Introduction

Interest in conjugated polymers (CPs) increased significantly in the last decades owing to their exceptional optical and semiconducting properties, giving rise to the emergent fields of plastic electronics1 and photonics.2 These materials combine the benefits of cost-effective processability, compatible with large-area deposition, with the distinctive mechanical features of plastics. The latter opens up ways for designing/fabricating laser geometries of virtually any shape, not possible with rigid inorganic semiconductor materials. In pursuit of this perspective, nanoimprint lithography (NIL)3,4 has been used in the past to replicate different periodical surface corrugations, on thermoplastic substrates5 or on the active conjugated layers themselves.6 Standing out from other patterning techniques, NIL faithfully reproduces geometric surface patterns with sub-5 nm feature size resolution7 and with high throughput that can be scaled up via roll-to-roll8,9 for low-cost mass production in a large variety of applications.10

The photoluminescence spectra of CPs can be chemically tuned, enabling optically pumped CP lasers with emission across the visible and near infrared spectrum.11,12,13 These highly luminescent materials have been incorporated into a variety of resonator geometries like microrings,14 photonic crystals,15 distributed Bragg reflectors (DBR),16 and distributed feedback (DFB) lasers.17 DFB lasers make use of periodic wavelength-scale nanostructures to backscatter photons within the plane of propagation. Constructive interference of the backscattered photons occurs when the grating periodicity (Λ) fulfills the Bragg law:

$$m\lambda = 2n_{\mathrm{eff}}\Lambda,$$
(1)

where m is the diffraction order, neff is the effective refractive index of the waveguiding medium, and λ is the emission wavelength. All DFB lasers presented in this work are optimized to operate with the second order of diffraction (m = 2), with the first order of diffraction providing outcoupling of laser emission perpendicular to the plane of propagation. A wide range of applications are known for DFBs, for example in lab-on-a-chip devices for biomedical analysis,18 information technology,19 and sensing.20 Owing to their highly luminescent and flexible properties, CPs have been applied in the past for the development of mechanically flexible DFB lasers obtained with different fabrication methods.21,22,23,24,25,26 In this work we report on DFBs based on a very simple low-cost and easily scalable method resulting in samples of high optical quality and large transparency. DFB structures were fabricated by nanoimprinting gratings of short (SP, Λ = 287 nm), medium, (MP, Λ = 350 nm), and long period (LP, Λ = 416 nm) on cellulose diacetate (CdA) films and subsequent coating with PFO, F8BT, and a blend of F8BT and P3HT, respectively. For this purpose, silicon master molds were developed by electron beam lithography and subsequent reactive ion etching. The imprinted CdA substrates faithfully reproduce the ones of the master molds, demonstrating the great accuracy of the fabrication route followed. To the best of our knowledge, CdA has not been reported as grating substrate in DFB structures, although it has been used in DBRs27,28 and as a host for laser dyes.29 It exhibits several advantageous features: it is cheap, readily available, flexible, and transparent even upon annealing, since it does not crystallize like, for instance, polypropylene. Moreover, CdA presents other advantages, like being compatible with several organic solvents, it is obtained from renewable wood pulp sources and it is biodegradable. The relatively low refractive index of CdA (n ≈ 1.475) leads to a 0.2–0.3 index mismatch with most CPs, sufficient to provide substantial vertical light confinement.

Results and discussion

In this work we have chosen PFO, F8BT, and a blend of P3HT and F8BT, which are well known CP systems exhibiting optical gain in the blue,30 green–yellow31,32 and red33 spectral regions, respectively. Superimposed photoluminescence and absorption spectra of the three CP media are illustrated in Fig. 1a. Unpatterned CdA exhibits large transmission (>80%) in the visible. The transmission drop below 400 nm and the dip at 280 nm (dashed line in Fig. 1b) is assigned to the nπ* transition of the acetate keto group as previously described.34

Fig. 1
figure 1

a Normalized absorption and photoluminescence spectra of PFO (blue solid line and dot line), F8BT (green solid line and dot line), and P3HT:F8BT (red solid line and dot line). b Transmission spectra at normal incidence of LP (red solid line), MP (green solid line), and SP (blue solid line) gratings imprinted on CdA foils and of the unpatterned CdA substrate (pink dashed line). The arrows point at the transmittance dips, corresponding to the wavelengths of 278, 350, and 416 nm, respectively. c PFO (right), F8BT (middle), and P3HT:F8BT (left) DFB structures illuminated by a UV lamp. d Diffraction pattern of a naked CdA grating when exposed to white light

