Energy-band diagram of metal/ Lu 2 O 3 /silicon structures G. Seguini, E. Bonera, S. Spiga, G. Scarel, and M. Fanciulli Citation: Applied Physics Letters 85, 5316 (2004); doi: 10.1063/1.1828600 View online: http://dx.doi.org/10.1063/1.1828600 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/85/22?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Luminescent properties and electronic structures of rare earth and alkaline earth borates of RE Ba 3 B 9 O 18 ( RE = Lu , Y ) J. Appl. Phys. 101, 023501 (2007); 10.1063/1.2409284 Anomalous photoconductivity in gamma In 2 Se 3 J. Appl. Phys. 100, 033707 (2006); 10.1063/1.2219002 Energy-band parameters of atomic-layer-deposition Al 2 O 3 In Ga As heterostructure Appl. Phys. Lett. 89, 012903 (2006); 10.1063/1.2218826 Energy band alignment at the (100) Ge/HfO 2 interface Appl. Phys. Lett. 84, 2319 (2004); 10.1063/1.1688453 Temperature dependent spectroscopic studies of the electron delocalization dynamics of excited Ce ions in the wide band gap insulator, Lu 2 SiO 5 Appl. Phys. Lett. 83, 1740 (2003); 10.1063/1.1603337
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: 193.205.155.112 On: Fri, 26 Sep 2014 09:35:45
APPLIED PHYSICS LETTERS
VOLUME 85, NUMBER 22
29 NOVEMBER 2004
Energy-band diagram of metal/Lu2O3/silicon structures G. Seguini,a) E. Bonera, S. Spiga, G. Scarel, and M. Fanciulli Laboratorio Nazionale MDM-INFM, Via C. Olivetti 2, 20041 Agrate Brianza (MI), Italy
(Received 9 August 2004; accepted 14 October 2004) Internal photoemission spectroscopy has been used to determine the band alignment in Lu2O3 based metal-oxide-semiconductor structures. The Si/Lu2O3 interface conduction- and valence-band offsets were determined to be 2.1± 0.1 and 2.6± 0.1 eV, respectively. The energy barrier for electrons at the Al/Lu2O3 interface is 2.4± 0.1 eV. The value of the Lu2O3 transport band gap, obtained by photoconductivity measurements, was found to be 5.8± 0.1 eV. Optical absorption spectroscopy gave a value of 4.89± 0.02 eV for the Lu2O3 optical band gap. © 2004 American Institute of Physics. [DOI: 10.1063/1.1828600] The scaling down of modern nanoelectronic devices based on a metal-oxide-semiconductor (MOS) structure motivated an intense activity on high dielectric constant materials.1 Although a large number of compounds has been investigated, suitable candidates have not been identified yet. The difficulties to substitute SiO2 are related to the many requirements that the new gate oxide material must fulfill. To reduce direct tunneling, a dielectric constant 共兲 higher than the one of SiO2 共 = 3.9兲, allowing the use of thicker films with the same capacitance, is not the only key parameter. The thermodynamical stability and the band alignments are among the issues affecting the reliability properties of complementary MOS devices. Band offsets lower than 1 eV make the oxide not suitable for gate applications.1 Internal photoemission (IPE) spectroscopy is a powerful technique to directly determine the band alignment in MOS structures.2 IPE has been employed to measure the electron energy barrier between the valence band of (100) Si and the conduction band of several oxides, such as SiO2 共4.25± 0.05 eV兲,3 Al2O3 共3.25± 0.08 eV兲,3 ZrO2 共3.1± 0.1 eV兲,3 and HfO2 共3.1± 0.1 eV兲.4 Rare-earth (RE) oxides are among the candidates as gate dielectrics. In particular, Lu2O3 (lutetia) is proposed because, despite the moderately high dielectric constant around 12, it is predicted to be thermodynamically stable on Si,5,6 and it is expected to have a good conduction-band offset (CBO) with Si. The latter property is related to a large band gap due to the completely filled 4f shell of Lu, and because of its 2:3 metal:oxygen stoichiometry ratio which lowers the charge neutrality level.7 Recently, Lu2O3 films have been grown on Si(001) by atomic layer deposition (ALD).8 In this work we report on the full band alignment characterization of Lu2O3 based MOS structures by IPE spectroscopy. The transport band gap 共EG兲 of Lu2O3 was measured by photoconductivity (PC) measurements and compared with the value obtained by adding the band offsets to the silicon band gap. Finally, optical absorption spectroscopy was used to determine the optical band gap 共EOPT兲. The Lu oxide layers were deposited by ALD in a F-120 ASM-Microchemistry reactor at 360 °C using 兵关C5H4共SiMe3兲兴2LuCl其2 and H2O as precursors.8 Both p- and n-type (100)Si 4 in. wafers with resistivity = 1–5 ⍀ cm covered with 1-nm-thick chemical silicon oxide were used as a)
Electronic mail:
[email protected]
