Photovoltaic effect on the conductive atomic force microscopic characterization of thin dielectric films

(1)

Photovoltaic effect on the conductive atomic force microscopic characterization of thin

dielectric films

M. N. Chang, C. Y. Chen, M. J. Yang, and C. H. Chien

Citation: Applied Physics Letters 89, 133109 (2006); doi: 10.1063/1.2357873 View online: http://dx.doi.org/10.1063/1.2357873

View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/89/13?ver=pdfcov

Published by the AIP Publishing

Articles you may be interested in

Effect of reduction of trap charge carrier density in organic field effect transistors by surface treatment of dielectric layer

J. Appl. Phys. 114, 224504 (2013); 10.1063/1.4842856

Conductive atomic force microscopy studies on dielectric breakdown behavior of ultrathin Al 2 O 3 films

Appl. Phys. Lett. 98, 092902 (2011); 10.1063/1.3560307

Direct observation/characterization of spatial distribution of current leakage spots in zinc oxide/aluminum nitride thin film precursor field effect transistor structures using conducting atomic force microscopy

Appl. Phys. Lett. 93, 093510 (2008); 10.1063/1.2975374

Nanoscale morphological and electrical homogeneity of HfO 2 and ZrO 2 thin films studied by conducting atomic-force microscopy

J. Appl. Phys. 97, 074315 (2005); 10.1063/1.1885166

Evolution of leakage paths in Hf O 2 Si O 2 stacked gate dielectrics: A stable direct observation by ultrahigh vacuum conducting atomic force microscopy

Appl. Phys. Lett. 86, 063510 (2005); 10.1063/1.1862779

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 140.113.38.11 On: Wed, 30 Apr 2014 07:42:22

(2)

Photovoltaic effect on the conductive atomic force microscopic

characterization of thin dielectric films

M. N. Changa兲and C. Y. Chen

Division of Nano Metrology, National Nano Device Laboratories, Hsinchu 30078, Taiwan

M. J. Yang

Nano System/Device Technology Research Cluster, National Nano Device Laboratories, Hsinchu 30078, Taiwan

C. H. Chien

Department of Electronics Engineering, National Chiao Tung University, Hsinchu 30050, Taiwan

共Received 17 January 2006; accepted 16 August 2006; published online 26 September 2006兲 The authors have used front-wing conductive probes to investigate the photovoltaic effect on the conductive atomic force microscopic共C-AFM兲 characterization of thin dielectric films. The surface photovoltage induced by the laser beam of an atomic force microscope can enhance the electrical field across the studied dielectric film, decreasing the onset voltage of the leakage current, resulting in a modified C-AFM image with a larger current distribution. Moreover, the experimental results also revealed that the influence of the photovoltaic effect on C-AFM would be more significant for dielectric films that are grown on a substrate with a higher carrier concentration. © 2006 American

Institute of Physics. 关DOI:10.1063/1.2357873兴

Electrical scanning probe microscopy 共E-SPM兲 is one promising method for studying the electrical properties of a nanometric area in electronic materials.1–5 Among the E-SPM techniques, conductive atomic force microscopy 共C-AFM兲 and scanning capacitance microscopy 共SCM兲 are two well-known techniques for the characterization of thin dielectric films. In contrast to SCM, the C-AFM technique can also provide the local current distribution, and has been widely used to investigate the nanoscale current-voltage 共I-V兲 characteristics as well as the breakdown mechanism of thin dielectric films.6–8 Typically, an atomic force micro-scope共AFM兲 is equipped with a current sensor to synchro-nously provide current images to the corresponding topo-graphic images. However, it has been revealed that an AFM laser beam can induce photoperturbation, which leads to dif-ficulty in doing the SCM characterization of the carrier dis-tribution and electrical junctions.9For instance, photopertur-bation can lead to false SCM images and distorted differential capacitance versus bias curves.10,11Since C-AFM and SCM are based on the same AFM apparatus, it is thus natural to ask whether C-AFM also suffers from the same photoperturbation problem. Recently, Chang et al. found that a conductive probe with a front-wing共FW兲 cantilever, which provides an effective shadowed area that fully covers the scanned region, is a practical approach for nonphotoper-turbed SCM characterizations.12 In this present work, we have employed FW conductive probes to perform nonphoto-perturbed C-AFM measurements and to qualitatively study the influence of the photovoltaic effect on the C-AFM characterization.