The transmission spectra of SP, MP, and LP patterned CdA display additional losses at wavelengths located below 420, 531, and 605 nm, respectively, caused by diffraction. These limit wavelengths stand for the grating cut-off values, i.e., the appearance of first order diffraction under 90° on the CdA side of the nanoimprinted grating (\(\frac{\lambda }{n} \approx \Lambda\)). Likewise, dips are observed at wavelengths marked by arrows, (278, 350, and 417 nm, respectively), likely due to first order diffraction under 90° at the air side of the grating (λ ≈ Λ). CP films were spin coated on top of these nanoimprinted CdA substrates. In Fig. 1c, the homogeneity and luminescent properties of these films under UV illumination is illustrated. Such good homogeneity is a consequence of the high surface energy of the CdA substrates,35 which are readily wetted by the CP solution in toluene. Figure 1d depicts a diffraction pattern obtained upon illuminating an uncoated CdA grating with white light.

The fidelity of the pattern transfer after the imprint process was assessed by comparing the atomic force microscopy (AFM) topography images of corrugated silicon and CdA substrates (Fig. 2). The grating depth profile for each mold is also shown. The fact that there are no major differences between the silicon master mold and the CdA grating indicates the high fidelity of the soft patterning. A cross-sectional profile across ten grooves in each surface provides local periods of 275 (283), 351 (354), and 403 (403) nm for the SP, MP, and LP Si (CdA) corrugated surfaces, respectively, confirming less than 3% of variation in periodicity in all cases. The surface profiles extracted from each AFM image indicate similar modulation depths of the grating structures in Si and CdA of about 120 nm for the SP, 140 nm for the MP, and 250 nm for the LP gratings, respectively.

Fig. 2
figure 2

AFM topography images and profiles along the lines indicated in the images of grating structures fabricated in Si and in CdA films, with SP (a, b), MP (c, d), and LP (e, f)

Figure 3 illustrates front views of CdA SP, MP, and LP gratings (Fig. 3a–c, respectively) and a side view of a CdA SP grating (Fig. 3d) obtained by scanning electron microscopy (SEM).

Fig. 3
figure 3

SEM images of gratings structures fabricated in CdA films, with a SP, b MP, and c LP. d Cross section of an LP CdA grating

Note that the height of the modulation in Fig. 3d (247 nm) is compatible with the height profile measured by AFM. The quality of the nanoimprinted gratings on CdA was further evaluated employing grazing incidence small angle X-ray scattering (GISAXS), a technique that has been employed previously for the characterization of gratings prepared by different methods.36,37 Unlike AFM, the 2D GISAXS patterns provide statistical information averaged over a larger surface area covered by the footprint of the X-ray beam. In our experiments, the very low incident angle (0.3–0.4°) and the cross-section area of the X-ray beam (1.3 mm × 0.3 mm) implied a probing area of ~70 mm2. The top row of the figure shows the corresponding X-ray diagrams obtained from the Si gratings. The patterns show an array of rods that correspond to different orders of scattering located at horizontal scattering angles ω, which are reciprocally related to the pitch of the periodic nanostructures. These rods intersect the Ewald sphere on a series of dots distributed along a semicircle. Such GISAXS patterns are characteristic for scattering from gratings at very low incidence angles.38,39 The semicircular pattern characteristic of Si gratings vanishes in the CdA replica, most likely due to its reduced planarity, being instead composed of vertical stripes. Figure 4b shows the 1D profiles obtained after integration of the vertical scattered intensities along the whole ω range. For the sake of clarity, only one half of the curve is presented. The curves were shifted to correct the pixel position according to the origin (ω = 0), estimated by the symmetry of the reflections on both sides of the beam stop. Despite of being qualitatively different, the profile obtained from the CdA grating patterns reproduce rather well the profile of the corresponding Si molds, confirming that the periodicity of the structure is faithfully transferred to the polymer substrate over a large area during the imprinting process.

