Chapter 2 Literature Review:
3.3 Fundamental of characterization techniques
3.3.5 X-ray electron spectroscopy (XPS)
X-ray photoelectron spectroscopy (XPS), also known as electron spectroscopy for chemical analysis (ESCA), is one of the most powerful and common chemical analysis techniques, especially for surface and interface analysis. XPS is based on the photoelectric effect in which the binding energy (EB) of a core-level electron is overcome by a sufficient impinging soft X-ray photon, and the core-level electron is excited and rejected from atom, called photoelectron (Figure 3.9) [17]. Determining the kinetic energy of photoelectron, i.e., binding energy EB will give meaningful chemical information of an analyzed sample.
When a photon impinges upon an atom, one of following phenomena may happen: (1) photon can pass through with no interaction, (2) photon is scattered by an atomic orbital electron, and (3) photon interacts with an atomic orbital electron with total energy transfer to electron, leading to electron emission from atom (Figure 3.9, right). If the photon is scattered, the phenomenon is referred to as “Compton scattering”. If the photon interacts with the electron, the phenomenon describes the photoemission process, a basic of XPS. To let the core-level electron emits from atom, the impinged photon energy, hn needs to be higher than the electron binding energy. EB. Electrons emitted from atoms by this process are called photoelectrons. The kinetic energy of a photoelectron, KE is simply described by the Einstein’s equation:
EB = ℎ𝜈 − 𝐾𝐸 (3-5)
One can measure the kinetic energy of photoelectron the binding energy is obtained according to equation (3-5). The value of EB will provides valuable information about photo-emitting atom. The emission of core-level electron will result in the rearrangement of atomic
29
orbitals and the emission of Auger electron or X-ray photon as described in Figure 3.10 [17].
Binding energy of the ejected photoelectron depends on the final state configurations after photoemission [17, 18].
The concept of the binding energy of an electron in an atom requires elaboration. A negatively charged electron will be bound to the atom b the positively charged nucleus. The closer the electron is to the nucleus, the more tightly it is expected to be bound. Binding energy will vary with the type of atom (i.e., a change in nuclear charge) and the addition of other atoms bound to that atom (bound atoms will alter the electron distribution on the atom of interest). Different isotopes of a given element have different numbers of neutrons in the nucleus, but the same nuclear charge. Changing the isotope will not appreciably affect the binding energy. Weak interactions between atoms such as those associated with crystallization or hydrogen bonding will not alter the electron distribution sufficiently to change the
measured binding energy. Therefore, the variations in the binding energy that provide us with the chemical information content of XPS are associated with covalent or ionic bonds between atoms. These changes in binding energy are called binding energy shifts or chemical shifts.
Figure 3.11 presents a simplified schematic diagram of an X-ray photoelectron spectrometer. The photons generated from the X-ray source impinge upon the sample, resulting in the ejection of photoelectrons from sample. The photoelectrons are collected by electron optics and directed into an electron energy analyzer where they are sorted by energy.
The number of electrons per energy interval is then transduced to a current by an electron detector. The photocurrent is subsequently converted and processed into a spectrum by suitable electronics. The experiment is typically performed under ultra-high vacuum (UHV) conditions, about 10-9-10-11 torr. This high vacuum is needed in order to maintain sample surface integrity (the surface gas adsorption) and minimize the scattering of photoelectrons by others gas molecules [18]. Due to the relatively short inelastic mean free path in the irradiated material and the typical kinetic energies possessed by the photoelectrons, only the
30
photoelectrons produced in the top several mono-atomic-layers of the sample are observed as their characteristics energies. Thus, the XPS is typically useful for surface and interface analysis.
31 Figure 3.1 A typical scheme of a CVD reactor
Figure 3.2 Schematic of UHVCVD system
Main chamber Load lock
SiH
4
Turbo pump MFC Turbo pump
Mechanical pump Mechanical pump
Transfer rod
GeH
4
32
Figure 3.3 Plot of the excitation volume generated by the SEM electron beam
Figure 3.4 A schematic of transmission electron microscope
33 Figure 3.5 A schematic of atomic force microscopy
Figure 3.6 Dependence of interatomic force on tip-sample separation
34
Figure 3.7 Schematic band diagrams for the photoluminescence processes
Figure 3.8 Schematic of an emission experiment of PL measurement
35
Figure 3.9 Surface irradiated by sufficient energy X-ray photon beam will emit photoelectrons: phenomenon (left) and principle schematic (right) [17]
Figure 3.10 (a) The X-ray photon transfers its energy to a core-level electron leading to photoemission from the n-electron initial state. (b) The atom, now in an (n-1)-electron state, can reorganize by dropping an electron from a higher energy level to the vacant core hole. (c) Since the electron in (b) dropped to a lower energy state, the atom can rid itself of excess energy by ejecting an electron from a higher energy level. This ejected electron is referred to as an Auger electron. The atom can also shed energy by emitting an X-ray photon, a process called X-ray fluorescence.
