Nanoscale Thermometry of Plasmonic Structures via Raman Shifts in Copper Phthalocyanine

Temperature measurements at the nanoscale are vital for the application of plasmonic structures in medical photothermal therapy and materials science but very challenging to realize in practice. In this work, we exploit a combination of surface-enhanced Raman spectroscopy together with the characteristic temperature dependence of the Raman peak maxima observed in β-phase copper phthalocyanine (β-CuPc) to measure the surface temperature of plasmonic gold nanoparticles under laser irradiation. We begin by measuring the temperature-dependent Raman shifts of the three most prominent modes of β-CuPc films coated on an array of Au nanodisks over a temperature range of 100–500 K. We then use these calibration curves to determine the temperature of an array of Au nanodisks irradiated with varying laser powers. The extracted temperatures agree quantitatively with the ones obtained via numerical modeling of electromagnetic and thermodynamic properties of the irradiated array. Thin films of β-CuPc display low extinction coefficients in the blue-green region of the visible spectrum as well as exceptional thermal stability, allowing a wide temperature range of operation of our Raman thermometer, with minimal optical distortion of the underlying structures. Thanks to the strong thermal response of the Raman shifts in β-CuPc, our work opens the opportunity to investigate photothermal effects at the nanoscale in real time.


S1. Focal spot size
Plasmonic photothermal effects are strongly dependent on the laser beam profile and focal spot size 1 , which can be measured by the well-known knife-edge method [2][3][4][5] . The knifeedge method is often used as a standard and precise method for Gaussian laser beam characterization. A USAF resolution test target from Thorlabs was mounted on the translation stage and moved across the beam in 0.2 μm steps while monitoring the transmitted light intensity with a power meter. To find the focal position, the intensity was measured at different Z depths. The beam widths were obtained by taking the derivative of the intensity profile at various depths and fitting it with a Gaussian function. The minimum beam width indicates the focal position. Our SERS experiments were done in a cryostat system, so we measured the laser beam width with a cryostat window above the sample, as shown in Figure S1. The full width at half maximum (FWHM) of the beam, Wy, is 2.46 um ± 0.04 μm with the cryostat window. Figure S1. Laser beam profile obtained along the Y-axis of motion and fitted to a Gaussian function. Figure S2 shows the absorption spectra of as-deposited and annealed CuPc films, both with and without an underlying Au NP array. The absorption peaks of CuPc originate from the molecular orbitals within the aromatic system 6,7 . Three absorption peaks are observed and belong to π-π* transitions 6,7 . In the near-UV region (300-350 nm), a single peak appears and is assigned to the Soret band (or B-band). The absorption band in the visible region (600-800 nm), corresponds to the Q band and has a doublet due to Davydov splitting [6][7][8] .

S2. UV-Vis spectra
As can be seen, the position and the relative intensity of the Q-bands are different for asdeposited (black line) and annealed (red line) CuPc films on a glass substrate. The band red-shifts upon annealing because of the structural modification. For as-deposited CuPc thin film, the intensity of the higher energy peak (at 619 nm) is larger than that of the lower energy peak (at 694 nm), as expected for α-CuPc 6-8 . For annealed CuPc thin film, this intensity ratio is inverted, indicating the formation of β-CuPc 7,8 . We also studied the absorption spectrum of a 20 nm thick CuPc film as-deposited on a Au nanodisk array (blue line) and after annealing (green line). A similar red-shift and intensity change is observed upon annealing, indicating the formation of the β-phase.

Previous investigations of the Raman tensors and polarized Raman spectra of oriented
CuPc single crystals have shown that the low wavenumber region (< 200 cm -1 ) results from the lattice vibrations and is defined according to the crystal symmetry C2h 9 , whereas the high wavenumber region (> 200 cm -1 ) corresponds to intramolecular modes and is defined following the molecular symmetry D4h 9 . Figure S3 plots the Raman spectra of asdeposited and annealed CuPc films with 60 nm thickness measured at room temperature.

S5
S4. XRD Figure S4 shows the XRD pattern measured over a 300 nm thick CuPc film evaporated onto a glass substrate and annealed at 300 °C. The peaks at 6.96° and 9.16° can be attributed to β-CuPc 10,12,13 . Figure S4 The XRD patterns of a 300 nm thick annealed CuPc film on a glass substrate.

S5. Absorption spectra of β-CuPc at varying temperatures
The thermal stability of β-CuPc films was tested via temperature-dependent UV-Vis spectroscopy. As shown in Figure S5, the spectral position and line shape of the peaks do not change, suggesting that the sample is not undergoing any phase change in the 300-495 K temperature range (see also Fig S2). Figure S5. UV-Vis absorption spectra of a 60 nm thick β-CuPc film on bare glass, measured from 300 K to 495 K in vacuum in ascending order with a heating rate of ~20 K/min.

S6. Choice of proper laser power
As a thermometer, the calibration would be done for β-CuPc film on Au nanodisks by temperature dependent Raman measurements. During the calibration process, any fluctuation in the thermal environment could introduce extra errors. Firstly, we considered the local heating effect from the laser by the photon absorption and the accumulated irradiation time. Therefore, we compared the influence of different laser powers and irradiation times on the Raman peak shift. Figure S6

(b)
Below are the parameters used to fit the three trends in Figure 2(b) using equations (1) and (2) Figure S7. Raman spectra of β-CuPc on Au nanodisks measured at 296 K before (black) and after (red) heating to 500 K.

S11 Optical and heat-transfer modelling under Gaussian beam illumination
The spatial electric-field enhancement (E(x,y,z)) and spatial heating profile (q(x,y,z)) of Au-CuPc arrays was modelled as follows. First, finite difference time domain (FDTD) simulations were conducted using Lumerical software to solve Maxwell's equations for the nanoparticle geometry, under the experimentally determined 532 nm Gaussian beam illumination (2.4 µm FWHM; 2.05 µm 1/e 2 beam waist radius, see Figure S1) S16 S12. Thermal transfer model Heat transfer through convection was neglected as the experiments were carried out in (partial) vacuum, i.e. the vacuum was modelled with 0.1% of the thermal conductivity of air. A cubic geometry was designed of 10×10×10 µm with the 32×32 particle Au-CuPc array at the center, on glass substrate and in partial vacuum (1 mbar), see Figure S12a.  S18