To investigate the influence of photoperturbation on C-AFM, various samples with an insulator-on-semiconductor structure were prepared. Samples 1 and 2 consisted of a lightly doped p-type具100典 Si wafer covered with thermally grown SiO2 thin films 2.5 and 5 nm in thickness,

respec-tively. The carrier concentration of the lightly doped sub-strate was about 5⫻1015cm−3. Sample 3 had the same

p-type Si substrate, but with a 4-nm-thick HfO2film

depos-ited on the substrate by atomic vapor deposition共AVD兲 using an AIXTRON Tricent system. After the AVD process, sample 3 was annealed at 900 ° C in ambient N2. Sample 4 consisted

of a 2.5-nm-thick SiO₂thin film that was thermally grown on a heavy-doped p-type Si wafer. The carrier concentration of the heavy-doped substrate was about 7⫻1017_cm−3_{. The}

thicknesses of the dielectric films were confirmed by a JEM 2010F high resolution transmission microscope operated at a 200 kV accelerating voltage.

All the AFM and C-AFM measurements were performed in an environment with well-controlled temperature and hu-midity, using an NT-MDT Solver P47 scanning probe micro-scope. The wavelength of the AFM laser ranged from 620 to 690 nm and the output power was 0.9 mW. To sig-nificantly reduce the current noise level of the C-AFM, we used a homemade shielding box for the scan unit. FW con-ductive probes with Cr–Co coated silicon tips共produced by MikroMasch兲 were used to scan the sample surface. These FW conductive probes allowed us to fine tune the photoper-turbation level without problems in measuring topographic images.12The force constant of the cantilevers was less than 4 N / m. The back side of the cantilever was coated with an 80-nm-thick Cr–Co layer. To make the I-V measurements, negative biases were applied to the samples, since the I-V data would be unstable due to negative tip biases.5More than 50 surface sites on each C-AFM sample were measured un-der both photoperturbed and nonphotoperturbed conditions because repeated measurements at one point on the dielectric film could induce modified I-V curves.

Figure 1 shows four groups of the I-V curves obtained from samples 1 and 2. It is obvious that the photoperturba-tion has significantly reduced the onset voltage for both samples, implying that it is easier to detect leakage current from a sample surface with photoperturbations. In Fig. 1, we can also see that sample 1 had a higher onset voltage shift

a兲_{Electronic mail: [email protected]}

APPLIED PHYSICS LETTERS 89, 133109共2006兲

(3)

than sample 2 for the same change in illumination. This re-sult indicates that the influence of photoperturbation on C-AFM measurements depends on the thickness of the stud-ied dielectric film. We attributed this result to the photovol-taic effect induced by the photoperturbations from the AFM laser beam. The total voltage drop Vtotalduring C-AFM

mea-surement, for a sample without photoperturbations, can be generally described by

Vtotal= Vox+ Vd, 共1兲

where Voxis the voltage drop of the dielectric film and Vdis

the voltage drop in the surface depletion region. At a con-stant negative sample bias Vs with respect to the tip

elec-trode, the electric field across the depletion region separates photogenerated electron-hole pairs and the minority carriers diffuse toward the illuminated surface, thus inducing the sur-face photovoltage that results in the potential change across the depletion region.13From the point of view of a circuit in equilibrium, the photovoltage will lead to the redistribution of voltage drops in the dielectric film and the depletion re-gion, since the total bias remains constant. In other words, the photovoltaic effect can increase the voltage drop in the studied dielectric film and enhance the electrical field across the film. According to the above deduction, we can modify Eq.共1兲 for a photoperturbed C-AFM measurement as

