JOURNAL OF APPLIED PHYSICS

VOLUME 89, NUMBER 6

15 MARCH 2001

The online versions of this article have been corrected as per the Erratum, J. Appl. Phys. 89, 5801 (2001).

Photocurrent mapping with submicron resolution on the silicon-electrolyte junction by using near-field optics Heinrich Diesinger,a) Ahmad Bsiesy, Roland He´rino, and Serge Huant Laboratoire de Spectrome´trie Physique, Universite´ Joseph Fourier (Grenoble 1), CNRS (UMR 5588), BP 87, 38402 Saint Martin d’He`res Cedex, France

共Received 13 September 2000; accepted for publication 20 December 2000兲 We have developed a technique allowing photocurrent 共PC兲 mapping of silicon surface in contact with an electrolyte which offers an unprecedented spatial resolution. The photocurrent is generated by near-field optics using an optical tip with a 100 nm diameter aperture as an illumination source. The comparison between topographic and photocurrent mapping of SiO2 /Si mesas is used to demonstrate the feasibility of such a technique. Topographic and PC images show 300 nm lateral resolution. It is shown that this resolution is topography limited, i.e., determined by the tip-topography interaction. Indeed, PC mapping on topography-less patterned porous silicon/silicon samples shows that the lateral resolution can be as good as 100 nm, limited by the aperture size. © 2001 American Institute of Physics. 关DOI: 10.1063/1.1350415兴

microns14 due to the diffraction-limited resolution of far-field optics. The aim of this article is to evidence submicron resolved photocurrent mapping on semiconductor surfaces in contact with an electrolyte and to investigate the factors that limit the lateral resolution.

INTRODUCTION

With the ever increasing complexity of integrated circuits, there is a growing need for semiconductor surface analysis techniques with high 共submicron兲 lateral resolution. Photocurrent 共PC兲 mapping using near-field optics has been demonstrated to be a powerful tool towards achieving this purpose. As an example, the optoelectrical activity of individual structural defects has been successfully studied using this technique.1 However, the measurement of the PC induced by the optical near-field probe requires to coat the top of the studied surface with a conducting electrode transparent at the light wavelength.2 This implies a probe-semiconductor distance higher than the coating electrode thickness, which may limit the lateral resolution. In contrast to the solid electrode, the use of an electrolytic contact should allow for an approach as close as some tens of nanometers and avoids an extended preparation of the sample prior to its analysis. Moreover, the electrolyte-semiconductor interface presents a very low surface recombination velocity.3 The electrolytic contact has been widely used to investigate the semiconductor optoelectrical properties.4–7 These photoelectrochemical experiments can yield the minority carrier diffusion length giving access to physical properties of the semiconductor such as surface and bulk damage, contaminants, etc. But only an overall photoresponse without any spatial resolution has been obtained. Laterally resolved photoresponse measurement has been performed by scanning a collimated laser beam across the sample surface,8,9 allowing to map the optoelectrical activity of the semiconductor surface. This technique was used to reveal areas of high recombination efficiency such as grain boundaries,10 localized defects,8,11 and impurity sites.12,13 However, the best spatial resolution obtained in these experiments remains in the order of a few

EXPERIMENT

The optical tips are tapered single-mode optical fibers obtained by pulling while being heated with a CO2 laser. The tapered end of the optical fiber is then aluminum coated to form a subwavelength aperture at its apex.15 The far-field radiation angular distribution transmitted by the optical tip is measured using an experimental setup similar to that developed by Obermu¨ller and co-workers.16 This allows us to verify the absence of radiation leakage through the metal coating and yields the aperture diameter.17 Only optical tips with subwavelength apertures of about 100 nm are used in this work. Moreover, the radiation characteristic of each optical tip is investigated before and after its use in photocurrent mapping to ensure that no modification of the optical resolution has occurred during the experiment. The tipsample separation distance is controlled using a nonoptical shear force detection system based on the oscillation damping of a piezoelectric tuning fork18 designed to operate at 32 768 Hz. The optical tip is glued along one arm of the tuning fork and allowed to protrude by a constant length of 3 mm. The resonance frequency of such a system is only slightly shifted in comparison to the isolated tuning fork. The Q factor is typically of the order of 600 in the air and 300 as the tip is immersed in the electrolyte. The piezoelectric signal is used in an electronic feedback loop in order to regulate the tip-sample distance via a vertical piezoelectric element 共piezo-z兲 on which the sample is mounted. We used a commercial atomic force microscopy electronic control unit with a homemade sample stage including the tuning fork-optical fiber system and the piezo-z element mounted on an X – Y

a兲

Electronic mail: [email protected]

