Simple and accurate spectra normalization in ion beam analysis using a transmission mesh-based charge integration

NIM B Beam Interactions with Materials & Atoms Nuclear Instruments and Methods in Physics Research B 243 (2006) 392–396 www.elsevier.com/locate/nimb Simple and accurate spectra normalization in ion beam analysis using a transmission mesh-based charge integration M. El Bouanani a, P. Pelicon b,*, A. Razpet b, I. Čadež b, M. Budnar b, J. Simčič b, S. Markelj b a b Laboratory for Electronic Materials and Devices, Department of Materials Science and Engineering, University of North Texas, Denton, TX 76203-5310, USA Microanalytical Center, Department of Low and Medium Energy Physics, Jožef Stefan Institute, P.O. Box 3000, SI-1001 Ljubljana, Slovenia Received 27 December 2004; received in revised form 25 July 2005 Available online 2 November 2005 Abstract Accurate and reproducible determination of the number of impact ions is essential for quantitative IBA measurements. Herewith we present an in-beam charge-collection device, consisting of a tungsten mesh enclosed by two negatively biased annular electrodes and a shaping slit. The charge-collection efficiency was measured as a function of aperture bias and the reproducibility of charge collection at different bias voltages studied. Scanning transmission ion microscopy (STIM) was used to check the effect of beam scattering at the mesh on its energy distribution. The effect of the device on the primary beam energy distribution was calculated for the beams typically used in Rutherford backscattering spectroscopy (RBS) and elastic recoil detection analysis (ERDA). Excellent characteristics of the device were demonstrated.
2005 Elsevier B.V. All rights reserved. PACS: 82.80.Yc; 68.49.Sf Keywords: Beam charge normalization; Rutherford backscattering spectroscopy (RBS); Elastic recoil detection analysis (ERDA); Ion implantation 1. Introduction Highly accurate measurement of the number of impact ions during the acquisition of spectra is a key issue for the precision of ion beam analysis (IBA). IBA beam normalization methods could be classified according to their principle of operation as follows: A: Charge integration • Total beam charge integration from an isolated chamber enclosing the sample. The chamber acts a Faraday cup. * Corresponding author. Tel.: +386 1 5885 294; fax: +386 1 5885 377. E-mail address: primoz.pelicon@ijs.si (P. Pelicon). URL: http://www.rcp.ijs.si/~mic/ (P. Pelicon). 0168-583X/$ - see front matter
2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2005.09.002 • Total charge integration from a positively biased sample for suppression of secondary electrons [1]. • In-beam charge integration devices that intercept a fraction of the primary ion beam. Integrated charge is proportional to the accumulated charge on the sample [2]. B: Secondary particle and photon emission • From the sample. • From in-beam chopping devices [3]. • From in-beam media, i.e. X-ray yield from argon or an exit foil used in in-air proton-induced X-ray emission (PIXE) analysis. The methods of the first group (A) use charge integrators for determination of the number of ions impinging on sample during the measurement. The process of M. El Bouanani et al. / Nucl. Instr. and Meth. in Phys. Res. B 243 (2006) 392–396 secondary electron emission, which increases the positive current readout at the irradiated sample, is strongly dependent on the beam-interacting surface and the intensity of local electric and magnetic fields. The stability of charge integration during a particular measurement may also be affected due to the beam-induced changes at the irradiated sample surface. The known effect of beamassisted hydrocarbon deposition at the beam-interacting surface strongly depends on the vacuum conditions. The reproducibility of the method therefore depends on the efficiency of secondary electron suppression. The normalization methods based on charge integration (group A) at beam currents below 1 nA, as is the case for the ion currents in nuclear microprobes, may yield significant error due to contribution of stray currents. The methods of the group B are in general less restricted regarding the beam intensity. The accumulation rate of normalizing events should be matched to a particular application in order to avoid excessive statistical error or significant dead time. The type of device discussed in this work belongs to the class of in-beam charge-collection devices (group A) and has been, according to our knowledge, suggested and implemented for standard IBA characterization for the first time at the University of North Texas (UNT) by El Bouanani. The original device in use at UNT is made of a tungsten mesh that is completely enclosed in negatively biased cylinder electrodes with an entry and exit aperture slits attached. A fine mesh on the annular holder, surrounded by the negatively biased cylinders, is used to collect fraction of charge that is proportional to the total charge of the transmitted beam. The entire transmitted beam through the mesh charge-collection device then passes to the sample without further obstructions. Due to high wire density, the used mesh takes into account the beam intensity variations over the entire beam cross-section. 