Vs= Vox

⬘

+ Vd

⬘

=共Vox+ Vpv兲 + 共Vd− Vpv兲, 共2兲

where V_ox

⬘

and Vd

⬘

are the voltage drops in the dielectric film

and the surface depletion region with photovoltaic effect, respectively, and Vpv is the photovoltage induced by

photo-perturbation. From Eqs. 共1兲 and 共2兲, the photovoltage-induced change in the electrical field across the dielectric film can be described by

Een− Eox= Vox+ Vpv tox −Vox tox =Vpv tox =⌬E, 共3兲 where E_enand E_oxare the photoenhanced electrical field and the original electrical field of the dielectric film, respectively,

tox is the thickness of the dielectric film, and ⌬E is the

change in the electrical field induced by Vpv. Due to the

existence of a positive ⌬E, the onset voltage could be re-duced. Equation共3兲 also indicates that the photoperturbation-induced change in the onset voltage would be smaller for a

thicker dielectric layer, which is in agreement with the re-sults shown in Fig. 1.

Figures 2共a兲 and 2共b兲 show current images of sample 2 at a sample bias of −6.7 V, with and without photoperturbation, respectively. It is apparent that, due to the enhancement of the electrical field across the SiO₂ film, the photovoltaic ef-fect causes more significant current leakage on the sample surface. Figure 2共c兲 shows the statistical results for current distribution corresponding to Figs. 2共a兲 and 2共b兲. The full

FIG. 1.共Color online兲 I-V curves obtained from samples 1 and 2, with and without photoperturbations. Although the illumination change is the same, sample 1 exhibits a higher onset voltage shift than sample 2 does.

FIG. 2.共Color online兲 Current images 共1⫻1␮m2_{兲 of sample 2 at a sample}

bias of −6.7 V共a兲 with and 共b兲 without photoperturbation; 共c兲 the statistical results of the current distributions corresponding to共a兲 and 共b兲.

FIG. 3.共Color online兲 Current images 共1⫻1␮m2_{兲 of sample 3 at a sample}

bias of −5 V共a兲 with and 共b兲 without photoperturbation; 共c兲 the statistical results of the current distributions corresponding to共a兲 and 共b兲.

133109-2 Chang et al. Appl. Phys. Lett. 89, 133109共2006兲

(4)

widths at half maximum 共FWHMs兲, with and without the photovoltaic effect, are 75 and 35 fA, respectively. The FWHM value can respond to the uniformity of current tribution on the sample surface. A more uniform current dis-tribution will exhibit a smaller FWHM value. It is evident that the maximum distribution of the current signal increases with the photoperturbation level. The FWHM also shows the same trend. The statistical results in Fig. 2共c兲 reveal that the current distribution is able to sensitively respond to the illu-mination intensity, even if the current signals are very small. Figures 3共a兲 and 3共b兲 show the current images of sample 3 at a sample bias of −5 V, with and without photoperturbations. Since the photovoltaic effect can enhance an electrical field across a dielectric film, this photovoltaic effect may highlight the difference between areas in an HfO2 film with different

breakdown strengths. Figure 3共c兲 shows the statistical results for current distributions corresponding to Figs. 3共a兲 and 3共b兲. The FWHMs, with and without the photovoltaic effect, are 1.246 and 0.13 pA, respectively. This indicates that the pho-tovoltaic effect may induce a significantly wider range of current distributions during C-AFM measurements when the dielectric film has the areas with weaker breakdown strength. From Figs. 2共c兲 and 3共c兲, one can expect that the FWHM value can also be a uniformity indicator of breakdown strength for thin dielectric films.

