Bimetallic copper palladium nanorods: plasmonic properties and palladium content effects

Cu is an inexpensive alternative plasmonic metal with optical behaviour comparable to Au but with much poorer environmental stability. Alloying with a more stable metal can improve stability and add functionality, with potential effects on the plasmonic properties. Here we investigate the plasmonic behaviour of Cu nanorods and Cu–CuPd nanorods containing up to 46 mass percent Pd. Monochromated scanning transmission electron microscopy electron energy-loss spectroscopy first reveals the strong length dependence of multiple plasmonic modes in Cu nanorods, where the plasmon peaks redshift and narrow with increasing length. Next, we observe an increased damping (and increased linewidth) with increasing Pd content, accompanied by minimal frequency shift. These results are corroborated by and expanded upon with numerical simulations using the electron-driven discrete dipole approximation. This study indicates that adding Pd to nanostructures of Cu is a promising method to expand the scope of their plasmonic applications.


Mitigations made to STEM-EELS analysis for Cu-CuPd NRs to eliminate artefacts caused during acquisition by C contamination
The STEM-EELS spectrum images were acquired individually for each NR.In each case, the NR was scanned along the short axis (perpendicular to the rod axis) before being scanned for the next row along the long axis (parallel to the rod axis, see Figure S13).Over time, each sample experienced considerable C contamination.Thus, data acquired at the later stage of scanning probed an increasingly more contaminated NR.C contamination was difficult to avoid as NRs were capped with OLAM and grafted with PS-PEHA for colloidal stability.
Neither baking nor plasma cleaning were applied as they posed a risk of oxidising the NRs.
The effect of C contamination is observed through various parameters.First, we find that LSPR energies of the part of the NR scanned later was redshifted compared to the earlier scans of the same NR (Figure S13), consistent with a dielectric environment with added C around the NR.We also observed a weak broadening of the LSPR linewidth at later scans (Figure S13).Lastly, the HAADF images taken after the EELS scan showed lower contrast compared to the images taken before the scan.This arises from the fact that the extra carbon layer becomes the entrance plane for the electrons during the scan.This causes the beam to broaden while it travels through the carbon, and by the time it reaches the NR, the imaging probe is 'blurred,' resulting in less contrast on the actual object of interest.One might wonder why the additional scattering from deposited C does not increase the contrast in HAADF images; this is because C scatters weakly and at medium angles rather than into the HAADF detector.
The NMF analysis involves a global fit to the spectrum image, and with LSPR energies being redshifted in the part of the image scanned later, the linewidths of each mode were broadened by the shifted amount.For Cu NRs, C deposition, and thus this shift, was insignificant and did not influence the NMF analysis, apart from the negligible systematic increase on the FWHM of LSP modes.Cu-CuPd NRs experienced more significant shifts to their LSPR energies, possibly due to the presence of leftover Pd precursors.For most Cu-CuPd NRs, NMF factorisation identified parts of the same mode (mainly the dipolar mode) as individual factors: one factor containing the dipolar mode in the part of the image acquired first, and another containing the part acquired towards the end of the scan.Decreasing the number of decomposition factors to combine both parts of the mode into a single factor was possible and the NMF spatial loadings (Figure 5B) were extracted using this method.However, extracting LSPR energies and FWHM from the combined spectral factors was misleading, as the resulting LSPR energies were averaged, and linewidth broadened.As such, a different approach was taken for the extraction of LSPR energies and their FWHM.We cropped the spectrum image in half, perpendicular to the rod axis, to contain only half of the NR that was scanned first.EELS point spectra (Figure 5A) were extracted near the tip of the NR in this cropped image.NMF decomposition was performed on this half of the image to extract LSPR energies and corresponding FWHMs (Figures 6A and 6B).

Figure
Figure S2.UV-Vis-NIR extinction spectra of (A) Cu NRs of labelled lengths and (B) Cu-

Figure
Figure S6.HAADF-STEM and STEM-EDS summed over 8 x 8 pixels as indicated by A and

Figure
Figure S8.STEM-EELS NMF spectral components for the three representative Cu NRs of

Figure S13 .
Figure S13.Effect of C contamination over an EELS scan of a Cu-CuPd NR (22%, 64 nm

Table S1 .
Dimensions of Cu NRs used for STEM-EELS analysis.

Table S2 .
Dimensions of Cu-CuPd NRs used for STEM-EELS analysis.