2. The mesh based charge-integration device The schematic drawing of the mesh charge-integration device constructed and in use at Jožef Stefan Institute (JSI) is shown in Fig. 1. The annular electrodes are separated by insulating ruby balls. The walls of the electrodes are shaped to shield the insulating ruby balls from the beam trajectory to avoid their charging. The first electrode in the stack is grounded and contains the mounting for the replaceable tantalum beam-shaping slit. The mesh is mounted between two annular elements. It is encompassed by two negatively biased electrodes to suppress the emission of secondary electrons emitted from the irradiated area of the mesh. Two electric connections are led through coaxial vacuum feedthroughs: the mesh contacts the charge integrator and the bias contacts the negative pole of the DC voltage power supply. The mesh in use consists of woven tungsten wires each with a diameter of 38 lm (0.0015 in.) and a density of 3.2 lines/mm (80 lines/in.) with 77.4% transmission. The mesh charge-integration device is mounted at the entrance of the RBS/ERDA scattering chamber at the 10 beamline of the 2 MV JSI tandem accelerator. The beam profile along the beamline is defined by an electrostatic quadrupole lens which is positioned at the accelerator exit. By means of this lens, the beam is focused on the shaping slit of the mesh charge-integration device. The shape and the size of the slit is chosen to obtain a desirable beam spot on the sample. In the case of simultaneous 30 mm 8 mm BEAM Replaceable shaping slit Insulating ruby balls 393 Charge int. Neg. bias Fig. 1. Construction of the in-beam mesh charge collector. The tungsten mesh is inserted between the two central electrodes. 394 M. El Bouanani et al. / Nucl. Instr. and Meth. in Phys. Res. B 243 (2006) 392–396 3. Effect of bias on the charge-collection stability and efficiency The effect of the bias voltage applied to the electrodes enclosing the mesh on the charge-collection efficiency (Eq. (1)) was measured using the RBS signal from a sample consisting of a thin film of TaN on Si [4] with a known Ta areal density. A 4.2 MeV 7Li2+ beam was used [5] and the integrated number of counts in the tantalum peak per constant integrated charge on the mesh was measured. Using the same sample and detector setup, we performed a series of measurements under different beam intensities in different runs over a period of one month. Three different bias voltages were evaluated. With a negative bias voltage higher than 500 V the charge integration was extremely reproducible, which is not the case for the low negative bias of 20 V (Fig. 2). Similar tests with the device built at UNT and in use at the UNT RBS setup demonstrated the reproducibility better than 1% with the He+ beam using the energies between 1 and 2 MeV. Further RBS measurements at JSI on the TaN/Si sample revealed a strong dependence of the Ta RBS yield per unit of mesh-integrated charge on the bias voltage (Fig. 3). To describe the observed results, the meshintegrated charge Qint can be presented as a sum of the following contributions: Qint ¼ AN beam Ze0 þ MAN beam e0
SN beam e0 . ð1Þ No. of counts/0.1 microcoulomb The quantity A is the relative area consisting of the mesh wires that blocks the beam and equals 1
T, where T stands for the mesh transmission. For this particular study A equals 0.226. Ze0 is the charge of the ions in the beam 18000 16000 14000 20 V 4 3 Charge collection efficiency measurement of RBS and ERDA spectra, the sample is tilted by 75 angular degrees and the shaping slit used is 1 mm wide and 4 mm high. This yields a square-like beam spot on the sample with the approximate size of 4 · 4 mm2. 2 1 1 10 100 1000 Negative bias [V] Fig. 3. Measured charge integration efficiency D (Eq. (2)) as a function of the negative bias at the suppression electrodes. The error bars represent the statistical uncertainty. used for the analyses. The first term in the sum (Eq. (1)) is the charge integrated from the stopped fraction of the primary beam consisting of Nbeam ions. The second term represents the effective positive charge collected from the mesh due to the escape of secondary electrons, where M describes the number of emitted secondary electrons per ion hitting the mesh. The third, negative term accounts for the contributions of positive ions emitted from the mesh and electrons hitting the mesh. Both contributions are assumed