1. Introduction

Femtosecond laser-induced periodic surface structures (LIPSS) have attracted much attention as an efficient surface texturing technique since 1965 when they were observed on surfaces irradiated by newly invented lasers [1–3]. LIPSS typically exhibit periodic ripples, which can be classified into near-subwavelength ($0.4\lambda <\varLambda <\lambda$) and deep-subwavelength ($\varLambda < 0.4\lambda$) structures depending on their periods [4]. Since these periodic structures can change the surface morphology as well as the optical, tribological, wettability, and electrical properties of materials, they are used for biomimetic surfaces [5,6], biosensing [7], wettability modulation [8], micro-optics [9–11], industrial anti-icing [12,13]. However, due to complex formation processes occurring over a few orders of magnitude in time and space after ultra-short sub-1 ps laser energy delivery, a better control for the fabrication of nanoscale periodic structures is needed.

For near-subwavelength ripples, the early explanation was that they originate from the interference between surface scattered electromagnetic waves and incident light, and their period is expressed as $\varLambda =\lambda /(1\pm ~\sin \theta )$, where $\theta$ is the angle of incidence and $\pm$ sign depends on the polarization s-/p-pol. [14,15]. In 2013, Ilday et al. reported a method to create metal oxide periodic structures using dewetting of molten nano-film and showed uniform large-area near-subwavelength patterns [16]. Regardless complexities of hydrodynamic motion, thermal quenching, and heat deposition/diffusion, it appears that even the initial energy deposition step in ripple formation needs revisiting of the scattering model [4,17]. It was suggested that the scattering model should be modified to take into account the effect of surface plasmon polaritons (SPPs), considering that the ripples are induced by SPPs-laser interference and subsequent ripple-assisted interference [4,18,19]. The SPPs model has been widely accepted and used to explain the regularity of ripples and the competition between near-subwavelength and deep-subwavelength structures [18,20]. In recent years, a number of remarkable works inspired by the SPPs model have emerged aiming at fast, large-area nanoimprinting of subwavelength and deep-subwavelength structures [20–25]. However, precise modulation of the period of deep subwavelength structures at the nanoscale has not been reported. Current methods to modulate the periods include changing the pulse number [26,27], pulse energy [26,27], laser wavelength [27,28], and material (actually it’s the refractive index) [25]. The first two are applied to control the periods before reaching the stable value, at which stage the ripple uniformity tends to be identical. [18,25] The last two are usually fixed to specific experimental conditions and are not practical. Stamkevic et al. reported a noteworthy approach. with the demonstration of heat diffusion changes along and across the ripple pattern coupled with the scanning direction of a tightly focused linearly polarized beam. Such anisotropic thermal diffusion changes the period and width of the nanograting for different scanning directions. This anisotropy is mainly manifested in two scanning directions, 0 and $\pi /2$ and it is difficult to achieve continuous modulation of the period. This limits the practical application of this method [29], which was demonstrated on the surface of a volumetric sample rather than a coated film. Another unique study was reported by Ionin et al. They developed a technique to manipulate the nanotopology and chemical components of titanium surfaces using forward and backward scanning of an asymmetric semicircular shaping beam. And this anisotropic laser scanning is well explained using the variation of the plasma resonance period with different irradiation fluxes [27,30,31].

Here, we demonstrate a femtosecond laser (343 nm/280 fs) non-reciprocal ripple writing via asymmetric optical far-field with the beam focused through a reflective objective lens. The period of the ripples changes along the opposite scanning directions with control within a 47 - 112 nm range on the ITO 100-nm-thick film on glass. The full electromagnetic model based on the electric field distribution around the induced structures at different stages of their evolution was adopted to illustrate the near-field induced dielectric breakdown guided and controlled by the optical far-field. The non-reciprocity of patterning in two directions up and down observed experimentally is also theoretically predicted. The model-predicted patterning of ripples was made with the period control via the scanning direction (up and down) using a simple beam shaping technique by an aperture mask at the pupil of an objective lens.

2. Materials and methods

ITO film of 100 nm thickness deposited onto glass substrates using magnetron sputtering was used in this study (acquired from Jinan Delta Optoelectronic Technology Co.).

