Impact of Premetallization Surface Preparation on Nickel-based Ohmic

State University, University Park, Pennsylvania 16802, United States. ACS Appl. Mater. Interfaces , 2016, 8 (50), pp 34802–34809. DOI: 10.1021/a...
0 downloads 12 Views 5MB Size
Research Article www.acsami.org

Impact of Premetallization Surface Preparation on Nickel-based Ohmic Contacts to Germanium Telluride: An X‑ray Photoelectron Spectroscopic Study Haila M. Aldosari, Hamed Simchi, Zelong Ding, Kayla A. Cooley, Shih-Ying Yu, and Suzanne E. Mohney* Department of Materials Science and Engineering and Materials Research Institute, The Pennsylvania State University, University Park, Pennsylvania 16802, United States ABSTRACT: Surfaces of polycrystalline α-GeTe films were studied by X-ray photoelectron spectroscopy (XPS) after different treatments in an effort to understand the effect of premetallization surface treatments on the resistance of Ni-based contacts to GeTe. UV−O3 is often used to remove organic contaminants after lithography and prior to metallization; therefore, UV−O3 treatment was used first for 10 min prior to ex situ treatments, which led to oxidation of both Ge and Te to GeOx (x < 2) and TeO2, respectively. Then the oxides were removed by deionized (DI) H2O, (NH4)2S, and HCl treatments. Additionally, in situ Ar+ ion etching was used to clean the GeTe surface without prior UV−O3 treatment. Ar+ ion etching, H2O, and (NH4)2S treatments create a surface richer in Ge compared to the HCl treatment, after which the surface is Te-rich. However, (NH4)2S also oxidizes Ge and gradually etches the GeTe film. All treated surfaces showed poor stability upon prolonged exposure to air, revealing that even (NH4)2S does not passivate the GeTe surface. The refined transfer length method (RTLM) was used to measure the contact resistance (Rc) of as-deposited Ni-based contacts to GeTe as a function of premetallization surface preparation. HCl-treated samples had the highest Rc (0.036 ± 0.002 Ω·mm), which was more than twice that of the other surface treatments. This increase in Rc is attributed to formation of the Ni1.29Te phase at the Ni/GeTe interface due to an abundance of Te at the surface after HCl treatment. In general, treatments that resulted in Ge-rich surfaces offered lower Rc. KEYWORDS: GeTe, phase change material, contact, X-ray photoelectron spectroscopy, surface

1. INTRODUCTION Germanium telluride (GeTe) is a phase-change material (PCM) that has gained recent attention due to its incorporation as an active material for radio frequency (rf) switches,1,2 as well as memory and novel optoelectronic devices. Although it has not been reported that contact resistance seriously limits the performance of phase-change memory, it contributes a parasitic resistance that limits the cutoff frequency of rf switches fabricated from phase-change materials. Previous studies suggest that contact resistance is roughly 25−50% of the on-state resistance of the device.1,3 Reduction of contact resistance is therefore critical for reducing the on-state resistance in order to meet the requirements of high-frequency rf applications. Previously, a Ti/Pt/Au stack with Rc = 9.85 × 10−3 Ω·mm was used for making GeTe-based rf switches.3 In another study, Chua et al.4 used in situ Ar+ plasma precleaning to achieve low specific contact resistance (ρc) using 50 nm Ni, W, or TiW, followed by 150 nm Al, as the metal contact to both crystalline and amorphous GeTe. The lowest ρc of 8.6 × 10−9 Ω·cm2 was reported for the as-deposited Ni/Al contact to crystalline GeTe, as measured by circular transfer length method (CTLM) test structures.4 Therefore, Ni contacts are interesting for further study. In addition, © XXXX American Chemical Society

knowledge about the GeTe surface and ohmic contacts to GeTe can be a valuable starting point for studies on other ternary alloys containing Ge and Te. In this work, we investigated the role of premetallization surface preparation procedures on the resistance of Ni-based contacts to crystalline GeTe. X-ray photoelectron spectroscopy (XPS) was used to characterize the GeTe surfaces after in situ and ex situ premetallization surface preparation procedures and to study the stability of the resultant surfaces in air. Finally, the dependence of the resistance of Ni-based contacts on different surface treatments prior to metallization was studied. The GeTe surface is prone to oxidation in air, starting with the oxidation of Ge, followed by a slower oxidation of Te.5−7 Optimal surface preparation procedures, which could play an important role in facilitating low Rc, are still being clarified. In many semiconductors, surface preparation procedures prior to metallization can promote low contact resistance.8,9 We chose in this study to investigate in situ Ar+ plasma treatment of GeTe Received: June 30, 2016 Accepted: November 28, 2016 Published: November 28, 2016 A