36
Figure 3.11 Schematic design of an X-ray photoelectron spectrometer
37
Chapter 4
Results and Discussion
In this chapter, the properties of epitaxial Ge grown on In0.5Ga0.5P/GaAs are discussed.
The effect of surface coverage on incubation will be demonstrated. Growth mode of Ge on In0.5Ga0.5P (100) is calculated and shown by various characterization methods. The
discontinuous variation of the surface roughness is discussed as well.
And then, high crystal quality of Ge epitaxial films are examined by HR-TEM. Direct band gap emission (0.8 eV) of this structure was detected by photoluminescence.
4.1 Incubation Time and Growth Rate
The same value of the growth rate at 28.9 nm per minute and the incubation time (Tinc) of 29.4 and 38.9 minutes for InGaP surface indium coverage of 27.82% and 45.84%,
respectively, were measured for the Ge epitaxial layer as shown in Figure 4.1. In our previous study [5,19], various incubation times of 14.4, 14.2 and 8.3 minutes were found of the Ge growth on GaAs (100), (110), and (111)A substrates, respectively, at the growth temperature of 600°C. The main cause for the different incubation times lies in the surface configurations of the GaAs substrates. As the Ge-Ga dimer has a lower energy state than the Ge-As dimer [20], the Ge atoms would easily bond with Ga atoms instead of As atoms. As a consequence, longer incubation times is needed for As desorption on the GaAs (100) and (110) substrate with 50% Ga atom-terminated and 50% As atom-terminated on the surface to form Ga-rich surface and, thus, to enhance the bonding between Ge and Ga atoms; whereas the shorter incubation time is needed for GaAs (111)A substrate whose surface configuration is 100% Ga atom-terminated. Bai et al. has further indicated that these incubation times were effected from the beginning of Ge deposition by the higher formation energy of Ge-As bond [21]. In compare with the Ge grown on InGaP layers, longer incubation times are measured at high In
38
coverage of InGaP layers. According to the report by Luo et al. [22], the Ge-In bond has greater formation energy than the Ge-As bond, which would similarly result in longer incubation time for the Ge growth. A much longer incubation time than Ge grown on GaAs substrates was measured for Ge grown on InGaAs in their report. In our study, the Tinc of Ge grown on InGaP layer is relatively higher than that of Ge on GaAs. As the In coverage of the InGaP layer increases, the incubation time increases as well.
The XPS results further summarized the composition ratio of the InGaP layer surfaces in Figure 4.1. The InGaP surfaces have In coverage of 27.82% and 45.84%, Ga coverage of 53.24% and 5.12%, and P coverage of 18.94% and 49.04%, respectively. The In atoms on the surface impeded the Ge adatom attachment to Ga sites similar to the effect of surface As atoms during the Ge on GaAs growth. As a consequence, the higher content of In on the surface, the longer the incubation time is.
On the other hand, much higher growth rate was obtained when the source gas flow rate was doubled. Figure 4.2 shows an enhancement of the growth rate from 28.9 nm/min up to 50.0 nm/min at gas flow rate of 10 and 20 sccm, respectively, with almost none influence on the incubation time due to the growth on the sample InGaP substrate. As a result, the
incubation time is only dependent on surface coverage of substrates instead of growth rate in this study.
4.2 Growth Mode of Ge on In0.5Ga0.5P (100)
The growth modes of Ge can be explained by the change of surface energy, so called the thermodynamic theory of capillarity [23]. An illustration of the basic processes of vapor deposition on a surface of a substrate is shown in the Figure 4.3. Island formation is assumed when atoms and molecules are impinging on the substrate. Young’s equation between the interfacial tensions at equilibrium yields
𝛾𝑠𝑣 = 𝛾𝑓𝑠+ 𝛾𝑣𝑓𝑐𝑜𝑠𝜃 (4-1)
where γvf is the interfacial tension between the vapor and the film, γfs is the interfacial tension
39
between the film and the substrate, and γsv is the interfacial tension between the substrate and the vapor.
