Collin
Rowe
b,
Sathish
Kumar Shanmugham
a,
Grzegorz
Greczynski
a,
Lars
Hultman
a,
Arnaud
le Febvrier
a,
Per
Eklund
*a and
Ganpati
Ramanath
*ab
aThin Film Physics Division, Department of Physics, Chemistry and Biology (IFM), Linköping University, SE-58222 Linköping, Sweden. E-mail: per.eklund@liu.se; Ramanath@rpi.edu
bRensselaer Polytechnic Institute, Department of Materials Science and Engineering, Troy, NY 12180, USA
First published on 14th March 2024
Integrating interfacial molecular nanolayers (MNL) with inorganic nanolayers is of interest for understanding processing-structure/chemistry correlations in hybrid nanolaminates. Here, we report the synthesis of Co/biphenyldithiol (BPDT)/Co nanolayer sandwiches by metal sputter-deposition and molecular sublimation. The density and surface roughness of the Co layers deposited on the native oxide are invariant with the Ar pressure pAr during deposition. In contrast, the Co layer roughness rCo deposited on top of the BPDT MNL increases with pAr, and correlates with a higher degree of Co oxidation. Increased roughening is attributed to MNL-accentuated self-shadowing of low mobility Co atoms at high pAr, which consequently increases Co oxidation. These results indicating MNL-induced effects on the morphology and chemistry of the inorganic layers should be of importance for tailoring nanolayered hybrid interfaces and laminates.
Hybrid multilayer composites with alternating organic and inorganic nanolayers are typically synthesized by combining atomic and molecular layer deposition.17,18 Since low-temperature deposition of inorganic layers is necessary to preserve the integrity of the MNL during synthesis of nanolayered inorganic–organic composites, there is interest in harnessing energetic particles during deposition to get good quality inorganic layers. While MNLs can be formed by wet-chemical self-assembly on inorganic surfaces,19–21 combining MNL formation with inorganic film deposition methods entailing vacuum breaks and repeated substrate loading/unloading compromise control over the interface chemistry and are not easily scalable.
Here, we report the synthesis of Co/biphenyldithiol (BPDT) MNL/Co nanolayer thin-film sandwiches using a combination of metal sputter-deposition and sublimation of the BPDT molecules without atmosphere exposure. Co nanolayers deposited on top of a BPDT MNL exhibit a higher roughness than Co nanolayers deposited directly on SiO2, i.e., without an MNL. Furthermore, MNL-induced roughening correlates with a higher degree of cobalt oxidation, both of which are accentuated at higher Co deposition pressures. This knowledge would be essential for the design and synthesis of inorganic–organic nanolaminates with emergent properties, e.g., transparent electrical conductors,7 flexible thermoelectrics,22,23 and smart materials with mechanical bandgaps.4,9
The Co film deposition rate and time were determined for different Ar pressures in the 0.20 ≤ pAr ≤ 3.2 Pa range. The substrates were placed on an electrically floating holder placed 140 mm from the Co sputter target. The first Co (bottom) layer was sputter deposited using 15 W dc power plasma. For BPDT MNL formation on the first (bottom) Co layer, the Co/SiO2/Si structures were transferred to the load-lock and exposed to BPDT flux from a crucible containing ∼100 mg BPDT powder (95% purity, Aldrich). The load-lock pressure was held above 0.4 Pa during transfer. BPDT molecules were sublimed for 30 minutes at ≲0.3 Pa (ref. 26) and with the crucible temperature at Tsublimation = 45 ± 5 °C. The BPDT/Co/SiO2/Si structures were transferred back to the main chamber for sputter-depositing the Co overlayer on the BPDT MNL. For all Co/MNL/Co sandwiches, the targeted thicknesses of the bottom and top Co layers tCo were 8 nm, and 4 nm, respectively.
