Low Temperature Thermal Atomic Layer Deposition of Cobalt Metal

Jan 21, 2016 - Jiyeon Kim , Tomi Iivonen , Jani Hämäläinen , Marianna Kemell , Kristoffer Meinander , Kenichiro Mizohata , Lidong Wang , Jyrki Räi...
63 downloads 24 Views 576KB Size
Communication pubs.acs.org/cm

Low Temperature Thermal Atomic Layer Deposition of Cobalt Metal Films Joseph P. Klesko, Marissa M. Kerrigan, and Charles H. Winter* Department of Chemistry, Wayne State University, Detroit, Michigan 48202, United States S Supporting Information *

obalt metal films have important applications as magnetic materials, precursors to CoSi2 contact materials, and liners and caps of copper features in microelectronics devices.1−6 These films have typically been grown by physical deposition7−9 or chemical vapor deposition (CVD) from various molecular precursors.10−15 The nanoscale features in future microelectronics devices require the growth of cobalt metal by atomic layer deposition (ALD) to meet thickness and conformality requirements.16−18 ALD intrinsically enables subnanometer control of film thicknesses because of the selflimited growth mechanism.16−18 In particular, cobalt is being increasingly explored as the metallization liner in sub-10 nm logic nodes,6 and growth by ALD may be required. Plasmaenhanced ALD of cobalt metal films has been reported from a number of organometallic precursors with various reducing coreactants.19−26 However, plasma processes can cause substrate damage from highly reactive species and can also result in poor conformal coverage of high aspect ratio features due to surface recombination of hydrogen atoms.27 Reports of thermal ALD processes for cobalt metal are limited. Co(iPrNCMeNiPr)2 was used with H2 at 350 °C with a growth rate of 0.12 Å/cycle.28 This precursor was subsequently employed for cobalt metal ALD at 300−350 °C with H2 or NH3.29−31 Although self-limited growth was demonstrated,28 the temperatures used for these studies are significantly higher than the decomposition temperature of Co(iPrNCMeNiPr)2 (215−225 °C),32,33 suggesting a significant CVD component in the film growth. Cobalt metal films were deposited from (2-tertbutylallyl)Co(CO)3 and Me2NNH2 at 140 °C; however, selflimited ALD growth was not demonstrated.34 Self-limited cobalt metal ALD growth was reported using Co((Me) (iPr)COCNtBu)2 and BH3(NHMe2) at 180 °C, but the growth rate was low (0.07 Å/cycle) and growth only occurred on a ruthenium substrate after a nucleation step.35 Existing ALD processes for cobalt metal are limited by low growth rates and high deposition temperatures that likely include a significant CVD growth component. Herein, we describe the thermal ALD of cobalt metal films using bis(1,4-ditert-butyl-1,3-diazabutadienyl)cobalt(II) and formic acid as precursors. This process occurs with a high growth rate of 0.95 Å/cycle within an ALD window of 170−180 °C and provides high purity, low resistivity cobalt metal films. We recently reported a three-step copper metal ALD process employing Cu(dmap)2 (dmap = 1-dimethylamino-2-propoxide), formic acid, and hydrazine.36 In this process, Cu(dmap)2 does not react with hydrazine, but the intermediate copper formate is readily reduced by hydrazine to copper metal. Accordingly, similar three-step processes were considered for

C

© XXXX American Chemical Society

cobalt metal ALD. We recently described the synthesis and properties of bis(1,4-di-tert-butyl-1,3-diazabutadienyl)cobalt(II) (1, Chart 1)37 and 1,4-bis(trimethylsilyl)-1,4-dihydropyrazine Chart 1. Structures of Precursors 1−3

(2, Chart 1),38 which represent a promising cobalt precursor (1) and a reducing coreagent (2) for ALD. We envisioned a three-step cobalt metal ALD process employing 1, formic acid, and 2, wherein the intermediate cobalt(II) formate layers would be reduced by 2 to afford cobalt metal. Such a growth experiment was conducted with a substrate temperature of 180 °C, as described in the Supporting Information.39 X-ray photoelectron spectroscopy (XPS) analysis revealed a 97.1% pure cobalt metal film. A binary process using 1 and 2 did not produce any films at 180 °C. Surprisingly, however, a process with 1 and formic acid afforded low resistivity cobalt metal films over a wide temperature range, and was further investigated as described below. A complete ALD growth study was carried out using 1 and formic acid.39 Precursor pulse lengths, substrate temperatures, and the number of cycles were varied to assess the growth behavior. The growth rate was probed as a function of the pulse lengths of 1 and formic acid at 180 °C, using 1000 cycles and 10.0 s N2 purges after each precursor pulse. Film growth on ruthenium, platinum, palladium, and copper substrates afforded 83−120 nm thick films. No films were observed visually on silicon(100), silicon-H, and silicon dioxide substrates after 1000 cycles, but these substrates were not investigated further. Ru (13 nm)/TaN (2 nm)/SiO2 (100 nm)/Si substrates were chosen for the ALD study. Figure 1a shows a plot of growth rate versus pulse length of 1, with a growth plateau at ≥3.0 s and a saturative growth rate of 0.95 Å/cycle. A similar plot of growth rate versus pulse length of formic acid shows saturative behavior at ≥0.1 s with the same growth rate.39 On the basis of these data, a pulse sequence of 1 (5.0 s)/N2 purge (10.0 s)/ formic acid (0.2 s)/N2 purge (10.0 s) was used for all Received: September 8, 2015 Revised: January 8, 2016

A

DOI: 10.1021/acs.chemmater.5b03504 Chem. Mater. XXXX, XXX, XXX−XXX

Communication

Chemistry of Materials

RMS surface roughness of 1.24 nm, indicating a very smooth surface. Figure 2 shows the XPS cobalt 2p ionization region of a 95 nm thick film grown at 180 °C. The as-deposited film (red

Figure 2. XPS cobalt 2p ionization region of a 95 nm thick film grown at 180 °C showing binding energies at 2 min intervals of argon ion sputtering.

trace) revealed the presence of surface cobalt(II) oxide, which likely forms upon exposure of the film to air. After 1 min of sputtering with 3 keV argon ions, only cobalt (58.3%) and oxygen (41.7%) were detected in the film. Continued sputtering resulted in ionizations that exactly matched the known values for cobalt metal after 2 min and a film composition consisting of 91.6% cobalt metal after 8 min. The remainder of the film was oxygen, with carbon and nitrogen levels below the detection limits (