For island growth mode, as referred to the Volmer-Weber growth, θ>0, which yields 𝛾𝑠𝑣 < 𝛾𝑓𝑠+ 𝛾𝑣𝑓. For layer growth mode, as referred to the Frank-van der Merwe growth, the deposit wets the substrate and θ=0, hence 𝛾𝑠𝑣 = 𝛾𝑓𝑠+ 𝛾𝑣𝑓. Otherwise, the 2D-3D mixed growth mode, as referred to the Stranski-Krastanov growth, fulfills the inequality 𝛾𝑠𝑣 > 𝛾𝑓𝑠+ 𝛾𝑣𝑓.
Tang et al. has reported that the interfacial energy (or tension) is negligible (γfs=0) when the lattice constants are almost the same [19]. The surface energy of Ge (100) and In0.5Ga0.5P (100) are 1.02 and 0.996 J/m2, respectively. Therefore, the growth mode of Ge epitaxial on InGaP layer is the Volmer-Weber growth.
The pattern of the Volmer-Weber growth was further demonstrated by the top-viewed SEM images and cross-sectional TEM images in Figure 4.4 and Figure 4.5, respectively. The 3D Ge islands would form to reduce the total surface energy at the early stage, and started to merge into a film later on. Finally, the Ge epitaxial film was observed.
4.3 Surface Roughness
According to the Figure 4.6 of surface roughness versus growth time, an abrupt up-step into a discontinuity at growth time between 45 and 50 minutes is shown because of the lack of surface diffusion. As soon as the InGaP surface was fully covered by Ge, the growth rate increased by almost 3 times, 10.0 nm/min to 28.9 nm/min afterward, due to the lower Ge-Ge formation energy as compared with Ge attachment to InGaP, thus, the high growth rate gave rise to the reduction of surface diffusion. As a consequence, rough surface morphology of Ge film was detected.
Porsche et al. has reported that the amount of islands would be increased for growth conditions with reduced surface diffusion, such as low temperatures or high growth rates [24].
Under these conditions, it was found in Figure 4.7 that the values of surface roughness were
40
higher all the time. Therefore, in order to have low surface roughness for thick Ge film, high temperatures and low growth rates, i.e. by decreasing gas flow rate, are recommended.
4.4 Crystal Quality
The good single crystal quality of Ge epitaxial film with thickness of 190 nm and islands with height of 40 nm was characterized by cross-sectional images of high-resolution
transmission electron microscopy (HR-TEM). It also showed good interface quality between Ge and InGaP. Figure 4.8 shows a cross-sectional TEM micrograph of Ge deposited on In0.5Ga0.5P/GaAs substrate at 500°C. There is no appearance of any threading dislocation, which is expected. The lattice mismatch between Ge and In0.5Ga0.5P is extremely small, which is 0.09% at room temperature and about 0.1% at growth temperature. The lattice constant of In0.5Ga0.5P is 5.653 Å , calculated by InP and GaP with lattice constants of 5.869 Å and 5.451 Å , respectively, at room temperature. The diffraction pattern of Ge film, punctiform
diffraction spots were characterized because of the consistency of the lattice.
4.5 Interdiffusion
Interdiffusion between III-V and other group materials, i.e. germanium, has been a key challenge for the reliability and performance of devices. TEM-EDS line concentration profile of indium(purple), arsenic (blue), germanium (green), gallium (light green), and phosphorous (red) along the line drawn across the Ge/In0.5Ga0.5P/GaAs structure was characterized in Figure 4.9. The profile has clear separation at the edge of each interface. Sharp interface with only a few nanometers of interdiffusion is demonstrated.
For the fabrication of Ge p-channel MOSFETs, the two-dimensional hole gas (2DHG) will form in the Ge layer near edge of bottom interface; therefore, good interface
characteristics including minimal interdiffusion and low defect density are of immense importance.
