The uniformity, defined as the range of thickness divided by 2 times of the average thickness, of as-deposited film thickness measured with N&K analyzer is between 4% and 5%
for both Hf and HfO2 films at all thickness. Fig.6-3 shows that the surface roughness of 2 nm thick Hf film measured by AFM is 0.253nm, which is very close to that of original Si surface and is much smaller than the thickness of deposited film. The continuity of films was further characterized with plane-view SEM inspection, no pin holes and agglomeration phenomenon were observed. All of these results confirm the homogeneity of the deposited films.
Fig.6-4 shows the HFCV curve of the sample H73. The curve looks normal but the calculated effective K-value is only 8.5, which is much lower than the typical K-value of HfO2 film. To check if the deposited film is normal, TiN/Hf/HfO2/Hf/TiN structure was prepared. The thickness of dielectric is 12.5 nm from the cross-sectional TEM inspection. The exact K-value of HfO2 film was calculated to be 27.9 and this value is consistent with the reported K-value of HfO2, 20~30. To understand why the effective K-value of sample H73 is so low, high resolution TEM was employed to inspect the actual sample structure. Fig.6-5 shows the cross sectional TEM micrograph of sample H73. Because of the poor adhesion between Pt and HfO2, the Pt layer peeled-off during sample preparation. A thick interfacial layer of 3.2 nm thick between HfO2 and Si substrate is clearly observed. This observation is consistent with that reported previously [27]. Because HfO2 was deposited immediately after HF-dip step, there should not be such an unusual thick native oxide on the Si surface. In fact, the native oxide thickness measured on Si wafer just after HF dip process by N&K analyzer is less than 0.3 nm.
The interfacial layer may be either silicon dioxide or Hf-silicate. To clarify what material it is, the K-value of the interfacial layer was estimated using the thickness of interfacial layer and the thickness of HfO2 layer measured from Fig.6-5, and the dielectric constant of HfO2 is assumed to be 27.9. The estimated K value of the interfacial layer is 3.8, which indicates that
Since the sample shown in Fig.6-5 did not experience PDA, the thick interfacial oxide must be grown during the HfO2 deposition period. It should be noted that the PDA process used in this work only produce a SiO2 layer thinner than 0.5 nm on bare Si. Even if the sample experienced PDA, the 3.2nm thick SiO2-liked IL can not be attributed to the PDA process.
Up to now, a fact can be sure is that the thick oxide layer is formed during the reactive sputter process. Why reactive sputter introduces such a thick oxide layer? The temperature during sputtering is only 100℃ and the content of oxygen is much lower than that in typical oxidation furnace. The partial pressure of oxygen during sputtering is kept at 0.5 mTorr and the deposition time is shorter than 15 min. So it is impossible that this thick SiO2 layer comes from the oxidation of the Si substrate by reacting with oxygen molecule under such a low thermal budget and short period of time.
In order to investigate the origin of interfacial SiO2, optical emission spectroscopy (OES) was employed to detect the chemical state in sputter chamber during film deposition. Fig.6-6 shows the OES spectrums for various gas conditions. Only O-radical instead of O2-radical is detected in the sputtering chamber, while in typical O2 plasma system, O2-radical is the main radical detected [28]. Based on the observation, an O-radicals enhanced oxidation model is proposed to explain the thick interfacial oxide layer. The injected oxygen molecules are excited into ions and radicals. The oxygen related species in the plasma system may include O-ion, O2-ion, O-radical, O2-radical, etc. O2-radicals may collide with the other particles and decompose into O-radicals and/or O-ions. Some O-radicals and O2-radicals diffuse randomly to Si surface and form an interfacial oxide layer at the initial deposition stage. Ions are accelerated toward target and after bombarding the target, all of the O2-ions decompose into O-ions or O-radicals. Some reflected O-radicals move toward wafers with high energy.
Because O-radical has small radius and is highly reactive, it penetrates through the HfO2/Hf stack rapidly and during the penetration, it not only oxidized the bottom Hf layer but also reacts with the Si substrate to form SiO2. Therefore, a very thick SiO2 layer is formed during the reactive sputtering process. The oxidation due to the other O-contained species can not be ruled out totally. Since O-radical has the highest activity and the intensity of O-radical is much higher than the other species, it is believed that O-radicals play the major role.
Based on the mechanism of formation of thick interfacial SiO2, a series of HfO2/Hf stack
top HfO2 layer, thicker Hf layer results in higher effective K value. On the other hand, for the same thickness of bottom Hf layer, thicker HfO2 layer results in lower effective K value.