DOI: 10.1021/acsami.6b07412 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

at 284.6 eV (adventitious carbon peak) as the reference except for the Ar+ ion etching treatment, wherein the Ge0 component of the Ge 2p3/2 at 1217.7 eV was used due to the absence of the C 1s peak. The background due to inelastic scattering was subtracted by the method of Shirley.15 Peaks were fit with Gaussian−Lorentzian curves satisfying the following constraints: (a) doublet intensities have ratios of 3:2 and 2:1 for the electrons from d and p orbitals, respectively; (b) each doublet has equal full width at half-maximum (fwhm); and (c) the spin orbit splitting of the doublets is matched with the database.16,17 In each case, the residual standard deviation used to evaluate the difference between the actual data and fit envelopes were also minimized to get the best fit. For the electrical measurements, “refined” transfer length method (RTLM) test structures18 were patterned with nominal gap spacings of 0.6−20 μm as shown in Figure 1a. Prior to lithography, samples were

prior to metallization, which was reported by other researchers, along with wet chemical methods. We investigated treatment with an aqueous solution of (NH4)2S because (NH4)2S has been used to passivate the surface of III−V semiconductors, such as GaAs,10 and it was effective for reducing the resistance of contacts to p-type InGaSb.11 It has also received attention as a premetallization surface treatment for GaSeTe12 in a study that showed via XPS that (NH4)2S dissolved the native oxide. We included HCl treatment in this study because HCl dissolves germanium oxide and leaves a Cl-terminated surface on the elemental semiconductor Ge.13 We also included deionized water as a control treatment and because our XPS data, presented later, suggested that it might provide a surface with little oxide coverage.

2. EXPERIMENTAL SECTION Polycrystalline p-type α-GeTe with a high hole concentration of ∼1020 cm−3 (110 nm thick), provided by Northrop Grumman Electronic Systems, was sputtered on a Si substrate with a 100 nm thick amorphous Si3N4 barrier layer for this study. Details of the sputtering process and film characterization can be found in their prior work.3 Samples were degreased in acetone, isopropyl alcohol (IPA), and deionized (DI) water for 5 min with ultrasonication for 10 s each time and then blown dry with compressed nitrogen. After degreasing, the samples were oxidized inside a PR-100 UV−ozone photoreactor (UVP Inc., San Gabriel, CA) for 10 min at 1 standard liter per minute (SLPM) to remove organic contaminants,14 although it does cause oxide to grow on the surface. Afterward, the samples were treated with ammonium sulfide [(NH4)2S], hydrochloric acid (HCl), or DI water with the intent of removing the oxide and passivating the surface. For (NH4)S2-treated samples, a dilution of (100:1) H2O/22−24% (NH4)S2 was used for 30 s. For HCl-treated samples, a dilution of (10:1) H2O/37% HCl was used for 120 s. After each treatment, the sample was rinsed with DI water for 15 s and blown dry with N2. When DI water treatment was used as the treatment, it lasted for 5 min. The samples were then immediately loaded into the XPS vacuum system (within 5 min). Note that other concentrations were initially tested to ensure that the chemicals would not attack and completely etch the semiconductor. For example, using a very highly concentrated (10:1) H2O/(NH4)2S solution would etch the semiconductor in a few seconds. In addition, in situ Ar+ ion etching was conducted on degreased samples. Surface preparation procedures used in the work are summarized in Table 1.

Figure 1. (a) SEM micrograph of RTLM test structures. (b) Metallization stack. (c) Resistance measurement setup for RTLM test structures and schematic for probe placement used in this study. Note that the extensions to the top and bottom probe pads, shown in the SEM image but not included in panel c, were not used. degreased as described above and heated in air for 3 min at 115 °C to remove moisture. RTLM structures were fabricated by spin-coating the sample with a dual-layer liftoff resist stack followed by i-line optical exposure, first-stage development, O2 plasma ashing, deep UV flood exposure, and second-stage development. After patterning, the samples were treated with the different surface treatments mentioned above (same conditions) except for the in situ Ar+ plasma sample. Afterward, samples were loaded in an e-beam evaporation system with base pressure