Co film thickness and density were determined using Rutherford backscattering spectroscopy (RBS) and X-ray reflectivity (XRR) measurements. RBS data was acquired using a 2 MeV He2+ beam incident at 5° to the surface normal, with the detector placed at a 170° scattering angle, and fit using SIMNRA simulations27,28 of model structures. XRR measurements were carried out in a Rigaku Smartlab diffractometer with Cu Kα radiation, and the data was analyzed using the PANalytical X'Pert Reflectivity software.29 The top Co layer roughness rCo in the Co/MNL/Co sandwiches were measured by tapping-mode atomic force microscopy (AFM) with a Bruker Dimension 3100 instrument, and the data was analyzed with the WSxM software.30
X-ray photoelectron spectroscopy (XPS) was used to study BPDT MNL formation in a Kratos Ultra DLD instrument (<1 × 10−7 Pa base pressure) equipped with a monochromatic Al Kα source. The pass energy was set to 160 eV for the survey scans and to 20 eV for higher energy resolution scans. Charging-induced shifts were corrected using the Co Fermi edge as the reference, and MultiPak31 software was used for analysis. Raman spectra were obtained in a home-built system with a ∼0.8 μm spot from a ∼0.5 mW 532 nm solid-state laser, and a ∼5.5 cm−1 resolution detector.
SIMNRA fits of RBS spectra for Co/SiO2/Si model structures (e.g., see Fig. 2) yield an areal Co film density ACo = 72 × 1015 atoms cm−2. Dividing ACo with tCo obtained from XRR analyses yields a Co film density ρRBSCo = 8.7 ± 0.1 g cm−3, which is within 1% of the density ρXRRCo. Additionally, dividing ACo with ρXRRCo yields tCo = 8.0 ± 0.6 nm, which agrees well with tCo determined from Kiessig fringes by XRR. These results confirm that the extracted values of tCo and ρCo are self-consistent, with no observable changes in ρCo.
Fig. 2 A representative RBS spectrum (points) and the SIMNRA simulation (line) from a Co film deposited at pAr = 0.8 Pa (6 mTorr). |
XPS spectra from Co films exposed to BPDT fluxes revealed S 2p core-level peaks33,34 (Fig. 3a). Surprisingly, even as-deposited Co films showed S 2p peaks, albeit of lower intensity. This result indicates inadvertent BPDT exposure during sample residence and exchanges in the load-lock either directly from the sublimation cell or refluxing from the load-lock chamber walls. Given the low BPDT vapor pressure,26 this outcome was unexpected, and underscores the importance of using a shutter to physically isolate the molecular source during hybrid nanocomposites synthesis.
The O 1s contribution35–37 associated with Co–O bonding is ∼4-fold lower in Co films intentionally exposed to BPDT than in Co films with inadvertent BPDT exposure (Fig. 3b). This result indicates that BPDT MNL formed by intentional sublimation curtails Co oxidation considerably more than BPDT adsorption through inadvertent exposure, suggesting that inadvertent BPDT fluxes in our experiments do not result in continuous BPDT MNLs.
The areal density of the top Co layer AtopCo was determined by fitting RBS data from Co/BPDT/Co sandwiches (Fig. 4a) using tCo = 8.1 nm and ACo = 72 atoms × 1015 cm−2 for the bottom Co layer. These results show that the top Co layer ACo monotonically decreases with increasing pAr (Fig. 4b). Our XRR results indicate that the Co density ρCo for the bottom layer is invariant for 0.2 ≤ pAr ≤ 3.2 Pa. If ρCo is similarly invariant for the top Co layer, and ρtopCo = ρbottomCo = 8.8 g cm−3, the top-layer Co thickness ttopCo = ρRBSCo/AtopCo is estimated to be in the 3.1 ≤ ttopCo ≤ 4.7 nm range. The mean ttopCo = 3.9 ± 0.7 nm, which is close to the intended ttopCo = 4 nm.
AFM measurements of the top Co layer in the Co/BPDT/Co sandwiches (Fig. 5a) show a monotonic increase in root-mean-square roughness rCo with pAr (Fig. 5b). Since the tbottomCo values are similar (within ±1.6 nm) with only a slight decrease with pAr, the observed rCo increase is clearly not a film thickness effect. The pAr–rCo correlation seen in the top Co layer is absent in the bottom Co layer, as described earlier in section 3.1, and plotted in Fig. 5b for comparison. We thus infer that the higher roughness of the top Co layer in Co/BPDT/Co structures is due to the presence of the underlying BPDT MNL.