The all-solid-state femtosecond laser with wavelength of $\lambda = 343$ nm and pulse duration of $\tau _p = 280$ fs and repetition rate of $f = 200$ kHz (Pharos, Light Conversion) was used for ripple patterning of ITO-on-glass. A 5-mm-diameter linear-polarized beam passes first through a motorized attenuator consisting of a polarizing beam splitting prism (PBS) and a motor-driven half-wave plate (Fig. 1(a)). Then, a beam expansion system consisting of two convex lenses expands the beam diameter to 10 mm, matching the entrance diameter of the galvo scanner unit (Sunny Inc.). Then, the beam passes through a motor-driven half-wave plate for polarization orientation and enters the galvo scanner. The galvo scanner acts as a positioner to deflect the beam in the XY-plane (linked to the plane of the sample’s surface). The outgoing beam is focused on the sample’s surface by a reflective Cassegrain objective lens ($40\times$, the numerical aperture $NA = 0.5$, THORLABS) after passing through a 4F beam delivery system. The sample’s position for focusing was controlled by a Z-axis piezoelectric nanopositioner (PI, Physik Instrumente). Our imaging system uses transmission illumination and a high-definition CCD camera (Nexcope Inc.) (Fig. 1(a)).

 

Fig. 1. (a) Schematics of the femtosecond laser direct writing system used in experiments($\lambda = 343$ nm, $\tau _p = 280$ fs, repetiton rate $f= 200$ kHz). (b) The intensity distribution of the focused spots which was calculated by Fraunhofer diffraction. (c) SEM (scanning electron microscopy) image of the laser-ablated structure on ITO film. Scale bar $5~\mu m$.

Download Full Size | PDF

The paper mask was designed as a sector with a diameter of 12 mm and an arc angle of $330^\circ$. Then laser ablation is conducted along the border of a sector forming a mask. After careful alignment, the mask was placed directly at the pupil of the objective lens.

The 3D positioning of the irradiation point on the sample for laser processing was defined with a resolution of 1 nm. At each irradiation point, additional variables such as shutter, polarization, and pulse energy were controlled for experimentation. The exposure time at each coordinate defined the number of pulses per spot and a cumulative exposure dose. With this laser fabrication system, precise control of scanning and exposure conditions were achieved. Herein, a point-by-point writing strategy is used for the point and line patterning as well as large-area texturization. The overlap of pulses was controlled by defining exposure time at the chosen coordinate rather than by scan speed. This was important for the analysis of non-reciprocity origin which depended on the writing direction (variability on the speed can be excluded for analysis of the mechanism).

The pulse energies were measured from the front of the objective lens using a handheld laser power meter (843-R and 919P-040-50, Newport). The transmittance of objective lens is 74%.

The surface morphology of the irradiated samples was analyzed by field-emission scanning electron microscopy (SEM, JEOL).

3. Results and discussions

3.1 Laser-induced structures (statics): pulses per spot

The reflective objective lens’ unique structure distinguishes it from conventional optical objectives lens. It consists of a circular convex mirror and an annular concave mirror, where the convex mirror is held in place by three curved vanes. This structure causes the input beam to be equally divided into three output beams so that the collection of three beamlets occurs when it is tightly focused. We first calculated the intensity distribution at the focal point using the Fraunhofer diffraction integral, as shown in Fig. 1(b). Most of the light intensity is distributed in the center of the focused spot. In order to observe the triangular side lobes, the pulse energy was gradually increased while adjusting the Z-position of the focus. In such cases, the surface ablation threshold was also surpassed at the locations feather away from the center of the focal region. The results are shown in Fig. 1(c), and it can be clearly observed that the focused spot of the reflective objective lens is composed of three adjacent triangular regions with a characteristic size of about 10 $\mu m$. A Newtonian ablation ring structure with a characteristic size of about 2.5 $\mu m$ due to interference is discernible (Fig. 1(c)).

A morphological study was perfo rmed to establish the focus features of the near-threshold ablation pattern. Figure 2 shows the $20\times 20$ exposure matrix of the array fabricated by changing the pulse number PN (1 to 20) and pulse energy PE per spot (conditions are shown in Table 1), and the diameter of focused spot is $D=1.22\lambda /NA=836.92$ nm (laser flux affects the effective irradiated area). The effects of pulse number $PN$ and pulse energy $PE$ on the induced structures were analyzed. In the near-threshold case, with pulse energy from 48.5 to 55.4 nJ ($PE =$1-to-7), the increase in the pulse number $PN$ causes the induced structure to be stretched from a nanohole into a nanogroove (perpendicular to the polarization $E$-field, has been systematically studied by our previous work [32]), and then continues to grow into multiple nanogrooves (along the polarization $E$-field) until it fills the focal spot-covered area. Take the induced structures in the green boxed region in Fig. 2 as an example (PE=5), obviously the accumulation of pulses will increase the number of nanogrooves in the same area and reduce the period $\varLambda$ of the ripple pattern. When the number of pulses is fixed, an increase of pulse energy enlarges the central groove (see blue boxed region in Fig. 2). The initial structure ablated by the action of 1-2 pulses constitutes the basis for forming of multiple nanogrooves. The high pulse energy significantly increases the size of the initial structure , upon which subsequent pulses drive the structure growth, resulting in the nanoripples with a wide central nanogroove (see yellow boxed region in Fig. 2). Increasing the pulse energy expands the area of the focused spot where the threshold is crossed, thus increasing the number of nanogrooves. Ablated structures formed at high PE and PN (see the red boxed region in Fig. 2) show that the ripple pattern has been completely ablated and only the perimeter of the ablated pit shows the ripple structure.

 

Fig. 2. SEM image of the exposure matrix of laser-induced structures on ITO-on-glass with different exposure conditions. The scale bar is $1~\mu$m, polarization marker show the orientation of the linearly polarized beam; see Table 1 for the pulse energy values. The highlighted areas are discussed in the text.

Download Full Size | PDF

Table 1. Pulse energies used for Fig. 2.

View Table

The asymmetric triangular focused spot significantly affects the distribution of the nanogrooves. Figure 3 shows the nanoripples induced by the triangular focused spot (corner located on the top) at the pulse energy $E_p=53.1$ nJ. It can be observed that the spacing of the nanogrooves near the top corner of the triangular-focused spot is larger, while the spacing of the nanogrooves near the bottom edge of the triangular-focused spot is smaller. With a high number of pulses, there is a probability that a new nanogroove will form between two widely spaced nanogrooves (see Fig. 3(d)), this phenomenon is called cleavage.

 

Fig. 3. SEM image of nanoripples induced by (a) 9 pulses, (b) 11 pulses, (c) 16 pulses, and (d) 20 pulses at pulse energy "6" of $E_p=53.1$ nJ

Download Full Size | PDF

3.2 Non-reciprocal nano-ripples at different scanning directions (dynamics)

Groups of lines were written on the ITO film with different scanning directions under varying laser exposure conditions. The laser beam was linearly polarized along the scanning direction (E-field marker in Fig. 4). The scanning direction was up and down (Fig. 4) , up is the scanning from the bottom edge of the triangular focused spot to the top corner and down is the opposite direction. The defined scan step was $\Delta y = 100$ nm. The period of the ripples written by scanning in the up direction was designated $\varLambda _{up}$ and the period along the down direction was designated $\varLambda _{down}$, and the values were measured from the Fast Fourier Transform (FFT, performed by ImageJ) image of the corresponding ripples. First, it was found that the period of the ripples strongly depends on the scanning direction (Fig. 4). The $\varLambda _{up} = 60\sim 112$ nm under different laser exposure conditions, while $\varLambda _{down}~=~48\sim 80$ nm under the same conditions. Their difference $\Delta \varLambda$ had a maximum value of 33 nm (Fig. 4(a)) and a minimum of 9 nm (Fig. 4(h)); the uncertainty of the ripple’s period definition by SEM image analysis using the Fast Fourier Transform (FFT) was extracted as the FWHM of the spatial frequency and was about $\pm 4$ nm. This direction-dependent, hence, non-reciprocal writing was further investigated.

