Greyscale 2D nanograting fabrication by multistep nanoimprint lithography and ion beam etching

The application of nanopatterned electrode materials is a promising method to improve the performance of thin-lm optoelectronic devices such as organic light-emitting diodes and organic photovoltaics. Light coupling to active layers can be enhanced by employing individual nanopatterns specically tailored to the device structure. During the development process typically a range of different nanopatterns need to be evaluated. Fabrication of each of these nanopatterns using electron-beam lithography is time and cost intensive, particularly for larger scale devices, due to the serial nature of electron beam writing. Here, we present a meth-od to generate nanopatterns of varying depth with different nanostructure designs from a single one-dimensional grating template structure with xed grating depth. We employ multiple subsequent steps of UV nanoimprint lithography and ion beam etching to fabricate greyscale two-dimensional nanopatterns. After each imprint step, the imprint resist is cured and etched to maintain the structural conformity. In this work we present variable greyscale nanopatterning of the widely used electrode material indium tin oxide. We demonstrate the fabrication of periodic pillar-like nanostructures with different period lengths and heights in the two grating directions. The patterned lms can be used either for immediate device fabrication or pattern reproduction by convention-al nanoimprint lithography. This parallel processing approach promises cost-ecient large-scale nanopattern variation for the device development process.


Introduction
The performance of thin-lm optoelectronic devices, such as organic and perovskite light-emitting diodes (OLEDs) and organic photovoltaics (OPV), has been increasing steadily over the last decades [1][2][3][4] . Their e ciencies however are still limited by losses inherent to the device architecture. In a default device stack more than 80% of the light cannot be utilized because of coupling to substrate modes, waveguided modes in the active layers or plasmonic modes at metal interfaces [5][6] . Various methods to access the light trapped in these modes such as the application of high-index substrate materials or the addition of one or multiple extraction lenses have been proposed [7][8] . Another promising approach is the introduction of a periodic or aperiodic nanopattern to the interfaces in the layer stack [9][10][11][12][13][14][15] . The nanostructure reduces total internal re ection and enables the outcoupling of quasi-guided modes [16][17] . Additionally, charge carrier injection may bene t from local eld enhancement effects at nanopatterned electrodes [18][19] . To achieve a signi cant increase in coupling e ciency, the geometrical dimensions of the employed nanopattern must be adjusted to the particular device design. Suitable parameters can be determined via simulation of the layer stack 20 . In an LED the maximum increase in outcoupling e ciency should match the emission maximum whereas the maximum increase in coupling to the active layer in a solar cell should coincide with the absorption peak. Aside from enhancing the overall e ciency, periodic nanopatterns also allow for directional coupling of selective wavelengths [21][22] and may facilitate nanoparticle alignment during device fabrication [23][24] . In order to comply with speci c requirements for different device architectures, a process to generate nanopatterned electrode lms for device development must enable the creation of different shapes and pattern designs with adjustable structural heights. This effectively requires a highly-controllable greyscale patterning method. Even though multiple greyscale electron beam lithography (EBL) techniques have been developed [25][26][27][28] , they are usually complex, time-consuming and not suitable for large-scale patterning. Nanoimprint lithography (NIL) on the other hand is a low-cost process allowing for high-throughput reproduction of multiscale patterns 29 .
However, the obtainable nanopatterns are essentially restricted to the geometrical shape and dimensions of the imprint template [30][31] . In this work we present a manufacturing method to introduce nanopatterns of variable shape and height into a transparent conductive indium tin oxide (ITO) lm on a glass substrate. As ITO is the most commonly used electrode material for thin lm optoelectronics, the resulting samples can be directly used for device fabrication. Additionally, they may also be utilized as imprint templates for simple reproduction of the generated pattern by NIL. We are able to create arbitrary twodimensional greyscale patterns by performing multiple successive nanoimprint steps employing the same or different template patterns in each step. In contrast to other works applying multiple nanoimprint steps to achieve multi-dimensional patterns 32 , we cure and etch the imprint resist after each imprint step.
In doing so, we are able to maintain structural conformity to the desired pattern and to obtain greyscale features.

Experimental
The different stages of the nanopatterning process are depicted in Fig. 1. One or more initial master templates patterned by electron beam lithography, laser interference lithography, or other techniques are required as a basis for the resulting nanopattern. We used commercially available glass templates holding periodic one-dimensional nanogratings with period lengths ranging from 400 to 600 nm and a grating height of 140 nm. From the master templates we reproduced negative imprint templates from polydimethylsiloxane (PDMS), which we applied for up to three successive imprints. Subsequently, we employed a standard UV nanoimprint process described previously to impress the positive pattern into an imprint resist layer 33 . Patterning was performed on ITO coated glass substrates (Lumtec, speci ed ITO layer thickness 140 nm ± 20 nm). We chose the UV-curable imprint resist Amonil MMS4 (Amo GmbH), which is an organic-inorganic composite containing zirconium alkoxides. After reproduction of the onedimensional pattern in the resist, we transferred the structure into the ITO layer by ion beam etching (IBE) using an IBE system PC3000 (Oxford Instruments). We employed a purely physical ion etching technique using only argon (Ar) plasma as the etchant in order to achieve a highly anisotropic etching. Choosing an RF power of 500 W and a beam current of 350 mA at a gas pressure of 8 torr allowed for a precisely controllable etching process while maintaining reasonable processing time. Following the pattern transfer we removed any residues of the imprint resist with tetramethylammonium hydroxide using a commercial photoresist stripper (TechniStrip P1316, MicroChemicals). By repeating the imprint and etching process rotating the imprint template by an angle for each imprint step we obtained two-dimensional greyscale patterns in the ITO.

Results And Discussion
The maximum structural depth for the rst etching step is determined by the height of the imprint template and the etching rate ratio between the imprint resist and the ITO layer. To perform an etching rate analysis at the chosen etching parameters, we prepared several planar samples with and without the imprint resist. We etched partially masked samples for different times before measuring the resulting height difference with a pro lometer. The measurements shown in Fig. 2 indicate constant etching rates over the entire process duration corresponding to an etching rate ratio of 1.2. Despite the low selectivity of the etching process (which is expected for a purely physical etching), it is suitable to transfer the grating pattern into the underlying ITO layer. However, as the ITO etching rate is lower than the etching rate of the imprint resist the maximum structural depth possible is only 116 nm (83% of the height of the imprint template).
The actual etching time for each sample was chosen according to the desired etching depth assuming the etching rates presented before. In order to account for potential thickness variations of the imprint resist during deposition, the etching process was additionally tracked in-situ by time-of-ight secondary ion mass spectrometry (TOF-SIMS), as shown in Fig. 3 By repeating the imprint and etching process rotating the imprint template by an angle for each imprint step, we obtain two-dimensional greyscale patterns in the ITO. In order to determine the etching time for subsequent patterning steps, the existing nanograting has to be taken into account. During coating, the imprint resist lls the voids of the underlying pattern while partly preserving the structure through the layer. However, application of the imprint template forces a new pattern onto the surface, causing differences in resist thickness, which result from both the existing pattern underneath the resist and the newly imprinted grating. As the etching rate of the imprint resist is higher than the etching rate of the ITO, this effect nally leads to attening of the existing pattern during successive etching steps. While the attening may be utilized to easily obtain greyscale features, one can also diminish it by choosing the etching time accordingly. The nal shape of the resulting nanopattern is determined by the number of the performed imprint and etching steps as well as the angles in which the imprint templates are applied. One can simply obtain square pillar patterns by using the same imprint template holding a 1D grating structure and rotating it by 90° for the second imprint step. Other designs (such as rectangular or rhomboid shaped pillars) can be fabricated by using multiple 1D templates with different period lengths or applying the PDMS templates at different angles. Examples for various different greyscale pattern designs fabricated with our method are presented in Fig. 4. The generation of more complex patterns such as isosceles triangles requires more than two imprint steps and therefore exact alignment of the imprint templates. For this work, all alignment was done manually, hence preventing the fabrication of those patterns. However, as alignment precision below 100 nm has been shown to be feasible for NIL 34 , we expect our method to be suitable for the generation of arbitrary periodic and aperiodic patterns from appropriate master templates.
Since the generation of complex multi-dimensional patterns from one-dimensional master templates requires multiple imprint and etching steps, exact reproducibility between different samples may be di cult to achieve. We therefore envision the main application of our method to be the fabrication of new master template patterns. Subsequently, one may utilize these master templates for pattern duplication via highly reproducible standard nanoimprint lithography techniques. Generating new variable master templates by nanoimprint lithography and ion beam etching using only one or few existing templates is especially bene cial in cases where greyscale EBL may not be available or not feasible due to size constraints. In order to show the applicability of our method to this work ow, we covered the etched ITO gratings with an anti-sticking layer (BGL-GZ-83) and used them as a mold for PDMS imprint templates.
With these PDMS stamps we performed a standard UV-NIL process yielding a nanopatterned layer of imprint resist on a glass substrate, which was coated with a thin silver layer for SEM imaging. A comparison of the employed ITO templates and the resulting imprinted nanopatterns is presented in Fig. 5. Both the one-dimensional grating pattern (a) and the two-dimensional pillar pattern (c) show very good conformity to the period length and grating depth of the templates (b) and (d) con rming that the etched ITO samples can indeed be used as master templates for pattern replication. The imprinted samples exhibit higher surface roughness and less distinct pattern features than the templates. We attribute this mainly to the silver coating on top of the resist and not to be a result of the imprint process itself. This is due to the fact that silver may form rough lms especially when deposited at the low rates necessary for the fabrication of very thin layers 35 , whereas we did not observe roughening of the pattern during any other imprints using the same template material and imprint resist.

Conclusion
In conclusion, we have demonstrated a method to generate variable two-dimensional greyscale nanopatterns in indium tin oxide using only a one-dimensional nanoimprint master template. The patterning process consists of multiple successive steps of nanoimprint lithography and ion beam etching. Greyscale features are obtainable because we cure and etch the imprint resist after each imprint step allowing individual etch durations for each etching step. Additionally, we employed secondary ion mass spectrometry to track the etching progress and precisely control the etching depth yielding two-dimensional patterns with variable feature height. The patterns generated by this method can subsequently be used as master templates for reproduction. We demonstrated this work ow by reproducing multiple of our etched patterns using a standard UV nanoimprint lithography technique. Finally, we believe our method to be bene cial for device fabrication and prototyping especially in lab environments where techniques such as greyscale EBL are either not available or not desirable due to Overview over the nanopatterning process work ow. Multiple steps of nanoimprint lithography and ion beam etching lead to a multi-dimensional nanopattern with variable feature height.    Scanning electron microscopy (SEM) images of greyscale two-dimensional nanopatterns fabricated with the method described above: (a) rectangular pillar pattern with period lengths 400 nm and 600 nm, (b) rhomboid pillar pattern with period length 500 nm at a 40° angle, (c) square pillar pattern with period length 600 nm. (d, e) Atomic force microscopy (AFM) measurement of the pattern depicted in (c). The grating depth along grating directions I and II is approximately 90 nm and 40 nm respectively. Figure 5