substrates for IPE and PC. No contribution to the energy barriers determination is expected for such a thin interfacial layer.3 For the optical absorption measurements Lu2O3 was grown on quartz. The as-grown films are amorphous. X-ray reflectivity (XRR) measured, for the sample investigated in this work, a thickness of 7.9± 0.1 nm. Capacitance-voltage 共C-V兲 measurements determined the dielectric constant of Lu2O3 to be 12± 1.8 More details about the deposition and the structural, morphological, and electrical characterization are reported elsewhere.8 For IPE and PC measurements MOS capacitors were fabricated by thermal evaporation at 10−6 mbar of semitransparent electrodes (15-nm-thick aluminum or gold dots with an area of 0.5 mm2) through a shadow mask. The back contact was realized with eutectic InGa. Neither postdeposition nor postmetallization annealings were performed. A 150-W Xe arc lamp was used as the radiation source for IPE, PC, and optical absorption measurements. The experiments were performed at room temperature and in air. The photon spectral range was h = 2.5–6.0 eV with a resolution of 0.02 eV. For IPE and PC, the quantum yield 共Y兲 was obtained normalizing the photocurrent with respect to the incident photon flux measured using a silicon photodiode. The spectral thresholds ⌽共V兲 were determined by extrapolating the quantum yield curves to zero. Using the flatband voltage obtained from the C-V curves and the thickness measured by XRR, it was possible to determine the oxide internal electric field 共Eox兲 as a function of the applied voltage V. This procedure allows extrapolating in a Schottky plot the spectral threshold at Eox = 0, which corresponds to the energy barrier.2 In Fig. 1, Y 1/3 versus the photon energy for the Al/ Lu2O3/n-Si stack is depicted with the positive applied voltage as a parameter. Voltage-dependent thresholds around 3 eV can be observed. The applied voltages were chosen high enough to suppress electron IPE from the aluminum electrode.9 As similar curves are obtained using a gold metallization, we conclude that the thresholds are due to electron emission at the interface between the silicon substrate and the oxide. The change in slope around 3.4 eV is due to a direct internal transition across the silicon band gap and represents another independent proof of the assignment of silicon as the emitter.2 IPE theory predicts that for a semiconductor emitter Y共h兲 follows a cubic power law.10 The extrapolation at Eox = 0 of the voltage-dependent threshold
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: 0003-6951/2004/85(22)/5316/3/$22.00 5316 © 2004 American Institute of Physics 193.205.155.112 On: Fri, 26 Sep 2014 09:35:45
Seguini et al.
Appl. Phys. Lett., Vol. 85, No. 22, 29 November 2004
FIG. 1. Cube root of the IPE yield as a function of photon energy for the Al/ Lu2O3/n-(100)Si sample at different positive applied voltages (V): (䊐) 1.1, (䉯) 1.2, (䉰) 1.3, (〫) 1.4 (䉮) 1.5, (䉭) 1.6, (䊊) 1.7. The lines represent the linear fitting. The arrow indicates the energy of the direct optical transition in silicon.
values gives a barrier of ⌽e = 3.2± 0.1 eV attributed to electronic transitions from the silicon valence band to the oxide conduction band. The contribution from electrons in the silicon conduction band is negligible, considering the low doping of the substrate.3 Subtracting to ⌽e the silicon band gap (1.1 eV), we obtained a CBO of 2.1± 0.1 eV. As already reported for Al2O3 and ZrO2, for photon energies lower than the transition threshold, there is a subbarrier emission which is related to energy levels within the gap of amorphous materials.3 We show in Fig. 2 the Fowler plot Y 1/2共h兲 for the Al/ Lu2O3/p-Si sample with negative applied polarization. In this case, the lowest spectral threshold probes the metal/oxide interface and gives a value of the energy barrier between the aluminum Fermi level and the conduction band of lutetia. The value obtained for this barrier is 2.4± 0.1 eV. The same value is measured in inversion on a Al/Lu2O3/n-Si sample. The change in slope at about 3.7 eV has been attributed to the transition of holes from the conduction band of silicon to the lutetia valence band similarly to what has been reported for HfO2 on Si,4 and Ge.11 The extrapolated value for this barrier is ⌽h = 3.7± 0.1 eV. This value is field independent, as already reported for hole emission.4 Subtracting the silicon band gap to ⌽h, a valence-band offset (VBO) of 2.6± 0.1 eV can be determined. The yield reduction below 4.5 eV, also shown in Fig. 2, is related to a high-symmetry transition in silicon. The energy barrier between the Fermi level of gold and the conduction band of lutetia was measured (data not shown) to be around 3.6 eV. The uncertainty on this value is due to the overlap with ⌽h.