41 4.6 Photoluminescence
N-type doping of Ge will compensate the 0.136 eV difference in energy between Г and L valleys. Figure 4.10 is the room temperature PL infrared emission of 1.8 μm thick Ge
epitaxial layer on In0.5Ga0.5P/GaAs substrate with 330-mW PL incident laser power. The PL peak at 0.8 eV indicates the electrons in the Г valley recombine with holes in the valence band that makes the direct band-gap emission occur. The emission at the range of 650 to 750 meV is also detected in the spectrum that revealed indirect emission and had much lower intensity.
The detection of direct emission in the PL measurement indicates that the defect density of the Ge film is very low. Otherwise, non-radiative recombination due to the large number of defects would decrease the intensity of the L emission.
42
Figure 4.1 (top) Plot of Ge film thickness versus growth time on different surface coverage InGaP layers. The incubation time are 29.4 and 38.9 minutes for indium coverage of 27.82%
and 45.84%, respectively. (bottom) The analysis of composition ratio on InGaP surfaces revealed by XPS
43
Figure 4.2 Plot of Ge film thickness versus growth time with GeH4 gas flow rate of 10 and 20 sccm, respectively, during growth
Figure 4.3 An illustration of the basic processes of vapor deposition on a surface of a substrate
44
Figure 4.4 Top view surface morphology of Ge grown on In0.5Ga0.5P/GaAs characterized by SEM at growth time of (a) 20, (b) 38, and (c) 45 minutes
45
Figure 4.5 Cross-sectional view surface morphology of Ge grown on In0.5Ga0.5P/GaAs characterized by TEM at thickness of (a) 40, and (b) 190 nm
46
Figure 4.6 Plot of surface roughness and film thickness of Ge versus growth time
Figure 4.7 Plot of surface roughness of Ge versus growth time in different growth conditions
47
Figure 4.8 Cross-sectional high-resolution TEM image and diffraction pattern of 190 nm Ge epitaxial layer
Figure 4.9 High-resolution TEM microstructure and the EDS line scan profile across two interfaces of Ge/In0.5Ga0.5P/GaAs structure
400
300
Ge InGaP
GaAs
200
100
0
0 0.5 1.0 Position (µm)
Counts
48
Figure 4.10 Room temperature photoluminescence infrared emission from the structure of 1.8 μm Ge film on In0.5Ga0.5P/GaAs substrate. The direct band gap emission occurs at 0.8 eV.
0.60 0.65 0.70 0.75 0.80 0.85 0.90
0.00E+000
600 650 700 750 800 850 900
Energy (meV)
49
Chapter 5
Conclusions
High quality epitaxial Ge films were successfully grown on In0.5Ga0.5P/GaAs (100) by UHVCVD, as confirmed by TEM. This is the first study of Ge grown on InGaP layer to provide with some characteristics of this structure. A longer incubation time is needed for high indium surface coverage of InGaP. The growth mode of Ge on In0.5Ga0.5P/GaAs (100) is the Volmer-Weber growth, calculated by the thermodynamic theory of capillarity, as well as examined by top-viewed SEM and cross-sectional TEM images. With continuous growth of Ge on InGaP, rough surface would formed as soon as the InGaP surface is completely covered by Ge because the Ge-Ge attachment has lower adatom bonding energy than the formation energy of bonds between Ge and InGaP, resulting in the intensively enhancement of growth rate as well as surface roughness. The main reason is that the surface diffusion is reduced because of high growth rate. As a consequence, for good surface morphology of the Ge film, growth conditions of high temperature and low growth rate are suggested.
The Ge epitaxial film on In0.5Ga0.5P/GaAs (100) has shown sharp interface with
interdiffusion depth as low as the requirement of device applications [25]. And a direct band gap emission at 0.8 eV was detected by PL. This structure studied is useful for the future integration of Ge p-channel and III-V n-channel MOSFETs on the same GaAs template for beyond Si-CMOS logic applications.
50
Reference
[1] S.E. Thompson, M. Armstrong, C. Auth, S. Cea, R. Chau, G. Glass, T. Hoffman, J.
Klaus, M. Zhiyong, B. Mcintyre, A. Murthy, B. Obradovic, L, Shifren, S. Sivakumar, S.
Tyagi, T. Ghani, K. Mistry, M. Bohr, and Y. El-Mansy, IEEE Electron Device Lett. 25, 191 (2004)
[2] T. Akatsu, C. Deguet, L. Sanchez, F. Allibert, D. Rouchon, T. Signamarcheix, C.
