FeRAMs

To characterize ferroelectric materials usually the dependence of the polarization on the applied voltage is measured by means of a Sawyer-Tower circuit or by recording the current response to a voltage step. The / (V/)-hys(crcsis curve is used to determine the remanent polarization and coercive voltage, respectively coercive field. These two parameters are of critical importance to the design of external circuits of FeRAMs. 

To emulate the operation of the FeRAM cell of the integrated circuit the measurement setup has to generate pulses of both polarities. The Shunt method as it is described in Section 3.2.2 is useful to exclude the influence of the sense capacitor and to reach high speed. 

Ferroelectric thin films considerably gain in interest within the last couple of years due to their potential application in nonvolatile random-accessmemory devices (FeRAM). Among potential candidates, PbZr. n i, (>> (pzt) is one of the most promising materials because of its large remanent polarization and low coercive field. However, pzt is also well known for its poor fatigue behavior on metal electrodes [1,2] and occurrence of size effects [3-5] which are well due to the ferroelectric/electrode interface properties [1-5]. 

The progress in the development and integration of ferroelectric memories (FeRAM) leads to increasing demand for electrical characterization of sub-micron structures. This article will point out the measurement problems arising from the reduction of the ferroelectric capacitor size e.g. from memory cells or nanostorage devices. Procedures and solutions are presented to overcome these problems and to increase further the resolution and speed of ferroelectric characterization to be ahead of the technological demand. 

A scheme of a typical memory cell of an integrated 32 Mbit Chain FeRAM device is shown in Figure 17.2 [2], Here the cell capacitor has an area of 0.49 /nm2, but sizes down to 0.09 /nm2 for integrated capacitor area used in devices. The fabrication of these devices can only be done by memory manufacturers since the whole processing has to be performed, including CMOS, ferroelectric capacitor and electrodes, and metalization to get a functional device. To perform measurements on these capacitors, the device must have the possibility to access their bottom and top electrodes, e.g. through the CMOS circuit or through additional contact pads. 

Figure 17.16 Hysteresis pulse measurement of a single FeRAM cell capacitor (0.49 /um2 PZT)...

Fatigue in FeRAM macroscopic results invoking nano scale features... 

There are two main approaches or configurations for the AFM-assisted detection of the local piezoelectric activity (pfm) in ferroelectric thin films for ferroelectric memory applications (FeRAM). The most used one was introduced in the early 90s and uses a conductive afm-tip as both top electrode and sensor for the induced vibration [2-4]. The second and more recent one [15,16], uses a normal metallic thin top electrode to apply the electric field and the vibration signal is detected by the AFM-tip above the top electrode. Both approaches present numerous advantages and disadvantages, as widely discussed in the literature, and are quite complementary. 

Figure 5.53(a) indicates the situation for a passive FeRAM (ferroelectric random access memory). If the capacitor is in the +P state (say a T) and the application of two V/2 voltages is sufficient to switch the polarization to the —P state, there is a resulting large current pulse. However were the cell in the —P state (a 0 ) then the current pulse would be very much smaller. If the state of the cell is such that the polarization is switched then a follow-on pulse must be applied to return it to its original state. 

Fig. 5.53 Schematic of FeRAM matrix (a) a matrix of memory cells (b) each memory cell coupled with a switching transistor to more closely define the threshold switching voltage.

For FeRAMs materials attracting interest are PbZrn-v Tiv03 (with. v 0.6) and the Aurivillius layer-structured perovskite, SrBi2Ta209 (SBT). 

The stability of the ferroelectric state as crystal size is reduced to typical film thicknesses (<100nm) is a shared interest between those working to reduce dielectric layer thickness in multilayer capacitors to maximize volumetric efficiency and those concerned with thin ferroelectric films for FeRAMs. There is evidence [26] for the ferroelectric state being stable to grain sizes as small as 40 nm, at least. 

There is a growing industrial demand for powders specially tailored for DRAM and FeRAM technologies. These are produced by chemical routes (see Section 3.4) with tight control exercised over composition. The general experimental approaches to forming thin films are summarized in Section 3.6.9 and in [25]. 

The ideal route would be one in which the pyroelectric detector material is laid down in thin film form by a route compatible with the production of the silicon ROIC. There are obvious parallels with the development of FeRAMS (see Section 5.7.5) and the substantial effort now devoted to their development will have a positive impact on the manufacture of pyroelectric arrays. Challenges he in the requirement to process the deposited films at temperatures not too high for the underlying integrated circuit, and the need to engineer the temperature diffusion characteristics within the element and its surroundings so as to optimise image definition. 

Dynamic Random Access Memory FeRAM = Ferroelectric Random Access Memory HCP = Hexagonal close packed HREM = High-resolution electron microscopy HTB = Hexagonal tungsten bronze MPTBh = Monophosphate tungsten bronzes with hexagonal tunnels MPTBp =... 

Ferroelectrics are high dielectric materials that are easily polarized in an electric field and can remain polarized to some degree after the field is removed. Such properties make them ideal candidates for computer memory applications and they have been used in the form of thin films as ferroelectric random access memories (FeRAMs) and as high permittivity dielectrics for Dynamic Random Access Memory DRAMs. They have also been looked at as a replacement for silicon dioxide in certain MOS applications. 

There is a different application for ferroelectric materials in solid state memory devices the ferroelectric RAM (FeRAM). However, in this chapter, the FERAM and related issues will not be considered because the subject would be too diverse. 

The ferroelectric effect is an electrical phenomenon. Parhcular materials, including the ternary oxides (Ba,Sr)Ti03, Pb(Zr,Ti)03 and (Bi,La)Ti03, exhibit a spontaneous dipole moment which can be switched between equivalent states by an external electric held. Ferroelectric thin hlms are of importance for the production of nonvolahle ferroelectric random access memory devices (FeRAM) °. Two possibilities to synthesize such mixed metal oxides are given by the CVD and ALD methods. Table 10 shows the preparation methods of such materials synthesized from metal enolates recently. 

Nonvolatile FeRAM devices ntilize either PZT or SET derivatives. In low density memory FeRAM prodncts, CSD is frequently nsed as the deposition method. For high-density 4- or 32-Mbit FeRAM prototypes, CSD is still used by industry to fabricate ferroelectric PZT thin-fihn capacitors, although gas phase methods like MOCVD have advantages due to the potential for conformal coverage of small three-dimensional strnctures. 

Thin films of BST have been the most widely studied dielectric for ferroelectric DRAMs (FRAMs or FeRAMs). The highest capacitance reported for a BST dielectric is 145fF/ j,m, which was achieved with a 20 nm film of a material having k = 325. Prototype BST capacitor DRAMs were first reported in 1995, but have not been widely used commercially because of the advances in other storage technologies. 

S.D., and Lee, J. A new ferroelectric material for use in FERAM Lanthanum-substituted bismuth titanate. 

---