Fig. 4
figure 4

a 2D GISAXS patterns corresponding to the Si gratings (top row) and to the replica prepared by NIL in CdA foil (bottom row). b 1D integrated X-ray intensity for the Si molds (black) and CdA replica (red) obtained by integrating the intensity along the vertical direction of the GISAXS. In order to facilitate the comparison between the periods of the Si mold and the CdA copy only one half of the curve is presented

The periods of each grating were calculated as the mean of those obtained from different reflection orders, according to the equations described in the methods section. Table 1 provides a comparison of the periods obtained for the Si molds and the CdA replica, respectively. The obtained values are in good agreement with those observed by AFM and SEM.

Table 1 Grating periods determined from GISAXS diffraction patterns of Si molds and the corresponding CdA replica

Figure 5 shows the emission spectra of (a) PFO, (b) F8BT, and (c) P3HT:F8BT films deposited on top of corrugated CdA substrates at excitation densities below and above the lasing thresholds, as well as the corresponding input–output curves of emission intensity vs. pump fluence.

Fig. 5
figure 5

Left column: Normalized intensity of a blue, b green–yellow, c red, emission from the CP DFB lasers below the threshold (dashed lines) and just above the threshold (solid lines). Right column: Variation of the output laser power as a function of pump fluence in d blue, e green–yellow, and f red emitting DFB structures. The red, green, and blue solid lines represent linear fits above the thresholds

The dotted lines represent the normalized emission spectra at pump fluence just below the respective laser thresholds. The blue, green, and red solid lines, on the other hand, represent lasing spectra for pump fluences just above the thresholds, detected in the direction perpendicular to the surface. These spectra are characterized by narrow peaks at 450, 576, and 676 nm, respectively, with linewidth values of 0.2, 0.4, and 0.3 nm. In all cases, the laser signals are emerging from dips surrounded by two fluorescence peaks. The dip is explained by the presence of a photonic stopband, caused by the distributed feedback (DFB) of the propagating modes in the corrugated films. The lasing thresholds of the plastic DFB lasers, (6.4 μJcm−2 for blue, 54.0 μJcm−2 for green–yellow, and 6.6 μJcm−2 for red emission, respectively), are extracted from the drastic variation in the slopes of the output emission intensities (see Fig. 5d–f). The thresholds obtained are comparable to those previously reported on similar blue,17,40 green–yellow,41,42 and red43,44 DFBs on nanostructured rigid substrates as well as on nanostructured photoresists.45,46 The slopes of the input–output curves for F8BT and F8BT:P3HT are 5.5 and 50 times lower, respectively, than the corresponding slope for PFO, which suggests superior optical gain properties in PFO. This is consistent with the larger gain coefficient obtained from ASE measurements in PFO (74 cm−1) with respect to F8BT (22 cm−1) using the variable slit method.47 Indeed, tuning of the emission of polyfluorenes to longer wavelengths via backbone functionalization results in a progressive increase of spectral overlap between stimulated emission and excited-state absorption absorption, which is detrimental for optical gain.48 Regarding F8BT:P3HT blends the lower slope efficiency could be explained as due to the lesser emissive properties of the blend respect to PFO and F8BT (24, 50, and 58% photoluminescence quantum efficiency values of F8BT:P3HT, PFO, and F8BT, respectively).47,49 Tunability of the emission from the DFB structures was achieved through control of the active layer thickness by varying the spin coating speed or the CP concentration in precursor solutions. The shift in DFB emission with thickness can be explained as due to the change in the effective refractive index of the planar waveguide, which increases toward the refractive index of the active medium when the film thickness increases. Changing the PFO and P3HT:F8BT film thicknesses from 142 to 232 nm and from 120 to 260 nm led consequently to a displacement of the emission peaks to longer wavelengths by up to 12 and 34 nm, respectively (Fig. 6).