Richtarch, A. Boussagol, V. Loup, F. Mazen, J.M. Hartmann, Y. Campidelli, L. Clavelier, F. Letertre, N. Kernevez, C. Mazure, Mater. Sci. Semicond. Process, vol.9, 444, (2006) [3] Yosuke Nakakita, Ryosho Nakane, Takashi Sasada, Hiroshi Matsubara, Mitsuru
Takenaka and Shinichi Takagi, IEEE international Electron Devices Meeting, pp. 1-4, (2008)
[4] Keisuke Yamamoto, Ryuji Ueno, Takeshi Yamanaka, Kana Hirayama, Haigui Yang, Dong Wang, and Hiroshi Nakashima, Applied Physics Express, vol.4, 051301, (2011) [5] S.H. Tang, E.Y. Chang, M. Hudait, J.S. Maa, C.W. Liu, G.L. Luo, H.D. Trinh, and Y.H.
Su, Appl. Phys. Lett., vol.98, 161905, (2011)
[6] Manijeh Razeghi, “The MOCVD Challenge: A survey of GaInAsP-GaAs for photonic and electronic device applications”
[7] Y. Dong, R.M. Feenstra, M.P. Semtsiv, W.T. Masselink, Applied Physics Letters 84, 227 (2004)
[8] W.T. Masselink, M. Zachau, T.W. Hickmott, and K. Hendrickson, J. Vac. Sci. Technol.
B10, 966 (1992)
[9] H.A. McKay, H. Chen, R.M. Feenstra, and P.J. Poole, J. Vac. Sci. Technol. B21, 18 (2003)
[10] M. Dubey, K.A. Jones, D.W. Eckart, L.M. Casas, and R.L. Pfeffer, Applied Physics Letters 64, 2697 (1994)
51
[11] S. Meyerson, E. Ganin, D.A. Smith, and T.N. Nguyen, J. Electrochem. Soc., vol.133 (6), p1232, (1986)
[12] D. Kruger, T. Morgenstern, R. Kurps, E. Bugiel, C. Quick, and H. Kuhne, J. Appl. Phys., vol. 75 (12), p7829, (1994)
[13] D.W. Greve, Physics of Thin Films, vol. 23, 1-82, (1997)
[14] V. Zela, UHV-CVD growth of Ge/Si nanostructures, 18-21, (2006)
[15] B.S. Meyerson, Chemistry, Physics and Device Applications, Proceedings of the IEEE 80, 1592-1608, (1992)
[16] J.S. Song, Y.C. Choi, S.H. Seo, D.C. Oh, M.W. Cho, T. Yao, and M.H. Oh, Journal of Crystal Growth 264, 98-103, (2004)
[17] J.C. Vickerman, and I.S. Gilmore, Editors, “Surface analysis: the principal techniques,”
John Wiley and Sons Ltd, ISBN 978-0-470-01763-0, 47-109, (2009)
[18] D.R. Vij, Editor, “Handbook of Applied Solid State Spectroscopy,” Springer, ISBN-13:
978-0387-32497-5, 486-507, (2006)
[19] S.H. Tang, C.I. Kuo, H.D. Trinh, E.Y. Chang, H.Q. Nguyen, C.L. Nguyen, and G.L. Luo, J. Vac. Sci. Technol. B 31, 021203, (2013)
[20] X.S. Wang, K.W. Self, and W.H. Weinberg, J. Vac. Sci. Technol. A 12, 1920, (1994) [21] Y. Bai, K.E. Lee, C.W. Cheng, M.L. Lee, and E.A. Fitzgerald, J. Appl. Phys. 104,
084518, (2008)
[22] G.L. Luo, Z.Y. Han, C.H. Chien, C.H. Ko, C.H. Wann, H.Y. Lin, Y.L. Shen, C.T. Chung, S.C. Huang, C.C. Cheng, and C.Y. Chang, J. Electrochem. Soc. 157, H27, (2010)
[23] O. Manasreh, Semiconductor Heterojunction and Nanostructures (McGraw Hill, Asia, 2005)
[24] J. Porsche, A. Ruf, M. Geiger, and F. Scholz, J. Cryst. Growth 195, 591-595, (1998) [25] Y.B. Cheng, C.K. Chia, Y. Chai, and D.Z.Chi, Thin Solid Films 522, 340-344, (2012)