0021-8979/2001/89(6)/3328/3/$18.00

3328

© 2001 American Institute of Physics

Downloaded 03 Aug 2006 to 193.48.255.141. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

JOURNAL OF APPLIED PHYSICS

VOLUME 89, NUMBER 6

15 MARCH 2001

FIG. 1. 8 ␮m⫻8 ␮m topographic 共a兲 and PC 共b兲 mappings of a SiO2 /Si array obtained simultaneously with an aluminum coated optical fiber tip.

scanner operated by the electronic control unit. The photocurrent is generated by using a He–Ne laser beam chopped at 400 Hz and then coupled into the optical fiber. The silicon-electrolyte junction is reverse biased by a potentiostat in a standard three-electrode setup. An ohmic contact is provided on the rear face of the silicon substrate by evaporation of aluminum at low pressure followed by thermal annealing. A toroidal rubber gasket is glued on the front of the sample allowing for filling up with the electrolyte. Two fine platinum wires are immersed in the electrolyte, one of which serves as a counter electrode and the second as a pseudoreference electrode. The electrolyte is made of saturated aqueous boric acid solution containing 0.2 mol/l potassium sulfate as a supporting electrolyte. The reverse-biased siliconelectrolyte junction yields a dark current density of 1 ␮A/cm2, i.e., several orders of magnitude above the expected photocurrent 共a few nanoamperes兲. In these conditions, the PC has to be extracted by using a lock-in amplifier with the chopping frequency as reference. The samples used in this study include an array of silicon dioxide islands obtained by thermal oxidation of a p-type (3⫻1016 cm⫺3 ) silicon substrate followed by optical photolithography, and a topography-less porous silicon pattern obtained by selective anodization of a p-type silicon substrate.

FIG. 2. Line profiles of the topography 共a兲 and PC 共b兲 images; on both profiles, the transition from the SiO2 island to the substrate takes about 300 nm.

somewhat distorted by topography measurements that are performed here with a tip coated with 100 nm aluminum on each side. Consequently, the transition from the SiO2 island to the silicon substrate takes 300 nm, which gives an idea of the lateral resolution in the topographic mode. In the PC measurement mode, the same result is observed; it takes 300 nm to switch from SiO2 to Si. These findings strongly suggest that the interaction tip topography might prevent the evidence of a lateral resolution better than 300 nm in the PC measurement mode. In order to avoid this topography effect a flat-surface sample has to be used. A topography-less sample can be obtained by selective formation of porous silicon on a patterned silicon substrate.19

RESULTS AND DISCUSSION

Figures 1共a兲 and 1共b兲 show, respectively, topographic and PC mappings simultaneously acquired on the SiO2 /Si array. The PC map is the negative of the topographic image indicating that the PC decreases when the optical fiber is scanned over the oxide layer. The PC arises from photocarrier generation in the space charge region at the silicon substrate surface followed by a charge exchange with electroactive species 共protons兲 in the electrolyte. Since the oxide patterns are not expected to block this charge exchange because of the important carrier diffusion length in the space charge zone, this PC contrast seems to be a near-field/farfield effect. Since the thickness of the oxide layer is 200 nm, the evanescent wave generated by the subwavelength tip cannot proceed to the underlying silicon substrate. In this case, charge carriers can only be generated by the weak far-field radiation, hence, the reduced PC intensity. Figure 2 represents two line profiles obtained by cutting the topography and the PC images along the same line. The topography line profile shows that the vertical edges of the SiO2 island are 0021-8979/2001/89(6)/3329/3/$18.00

FIG. 3. A 40 ␮m⫻40 ␮m PC image of a porous silicon 共PSi兲 pattern 共a兲, a 4 ␮m⫻4 ␮m PC image of the center region 共b兲 and a PC line profile across the border of the porous silicon pattern 共c兲; the PC amplitude switches within 100 nm, corresponding to the aperture of the optical tip. 3329

© 2001 American Institute of Physics

Downloaded 03 Aug 2006 to 193.48.255.141. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

3330

Diesinger et al.