 

Fig. 4. Laser-induced periodic ripples with different scanning directions under various laser exposure parameters. The polarization of the laser pulses was along the scanning direction. The red arrow indicates the up direction, and the yellow arrow indicates the down direction. The pulse numbers and pulse energies are shown on the SEM images. The scale bars are 200 nm. The step was $\Delta y = 100$ nm between exposure sites.

Download Full Size | PDF

Secondly, it was observed that the period of the ripples does not depend on the pulse energy. Figure 4(a,b,c,d) and Fig. 4(e,f,g,h) illustrate the dependence on the number of pulses at two pulse energies for the up and down writing. For $\Delta \varLambda _{up} = 3\sim 10$ nm and $\Delta \varLambda _{down} = 1\sim 4$ nm, this difference is close to the error margin of the measurement, but there was a discernable trend. The effect of pulse energy is reflected in the structural morphology. Figure 4(a,e) shows the effect of increased pulse energy on the probability of ablation, the enlarged lateral dimension of the modified region (The effective irradiation area of the focused spot is expanded from 460 nm to 560 nm) and a change in the duty cycle of the ripples. In summary, Fig. 5(a) shows that when the $PN$ and scanning direction are fixed, the increased pulse energy does not contribute to the change in ripple’s period. This is the same trend as observed for the single-site induced structure discussed in the previous section. The reason for the difference from other studies is that the films are easily removed by the high pulse energy and more complex structures cannot be observed.

 

Fig. 5. The period of laser-induced periodic ripples versus pulse energy (a) and pulse number (b); a marker orientation indicates the scanning direction. The error bar represents the standard deviation of the data estimated from five sets of ripples (each processing parameter was repeated five times).

Download Full Size | PDF

Finally, the period of the ripples is strongly influenced by pulse numbers. In Fig. 4(a,b,c,d) and Fig. 5(b), it is summarized that the period decreases nonlinearly with pulse numbers until it saturates. Numerous previous studies have shown that the period of nanoripples is limited by material and laser wavelength [26,27]. However, what is the minimum value of the nanoripple’s period under certain conditions has hardly been reported. We will study this interesting topic later.

3.3 Numerical simulation of the light intensity re-distribution

A full electromagnetic model has been developed using Comsol Multiphysics (Comsol Inc.) to explain the mechanism of non-reciprocal writing. The model is a three-layer structure of air(300 nm, $n_{air} = 1$), ITO film (50 nm, $n_{ITO}\equiv \sqrt {\varepsilon _{ITO}}=2.15$) and glass substrate ($50~nm$, $n_{glass} = 1.5$); a perfectly matched boundary condition was set at the very bottom to eliminate back reflection. A half-thinner ITO was used to speed up resource-demanding calculations that providing semi-quantitative insights into a light intensity distribution. An equilateral triangular window/aperture with a side length of 700 nm was set at the top of the air layer illuminated by a Gaussian beam, i.e., the focal region has a triangular shape (Fig. 6).

 

Fig. 6. Numerical simulation of the electric field distribution near the induced structures under in-situ pulsed irradiation. Induced structure with (a) low pulse number and (b) high pulse number. Laser beam linear-polarized along the y-axis.

Download Full Size | PDF

3.3.1 Static model

The laser-induced nanogroove is approximated as an elliptical cylinder and removed from the ITO film layer (the entire 50 nm thickness of ITO is replaced by air). Figure 6 displays the electric field distribution near the induced structure under in situ pulse irradiation, which is used for semi-quantitative analysis at the initiation conditions used for the linear scan. Here, the entire light exposure was similar to the case of the near-threshold ablation shown in Sec. 3.1. The first few pulses induce a nanohole at the center of the focused spot, which is subsequently stretched into a nanogroove in the direction perpendicular to the polarization (see Fig. 6(a), level-I). After the appearance of the level-I nanogroove, the originally Gaussian-distributed electric field is modulated as shown in Fig. 6(a); the green curve profile shows the $I=|E|^2$ at the central axis. It is revealed that the nanogroove has strong subwavelength enhancement areas along both ends of the x-axis. The electric E-field on both sides along the y-axis first weakens and then strengthens with the appearance of two level-II enhancement lobes along the polarization orientation. Due to the asymmetry of the focused spot, the level-II enhancement area on the top side of the triangle is located at $Y = 100$ nm while the level-II enhancement area on the bottom side of the triangle is located at $Y =-80$ nm. Thus, the level-II nanogroove appears at the corresponding position as in Fig. 6(b). Similarly, the level-III enhancement area on the top side of the triangle is located at $Y = 200$ nm and the level-III enhancement area at the bottom side of the triangle is located at $Y~= -160$ nm, respectively. The simulation data of these ideal structures are close to our experimental results, which illustrates the accuracy of the model.

