Additive manufacture of complex 3D Au-containing nanocomposites by simultaneous two-photon polymerisation and photoreduction

The fabrication of complex three-dimensional gold-containing nanocomposite structures by simultaneous two-photon polymerisation and photoreduction is demonstrated. Increased salt delivers reduced feature sizes down to line widths as small as 78 nm, a level of structural intricacy that represents a significant advance in fabrication complexity. The development of a general methodology to efficiently mix pentaerythritol triacrylate (PETA) with gold chloride hydrate (HAuCl4∙3H2O) is reported, where the gold salt concentration is adjustable on demand from zero to 20 wt%. For the first-time 7-Diethylamino-3-thenoylcoumarin (DETC) is used as the photoinitiator. Only 0.5 wt% of DETC was required to promote both polymerisation and photoreduction of up to 20 wt% of gold salt. This efficiency is the highest reported for Au-containing composite fabrication by two-photon lithography. Transmission Electron Microscopy (TEM) analysis confirmed the presence of small metallic nanoparticles (5.4 ± 1.4 nm for long axis / 3.7 ± 0.9 nm for short axis) embedded within the polymer matrix, whilst X-ray Photoelectron Spectroscopy (XPS) confirmed that they exist in the zero valent oxidation state. UV-vis spectroscopy defined that they exhibit the property of localised surface plasmon resonance (LSPR). The capability demonstrated in this study opens up new avenues for a range of applications, including plasmonics, metamaterials, flexible electronics and biosensors.

: The optical absorption spectra of pure DMAc, PETA in DMAc, DETC in DMAc and tetrachloroauric acid trihydrate in DMAc at different concentrations. In all spectra, no absorption band is observed at the laser wavelength of 780 nm, which implies that the photoinduced reactions were associated with exciting the chemicals by the two-photon absorption process. * As DMAc is used as a solvent and plays no role in polymerisation and photoreduction, the concentration of gold talked in this paper is calculated by the weight ratio of gold chloride hydrate vs the total weight of gold chloride hydrate, PETA and DETC. Figure S3: SEM image showing how the polymerisation threshold was defined according to the integrity of the lines. Please note the number written under each group of testing lines represents the percentage of laser power specified by Nanoscribe. The actual laser energy used is that number times 50 mW. In the sample shown here, the threshold is 48% x 50 mW = 24 mW. To minimize the error caused by interface finding (if a laser beam is focused beneath the substrate due to uneven surface, although the power is high enough for polymerisation, no polymeric lines can be drawn on substrate), a series of horizontal lines were printed beforehand using a fixed laser power. They acted as a supporting layer to ensure a level base above the substrate. The lines for testing were then fabricated on top of the horizontal lines in a direction rotated 90 o relative to the first lines, and at various laser powers. Figure S4: Raman spectra of (i) non-polymerised monomer (blue),(ii) PETA without Au, polymerised Formulation 11 (red), (iii) PETA with 5 wt% gold salt, polymerised Formulation 12 (green) and (iv) PETA with 10 wt% gold salt, polymerized Formulation 13 (purple) in the range 1550-1800 cm 1 . The spectra have been baseline-corrected and shifted on the vertical axis for viewing clarity. The polymerised samples were prepared under the same processing conditions (laser power of 42.5 mW, scanning speed of 5000 μm/s, hatching distance of 0.8 μm) to afford woodpile structures with 4 layers. Two distinctive peaks at 1635 and 1725 cm -1 are observed in all cases, associated with the carbon-carbon double bonds (C=C) and the carbon-oxygen double bonds (C=O) respectively 1 . During cross-linking, the bond order of the C=C bonds adjacent to the ester groups in the monomer is reduced as a consequence of the Michael addition of monomer units, yielding single C-C bonds and a corresponding reduction in the intensity of the vibrational mode associated with C=C bonds at 1635 cm -1 However, the C=O bonds do not participate in these reactions and thus no change in their intensity at 1725 cm -1 is observed. Therefore, by measuring the change in the intensity (as peak areas) ratio of the C=C and C=O bonds before and after polymerisation, the degree of polymer conversion (DC) can be estimated 1-3 .

Degree of polymer conversion (DC):
where, , and are the peak intensities in the Raman spectra related to the C═C and C═O groups in the polymerised structures and the non-polymerised liquid resin.