J. Appl. Phys., Vol. 89, No. 6, 15 March 2001

We used a silicon dioxide pattern as a mask against the electrochemical etching in a HF acid electrolyte that leads to porous silicon 共PSi兲 formation. Five seconds of electrochemical etching yielded a 40-nm-thick porous layer. After the porous silicon formation, the sample was left during 1 min in the HF electrolyte in order to chemically etch the SiO2 pattern. Figure 3共a兲 shows a 40 ␮m⫻40 ␮m PC image of the resulting porous silicon pattern. The porous silicon region exhibits a higher PC intensity than the bulk silicon. This behavior is rather surprising since it was established that the porous layer does not contribute to the photocurrent due to the absence of an important electric field in the porous structure.20 However, the porous silicon acts as an antireflection layer.21 The porous nature of the layer results in a refractive index lower than that of bulk silicon and in enhanced light transmission responsible for the observed high PC intensity on porous silicon regions. Figure 3共b兲 is a PC image of the inner square of Fig. 3共a兲. A line profile across the silicon/porous silicon border is represented by Fig. 3共c兲. It shows that the PC amplitude switches between different regions within a distance of 100 nm, corresponding to the optical resolution that can be expected given the value of the tip aperture. The comparison to the 300 nm obtained on the Si/SiO2 sample 共Fig. 1兲 indicates that the PC resolution can attain the maximal optical resolution only on flat surface samples. CONCLUSION

In summary, we have demonstrated the feasibility of photocurrent mapping on semiconductor surfaces using nearfield optics in an electrochemical environment. Compared to photocurrent mapping on semiconductors that are provided with a solid electrode, the lateral resolution is enhanced since the electrolytic contact does not obstruct the approach of the near-field probe towards the sample surface. However, if the observed photocurrent contrast is topography related, its resolution is limited by the topographic response that depends on the shape of the probe. On flat samples where the photocurrent contrast results from features other than topo-

graphic ones, the lateral resolution is determined by the aperture size itself, which is about 100 nm. Consequently, we expect an even better resolution if smaller apertures are used. This order of resolution makes the technique an interesting tool for the study of semiconductor surfaces including electrically active defects and contamination mapping and twodimensional carrier concentration profile. ACKNOWLEDGMENT

The authors would like to thank Yong Chen for the preparation of the silicon dioxide arrays. Q. Xu, M. H. Gray, and J. W. P. Hsu, J. Appl. Phys. 82, 748 共1997兲. J. W. P. Hsu, E. A. Fitzgerald, Y. H. Xie, and P. J. Silverman, Appl. Phys. Lett. 65, 344 共1994兲. 3 E. Yablonovitch, D. L. Allara, C. C. Chang, T. Gmitter, and T. B. Bright, Phys. Rev. Lett. 57, 249 共1986兲. 4 A. Etchebery, M. Etman, B. Fotouhi, J. Gautron, J. L. Sculfort, and P. J. Lemasson, J. Appl. Phys. 53, 8867 共1982兲. 5 A. M. Goodman, J. Appl. Phys. 53, 7561 共1982兲. 6 M. Saritas and H. D. McKell, J. Appl. Phys. 63, 4561 共1988兲. 7 J. Lagowski, A. M. Kontkiewicz, L. Jastrzebski, and P. Edelman, Appl. Phys. Lett. 63, 2902 共1993兲. 8 T. E. Furtak, D. C. Canfield, and B. A. Parkinson, J. Appl. Phys. 51, 6018 共1980兲. 9 M. A. Butler, J. Electrochem. Soc. 131, 2185 共1984兲. 10 M. A. Butler, J. Electrochem. Soc. 130, 2358 共1983兲. 11 S. Eriksson, T. Gruszecki, P. Carlsson, and B. Holmstro¨m, Thin Solid Films 269, 14 共1995兲. 12 V. Lehmann and H. Fo¨ll, J. Electrochem. Soc. 135, 2831 共1988兲. 13 S. A. McHugo, A. C. Thompson, I. Pe´richaud, and S. Martinuzzi, Appl. Phys. Lett. 72, 3482 共1998兲. 14 S. Kudelka and J. W. Schultze, Electrochim. Acta 42, 2817 共1997兲. 15 E. Betzig, J. K. Trautman, T. D. Harris, J. S. Weiner, and R. L. Kostelak, Science 251, 1468 共1991兲. 16 C. Obermu¨ller, K. Karrai, G. Kolb, and G. Abstreiter, Ultramicroscopy 61, 171 共1995兲. 17 A. Drezet, J. C. Woehl, and S. Huant 共unpublished兲. 18 K. Karrai and R. D. Grober, Appl. Phys. Lett. 66, 1842 共1995兲. 19 A. G. Nassiopoulos, S. Grigoropoulos, L. T. Canham, A. Halimaoui, I. Berbezier, E. Gogolides, and D. Papadimitriou, Thin Solid Films 255, 329 共1995兲. 20 B. Gelloz and A. Bsiesy, Appl. Surf. Sci. 135, 15 共1998兲. 21 S. Strehlke, D. Sarti, A. Krotkus, K. Grigoras, and C. Levy Clement, Thin Solid Films 297, 291 共1997兲. 1 2

Downloaded 03 Aug 2006 to 193.48.255.141. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp