Structural colors based on optical metasurfaces have been widely used in color filters, surface holograms, and anti-counterfeiting [1–5]. Vivid structural colors originate from the interaction of light with optical metasurfaces at the subwavelength scale, yielding salient features including high efficiency, environmental friendliness, long-term stability, high resolution, and a non-photobleaching nature compared with traditional pigments and dyes [1,6,7]. Optical metasurfaces comprise sophisticated nanostructures in a predesigned arrangement to achieve customized optical functions, and their fabrication is an extremely delicate process. Although electron-beam lithography [8], nanoimprinting lithography [9], and focused ion-beam milling [10] have been primarily exploited to create various metasurfaces and structural colors, their unaffordable cost and time consumption, as well as multistep procedure, hinder the generation of structural colors over large areas with high throughput for fine arts, printing, and technological applications [2,4,11].

Laser processing is another approach to obtaining structural colors because it is a cost-effective, rapid, flexible, and one-step treatment process [11–18]. Currently, laser coloring predominantly is based on random nanostructures and laser-induced periodic surface structures (LIPSSs) [8,11–18] produced by direct laser writing and plasmonic lithography, respectively. However, random nanostructures-enabled colors are plagued by poor resolution (typically 0.8–50 dpi), which is substantially lower compared with the case of optical metasurfaces (i.e., 100,000 dpi) [7]. The disorder of random nanostructures leads the laser spot size to determine the resolution of the created colors according to statistical parameters [11]. A small size laser spot is necessary to obtain the high resolution but is detrimental to high productivity [12,14], which causes a trade-off. The appearance of LIPSSs alleviated this trade-off because its feature size is independent of the laser spot size [11,12,19,20]. Unfortunately, the resolution of LIPSSs [21] was confined by the optical diffraction limit resulting from the invalidity of high-spatial-frequency LIPSS (HSFL) for structure colors [11,12,22]. Using femtosecond laser plasmonic lithography (FLPL) [13,20,22] can excite HSFL on some wide bandgap materials, which allows for nanostructuring far below the optical diffraction limit [13,20,22]. But HSFL is difficult to grow on low-loss metals (e.g., Au and Ag), which are the most popular materials in laser coloring with the potential for exhibiting vivid colors [12,15,18].

A primary obstacle for generating HSFL on Au is the tenuous electron–phonon coupling relatively expediting the hot-electron diffusion effect. Subsequently, energy with periodic distributions stemming from the interference between incident fs-laser and surface fields [23,24], cannot be delivered from electrons to neighboring lattices due to rapid hot-electron diffusion. Therefore, initial periodic energy distributions are interrupted, and growing HSFL becomes impossible [25]. In addition, surface plasmon polaritons (SPPs) excited on an Au surface are prone to loss of coherence [13,26], which further affects the adoption of HSFL on Au.

This study realized high resolution and productivity of color printing simultaneously through forming HSFL on Au by FLPL, which was attributed to the use of amorphous Au (a-Au) on a SiO2/Si substrate. Exploiting the capillarity of single-layer graphene (SLG), the chromotropic capability for the background colors of color printing was demonstrated after sandwiching SLG between a-Au and SiO2/Si to form Au/SLG HSFL. Formation of HSFL was attributable to the disordered lattice structures and high transmission of a-Au film, which enables suppressed hot-electrons diffusion and artificial “seed” pre-structures (i.e., LIPSS on a Si surface), respectively. Due to fs-laser-induced crystallization, the changed complex refractive index (n=n2+jk) of Au during the conversion from a-Au to crystalline Au (c-Au) endowed the proposed color printing with a wider gamut (approximately 55% sRGB) than general laser coloring (15%–35% sRGB), owing to the introduction of the photonic effect into the intrinsic plasmonic coloring mechanism. Finally, multi-level anti-counterfeiting along with customizable shapes with the assistance of isolation zones is demonstrated to illustrate the application potential in security marking.

Figure 1(a) shows a schematic FLPL process for high-resolution and high-speed chromotropic color printing incorporating Au/SLG HSFL. FLPL was conducted on a 50-nm-thick a-Au/SLG layer supported by a SiO2/Si substrate. The a-Au was deposited via thermal evaporation with a low deposition rate of 0.02nms−1 owing to the integrity of the crystal structure of Au depending on the evaporation rate [27]. A fs-laser beam with a wavelength (λ) of 520 nm was delivered to the surface of a-Au/SLG via an objective lens with a numerical aperture (NA) of 0.25. The number of fs-laser pulses per irradiated spot (N), scanning period, and scanning speed were fixed to 1408 (corresponding to a repetition rate of 500 kHz), 1.5 μm, and 250μms−1, respectively. The scanning period alludes to the distance between two adjacent scanning traces. Considering enhanced transparency of a-Au, fs-laser pulses first induced LIPSS at the Si surface passing through a-Au/SLG and SiO2 layers, which functioned as “seed” pre-structure to guide the stable and strong coupled SPP at the Au surface. Consequently, large-area Au/SLG HSFL was generated quickly (Fig. 7, Appendix A). According to different fs-laser fluences (F), the FLPL can generate wide-gamut colors as shown in Fig. 1(b) owing to the hybrid plasmonic/photonic mechanism originating from changes in n and structural morphologies of Au/SLG HSFL. The change in n can be attributed to the crystallization in Au, which transforms a-Au into c-Au with the reorganization of lattice gains during FLPL [28–30]. In addition, based on the crystallization, patterned Au exhibited corrosion resistance such that only background color (pristine a-Au/SiO2/Si) was varied from yellow to green (SiO2/Si) while preserving the printed colors after etching treatment. Moreover, the background color can be remotely altered and flexibly shaped due to the presence of SLG featuring capillarity to trap Au etchant [31], as shown in Fig. 1(a). The period of Au/SLG HSFL was identified to be 100 nm using the typical two-dimensional fast Fourier transform (2D-FFT) spectrum [Fig. 1(c)], derived from Fig. 7 (Appendix A). The height of Au/SLG HSFL is approximately 50 nm as shown in the transmission electron microscopy (TEM) images of the Au/SLG layer before [Fig. 1(d)] and after FLPL [Fig. 1(e)], respectively.

Except for lowering the fluence threshold for forming uniform HSFL, the SLG has no significant impact on the formation of ultimate HSFL structures, as shown in Fig. 8 (Appendix B, for details). Thus, for convenience SLG is omitted from the discussion of the operation mechanism. The formation of Au HSFL was based on modulation both of hot-electron diffusion and SPP on the Au surface via disturbing the lattice arrangement of Au and growing artificial “seed” pre-structures, respectively. As shown in Figs. 2(c1), 2(d1), 9(d), and 9(g) (Appendix C), disordered lattice arrangement in a-Au was confirmed, which was supposed to ensure the periodic energy modulation can be imprinted successfully on lattices at the a-Au surface through electron–phonon coupling [23,29,32,33]. Because the coherence of SPP on Au surface is susceptible to loss [13,26], utilizing a-Au can only yield periodic structures but not consistent and stable structures (see Appendix D for details). As Si is among the materials most prone to generating LIPSS with its top surface turning metallic under laser irradiation [34], growing LIPSS on Si has been used to guide other materials’ nanostructuring [34,35]. Similarly, a SiO2/Si substrate is used to form LIPSS initially at the SiO2/Si interface, which can guide the stable and strong SPP on the Au surface.

Specifically, the formation of Au HSFL can be summarized in three steps as illustrated in Fig. 2, dominated by a-Au, intermediate state Au (Inter-Au), and c-Au classified by the orderliness of the Au lattice with respect to N (see Fig. 9, Appendix C, for details). (1) Several fs-laser pulses were focused on a-Au; however, most were absorbed by the Si substrate [Fig. 11(a), Appendix E] after passing through the a-Au layer, owing to its negligibly small imaginary part of the dielectric constant (ε′′), as presented in Table 1 (Appendix E). Consequently, the SPP at the SiO2/Si interface (SPPSiO2/Si-∥) was excited [34,35] with a parallel orientation to the polarization (i.e., E) of the applied fs-laser beam, as displayed in Fig. 2(a1). Figure 12(a) (Appendix E) shows that the direction of the SPP at the air/a-Au interface (SPPair/a-Au-⊥) was perpendicular to E with a single void hemisphere as the surface defect, attributed to the non-metallic properties of a-Au [36] (real part of the dielectric constant, ε′>0). (2) With additional fs-laser pulses (N≥6), a-Au was converted to Inter-Au with a better arrangement of lattice structures owing to fs-laser-induced crystallization, as shown in Figs. 2(c2) and 2(d2). The self-organized reliefs as “seed” pre-structure on the Si surface were generated following the energy modulation of SPPSiO2/Si-∥ [Figs. 2(a2) and 2(b2)], whose period and depth were 320 and 15 nm, respectively. Therefore, the thickness of the SiO2 layer increased from 340 to 355 nm [Figs. 2(b1) and 2(b2)]. The generated reliefs strengthened the SPPSiO2/Si-∥ that interacted with SPPair/a-Au-⊥ to form stable coupled SPP [SPP-∥, Fig. 2(a1)] at the Inter-Au surface, which is parallel to E, as shown in Fig. 12(c) (Appendix E). The coupled SPP-∥ evolved into periodic structures that were randomly distributed in the top thin layer of Inter-Au (Fig. 13, Appendix E), considering the increased absorption of fs-laser energy by the Inter-Au layer [Fig. 11(c) and an increased ε′′ in Table 1]. (3) As N increased further, the classical (111) lattice plane of Au [37] with a plane spacing of 0.235 nm was observed, as shown in Figs. 2(c3) and 2(d3). From the 2D-FFT image of a larger lattice structure area, another lattice distance of Au (220) [38] was identified for N=1408 (Figs. 93f and 93i). Therefore, a-Au converted into c-Au, exhibiting a larger ε′′ (Table 1). As shown in Fig. 11(d), the pulses were mostly absorbed by the c-Au layer to excite SPPair/c-Au-∥ with the assistance of periodic structures previously formed during the Inter-Au period (Figs. 12 and 13, Appendix E). Finally, the uniformly covered Au HSFL was established following the energy modulation by SPPair/c-Au-∥. Owing to optothermoplasmonic and catalytic effects of Au on SLG, Au/SLG HSFL was generated following the addition of SLG between a-Au and the SiO2/Si substrate [32].

As shown in Fig. 3(a), a wide gamut (approximately 55% sRGB) can be achieved by adjusting F, which is wider than the typical value (15%–35% sRGB) of laser coloring [12]. Au/SLG HSFL formation was observed at F=1.97Jcm−2, and henceforth, evolved into homogeneous and stable structures for F=2.37Jcm−2, corresponding to the color change from yellow to red [Figs. 3(b) and 3(c)]. In the optical reflection spectra [Fig. 3(d)], a new peak was observed at λ=450nm and the original peak at λ=580nm red-shifted with increasing F. The decreased intensities and red shift of the two reflection peaks changed the reflected color from red to green, with boosted porosity in Au/SLG HSFL at F=3.07Jcm−2. The brown color gradually replaced the green after further increasing F. The reflection peaks continued to red shift and decrease in amplitude. This was interpreted as abundant scattering and localized surface plasmon resonance phenomena [14] with the gradual fragmentation of the Au/SLG HSFL and the formation of spherical nanoparticles [Fig. 3(c)].