Higher frequency of gas-phase collisions expected with increasing pAr could promote higher roughness due to lowered adatom mobility resulting from lower average energy of sputtered species impinging on the surface at shallower angles. However, since this effect is not seen for the bottom Co layer deposited on SiO2, we attribute the pAr–rCo correlation in the top Co layer to the presence of the BPDT MNL. It appears that the presence of an MNL accentuates the self-shadowing from low mobility of metal adatoms expected at higher pAr by virtue of the initial roughness caused by the BPDT layer. Thus, higher pAr leads to roughening for the top Co layer but not the bottom one grown on flat SiO2. These observations suggest that the nature and extent of the MNL influence on the inorganic overlayer morphology are likely to be dependent on MNL chemistry and morphology.
XPS spectra recorded from Co/BPDT/Co sandwiches (see Fig. 6, and ESI†) indicate a correlation between Co oxidation and pAr, and corroborate BPDT adsorption on the top Co layer surface due to unintentional BPDT exposure discussed earlier. Co 2p peaks contain metallic Co0 as well as oxidized Co+2 and Co+3 contributions38–40 (Fig. 6a). Assigning these non-zero oxidation states specifically to either Co–O or Co–S bonds is precluded by overlaps in Co–S and Co–O signatures. The normalized oxidized/metallic Co peak ratio (Fig. 6b) does not exhibit any discernable trend with pAr.
Fig. 6 (a) Co 2p, (c) O 1s and (e) S 2p spectra from Co/BPDT/Co nanolayers at θSD = 60°. In the Co 2p band, * represents Co satellite peaks. Semi-log plots of sub-band intensity ratios from (b) Co 2p, (d) O 1s and (f) S 2p plotted as a function of pAr. The total O 1s band intensity is also shown on a linear scale (d, right axis). Detailed sub-band fittings are given in ESI.† |
Intensities of both the O 1s peak (Fig. 6c) and the O–Co sub-band35–37 (Fig. 6d) increase with pAr, indicating that higher pAr is associated with a greater degree of oxidation of the top Co layer. Viewing this result alongside the pAr–rCo correlation for the top Co layer suggests a connection between Co layer roughness and the degree of Co oxidation to Co+2 or Co+3 states. The S 2p peak from adventitious BPDT are comprised of sub-bands from –SH, S–C, S–Co, and S–O moieties33,34 (Fig. 6e). The sub-band in the 162–164 eV regime is labeled as S–Co/S–C because the S–Co and S–C states were indistinguishable in our measurements limited by ∼1 eV spectral resolution in our measurements. The high-binding energy sub-bands from a small fraction of S–O bonds are not captured in the O 1s spectra because of vast differences in the relative intensities of O 1s and S 2p peaks (see Fig. 6c & e and survey spectra in ESI†). The oxidized/unoxidized S (Fig. 6f) ratio does not exhibit any discernable trend with pAr.
Variable take-off angle XPS measurements show that the C 1s peak attenuates the least and Co 2p peaks attenuates the most with increasing surface-to-detector angle θSD (Fig. 7a). These results indicate41 that a C layer covers the top Co layer, with oxygen and sulfur straddled in between. For all θSD, the Co 2p and O 1s intensity attenuation trends with pAr appear to be correlated (Fig. 7b), consistent with increased Co oxidation at high pAr. The C 1s band intensity attenuation is insensitive to pAr, and the S 2p band mimics this for pAr < 0.4 Pa. At higher pAr, the S 2p relative intensity is distinctly lower than that of O 1s. These results suggest that Co oxidation decreases with decreasing Co–S linkages at the Co/BPDT interface.
Raman spectra from Co/BPDT/Co sandwiches (Fig. 8) exhibit the following signatures of BPDT:24,42–44 CC stretching (1590 cm−1), C–C stretching (1288 cm−1) between, and C–H bending (1201 cm−1) within, the aromatic rings, and S–C stretching (1078 cm−1). The lack of an observable peak at 376.6 cm−1 indicates –SH moieties did not react with each other to form disulfides. These suggest that the sputter-deposition of a Co overlayer does not destroy the BPDT structure, consistent with MNLs being resilient to <∼10 eV projectiles.45
Fig. 8 Raman spectra from the Co layer before and after BPDT exposure, and from Co/BPDT/Co sandwiches plotted on a semi-log plot. For the bottom Co layer pAr = 0.8. |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3dt01910a |
This journal is © The Royal Society of Chemistry 2024 |