Fig. 6
figure 6

a Polarized emission output of a DFB based on F8BT. The pump polarization was fixed along the DFB lines, whereas the output intensity was filtered with a polarizer at different angles (α). b Normalized output spectra of PFO and P3HT:F8BT DFB structures above threshold. PFO concentrations: 10 mgml−1 (blue solid line line), 20 mgml−1 (blue dash line), 30 mgml−1 (blue short dot line), and 40 mgml−1 (short dash dot line). P3HT:F8BT concentrations: 10 mgml−1 (red solid line line), 20 mgml−1 (red dash line), and 40 mgml−1 (red short dash dot line). The corresponding PFO film thicknesses were 142, 178, 206, and 232 nm, as measured by AFM. The corresponding P3HT:F8BT film thicknesses were 120, 180, and 260 nm

Finally, we investigated the laser emission properties upon bending the substrate at three different angles: 41°, 61°, and 113° (Fig. 7). Increasing the bending angle leads to a progressive emission blue shift, (Fig. 7a). We tentatively assign this effect to an effective decrease in thickness of the waveguiding medium leading to the aforementioned shift of the lasing mode. Importantly, the lasing threshold values of bent DFBs did not differ substantially from that of an undeformed DFB (Fig. 7b), confirming that the optical and structural properties of the active layer do not deteriorate.

Fig. 7
figure 7

a Laser emission on DFBs based on F8BT upon 113° bending (red), 61° (blue), 41° (brown), and 0° (orange). The inset sketch illustrates a DFB under a β bending angle. b Integrated emission intensity as function of pump fluence for different bending angles: 113° (squares), 61° (circles), 41° (up-triangles), and 0° (down-triangles). Lasing thresholds as a function of bending angle are shown in the inset

In summary, we report a straightforward fabrication method of flexible DFB lasers that involves NIL on inexpensive transparent CdA substrates. By spin coating different CPs on periodically patterned CdA we have achieved optically induced vertical single-mode laser emission in the blue, green–yellow, and red wavelength range. The described procedure enables the mass fabrication of low-cost and large area diffractive elements on plastic for lighting, sensing, and lab-on-a-chip applications. The fact that these DFBs are deployed on thin and transparent substrates provides the option to pump them simultaneously by stacking them on top of each other, giving rise in this way to multicolored laser emission.

Methods

Fabrication of the 350 nm Si mold

Silicon master molds with periods of 278 and 416 nm were acquired from Lightsmyth Technologies Inc. The silicon master mold with a period of 350 nm was fabricated by electron beam lithography and reactive ion etching techniques. First of all the pattern was transferred to a PMMA layer of about 180 nm in thickness, directly spun onto the silicon substrate. The resist was exposed by using an Auriga SEM (Carl Zeiss) controlled by an ELPHY MultiBeam lithography system (Raith GmbH). After developing the sample in MIBK:IPA (1:3) and rinsing it in IPA, as usual, the sample went through a soft oxygen 50 W plasma process during 15 s to remove any possible residues of resist in the developed areas. The reactive ion etching was performed with an Oxford Instruments Plasmalab System 100. As the required depth was about 150 nm and the selectivity of silicon to PMMA is as high as 5:1 at about 20 °C, the etching process was performed at ambient temperature. A short run of 17 s, with 10 sccm of SF6 for etching and 63 sccm of CHF3 for sidewall passivation, 20 mTorr, 20 W of forward power, and 220 W for the ICP were enough to obtain the desired depth.

Nanoimprint of gratings on CdA substrates

The Si DFB master molds were initially treated with a fluorosilane through vapor deposition of 1H,1H,2H,2H-perfluorodecyl trichlorosilane (Alfa Aesar, MA) as release agent to reduce the surface energy, easing the demolding procedure. Next, the master mold was imprinted directly onto a CdA film by means of a Eitre 3 Nano Imprint Lithography System by Obducat. As the CdA film is flexible, it perfectly adjusts to surface irregularities, therefore leading to very homogeneous imprint nanopatterns. Throughout the nanoimprinting route, substrates were heated at 180 °C with an applied pressure of 40 Bar for 300 s. At the imprinting temperature the polymer softens and fills the mold cavities. The substrate was then cooled down to a temperature of 70 °C before the pressure was released and the imprinted substrate was removed. Three grating structure varieties were used: (i) Λ = 416 nm period, fill-factor 50%, and 250 nm etch depth, (ii) Λ = 350 nm, fill-factor 50%, and 150 nm etch depth, and (iii) Λ = 278 nm, fill-factor 50%, and 110 nm etch depth.

