 A4 A5 A6 Al A8 A9 A10 Bl B2 B3 B4 B5 Al annealed A5 annealed B3 polished B4 polished B5 polished

0.474 0.443 0.467 0.482 0.483 0.441 0.473 0.407 0.450 0.471 0.525 0.505 0.492 0.507 0.477

0.46(3) 0.45(3) 0.42(3) 0.48(3) 0.48(3) 0.49(3) 0.45(3) 0.34(3) OM3) O/%(3) 0.47(2) 0.46(2) 0.45(2) 0.45(2) 0.45(2) 0.45(3) 0.49(3) 0.52(2) OSl(2) 0.52(2)

mbar. Subsequen~y the samples were hydrogenated at 620 K for 24 hours at a hydrogen pressure of 900 mbar (The hydrogen was of 99.9997% purity). The cooling was carried out at constant pressure, reaching room temperature in approximately 10 min. Weighing before and after the hydrogenation gave the average H/Taratio in the different samples (see table 1). The diffusion length, I, was estimated by assuming a moderate concentration dependence of the diffusion rate (using D,, = 4.4 X 10v4 cm2/s and E,,, = 0.140 eV [6]), for 0 I H/Ta zc 0.5, and using the formulae: D = De exp( --E.&G),

(3)

I = (2I)t)“Z. (4) The resulting diffusion length over 24 hours at 620 K is 2 cm which is more than an order of magnitude larger than the sample thickness. Two sets of samples, which we refer to as Al-Al0 and Bl-B5, were prepared as described above. After the first series of profile measurements two of the samples (Al and A5) were heated in vacuum to 420 K at a base pressure of 1 x lo-’ mbar for 48 h, followed by cooling to room temperature before exposure to air. To investigate the effect of the surface roughness three samples (B3, B4 and 85) were polished after the first profile m~urements.

2.2. Profiling using the N-15 method The profiling measurements were made at the Tandem Accelerator Laboratory in Uppsala. The maximum terminal voltage is 6 MW which makes it possible to scan up to 18 MeV using 15N2+ ions. The analysing chamber is pumped with a cryo-pump giving a working pressure in the 10m9 mbar region, without baking. The 4.43 MeV gamma rays are detected in a NaI detector (12.5 x 15 cm) placed at a distance of 3 cm from the sample at an angle of O” with respect to the direction of the incident beam. The total count rate is about 12 counts/s for a beam current of 35 nA on TaH,,. The beam current is integrated in the chamber, which is electrically insulated from the beam line. This gives the total number of ions hitting the sample (Q in formula 1). Due to good thermal contact to the sample holder, no significant macroscopic increase of the sample temperature is believed to occur during analysis. The diameter of the beam at the surface of the sample is 2 mm and therefore with a typical current of 20-50 nA, a significant energy transfer occurs in the small volume penetrated by the nitrogen ions. No evidence has so far been observed for a phase transition to occur as a consequence of local heating by the beam, therefore we assume that the local temperature, at the beam spot, is the same as the macroscopic temperature of the sample. For calibration purposes we used a calibration sample, here referred to as Cal, prepared five years ago according to the description above [3].

3. Rem&s and discussion Weighing the sample before and after hydrogenation gives a reliable value of the bulk H/Ta-ratio. As the studied foils were 0.5 mm thick, the exclusion of surface corrections introduces an estimated error in the H/Taratio of less than O.l%, consequently polishing is not required for accurate determination of the hydrogen content by weighing. The atomic H/Ta ratios obtained both by weighing and profiling are given in table 1. To calculate the wn~ntration the following values were used: eH = 4.185 X lo-‘*

MeV cm2

(stopping cross section for hydrogen [7]), z ra = 6.100 X 1O-‘9 MeV cm2 (stopping cross section for tantalum [ 7]), PTaH(0.47)

=

15q3

@;icm3

([31).

Using Braggs rule [S] and the values above, the specific energy loss for TaH,,, at a beam energy of 6.5 MeV, was calculated. dE,/dx

(TaHo,4,) = 3.20 X lo3 MeV/cm

B. HjLhwswn

et al. / Hydrogen tantalum calibration standard

specificenergy loss for the TaH0,,7 samples increases by 6% as the energy increases frum 6.5 to 13.0 MeV. As a consequence, the calculated depths at an energy Qf 13 h%ev are shifted by 6%. Since a_naccurate d~t~~a~u~ of the depth distribution of the hydrogen in the bulk is not of interesk we do not correct for this. The hydrogen contents of the samples correspond to the c phase of the Ta-II system at room temperature, with the exception of A8 which is in the o--c: phase [Q]. This is the most probable explanatmn for the observed deviation of the measured concentration Qf sample A8 from that expected. The hydrogen distribution in this sample is inhomogenwus causing a significant deviation in the H/Ta-ratios as given in table 1, In a study of TaH, (0 s x I I) Westlake et al. IlO] observed microThe

