The realization of high-transition-temperature (high-T(c)) superconductivity confined to nanometre-sized interfaces has been a long-standing goal because of potential applications and the opportunity to study quantum phenomena in reduced dimensions. This has been, however, a challenging target: in conventional metals, the high electron density restricts interface effects (such as carrier depletion or accumulation) to a region much narrower than the coherence length, which is the scale necessary for superconductivity to occur. By contrast, in copper oxides the carrier density is low whereas T(c) is high and the coherence length very short, which provides an opportunity-but at a price: the interface must be atomically perfect. Here we report superconductivity in bilayers consisting of an insulator (La(2)CuO(4)) and a metal (La(1.55)Sr(0.45)CuO(4)), neither of which is superconducting in isolation. In these bilayers, T(c) is either approximately 15 K or approximately 30 K, depending on the layering sequence. This highly robust phenomenon is confined within 2-3 nm of the interface. If such a bilayer is exposed to ozone, T(c) exceeds 50 K, and this enhanced superconductivity is also shown to originate from an interface layer about 1-2 unit cells thick. Enhancement of T(c) in bilayer systems was observed previously but the essential role of the interface was not recognized at the time.

Focused ion beam (FIB) specimen preparation techniques have been successfully used for nearly all types of (scanning) transmission electron microscopy (S)TEM methods. However, the milling process from high energy Ga + FIB columns (e.g. 30 keV) can impart sufficient surface damage (i.e., ~ 20 nm per specimen side for Si) to impede quantitative high resolution STEM and TEM imaging. Previous results have shown that amorphization damage in Si and GaN can be removed by chemical wet polishing after FIB milling [1,2]. However, chemical polishing methods are material dependent and are difficult or impossible to use for complex multi-phase or multilayered specimens. Previous results have shown that better high resolution TEM images can be obtained by FIB polishing a 30 keV FIB prepared specimen with a 10 keV ion beam [1,3]. Low energy broad beam Ar + ion milling has also been used to reduce FIB damage for HRTEM [1,4]. In this paper, we discuss specimen surface quality improvements using a FIB column whose ion beam energy can be reduced to 2 keV, resulting in the lowest amount of FIB damage reported to date.A blanket wafer Si substrate specimen was prepared for TEM analysis using conventional 30 keV FIB milling via the in-situ lift-out and Omniprobe total release technique [5] using an FEI Strata 400S DualBeam instrument equipped with a Sidewinder FIB column. Lines were FIB milled into the specimen using beam energies of 30 keV, 5 keV, and 2 keV, and were used to assess the sidewall damage at these energies. Then the surface of the specimen was FIB polished with the Ga + ion beam at +/-85 o incident angle at 5 keV and then at 2 keV. The sidewall damage and surface quality of the specimen was observed using HRTEM imaging with an FEI Tecnai F20 equipped with a spherical aberration coefficient (Cs) image corrector operating at 200 keV. The specimen was plasma cleaned using a Fischione plasma cleaner after each FIB milling procedure prior to TEM observation. Ion-solid interactions predict that lighter incident ions will produce more surface damage than heavier incident ions [see chap. 2 in ref. 6]. This implies that Ga + ions will yield less specimen damage compared to similar energy beams from conventional Ar + ion mills, and thus, may negate the need for broad beam Ar + ion milling of FIB prepared specimens.

A liquid metal ion source (LMIS) of Ga + ions has been the mainstay of focused ion beam technology over the past 20 years or so [1]. While state-of-the-art Ga + ion FIB columns typically possess a resolution of < 5 nm, FIB-based transmission electron microscopy (TEM) specimen preparation techniques were initially developed using Ga + ion beams with much broader beams (i.e., ~ 50-100 nm resolution) [2,3].Inductively coupled plasma (ICP) sources [4] have recently been introduced into FIB columns. The ICP source yields worse ultimate resolution than the LMIS FIB within the low current regime, but greatly improved resolution in the high current regime [4]. Thus, the plasma FIB (PFIB) allows for faster removal rates of large volumes using a combination of higher beam currents plus larger mass Xe + ions which increase the sputter yield [5]. The PFIB beam size in the low current regime is similar to the beam sizes first introduced in a LMIS FIB. Therefore, it seems plausible that the PFIB could be used for TEM specimen preparation. Indeed, we present PFIB-prepared TEM results below. Since Xe is inert, Xe + ion bombardment may be preferrable for specimen preparation, particularly for cases where it is well known that deliterious Ga-rich intermetallic phases can precipitate on milled surfaces [7.8]. The high mass Xe + ions also allows for greater throughput via higher sputter yields, and yields less surface damage due to the smaller ion range of Xe + compared to Ga + in a given target. Using SRIM, the sputter yield and longitudinal range for 30 keV Xe + in Si at 0 o incidence is 2.9 atoms/ion and 24 nm respecitively, while that for Ga + at the same conditions is 2.2 atoms/ion and 28 nm respecitvely. SRIM plots showing the ion range and recoil motion of 30 keV Xe + and 30 keV Ga + in Si at 0 degrees incidence are found in FIG. 2. The reduced ion range of Xe + is consistent with the observation of less amorphous damage for Xe + over Ga + (despite the slight differences in ion energy herein). In summary, the PFIB can be used to prepare electron transparent specimens and Xe + ions have tremendous potential as an alternative to Ga + ions for TEM specimen preparation and other applications [9].

No abstract

No abstract