P. Xiong

Magnetic Properties of Atomic-Layer-Deposited Hafnium Dioxide

Unexpected magnetism was reported recently in undoped hafnium dioxide (HfO2) thin films grown by pulsed-laser-deposition (PLD) [1]. The results pose a challenge to the common understanding that magnetic order in an insulator requires the presence of cations with partially filled shells of d or f electrons. The Hf 4+ ion has a closed shell [Xe] 4f configuration. Some theoretical models, based on ferromagnetism driven by intrinsic point defects in HfO2 [1,2], have been proposed to explain the experimental results. However, there remain open questions regarding the origin of the results and even debate on whether the observed magnetism is an intrinsic effect. HfO2 has attracted enormous attention in microelectronic device field in the past few years since it is a leading candidate to replace SiO2 as the gate dielectric in future sub-100 nm CMOS devices. In particular, HfO2 gate dielectrics grown using atomic layer deposition (ALD) have been heavily studied, and they have demonstrated very encouraging electrical performance with outstanding uniformity and conformality. In this paper, we report for the first time a study of the magnetic property of this very important ALD electronic material. HfO2 films, 30 to 600 Å thick, were grown by ALD on p-Si (100) substrates. HfO2 films were deposited using alternating pulses of HfCl4 and H2O with N2 as a carrier gas. The ALD process was performed in an ASM Pulsar 2000 module at 300 C and a chamber pressure of ~1.5 Torr. The purity of the ALD HfO2 films is known to be extremely high, having been widely used in state-of-the-art CMOS device research, and this ALD reaction chamber has never been exposed to any magnetic materials. Figure 1(a) shows that the as-deposited ALD HfO2 layer is noncrystalline on SiO2, although there are signatures of ordering on a subnanometer scale. Figure 1(b) shows the same stack after annealing at 600 C for 30 s. The annealing was performed ex situ in a rapid thermal annealing chamber. The clearly visible HfO2 lattice fringes indicate crystallization into a well-ordered microstructure [3]. The size and the area of the crystallites are observed to increase with the annealing temperature. Measurement of magnetic moment vs. applied magnetic field is performed with a Quantum Design SQUID magnetometer at temperatures between 5 and 400 K. The magnetic field, up to 5T, is applied perpendicular to the plane of the sample. Figure 2 shows the raw magnetic moment of a 600 Å HfO2 film including the diamagnetic background from the Si substrate. The dashed lines indicate the diamagnetic contributions, which are subsequently subtracted. Figure 3 plots the magnetization of two 600 C-annealed HfO2 films of 300 Å and 600 Å in thickness. The coercive force is about 100 Oe at 5K. The apparent independence of the magnetic properties on the thickness clearly demonstrates that the observed magnetism is a bulk property of the films and not a mere reflection of the interface. In contrast, Figure 4 shows the strong dependence of the magnetic properties on the annealing temperature: Ms for the 600 C sample is about twice that of the 500 C sample. The difference is likely due to the better crystalline order resulting from higher annealing temperature, and it signals a high sensitivity of the observed magnetism to the film crystallinity. In Figure 5, we plot Ms as a function of the measurement temperature for the 600 Å film annealed at 600 C. This temperature dependence is consistent with a coexistence of two magnetic phases [4]: a paramagnetic part which shows up prominently at low temperatures in the form of a Curie tail and a ferromagnetic one whose ordering temperature is beyond 400 K. In summary, we have experimentally demonstrated that high-purity ALD HfO2 is magnetic. This discovery could offer a new mechanism to account for the degradation of carrier mobility on HfO2 based CMOS devices. It could also open the possibility to use this magnetic oxide as an element in novel spintronic devices.

FEASIBILITY OF MICRO-COILS

The feasibility of micro-coils of the so-called pancake geometry is investigated. Different materials are being compared for optimum conductor choice, including electrical conductivity, specific heat, magneto-resistance and mechanical properties. The modalities of operation are presented. The feasibility of manufacturing such coils is examined. Planar coils consisting of about 100 turns with conductor widths as small as 3 pm are being fabricated. The turns will be deposited either directly onto Si or onto SiOJSi. All patterns will be defined by photolithography. The results show that the maximum stress is very high for such micro-coils. The maximum stress may be reduced by adjusting the current density.