cracks as a result of the a-@ phase tram&ion. No such microcracI~ were observed for the W-T phase transition, When exposed to air a surface Iayer of Ta,O, is formed, which is extreme& @able [II] and prevents hydrogen from diffusing through the surface+ A concentration profile obtaiued from B3 and cal is shown in fig. 1. The high y-yield normalIy obtained at the surface is due to water adsorbed during exposure to the atmosphere (not shown in the figure). The surface peak is useful for the energy calibration of the analysing magnet (using the resonance at 13.35 MeV gives an extra point for that purpose). The low comoMration region, from the surface to a depth of about 10 nm, is due to the tantalum pentoxide which should not contain any hydrogen. Note that the stopping power for TaH,,, has been used for the whole sam#e. Then the hydrogen ~~~tra~o~ increases reaching the bulk value at a depth of approximately 3ocf nm. This concentration gradient is due to the varying thickness of the oxide layer combined with the surface roughness. Even though

259

the surface oxide layer is only some tens of monolayers thick it will appear extended due to the differem incident angles to the surface of the incoming ions. The distinction in the concentration gradient (region 2 in fig. 1) for the samples B3 and cat is explained by the differerzce in surface roughness_ any electron microscopy (SE&f) shows (fig. 2) that the surface roughness is in the pm range for 83, but is smoother for cal. That the concentration gradient

near the surface is due to the surface roughness was verified by comparing the profile measurements and the SEM-pictures for the unpolished and the polished sample (see figs, 2 and 3). In fig. 3, the concentration profile before and after polishing is shown for sample BJ. A drastic change in the concentration gradient and an increase of the measured hydrogen content in the bulk was observed after polishing. The increase is due to the smff ~~e~~tio~ depth of the “N ions (I Pm), having the same order of magnitude as the surface roughness. As a consequence a fraction (appro~tely 8%) of the incident ions had not penetrated the oxide layer when the resonance occurred for the unpolished sample. A similar concentration gradient cIose to the surface of the Nb hydride was reported by Pick et al. [IS!]. This behaviour was explained by the existence of a surface oxide layer (Nb,Os) and a layer containing dispersed oxide partijcles decreasing in size and number with increasing depth. Xn a study of Nb-hydrides we Qbserved similar tendencies; a concentration gradient dose to the surface and high stability induced by the surface oxide layer (see fig 4). The concentration gradient obt~~ is assumed to be related to a ~rn~natio~ of the oxide

layer at the surface and the surface roughness of the sample in the same way as for the tantalum samples described ahove.

0

Q

200

4QQ

800 dsptft

@?3)

800

1000

B. HjSrvarsson et al. / Hydrogen tantalum calibration standard

260

Measurements on Al and A5 showed that annealing 420 K for 48 h in vacuum has no effect on the concentration or the depth profile. at

3.1. Range of measurements and the off-resonance contribution The

off-resonance

cross

section

of the lH(15N,

ay )“C reaction was recently determined and the results

Fig. 2. Three SEM photographsof the sample surfaces. The magnificationis 1000 times. The marker bar in each photograph indi( :ates gradient near the surface. b) The 10 !rm. a) Sample B3 has a surface roughness in the pm range causing an artificial concentration surface is much smoother

for the sample cal. c) The surface

of sample B3 after polishing

has about the same roughness

as cal.

261

Fig. 2. Continued. 3.0 2.5

_.~_._._._.I_._.-._~_.-._.~_.___. I_._._.

&__-_?____r____r_-___-_,____~___I__

2.0

f

I

1.5

/ , :

.’

1 .o I

,’

0.5

.:’ 0

I

200

0

I

I

400

600

depth Fig. 3. Concentration

I

600

I

1000

(nm)

profiles obtained from B3 before (lower) and after (upper) polishing. Note the change in the concentration gradient near the surface.

5.0

_-

4.0

-

3.0

-

2.0

_

1.0

-

0

j /I’ 1

____

___-S--------w--_

I

I

1

! i : I I , , I

0

I 200

1 400

I 600

depth Fig. 4. Concentration

I 800

(nm)

profile for a Nb sample.

I 1000

_

262

B. Hjirvarsson

et al. / Hydrogen tantalum calibration standard

;_f___+._

3000-

:

9 2500.

;

I I I I I

‘S

m 2000. 2 ._ 2 1500.

:

I

5 c IOOO500’

l*.,_‘.’ -.* _*__. _..____<_*_*__*__

, cl,;;*.

, 7

.

, 8

.

, 9

.

, 10

.___.__

.--

.

.