to be proportional to the number of ions in the primary ion beam Nbeam with a proportionality constant Se0. Fig. 2 shows the measured charge-collection efficiency D as a function of bias voltage and is defined by D ¼ Qint =ðAN beam Ze0 Þ; ð2Þ as a function of bias voltage. The sharp decrease of collection efficiency in the negative bias region from 0 to 10 V is a result of strong suppression of slow secondary electrons. Suppression of the processes described by the third term in Eq. (1) is achieved with a much higher negative bias. At negative bias voltages over 600 V the charge collection efficiency does not depend on the bias voltage. In this region, the contributions from both the second and third term in Eq. (1) are completely suppressed and the measured charge corresponds to the charge of the ions stopped in the mesh. 300 V 800 V 4. Application of the mesh charge-integration device for RBS and ERDA spectra normalization 12000 10000 Consecutive runs Fig. 2. The stability of the device checked at bias voltages of
20,
300 and
800 V. The number of counts in the tantalum peak has been measured in the RBS spectrum of a thin film of TaN on Si. Error bars show one standard deviation of the number of accumulated counts. The normalization parameter P of the RBS/ERDA measurement, given as the number of ions hitting the sample during measurement Ni multiplied by the detector solid angle Xdet, is proportional to the integrated charge Qint on the mesh electrode: P ¼ N i Xdet ¼ K Qint . ð3Þ 395 M. El Bouanani et al. / Nucl. Instr. and Meth. in Phys. Res. B 243 (2006) 392–396 The proportionality constant K is determined by RBS measurements on standard reference samples. Thin film reference standards for RBS with a known area density of a deposited heavy element on a substrate consisting of a light element are the best choice. Thin film reference standards with a certified precision of better than 2% are available [6,7]. An alternative to these standards are thick pure amorphous targets with very well known stopping powers for the ions in use (e.g. 4He ions in Au). The geometrically determined value of the solid angle Xdet is used at the JSI setup only to obtain an approximate value for the number of impact ions Ni during measurement from the normalization parameter P. The resulting Ni is used for evaluation of beam damage effects during the measurement. Use of the number of impact ions Ni for normalization of the RBS/ERDA analysis would introduce an additional experimental uncertainty to the results. 5. Effect of the mesh on the primary ion beam The model presented in Section 3, which describes the charge-collection efficiency of the device, neglects processes at the mesh edges, where a part of the beam is subject to only partial stopping. This is expected to affect both the energy and the angular distribution of the primary beam. To estimate the fraction of the beam which is subject to partial stopping at the edges of the mesh wire, the finite range of the ions in the mesh material should be considered. If the radius of the wire is r and the range of the ion in the mesh material is R, then the radial distance p from the edge at which the ion could penetrate the wire is  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2 p ¼ r 1
ð1
ðR=ð2rÞÞ Þ . ð4Þ For the square mesh with the wire distance of D, the fraction of the beam at the sample, which undergoes the deceleration process in the mesh wires and has non-zero energy, is given by F ffi 4pðD
2r þ 2pÞ 2 ðD
2r þ 2pÞ . F ffi 4 ðp=rÞ. ðD=rÞ
2 ð6Þ Table 1 lists the calculations of the contaminated fractions of the incident beam based on this geometrical model (Eqs. 4 and 6) for some ion beam types and energies typically used in IBA analysis. The set of beams evaluated includes 2 MeV protons, typically used in proton-induced X-ray emission (PIXE), 2 MeV He ions frequently used for RBS and hydrogen-ERDA and beams used for heavy-ion ERDA. As expected, the contaminated fraction of the incident beam by the mesh increases with the increase in the range of ions in the mesh material. The case of 2 MeV protons exhibits the most significant fraction of contamination among the ion beams included in Table 1. For the other ion beams than protons that are typically used in IBA analyses, the contaminated fraction is in the order of one ion per thousand or lower. Direct measurement of the energy distortion of the incident ion beam through the tungsten mesh was performed using scanning transmission ion microscopy (STIM). A 2 MeV proton beam with a diameter of 1 lm and an intensity of ca. 100 ions/s was used for the STIM measurement. An area of 1 · 1 mm2 was uniformly scanned and the ion detector was positioned directly in the beam. Measured spectra of the beam with and without the mesh are shown in Fig. 4. The peaks in both spectra fit well a Gaussian distribution with a width r of 16.7 keV. To evaluate the energy distortion of the beam that is due to the mesh, examination of the collected STIM spectra without the mesh was first undertaken. The integrated yield I is defined as follows: Z E0