Because the final thickness of dielectric is different for different sample, the improvement of effective K value may be due to the reduction of interfacial oxide layer or simply due to the decrease of the thickness percentage of the interfacial oxide layer. Fig.6-8(a), (b), and (c) show the cross-sectional TEM micrographs of samples with HfO2/Hf stack of 7nm/1nm (sample H71), 7nm/5nm (sample H75), and 3nm/1nm (sample H31), respectively. The interfacial layer thickness depends on the HfO2/Hf stack structure. The correlation between interfacial layer thickness and bottom Hf thickness with the same upper HfO2 thickness is shown in Fig.6-9. By increasing the bottom Hf thickness to 5 nm, the thickness of interfacial oxide layer can be reduced to 1.3 nm, which is the same as that observed in the sample with only Hf layer (H03). This phenomenon implies that a 5 nm thick Hf is necessary to block the diffusion of O-radicals during reactive sputtering of HfO2. The origin of the 1-1.5 nm thick interfacial oxide layer is postulated to the surviving oxygen in the chamber or the traced oxygen impurity in Ar gas.
Correlation between interfacial layer thickness and upper HfO2 thickness is shown in Fig.6-10. Similar interfacial SiO2 thickness of about 3-4 nm was observed. This result indicates that the growth of interfacial oxide layer occurs at the early stage of HfO2 deposition.
With the increase of HfO2 thickness, oxidation is blocked.
6-4 Conclusions
In this work, the formation of interfacial SiO2 layer at HfO2/Si interface was studied comprehensively. It is observed that using physical vapor deposition technique, it is difficult to totally eliminate the formation of interfacial SiO2 layer. During reactive sputtering deposition of HfO2 layer, an interfacial SiO2 layer thicker than 3 nm would be grown. Such an unusual thick SiO2 layer is formed due to the enhanced oxidation of O-radicals generated in the sputtering chamber. Adoption of two-step deposition method, the thickness of interfacial SiO2 layer can be reduced only if the bottom Hf layer is thicker than 5 nm. However, the reduction of effective oxide thickness would be limited. Re-oxidation of Hf film sounds a
sputtering chamber must be well controlled.
References
[1]. B. Yu, H. Wang, C. Riccobene, Q. Xiang, M. R. Lin, Symp. on VLSI Tech. Dig., 90 (2000)
[2]. The International Technology Roadmap for Semiconductors (Semiconductor Industry Association, Sematech, Austin, 2001)
L. Kwong, Tech. Dig. IEDM, 141 (1999)
[9]. Donggun Park, Ya-chin King, Qiang Lu, Tsu-Jae King, Chenming Hu, Alexander Kalnitsky, Sing-Ping Tay, and Cia-Cheng Cheng, IEEE Electron Device Letters, 19, 441 (1998) Z. Yu, P. M. Zeitzoff, and Jason C. S. Woo, IEEE Tran. Electron Device, 46, 1537(1999) [14]. J. Robertson, MRS bulletin, MRS Bull. 27, 217 (2002)
[15]. H. F. Luan, S. J. Lee, S. C. Song, Y. L. Mao, Y. Senzaki, D. Roverts, and D. L. Kwong, IEDM Tech. Dig., 141 (1999)
[16]. B. H. Lee, L. Kang, W. J. Qi, R. Nieh, Y. Jeon, K. Onishi, and J. C. Lee, IEDM Tech.
Dig., 133 (1999)
[17]. S. P. Muraka and C. C. Chang, Appl. Phys. Lett. 37, 639 (1980)
[18]. Y. Kim, G. Gebara, M. Freiler, J. Barnett, D. Riley, J. Chen, K. Torres, J. Lim; B. Foran,
Symp. on VLSI Tech. Dig., 44 (2000)
[20]. H. J. Cho, C. S. Kang, K. Onishi, S. Gopalan, R. Nieh, R. Choi, E. Dharmarajan, and J.
C. Lee, Tech. Dig. IEDM, 655 (2001)
[21]. S. J. Lee, H. F. Luan, W. P. Bai, C. H. Lee, T. S. Jeon, Y. Senzaki, D. Roberts, and D. L.
Kwong, Dig. IEDM, 31 (2000)
[22]. C. Hobbs, H. Tseng, K. Reid, B. Taylor, L. Dip, L. Hebert, R. Garcia, R. Hegde, J. Grant, D. Gilmer, A. Franke, V. Dhandapani, M. Azak, L. Prabhu, R. Rai, S. Bagchi, J. Conner, S. Backer, F. Dumbuya, B. Nguyen, and P. Tobin, Dig. IEDM, 651 (2001)
[23]. E. P. Gusev et al., AVS Topical Conf. on Atomic Layer Deposition, May 14 (2001) [24]. M. Ritala and M. Leskela, Handbook of Thin Film Materials, H. S. Nalwa ed.,
Academic Press, Vol.1, Chap.2 (2001)
[25]. W. Zhu, T. P. Ma, T. Tamagawa, Y. Di, J. Kim, R. Carruthers, M. Gibson, T. Furukawa, IEDM Tech. Dig., 463 (2001)
[26]. Z. J. Luo, T. P. Ma, E. Cartier, M. Copel, T. Tamagawa, B. Halpern, Symp. on VLSI Tech. Dig., 135 (2001)
[27]. A. Callegari, E. Cartier, M. Gribelyuk, H. F. Okorn-Schmidt, and T. Zabel, J. Appl. Phys.