, 11

energy

n

, 12

-

i 1

_._-¤*

,

, 13

,

,

.

r

14

(MeV)

Fig. 5. Normalized yield versus energy for the resonances at 6.385 MeV and 13.35 MeV.

show that only minor corrections are needed in the 6.5-S MeV range [13]. The increase of the normalized yield induced by the off-resonance contribution at an energy of 8.5 MeV is expected to be of the order of 2%. This is in agreement with our results since we do not observe any significant increase in the normalized yield at energies below 10.5 MeV (see fig. 5). Therefore we conclude that in the energy range 6.385-10.5 MeV, no corrections are needed for the TaH, calibration samples. This illustrates the validity of the simplified form for the concentration calculation introduced in formula 2 above. The relative yield of the higher to the lower resonance is estimated to be 8.19 k 0.24 from our measurements, compared with the value of 8.4 reported in [2]. We are now preparing a detailed study of the cross section of the ‘H(15N, oly )'*C reaction at energies up to 15 MeV. This work is stimulated by the possibility of detecting low concentrations in thin films using backing materials such as W (dissolves negligible amount of hydrogen at room temperature), which have been extensively used for in situ UHV-studies of the Mg-Pd system [14].

gen profiling. For all other samples the H/Ta-ratios obtained from weight and profile measurements were found to agree within experimental uncertainties (see table 1). We conclude that hydrogenated tantalum is well suited as a calibration sample for hydrogen depth profiling measurements; it is insensitive to UHV-treatment and heating up to 420 K and it can also be stored under normal laboratory conditions for years without any significant loss of hydrogen. For investigations where an accuracy of better than 10% is not required, polishing is not necessary. By polishing the hydrogenated tantalum samples it has been estimated that the uncertainty is reduced to less than 5%. The homogeneous hydrogen distribution and well-defined hydrogen concentration makes the calibration procedure simple, and due to the high y-yield, the time needed for the calibration is minimized. A calibration can be performed in less than five minutes. Dr. Yvonne Andersson prepared the tantalum-hydride samples. The scarming electron microscopy of the hydrogenated tantalum was performed by Mr. Rein Kalm. The authors are grateful for their assistance.

4. Conclusion When using hydrogenated tantalum as a primary calibration standard an error in the weighing can occur due to fragmentation and loss of the metal during preparation. This effect can be traced by comparing the consistency within the series of samples prepared. According to this argument, the deviation of the hydrogen content of sample B5, measured by weighing, is an artifact. Samples prepared at the same time, at the same temperature and pressure, are expected to have the same stoichiometric composition as verified by hydro-

References [l] D.A. Leich and T.A. Tombrello, Nuci. Instr. and Meth. 108 (1973) 67. [2] F. Xiong, F. Rauch, C. Shi, Z. Zhou, R.P. Livi and T.A. Tombrello Nucl. Instr. and Meth. B27 (1987) 432. [3] L. Westerberg, L.E. Svensson and E. Karlsson Nucl. Instr. and Meth. B9 (1985) 49. [4] W. Rudolph, C. Bauer, K. Bra&oft, D. Granbore, R. Grotzschel, C. Heiser and F. Herrmann, Nucl. Instr. and Meth. B15 (1986) 508.

B. Hj&-varssm et al. / Hydrogen tantalum calihmtlon standard

[S] H.J. Whitlow, J. Keinonen, M. Hautala md A. Hautoj&i, Nwcl. Instr. and Meth. B5 (1984) SOS, f6] G. Alefeld and 1. Vijlkl, in: Hydrogen in Metals i, eds. G. Aiefeid and J. V5ikl (Spriager, Be&, 1978) p* 331. f7] J.F. %zgler, Haud&mk of Stopping Cross-sections &.x Energetic Tons in aif Ehments~ Vof 5 of The Stopping axed Rages of Ions in Matter (Pergamon, New York, F?gQ. fgf W-K. Chu, 3.W. Mayer and M-A. Nicolet, ~a~ks~atie~~~ Spectrometry (Academic Press, New York, 1978). f9] T. Schober and H. Weti, in: Hydrogen in MetaIs If, eds. G. Alefeld and J. v61kl (Springer, Berlin, 1978) p. 36.

263

[lo] D.G. Westlake and S.T. Ockers, J. Less-Common Metals 42 (1975) 258. 1111 M.P. Seahh, M.W. Holbourn, C. Ortega and 5.14. Davies~ Nucl. Xnstr. and Meth. B30 (19%) 140. $X2] MmA. Pick, A. Hanson, K.W. Jonq and A.N. GoXami, PfryL Rev, B26 (1982) 2%x3. [IS] KM. Horn and W.A. Lanford, FZucL Instr_ and Meth. B34 (1988) 1. fr4] J. R@II, 8. Hj&mrsson, T. Eriessorr, E. Karlsson, A. Krozer, and B. Kasemo, accepted for publication in J. Less-Common Metals.