Figure 4 shows four groups for the I-V curves of samples 1 and 4. Since the photovoltaic effect may increase with substrate doping level,14 the change in the electrical field across the studied dielectric films for dielectric films grown on a substrate with a higher carrier concentration will be

more significant. As a result, with the same change in illu-mination there is a bigger change in the onset voltage exhib-ited by sample 4 than by sample 1, implying that sample 4 is more sensitive to photoperturbations during C-AFM mea-surement. Indeed, obvious current fluctuations frequently oc-curred during the C-AFM measurement of samples such as sample 4.

To summarize, we employed FW conductive probes to investigate the photovoltaic effect during the conductive atomic force microscopic characterization of thin SiO2 and

HfO₂films. The photovoltaic effect induced by an AFM laser beam can lead to an enhanced electrical field across the di-electric film and hence result in false C-AFM images as well as modified I-V characteristics. Provided the illumination conditions are the same, the influence of the photovoltaic effect on the C-AFM measurements of an insulator-on-semiconductor structure is significantly dependent on the properties of the dielectric film and the carrier concentration in the semiconductor region. Thus, C-AFM measurements would be more stable and accurate without the photovoltaic effect. On the other hand, the photovoltaic effect can be em-ployed to study dielectric breakdown in nanometric areas provided that the photoperturbation level in the C-AFM can be quantitatively controlled.

1_{S. D. Wang, M. N. Chang, C. Y. Chen, and T. F. Lei, Electrochem.}

Solid-State Lett. 8, G233共2005兲.

2_{P. Eyben, D. Alvarez, M. Jurczak, R. Rooyackers, A. De Keersgieter, E.}

Augendre, and W. Vandervorst, J. Vac. Sci. Technol. B 22, 364共2004兲.

3_{M. Porti, M. Nafria, X. Aymerich, A. Olbrich, and B. Ebersberger, Appl.}

Phys. Lett. 78, 4181共2001兲.

4_{H. Ikeda, N. Kurumado, K. Ohmori, M. Sakashita, A. Sakai, S. Zaima, and}

Y. Yasuda, Surf. Sci. 493, 653共2001兲.

5_{S. J. O’Shea, R. M. Atta, M. P. Murrell, and M. E. Welland, J. Vac. Sci.}

Technol. B 13, 1945共1995兲.

6_{M. Porti, M. Nafría, M. C. Blum, X. Aymerich, and S. Sadewasser, Appl.}

Phys. Lett. 81, 3615共2002兲.

7_{H. Ikeda, T. Goto, M. Sakashita, A. Sakai, S. Zaima, and Y. Yasuda, Jpn.}

J. Appl. Phys., Part 1 42, 1949共2003兲.

8_{S. Kremmer, C. Teichert, E. Pischler, H. Gold, F. Kuchar, and M.}

Schatz-mayr, Surf. Interface Anal. 33, 168共2002兲.

9_{M. N. Chang, C. Y. Chen, W. W. Wan, and J. H. Liang, Appl. Phys. Lett.} 84, 4705共2004兲.

10_{M. N. Chang, C. Y. Chen, F. M. Pan, J. H. Lai, W. W. Wan, and J. H.}

Liang, Appl. Phys. Lett. 82, 3955共2003兲.

11_{S. Shin, J.-I. Kye, U. H. Pi, Z. G. Khim, J. W. Hong, S. Park, and S. Yoon,}

J. Vac. Sci. Technol. B 18, 2664共2000兲.

12_{M. N. Chang, C. Y. Chen, W. J. Huang, and T. C. Cheng, Appl. Phys. Lett.} 87, 023102共2005兲.

13_{D. K. Schroder, Semiconductor Material and Device Characterization,}

2nd ed.共Wiley, New York, 1998兲, p. 440.

14_{E. S. Yang, Microelectronic Devices, 1st ed.}_{共McGraw-Hill, New York,}

1988兲, p. 360. FIG. 4. 共Color online兲 I-V curves for samples 1 and 4 with and without

photoperturbations. Although the change in illumination is the same, sample 4 exhibits a bigger change in the onset voltage than sample 1 does.

133109-3 Chang et al. Appl. Phys. Lett. 89, 133109共2006兲