5r I¼ ðdn=dEÞ dE; ð7Þ 0 where E0 is the energy of the primary beam and the (dn/dE) refers to the measured STIM spectrum. The beam empty space ð5Þ mesh in the beam 1000 Table 1 Range R of ions in tungsten [8], penetration range given as the ratio p/r (Eq. (4)) from the edge of the tungsten wire in the mesh used (radius 19 lm) and the fraction of the beam F with altered energy due to positioning of the mesh in the beam (Eq. (6)) Beam R (lm) p/r (Eq. (4)) F (Eq. (6)) 1 14.3 4.25 3.5 0.98 3.1 3.8 4.6 7.4e
2 2.5e
3 4.25e
3 3.3e
4 3.3e
3 5.0e
3 7.3e
3 2.0e
2 6.9e
4 1.1e
3 9.1e
5 9.1e
4 1.4e
3 2.0e
3 H, 2 MeV He, 2 MeV 7 Li, 4.2 MeV 12 C, 2 MeV 19 F, 14.2 MeV 35 Cl, 35 MeV 127 I, 70 MeV 4 Linear trajectories of ions in the mesh material are assumed. Yield Since Eq. (5) stands for all the relevant cases where p/r  1 (Table 1), it could be rewritten as 100 10 1 100 200 300 Channel Fig. 4. Effect of the mesh on the energy distribution of the primary beam: a 2 MeV proton microbeam was scanned over the mesh area of 1 · 1 mm. The detector for scanning transmission ion microscopy (STIM) was positioned at 0. The result of the blank scan is included for comparison. 396 M. El Bouanani et al. / Nucl. Instr. and Meth. in Phys. Res. B 243 (2006) 392–396 contamination C is defined as measure of the ratio of I and the total number of detected ions Itot: C ¼ I=I tot . ð8Þ The spectrum of the blank scan revealed contamination of 2%, which is result of the extremely small object slit opening during the 0 STIM measurement. For the STIM spectrum of the mesh, a weak low energy tail is present in the spectrum and the fraction of contamination increases to 4%. This rise of contamination of 2% is in reasonable agreement of the prediction of the geometrical calculation (Table 1). The 2 MeV proton beam, studied here by STIM, is expected to exhibit the highest fraction of contamination among the beam types discussed in Table 1. The use of He ions or heavy ions results in a significantly lower beam contamination, which enables high sensitive ion beam analysis [5,9]. analysis (ERDA) and nuclear reaction analysis (NRA). The presented results encourage the use of the device also at the field of ion implantation. The sweeping of the beam, which is usually applied in the implantors to achieve the fluence uniformity over the large implanted areas, should be applied in this case after the beam final shaping and charge collection with the presented device in order to avoid shade effects. Acknowledgement This work at JSI was supported by the grant J1-32790106-01 from the Slovenian Ministry of Education, Science and Sport, by the IAEA Co-ordinated Research Project ‘‘Analysis of light elements in thin films, including depth profiling’’ and the bilateral project SLO-USA-2002/ 19. 6. Conclusion References The proposed mesh-based charge integration device in combination with known thin film standards is a simple and very effective solution for the accurate spectra normalization in ion beam analysis. It features reproducibility of better than 1%, exceptional stability and outstanding reliability when compared to the instrumentally more complex transmission Faraday cup [2]. The effects of the mesh on the primary incident beam were carefully evaluated and it was experimentally demonstrated that the distortion is negligible for most of the standard beams used in conventional ion beam analysis except for protons, where a thicker tungsten mesh would perform better than the one studied here. Its ease of construction and simple use provide great potential for applications in Rutherford backscattering spectroscopy (RBS), elastic recoil detection [1] F. Herrmann, D. Grambole, Nucl. Instr. and Meth. B 104 (1995) 26. [2] F. Pászti, A. Manuaba, C. Hajdu, A.A. Melo, M.F. Da Silva, Nucl. Instr. and Meth. B 47 (1990) 187. [3] L. Bartha, I. Uzonyi, Nucl. Instr. and Meth. B 161–163 (2000) 339. [4] W.A. Lanford, P. Pelicon, B. Zorko, M. Budnar, Nucl. Instr. and Meth. B 190 (2002) 410. [5] P. Pelicon, A. Razpet, S. Markelj, I. Čadež, M. Budnar, Nucl. Instr. and Meth. B 227 (2005) 591. [6] K.H. Ecker, A. Berger, R. Grotzschel, L. Persson, U. Watjen, Nucl. Instr. and Meth. B 175 (2001) 797. [7] G. Boudreault, C. Jeynes, E. Wendler, A. Nejim, R.P. Webb, U. Watjen, Surf. Interf. Anal. 33 (2002) 478. [8] J.F. Ziegler, SRIM program. Available from: <http://www.srim.org/>. [9] P. Pelicon, G.V. Ravi Prasad, M. El Bouanani, B.N. Guo, D. Birt, J.L. Duggan, F.D. McDaniel, in: Proceedings of 17th International Conference on the Application of Accelerators in Research and Industry, Denton, 2002, AIP Conf. Proc. 680 (1) (2003) 482.