PETA
42% PETA + 5%HAuCl4 35% PETA + 10%HAuCl4 27% As noted by Burmeister et al 2 , the hatching distances also influence the DC of the pure polymer, with the DC increasing with a corresponding reduction in the hatching distance. This relationship has been attributed to the effective increase in total exposure time and the influence of created radicals to the adjacent rods 2 . Here for composite fabrication, this relationship still applies. In addition, the presence of in situ generated Au nanoparticles can absorb laser energy and on occasions facilitate local burning through plasmonic heating effects. To minimise the influence of these factors on the DC, the samples shown in Figure S4 were prepared with a hatching distance of 0.8 μm, and adjacent rods were fabricated with a pause of 1 second interval. For general 3D feature fabrication, the hatching distance is normally less than 0.8 μm and without pause or with shorter pause between adjacent lines or layers. Then the actual achievable DC can be higher. (a) (b) Figure S6: SEM images illustrate how to prepare TEM sample. (a) A piece of helix structure prepared by simultaneous two-photon polymerisation and photoreduction was picked up by a micro-manipulator; and (b) was then transferred to a TEM sample holder.

XPS analysis
Samples were analysed using the Kratos AXIS ULTRA with a mono-chromated Al kα X-ray source (1486.6eV) operated at 10 mA emission current and 12 kV anode potential (120 W). Samples were mounted on a standard Kratos sample bar using double sided tape and pumped down to UHV overnight before insertion to the analysis chamber. A charge neutralizer filament was used to prevent surface charging. Hybrid-slot mode was used measuring a sample area of approximately 0.5 mm 2 , note that this is significantly larger than the size of the structure being analysed so a large proportion of the signal detected in the wide scan will come from the surrounding glass substrate material. The analysis chamber pressure was better than 5 x 10 -9 mbar. A wide scan at low resolution (1400 --5 eV binding energy range, pass energy 80 eV, step 0.5 eV, sweep time 20 minutes). High resolution spectra at pass energy 20 eV with step of 0.1 eV, sweep times of 10 minutes each were also acquired for photoelectron peaks from the detected elements. The high resolution spectra were charge corrected to the C 1s peak set to 285 eV. Casaxps (version 2.3.18dev1.0x) software was used for quantification and spectral modelling. Figure S9: A low resolution wide scan XP spectrum over the full energy range, 1400 --5 eV binding energy for the Au NP in polymer matrix. Labels indicate the principle photoelectron peaks detected. Since no Cl was observed in the wide scan spectrum, it is reasonable to assume that no gold chlorides from the starting material were present.

3D composite structures
(a) (b) Figure S10: Tilted SEM images showing the ring structure and pyramid structure shown in Figure 6(d) and (e). Please note the contrast was adjusted to show more detail of the surface morphology.

UV-vis spectroscopy:
Visible light of Thorlabs OSL 1-EC Fiber Illuminator was collimated and focus with an Olympus Plan 10x Objective. The focus spot size on the sample was ~200um. The sample was mounted on a XYZ translation stage with micrometre drives and placed at the focal point. Light after the sample was collected with another Olympus Plan 10x Objective and lens and then focused into an Andor Shamrock 303i spectrometer with a grating of 300l/mm and measured with an Andor Newton 940U CCD. Due to limited spectral coverage, one spectrum was obtained by combining two spectra centred at 560 nm and 700 nm respectively. Reference was measured at a position of substrate without polymer/gold structure and signal was measured at a position of substrate with polymer/gold structure. The spectrum is then calculated as, Four Au-containing samples and one none-Au sample were prepared for optical absorption analysis, each with a thin film of 1 mm x 1mm x 2 µm on glass substrate. The big thin film structure was made by stitching 100 (10x10 style) unit square structures, each with a size of 100 µm x 100 µm x 2 µm (due to the limited projection area of the Galvo scanner during laser fabrication). Therefore, the collected absorption signal came from 4-9 unit square structures, depending on the projection area of the light source on sample. For each sample we have examined different locations. As mentioned in the main text, local burning was sometimes observed, due to the plasmonic heating of the in situ generated Au nanoparticles. Although much effect has been tried to optimize the laser parameter setting to minimize this random effect, it is challenge to completely eliminate this phenomenon through 100 times unit structure repetition to reach mm areas. Therefore, minor shift in the location of the absorption peak and strength was sometimes observed for different samples and different examining area.