5317
FIG. 3. On the right-hand side the photoconductivity spectral dependence from the square-root quantum yield for the Al/Lu2O3/p-(100)Si sample at the negative applied voltage of ⫺1.5 V. The line indicates the linear fitting. On the left-hand side the optical absorption coefficient as a function of the photon energy for Lu2O3 grown on quartz. The line indicates the parabolic fitting.
The PC spectra allow the evaluation of the oxide transport band gap extrapolating Y 1/2共h兲 to zero. With this method we found EG = 5.8± 0.1 eV, as shown in Fig. 3. This value is independent of the bias polarity and of the doping type of the silicon substrate, proving that it is an oxide transition and not an interfacial one. The sum of CBO and VBO, measured by IPE, and of the silicon band gap gives the same value of EG. The CBO, VBO, and EG values determined in the present work can be compared with those (CBO= 1.88, VBO= 3, EG = 6 eV) obtained using the O 1s photoelectron spectra of Lu2O3 films deposited by electron beam evaporation.12 It is also noteworthy that, within the experimental error, lutetia shows a band gap and electron energy barriers at the metal (Au or Al)/oxide and silicon/oxide interfaces similar to those observed for hafnia.4 The measured CBO is, within the experimental error, identical to the one reported for Al2O3. Figure 3 shows the absorption spectrum for a film grown on a quartz substrate. As the reflectance R is negligible, the absorption coefficient ␣ was related to the optical transmittance T by ␣ = −共1 / t兲共lnT兲, where t is the film thickness. The upper part of the curve is well fitted with the Tauc formula,
␣共h兲 ⬀ 共h − EOPT兲2 , which gives EOPT = 4.89± 0.02 eV. The difference between EG and EOPT is about 0.9 eV, and can be related to tails of localized states within the oxide band gap.13 Figure 4 summarizes the band diagram of the lutetiabased MOS structure. It is not possible to determine exactly the position of the optical band gap with respect to the transport band gap, because the conduction- and valence-band densities of states are unknown.
FIG. 2. Square root of the IPE yield as function of photon energy for the Al/ FIG. 4. Energy-band diagram for Lu2O3 based MOS structures. The origin Lu2O3is/p-(100)Si of the energy scale is placed at the top of the silicon valenceDownloaded band. The to IP: sample at differentinnegative applied voltages (V):content The lines This article copyrighted as indicated the article. Reuse of AIP is subject to the terms at: http://scitation.aip.org/termsconditions. experimental error is ⫾0.1 eV. represent the linear fitting, (䊐) 1.5, (䊊) ⫺1.7, (䉭) ⫺1.9. 193.205.155.112 On: Fri, 26 Sep 2014 09:35:45
5318
The CBO (2.1 eV) and VBO (2.6 eV) values indicate that the band alignment at the interface between silicon and lutetia is symmetric. The absolute values are compatible with gate dielectric applications. This work was partly supported by the INFM PAIS Project REOHK and the Italian Ministry for Foreign Affairs. 1
Seguini et al.
Appl. Phys. Lett., Vol. 85, No. 22, 29 November 2004
G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89, 5243 (2001). 2 V. K. Adamchuck and V. V. Afanas’ev, Prog. Surf. Sci. 41, 111 (1992). 3 V. V. Afanas’ev, M. Houssa, A. Stesmans, and M. M. Heyns, Appl. Phys. Lett. 78, 3073 (2001).
4
V. V. Afanas’ev, A. Stesmans, F. Chen, X. Shi, and S. A. Campbell, Appl. Phys. Lett. 81, 1053 (2002). 5 D. G. Schlom and J. H. Haeni, MRS Bull. 27, 198 (2002). 6 L. Marsella and V. Fiorentini, Phys. Rev. B 69, 172103 (2004). 7 J. Robertson, J. Vac. Sci. Technol. B 18, 1785 (2000). 8 G. Scarel, E. Bonera, C. Wiemer, G. Tallarida, S. Spiga, M. Fanciulli, I. L. Fedushkin, H. Schumann, Yu. Lebedinskii, and A. Zenkevich, Appl. Phys. Lett. 85, 630 (2004). 9 P. V. Dressendorfer and R. C. Barker, Appl. Phys. Lett. 36, 933 (1980). 10 R. J. Powell, J. Appl. Phys. 41, 2424 (1970). 11 V. V. Afanas’ev and A. Stesmans, Appl. Phys. Lett. 84, 2319 (2004). 12 H. Nohira, T. Shiraishi, T. Nakamura, K. Takahashi, M. Takeda, S. Ohmi, H. Iwai, and T. Hattori, Appl. Surf. Sci. 216, 234 (2003). 13 J. Tyczkowski and R. Ledzion, J. Appl. Phys. 86, 4412 (1999).
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: 193.205.155.112 On: Fri, 26 Sep 2014 09:35:45
