Example 5: nonvolatile memory devices

THIN FILM SERVICES / APPLICATION AREA

Multi-component Dielectrics for Electronic Applications

Example: HIGH-TEMPERATURE PEROVSKITES AND HIGH-ION-CONDUCTION METAL OXIDES

APPLIED IN NON-VOLATILE MEMORY APPLICATIONS

MATERIALS / APPLICATION AREA: Whereas the body of dielectric theory itself is a highly advanced and powerful development in the physical sciences, the phase space and associated behaviors of various complex multi-component, nominally dielectric, oxide compounds, particularly under highly non-equilibrium formation conditions, is a physically complex area that, as a set of thin film materials processes, is still primarily an experimental field, despite the many advances in understanding the resultant insulating materials.

Researchers at Helicon have been heavily involved in process development for many multi-component/doped oxide materials in a host of applications involving very different types of phase development. Our work in such complex oxide phases have included ferroelectrics & piezoelectrics, fast ion-conductors, superconductors, opto-electronics, laser host materials, nonlinear optical materials, optically active materials, and others.

From a process and fabrication point of view, vapor deposition of such compounds in a particular, phase-pure, thin film form, inevitably each represent a case-specific set of challenges. Recorded history of this area reveals many false starts in the claimed thin film process successes; however, this area of thin film fabrication, as a whole, represents an area of both very high commercial value and intensive research, with some practical unifying considerations.

Having performed continual development in reactive vapor deposition of a wide variety of multicomponent compounds for well over 25 years, we have acquired considerable insight into both common and specific materials deposition characteristics in these types of thin film devices.

FIGURE 1: subsets of electrically polarizable material.

MEMORY MATERIALS APPLICATION EXAMPLES IN GENERAL: In Fig. 1, above is shown three subsets of electronic behavior accorded essentially to non-centrosymmetric crystalline dielectrics. The most narrowly specified of these subsets is the ferroelectrics; materials, which, below their curie tempurature, display a spontaneous polarization. These dielectric susceptibilities will also yield nonlinear optical coefficients and related nonlinear optical behaviors (optical harmonic generation, periodic poling capability, etc.) as well as pyroelectric imaging capabilities, none of which are addressed under this application topic.

FIGURE 2: The general trending requirement (arrows) for enabling a new era in processor capability: aspirations of the semiconductor industry to cross over into a new paradigm of nonvolatile "RAM-like" memory cannot be met unless both requirements of high read/write (generically, "switching") speeds and reduced cycling fatigue

are addressed by the prospective nonvolatile memory technology. Currently, prospects such as the newer, experimental ferroelectric compounds, the Re-RAM variants (such as Intel/Micron's "Optane"), as well as other current up-and-coming nonvolatile memory platforms, all perform in this intermediate range. Where, exactly, in this range any one of these newer technologies can reliably perform on a commercial (cost-effective) scale, is yet to be determined.

FIGURE 4: ferroelectric behavior of our ferroelectric films, deposited via purely reactive, dual-magnetron, high-rate sputtering from fast-quenched Bi-Sr intermetallic and tantalum metal targets; sequence:Pt/SrxBiyTaO/Pt/TiOx/SiO2/Si;

strontium bismuth tantalate (SBT) device fabricated by Helicon.

Figure 5: deposited Pt top-electrode pattern in experimental ferroelectric materials testing.

FIGURE 3: ferroelectric behavior of device fabricated comprising Pt/SrxBiyTaO/Pt/TiOx/SiO2/Si by Helicon. Demonstration of ferroelectric hysteresis by integrated ferroelectric films via ceramic target sputtering of SBT.

APPLICATION EXAMPLE (SBT): As demonstrated in the adjacent figures, this example demonstrates our depositing phase-pure super-lattice perovskite through multiple vapor deposition routes - namely both RF plasma and a high-rate, purely reactive quasi-DC sputtering - and also successfully fabricating multilayer nonvolatile memory cells that utilize these ferroelectric phases. While the more standard ferroelectrics we've studied, used in bulk applications and having greater dipole moments, such as lithium niobate, potassium niobate or even PZT, are relatively less challenging to fabricate by vapor deposition methods, they do not possess the switching speed or fatigue resistance of more recently explored materials, such as, in particular, certain super-lattice-structured perovskites represented here by stontium bismuth tantalate (SBT). SBT offers the promise of exceedingly high switching speeds (>20nsec) and excellent cycling fatigue characteristics (>10E-12) compatible with processor memory applications.

Within the subset of thin film ferroelectrics compatible with integrated memory utilized in the semiconductor industry, Helicon has conducted considerable wor

k in multiple compositions. This includes next-generation ferroelectrics such as (SBT), (BiT), as well as more traditional ferroelectrics such as lithium niobate (LN) and potassium niobate (KN). Of the various ferroelectric compounds, the superlattice perovskite, SrBiTaO represents both particular challenges and benefits, requiring higher temperature processing while including a more volatile component than do lead zirconate titanate (PZT/PZLT) compositions.In the mid-90's we developed a new method for reactive sputtering of ferroelectrics that enabled excellent control of stoichiometry. Until then reactive sputtering of more complex ferroelectric phases had seemed very attractive, but the claimed methods of success were inherently too unstable for reproducible manufacturing. In a multicomponent system such as SBT, we introduced and proved out the methodology of creating/isolating an intermetallic of two components, while leaving the third component as the second source in a dual-magnetron system. Roughly ten years later, our developed approach to multicomponent phase formation was subsequently discovered to be critical for the industrially succesful fabrication of CIGS semiconductors in sputtered thin film solar applications.

Figure 6: Pure reactive sputtering through a bimetallic, dual-metal-target set-up proven by Helicon in the 1990's for depositing thin films of ferroelectric phases; specifically the Strontium Bimuth Tantalate ferroelectric.

Target Technology: In these multicomponent phases, Helicon has considerable experience in target development programs, including both sintered ceramics and experimental intermetallic compositions. This includes both the broad physical understanding and expertise set that goes into designing the target approach, as well as the fabrication and baselining that is required to approach any qualification process.

Conclusion: Due largely to the process complexity/cost for both the thin film ferroelectric and Re-RAM materials, along with their materials interfaces, the cost-benefit picture for scaling these materials, as nonvolatile memory technologies, at least to the broadly disruptive market-penetration anticipated, is still a work in progress. For example, notably, Cypress Semiconductor now offers a PZT-based nonvolatile memory product that is primarily suited (economically) to more boutique applications, such as special mobile and defense applications. Alternately, Intel reports a roughly 5x "increase in performance" of its, relatively, costly Optane ReRAM variant over flash, which does not convincingly portend a sea-change in either storage or processing memory.

Over the last decade or so, the earlier expectations of previously-favored ferroelectric RAM (Fe-RAM) have become substituted, in some ways, by similarly-based expectations of the currently-favored, in its great many forms, resistivity-based memory circuits (Re-RAM) - sometimes uniformly grouped under the (historically, a technically and physically vague) terminology of "memristor." However, the introduction of non-volatile memory technology that can provide fundamentally greater cycling capability and switching speed on the commercial innovation scale of current floating-gate/flash memory, while perpetually claimed, is yet to be proven.

var gaJsHost = (("https:" == document.location.protocol) ? "https://ssl." : "http://www.");

document.write(unescape("%3Cscript src='" + gaJsHost + "google-analytics.com/ga.js' type='text/javascript'%3E%3C/script%3E"));

</script>

try {

var pageTracker = _gat._getTracker("UA-7559868-1");

pageTracker._trackPageview();

} catch(err) {}</script>

Report abuse