3.3.2 Dynamic model

The case of scanning along different directions is simulated. Figure 7(a) shows the induced structures scanned along the down direction. The focused-spot position was kept unchanged and the induced structures were shifted by 100 nm in the positive direction. Since the newly induced structures are all irradiated by the bottom edge of the triangular spot first, they will be preferentially generated in the level-III enhancement region (bottom side of Fig. 6(b)). It can be observed that the level-IV enhancement region is located at $Y = -140$ nm, which yields in estimate of the spacing/period of ripples $\varLambda _{down}=80$ nm.

 

Fig. 7. Numerical simulation of the electric field distribution near the induced structures under misaligned pulsed irradiation. Induced structures with laser beam shifted (a) down $100~nm$ and (b) up $100~nm$. Laser beam linearly polarized along the y-axis.

Download Full Size | PDF

Figure 7(b) shows the induced structures scanned along the up direction. The same strategy was adopted by keeping the spot unchanged and shifting the induced structures by 100 nm in the negative direction. Since the top angle of the triangular focused spot will act as the leading side along the scan path, new induced structures will appear in the level-III enhancement region (top side of Fig. 6(a)). As a result, the level-IV enhancement area located at $Y=200$ nm is formed. Then, one can estimate $\varLambda _{up} = 100$ nm. Although the presented model is relatively simple and neglects both, the nonlinear light-matter interaction and the stochastic light scattering occurring in realistic structures, and does not fully display the evolution of the nanoripples’ formation, it matches the experimental results and captures light redistribution. The localized near-field enhancement and redistribution of intensity due to nano-ablated structures are realistically captured by the model. In an iterative way of light distribution, the opening of nanogrooves upon asymmetric light irradiation along the scanning direction, the model provides useful insights into ripple formation. The feedback between the optical far-field enhancement and nano-ablation is consistent with the experimentally observed formation of LIPSS. Therefore the intensity distribution and shape of the optical far-field affect the period of the induced structure.

3.4 Writing of ripples during in-plane scanning

Next, the effect of ripple formation at different scanning directions was experimentally investigated. Figure 8(a) shows the induced structures written along a counterclockwise circular path at a fixed orientation of the linearly polarized beam (Fig. 8(a)) and at the spacing of 100 nm along circular scan trajectory. The period of ripples was measured every 45 degrees according to the angle of the azimuthal position. The results contain data of 8 scanning orientations plotted in the polar diagram (Fig. 8(b)). The period on the scan path can be divided into two parts, including the scan along the down direction from 90 to 270 degrees with period $\varLambda _{down} = 55.9 \sim 60.8$ nm and the scan along the up direction from 270 to 90 degrees with period $\varLambda _{up}=70.7\sim 76.5$ nm. The non-reciprocal nature is manifested here and is determined by the redistributed localized near-field as described above in the simulations section. The reason why there is a difference in the period only in the up-down direction of the scan as well as no significant difference in the left-right direction is that the electric E-field is polarized along the vertical direction. It is also noteworthy that both the maximum and minimum values of the period appeared on the diagonal 45-225-degree axis, which is also the position of the maximum line width. Similar phenomena were reported in previous studies [29], which attributed it to laser polarization affecting the light-matter interaction for writing ripples on the bulk sample. A more detailed study of the symmetry axis shift during the circular scan will be investigated separately.

 

Fig. 8. (a) Laser-induced multilayer counterclockwise ring structures, the scale bar is 5 $\mu m$. The enlarged views of (b) $0^\circ$ and (c) $270^\circ$ scanning direction, the scale bars are 200 nm. (d) Polar diagram of period and scanning direction ($\phi$, degrees).