The demonstrated wide gamut is attributable to the hybrid plasmonic/optical mechanism, corresponding to changes in nanoscale structures and n of Au [14,29,39], respectively. As shown in Fig. 4(a), the measured n2 and k of a-Au before the FLPL treatment differed from the typical values of Au [40] plotted in Fig. 4(b). Two groups of n2 and k values were considered to calculate the optical reflection spectra of the a-Au/SLG and c-Au/SLG HSFL on SiO2/Si using finite-difference time-domain (FDTD) simulations [39]. The HSFLs were reconstructed from the measured SEM images (Fig. 14, Appendix F) for calculations. Prior to FLPL treatment, the simulated result using n2 and k values of a-Au matched well with the measured reflection spectrum [Fig. 4(c)]; n2 and k of typical Au enabled a better agreement for the case of the c-Au/SLG HSFL [Fig. 4(d)]. The mismatched spectra in the wavelength region of 600nm<λ<900nm in Fig. 4(d), may be caused by dislocations and stacking faults in the lattice structures indicated by the white arrows in Fig. 2(d3); hence, the interband and intraband transitions of Au were inhibited [7,14]. Therefore, complex refractive index, n, altered with the restoration of the crystal structure in Au due to fs-laser-induced crystallization [41]. Additionally, the plasmonic phenomenon was underpinned by the fact that the energy was confined around boundaries of c-Au nanostructures, as observed in the electric field distributions on the surface [Fig. 4(e)] and cross-section [Fig. 4(f)] of the c-Au/SLG HSFL. Finally, the hybrid plasmonic/photonic mechanism was further interpreted by analyzing the absorption features of the proposed device (see Fig. 15, Appendix F, for details).

The color printing built on Au/SLG HSFL featured high production speed and printing resolution simultaneously, which benefited from the insensitivity of feature size to laser beam size in HSFL. For example, the period size and exhibited colors of the two kinds of HSFL structures generated by 0.25- and 0.85-NA objective lenses were identical, as depicted in Figs. 5(b) and 5(c). Both structures facilitated the distinguishable magenta points (approximately 200 nm) under the optical microscope, corresponding to a printing resolution of 127,000 dpi [6,7]. Therefore, the 0.25-NA lens with a bigger beam size was used to print the large-area “red sunset clouds” as shown in Fig. 5(a) for a low time expenditure. Analogously, the printing speed can be further expedited with a larger focused laser beam without concerning the loss of resolution.

Color alterations have been extensively employed in anti-counterfeiting labels, with the most distinctive feature unequivocally defining the identity of the goods [4,5]. We devised a three-tier anti-counterfeiting system built upon the customizable background color of our chromotropic color printing, alongside straightforward interrogation steps. The first level of anti-counterfeiting hinges on different reactions of background and printed colors to aqua regia. As depicted in Fig. 16 (Appendix F), the spacing between crystalline grains decreased [29] in c-Au owing to fs-laser-induced crystallization, resulting in improved corrosion resistance compared with the case of a-Au. Therefore, the integrity of HSFL and printed colors was maintained but the background color changed from yellow to green because a-Au was etched upon the aqua regia treatment for 2 min. Another contributing factor to enhanced corrosion resistance could be the improved hydrophobicity resulting from increased surface roughness with the presence of the HSFL [42]. The second anti-counterfeiting scheme leveraged the remote control of background color via the capillary trapping ability of SLG in aqua regia. As shown in Fig. 6(a), a small droplet of aqua regia can spread to other regions along the a-Au/SLG interface and effectively etch away the a-Au layer. Therefore, the pristine background color can be indirectly modified even at a location far from where the aqua regia was applied. In Fig. 17 (Appendix F), a 2-μL droplet of aqua regia was found to etch yellowish a-Au 1 cm away within 3 min, whereas the printed colors were preserved. Notably, this remote control of background color can be effectively prevented by destroying the SLG to introduce isolation zones [Fig. 6(b)]. As a consequence, the background color could be customized into desired arbitrary shapes, facilitating the third level of anti-counterfeiting. The isolation zones were prepared by fs-laser direct writing to enclose a designated region and entirely prevent aqua regia from spreading into the targeted areas. An increased number of structural defects in SLG in the isolation zones were inspected by the Raman spectra (Fig. 18, Appendix F). The surface morphologies of the Au/SLG hybrid layer in terms of FLPL and etching conditions were investigated using SEM [Figs. 6(c)–6(f)]. For the case of pristine a-Au/SLG subjected to drop-wise aqua regia treatment, a-Au was removed completely with the absence of the isolation zone [Fig. 6(c2)]. In contrast, the a-Au surface was negligibly damaged within the isolation zone [Fig. 6(c3)], implying that the isolation zone played a key role in protecting the pristine background color. Meanwhile, for the Au/SLG HSFL, the surface morphologies and exhibited colors were safely preserved under all etching conditions [Figs. 6(d)–6(f)]. It was consequently verified that the fabricated colors based on fs-laser-induced Au/SLG HSFL were immune to the etching solution.

