We have studied high-k gate stacks since 2000. We have presented our results on high-k and also learned a lot at ECS symposium. We gave a talk concerning doped HfO2 in order to achieve higher-k [1], and then also discussed a number of high-k properties including the interface dipole formation at high-k/SiO2 [2], advanced ALD process [3], the interface dipole formation mechanism [4, 5], high-k/Ge gate stacks [6], non-equilibrium PDA-induced HfO2 phase transformation [7], and the interlayer scavenging kinetics in HfO2 gate stacks [8]. In this talk, we would like to look back on our way, and discuss the origin of versatile properties of HfO2. The dielectric constant is important in high-k technology, and ferroelectric HfO2 recently reported is also exciting for functional device applications [9]. Why does HfO2 show such variety of properties? We would like to understand HfO2 in more detail. HfO2 has already been used in advanced CMOS. It is technically important whether a new higher-k material should be searched for or HfO2 should be further used. We have been interested in the structural phase control of HfO2 in order to enhance the dielectric constant. The 1st-principles calculation predicted that high temperature phase HfO2 such as cubic or tetragonal one had higher-k than amorphous or monoclinic one [10]. The monoclinic phase, however, is thermodynamically stable at Si processing temperatures, so usually cubic or tetragonal phase HfO2 is not available. Nevertheless, the significant enhancement of the dielectric constant in HfO2 was experimentally demonstrated through the stabilization of the cubic or tetragonal phase. The trick is the structural phase transformation by doping another metal-oxide to HfO2 [1]. The typical example is Y2O3-doped HfO2. In the ceramics community, it had been actually well recognized since long before that Y2O3 doping could induce the oxygen vacancy in ZrO2 to maintain the charge neutrality, which stabilized the cubic or tetragonal phase [11]. Furthermore, another way to realize the symmetric phase HfO2 was through a non-equilibrium thermal process of amorphous HfO2 [7]. In particular, the fast ramping-up in the RTA process enabled us to achieve the symmetric phase HfO2. It means that the nucleation rather than growth process is more important for implementing high symmetric phase HfO2. In addition, the size effect has also been well investigated thermodynamically for characterizing the structural phase [12]. Very unfortunately for us, we did not notice that HfO2 ferroelectricity came out by doping other oxides. An interesting point in the ferroelectric HfO2 is that the ferroelectric orthorhombic phase is not in the conventional phase diagram. Very recently we found that ferroelectric HfO2 was also made by anion doping into HfO2 [13]. The key is how to realize non-centrosymmetric phase HfO2 structure. Ferroelectric orthorhombic phase might be between monoclinic and tetragonal phases as a metastable state. This makes it more difficult to recognize such unexpected fabulous functionality easily. Materials are so challenging but so exciting.
Read full abstract