Download Full Size | PDF

3.5 Nanoscale control of ripple’s period

Based on the fact that the intensity distribution and the shape of the focused spot affect the induced structures, the manipulation of the focused spot by beam shaping is considered next to change of the ripple’s period. A simple and practical technique using a paper mask as an aperture was adopted. A galvo-scanner combined with a field lens was used to cut the desired shape out of paper. The obtained mask was placed at the entrance pupil of the objective Cassegrain lens. The mask was shaped so that only a part of the original beam was transmitted. The orientation of the mounting angle of the mask was adjusted to obtain four distinct cases shown in Fig. 9(a,b,c,d). The corresponding intensity distributions in the focal plane and the off-focus plane are shown in the images below. The optical far field of these numerical simulations does not exhibit sufficient asymmetry in the focal plane, but significant non-reciprocal writing can be observed from the results. And the optical far fields in the off-focus plane show a matched asymmetric intensity distribution. The possible reason is that the effective irradiation region of the shaped focused spot differs from the simulated results and it remains asymmetric thus driving asymmetric localized redistributed near-field ablation.

 

Fig. 9. The laser beam (white area) after shaping using a paper mask (blue area) are shown for different cases: (a) bottom, (b) top, (c) left and (d) right beams defined by the different mounting angles, respectively. The images below show numerical simulation of the corresponding intensity distribution in the focal plane and the off-focus plane. (e)-(h) SEM images of the ablated ripples corresponding to the shaped beams above at the pulse number $PN = 10$ per site and pulse energy $PE = 96$ nJ and 100 nm steps for site relocation along the scanning direction up and down. The red arrow indicates the up direction, and the yellow arrow indicates the down direction. The scale bars 200 nm.

Download Full Size | PDF

The ablated structures in the case 1 are shown in Fig. 9(e) with $\varLambda _{up} = 65$ nm and $\varLambda _{down} = 50$ nm. The mask position of case 2 is opposite to that of case 1. Hence, the ablated ripples have $\varLambda _{up}=50$ nm and $\varLambda _{down}=59$ nm (Fig. 9(f)), which have the opposite size relationship to the case 1. The mask of case 3 (Fig. 9(c)) is on the left side and the beam of case 4 (Fig. 9(d)) is on the right side. Thus, they form another opposite group. The induced structures of the case 3 (Fig. 9(g)) have $\varLambda _{up}=66$ nm and $\varLambda _{down}=76$ nm, and the induced structures of the case 4 (Fig. 9(h)) showed $\varLambda _{up}=72$ nm and $\varLambda _{down}=62$ nm. This experimentally verified phenomenon reflects the fact that the change in symmetry of the focused spot influences the effect of scanning direction on the ripple formation. The effect is not strongly pronounced, however, it was discernable above the uncertainty level of $\pm 4$ nm. Thus. it becomes possible to manipulate the period of ripples by customizing the shaped beam and changing the scanning direction. This method should not only work for the reflective objective lens, but is also expected to be applicable for the conventional optical objective lens and using masks or spatial light modulators. The virtue of the reflective objective lens used in this study was the elimination of a dispersion-induced chirp.

4. Conclusions

A simple, low-cost method to continuously modulate the period of ripples at the nanoscale via non-reciprocal writing is experimentally demonstrated and numerically validated. Deep-subwavelength ripples on ITO films with continuously variable periods from 47 to 112 nm were fabricated (uncertainty of ripples’ period definition was $\pm 4$ nm). It was revealed that the formation of deep-subwavelength ripples originates from the localized optical near-field caused by the optical far-field intensity re-distribution in the thin film as captured through numerical simulations. In addition, it is demonstrated that shape-tailored focal spots can effectively control the non-reciprocal writing process using a paper-masked beam. In principle, the non-reciprocal writing can be achieved by combining conventional optical objective lens with spatial light modulation techniques to arbitrarily modulate the induced structure with up to $40\%$ period change. It is revealed that non-reciprocity originates in the asymmetry of the focal spot for thin film ablation and ripple texturing. The non-reciprocal writing allows flexible period control during ripple writing for applications in complex design micro-optics, biomimetic surfaces, wettability modulation, etc.