The Au/SLG HSFL enabled via FLPL solves the trade-off between resolution and speed in the field of laser coloring, thereby achieving a wide-gamut, high-resolution, and high-speed, chromotropic color printing. Employing a-Au with disordered lattice structures and growing LIPSS on SiO2/Si substrate as artificial “seed” pre-structures ensured the formation of Au/SLG HSFL via overcoming the rapid hot-electron diffusion and unstable SPP on typical c-Au surface, respectively. The period of Au/SLG was 100 nm, thereby producing a high resolution of 127,000 dpi for the fabricated colors. Crystallization occurred during FLPL to convert a-Au into c-Au with a changed complex refractive index, which combined with the nanostructural plasmonic effect, resulted in a wide gamut of 55% sRGB. Furthermore, controllable background color and its distinct response compared to printed colors to etching solution were confirmed to work as three-level anti-counterfeiting tags, which are attributed to the capillarity of SLG and fs-laser-induced crystallization. Additionally, the proposed approach holds significance for the advancement of practical color printing, owing to its straightforward color design origins and the ease of sample manufacturing implementation.

On the Si surface, 0.3 and 0.02nms−1 deposition rates were used to deposit 340-nm-thick SiO2 and 50-nm-thick a-Au [26] layers via standard thermal evaporation, respectively. SLG was synthesized via chemical vapor deposition and transferred onto SiO2 before depositing a-Au [43]. The thickness of the SLG was estimated to be 0.4–0.5 nm. A fs-laser beam (250 fs, Lasernics, Republic of Korea) with a wavelength of 520 nm was irradiated on the surface of a-Au via a 10× and 0.25-NA objective lens. The scanning speed and period were set to 250μms−1 and 1.5 μm, respectively, using a computer-controlled translation stage (XYCV630-C-N, Misumi, Japan). Further details regarding the fs-laser system are provided in our previous work [44].

A field-emission scanning electron microscope (FE-SEM, Inspect F50) was used to capture all SEM images. A focused ion beam (Quanta 3D FEG) cutting method was used to prepare the TEM samples. Cross-sectional and high-resolution images of atomic structures were obtained using TEM (JEM-2100F) and analyzed using the Digital Micrograph Software (Gatan, USA) to obtain the 2D-FFT and IFFT results. The reflection spectra were captured using a spectrometer equipped with a reflection probe, which was designed to detect optical beams that were normally reflected off the samples. An ellipsometer (Ellipso Technology, Republic of Korea) was used to measure the complex refractive index of a-Au. A 100× objective lens was adopted to obtain the Raman spectra (FEX, NOST) of the SLG by focusing a 531 nm laser with a power of 0.3 mW.

For three-dimensional FDTD (Ansys/Lumerical, Canada) simulations pertaining to the optical reflection spectra [Figs. 4(c)–4(f)], the smallest mesh sizes for Au regions and SLG were set to 0.8 and 0.1 nm, respectively. The c-Au structural morphology was directly extracted from the SEM images of the Au/SLG HSFL. It is noted that periodic boundary conditions and perfectly matched layers were applied in the horizontal direction at the top and bottom boundaries, respectively.

For the FDTD simulations in connection with Figs. 11 and 12 (Appendix E), the mesh sizes for Au, SiO2/Si, and Si (metalized) relief regions were 3, 8, and 1 nm, respectively. A hemisphere made of “etch” material was embedded from the a-Au surface to mimic surface inhomogeneity. To emulate the actual processing conditions, a Gaussian beam exhibiting a pulse duration and wavelength of 250 fs and 520 nm, respectively, was used as the light source.

Acknowledgment. This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education, Ministry of Science, ICT, and Kwangwoon University in 2024.