Fabrication of all-polymer DFB

As starting materials, a flexible CdA film (refractive index: n = 1.470 at 676 nm, n = 1.475 at 576 nm, and n = 1.482 at 450 nm)50 of 300 µm thickness with flat surfaces provided by Clarifoil (Celanese, UK) was selected as a substrate. Highly luminescent conjugated polymers poly(9,9-dioctylfluorene-alt-benzothiadiazole) (F8BT) (American Dye Source, refractive index51 n = 1.76 at 576 nm, Mn = 33 kgmol−1, Mw = 217 kgmol−1); poly (9,9-dioctylfluorene) (PFO, American Dye Source, refractive index n = 1.81 at 450 nm,51 Mn = 40–150 kgmol−1), and a blend of poly(3-hexylthiophene) (P3HT) (BASF Sepiolid P200, >98% regioregularity, Mn = 10 kgmol−1) and poly(9,9-dioctylfluorene-alt-benzothiadiazole) (P3HT:F8BT) (refractive index43: n = 1.80 at 676 nm), were dissolved in toluene at a concentration of 20 mgml−1 and the resulting solutions were filtered with a 0.22 µm syringe filter. P3HT:F8BT solution mixtures were prepared upon mixing master toluene solutions of 20 mgml−1 regioregular P3HT with F8BT in a weight concentration of 20 mgml−1. All CP films were spin coated at 2000 rpm for 1 min onto periodically corrugated CdA substrates, resulting in average thicknesses of 180 nm for all cases.

Optical and structural characterization

A narrow-band OPO (Continuum Sunlite, bandwidth 0.075 cm−1, pulse width ≈3 ns) pumped by a seeded frequency-tripled Nd:YAG laser (Continuum Powerlite PL8000) and equipped with a frequency doubler (Continuum FX-1) was used for excitation. The wavelength range accessible with this OPO is 220–1700 nm. The pump beam was strongly attenuated for the experiments. The lasing emission arising perpendicularly from the surface of the DFBs was detected with a spectrometer (SP2500, Acton Research) equipped with a liquid nitrogen cooled back-illuminated deep depletion CCD (Spec-10:400BR, Princeton Instruments). For lasing measurements, the oblique incidence pump beam was passed through a calibrated diaphragm and focused using a 30 mm focal length spherical lens, with the sample placed out of focus. The emission was collected in free-space normal to the DFB surface. The pumping intensity was regulated with neutral density filters. UV–Vis characterization was performed using a Varian Cary 50 UV–Vis spectrophotometer. AFM images were obtained using a JPK NanoWizard II operating in dynamic mode and air conditions, with MikroMasch HQ:XSC11/Al BS cantilevers of 2.7 nm−1 force constant and 80 kHz resonance frequency. SEM characterization was carried out by means of a Carl Zeiss Auriga microscope operating at 3 kV.

GISAXS measurements

GISAXS measurements were performed at the BM26 DUBBLE Dutch–Belgian beamline of the European Synchrotron Radiation Facility.52 A sample to detector distance of 7322 mm was estimated using a rat tail sample as standard of calibration. The energy of the X-ray beam was fixed at 12 KeV (1.03 Å). The 2D scattering patterns were recorded by using a Pilatus1M detector with 172 µm pixel size. The slits alignment allowed for a beam size (horizontal × vertical) of 1.3 × 0.3 mm. A nominal incidence angle of αi = 0.35° was selected, allowing to average the recorded information over an area of ca. 70 mm2. In order to verify the periodicity of the gratings and the quality of the polymer replica, the samples were carefully aligned, with the lines parallel to the X-ray beam, since the appearance of the diffraction pattern is very sensitive to any change in this factor. GISAXS images were analyzed by using the FIT2D software. Basically, in a GISAXS pattern there exist two main scattering angles used to obtain information about structural correlations in the vertical and horizontal directions, α and ω, respectively. The information can be interpreted on the basis of three orthogonal scattering vectors, qx, qy, and qz, being qy = (2πλ−1) sin ω·cos α, the one which probes the correlations existing in the horizontal plane,53 that is, the grating period under study. Once the scattering vector qy has been calculated, the period, Λ, of the grating can be extracted from the relation Λ = 2π(qy)−1.