90, 6466(2001)
[28]. B. H. Lee, L. Kang, R. Nieh, W. J. Qi, and Jack C. Lee, Appl. Phys. Lett. 76, 1926 (2000)
[29]. C. S. Kang, H.-J. Cho, K. Onishi, R. Chio, R. Nief, S. Goplan, S. Krishnan, and Jack C.
Lee, Symp. on VLSI Tech. Dig., 146 (2002)
Table 6-1. Sample ID and film structures used in this work.
Sample ID H70 H71 H73 H75 H31 H51 H71 H91 H03
HfO2 (nm) 7 7 7 7 3 5 7 9 0
Hf (nm) 0 1 3 5 1 1 1 1 3
0 20 40 60 80 100 0
5 10 15 20 25 30
K=10
K=3.9 K=15
Effective Dielectric Constant
Thickness percentage of interfacial layer
Fig.6-1. Effective dielectric constant of HfO2/interfacial layer stack structure versus the thickness of interfacial layer with the dielectric constant of interfacial layer as parameter. The dielectric constant of HfO2 is assumed to be 27.
Fig.6-2. Schematic drawing of main process flow used in this work.
Fig.6-3. Atomic force microscopic image of 2 nm thick Hf film. The root-mean-square roughness is only 0.253nm.
-3 -2 -1 0 1 0.0
500.0p 1.0n 1.5n 2.0n 2.5n
Capacitance(F)
Voltage(V)
Fig.6-4. High frequency capacitance-voltage curve of sample H73. The curve looks normal but the calculated effective K-value is only 8.5.
Fig.6-5. Cross-sectional TEM micrograph of sample H73. A thick interfacial layer of 3.2 nm thick between HfO2 and Si substrate is clearly observed.
400 500 600 O*
O*
O* O*
O* O* Ar
Ar
O2:10 100W Ar/O2:24/8 100W
Ar/O2:24/8 300W
Intensity (A.U.)
Wavelength (nm)
Fig.6-6. Optical-emission-spectrums of various gas mixtures in the sputtering chamber.
Strong O-radical signals are detected in all cases.
(a)
(b)
Fig.6-7. Effective dielectric constant as a function of the thickness of (a) bottom Hf layer
Fig.6-9. Interfacial SiO2 layer thickness as a function of bottom Hf thickness with the same upper HfO2 thickness.
Fig.6-10. Interfacial SiO2 layer thickness as a function of upper HfO2 thickness with the same bottom Hf thickness.
Chapter 7
Electrical Characteristics of HfO2 Film Prepared by MOCVD
7-1 Introduction
As the dimensions of complementary metal oxide semiconductor (CMOS) devices are scaled into the nanometer regime, the equivalent oxide thickness (EOT) of the gate dielectric decreases steadily to thinner than 1nm [1]. Its leakage current under normal operation bias falls into the direct tunneling regime [2]. High-k materials are employed to increase the physical thickness of the gate insulator while maintaining the same EOT and gate capacitance, thus reduces significantly the tunneling leakage current. Although many high-k materials are proposed to replace silicon dioxide as gate insulator, HfO2 is the most promising one for its excellent advantages, such as a suitable dielectric constant (~25), high band-gap energy (~
5.9eV), suitable tunneling barrier height for both electron and hole (>1eV), and thermal compatibility with contemporary CMOS process [3-8].
As mentioned in previous chapter, many methods had been employed to deposit high-k materials, such as physical vapor deposition (PVD) [6], metal-organic chemical vapor deposition (MOCVD)[8], atomic layer deposition (ALD) [8], and jet vapor deposition (JVD) [9]. The advantages and disadvantages of PVD and MOCVD are listed in Table7-2. Although PVD is a simple technique for depositing new materials for evaluation in an academic organization, it may cause damage to the electrical devices and is not preferred by industries because of poor step coverage and thickness uniformity [6].
In this chapter, we used MOCVD system to deposit HfO2 film. For MOCVD system, it is known that precursors play a crucial role in determining process parameters, such as deposition temperature, vapor pressure, and contamination issues etc. In this thesis, the precursor is provided by the Axitron Co. and the details can not be disclosed at this moment.
Different gate electrode, such as Al, Ta-Pt, Pt and poly-Si, were employed to investigate the effect of gate electrode on HfO2. Different gas injection sequences, oxygen first and hafnium first, were study to simulate the two-step deposition method in PVD system and to reveal their
current-voltage (I-V) characteristics are measured to evaluate the film electrical properties.
Material and micro-structure analysis will be conducted in the next chapter.