Large-area Scanning Probe Nanolithography Facilitated by Automated Alignment and Its Application to Substrate Fabrication for Cell Culture Studies

Scanning probe microscopy has enabled the creation of a variety of methods for the constructive ('additive') top-down fabrication of nanometer-scale features. Historically, a major drawback of scanning probe lithography has been the intrinsically low throughput of single probe systems. This has been tackled by the use of arrays of multiple probes to enable increased nanolithography throughput. In order to implement such parallelized nanolithography, the accurate alignment of probe arrays with the substrate surface is vital, so that all probes make contact with the surface simultaneously when lithographic patterning begins. This protocol describes the utilization of polymer pen lithography to produce nanometer-scale features over centimeter-sized areas, facilitated by the use of an algorithm for the rapid, accurate, and automated alignment of probe arrays. Here, nanolithography of thiols on gold substrates demonstrates the generation of features with high uniformity. These patterns are then functionalized with fibronectin for use in the context of surface-directed cell morphology studies.


Introduction
Progress in nanotechnology is dependent on the development of techniques capable of efficiently and reliably fabricating nanoscale features on surfaces. 1,2 However, generating such features over large areas (multiple cm 2 ) reliably and at relatively low cost is a non-trivial endeavor. Most existing techniques, derived from the semiconductor industry, rely on ablative photolithography to fabricate 'hard' materials. More recently, lithographic techniques derived from scanning probe microscopy (SPM) have emerged as a convenient and versatile approach for the rapid prototyping of nanoscale designs. 3 SPM-based techniques are able to conveniently and rapidly 'write' any user-defined pattern. The most wellknown of these is dip-pen nanolithography (DPN), pioneered by Mirkin et al., 4 where a scanning probe is used as a 'pen' to transfer a molecular 'ink' to the surface producing features in a fashion analogous to writing. Under ambient conditions, as a probe is scanned across a surface the 'ink' molecules are transferred to the surface via a water meniscus that forms between the probe and the surface (Figure 1). DPN thus allows the nanolithographic deposition of a wide range of materials, including 'soft' materials such as polymers and biomolecules. 5 Related techniques using probes engineered with channels for fluid delivery, variously referred to as 'nanopipettes' and 'nano-fountain pens', have also been reported. 6,7,8 The main obstacle to the wider application of SPM-derived lithography is throughput, as it requires an excessively long time to pattern centimeter-scale areas with a single probe. Early efforts to address this issue focused on the parallelization of cantilever-based DPN, with both 'one-dimensional' and 'two-dimensional' (2D) probe arrays being reported for the lithography of centimeter-sized areas. 5,9 However, these cantilever arrays are produced through relatively complex multistep fabrication methods and are relatively fragile. The invention of polymer pen lithography (PPL) addressed this issue by replacing the standard SPM cantilevers with a 2D array of soft siloxane elastomer probes bonded to a glass slide. 10 This simple probe setup significantly decreases the cost and complexity of patterning large areas, opening up nanolithography side of the probes came into contact with the surface first, then adjusting the angle and repeating the procedure in an iterative manner until the difference in contact on each side of the probe was indistinguishable to the eye. As this alignment procedure relies on subjective visual inspection by the operator, reproducibility is low.
Subsequently, a more objective approach has been developed, consisting of a force sensor mounted beneath the substrate to measure the force applied upon contact of the probes on the surface. 12 Alignment was thus achieved by adjusting the tilt angles to maximize the force exerted, which indicated that all the probes were simultaneously in contact. This method showed that alignment to within 0.004° of the surface parallel was possible. This 'force feedback levelling' has now been implemented into fully automated systems in two independent reports. 13,14 Both use a triad of force sensors mounted either beneath the substrate or above the array and measure the amount of force exerted upon contact between the probe arrays and surface. These systems give high precision, reporting misalignments of ≤0.001° over a 1 cm length scale, 14 or ≤ 0.0003°over 1.4 cm. 13 These automated alignment systems also provide major savings in operator time and overall time taken to complete the lithography process.
One major application of high-throughput surface fabrication enabled by this technology is the generation of cell culture substrates. It is now well established that cell phenotype can be manipulated by controlling the initial interaction between cells and surface features, and that this can be enhanced at the nanoscale. 15 Specifically, scanning probe lithography methods have been shown to be a facile method to produce a variety of nanofabricated surfaces for such cell culture experiments. 16 For example, surfaces presenting nanoscale patterns of self-assembled monolayers and extracellular matrix proteins templated by PPL and DPN have been used to study the potential of nano-modified materials in material induced differentiation of stem cells. 17 This protocol describes the utilization of a modified atomic force microscope (AFM) system that enables large-area PPL. We detail the detection of force using multiple force sensors as the means of determining probe-surface contact, together with an algorithm that automates the iterative alignment process. Subsequent functionalization of these patterns with the extracellular matrix protein fibronectin and the culture of human mesenchymal stem cells (hMSC) are described, as a demonstration of PPL-fabricated surfaces applied for cell culture.
2. Place a 13 x 13 mm glass slide in a plastic screw-topped vial filled with 20 mL 2-propanol, then place the vial in an ultrasonic bath for 10 min to remove any large debris. Wash the slides by submerging the slides in fresh 2-propanol (100 mL) and dry under a stream of nitrogen gas. 3. Place a silicon master 18 into a 4 cm diameter petri dish and add sufficient degassed PDMS prepolymer mixture (from step 1.1) until fully covered. Typically, 100 µL is required for a 20 x 20 mm master. Place the master with the prepolymer mixture in a vacuum dessicator. Degas the polymer for a further 5 min to remove any gas bubbles formed during the transfer of the mixture. The O 2 plasma treatment of the glass slides (step 1.4 below) should be performed while the degassing is taking place. 4. Treat the glass squares with O 2 plasma (600 mTorr) at maximum RF power for 1 min to remove any organic contamination and to generate a uniform oxide layer on the glass for adhesion of the elastomer. 19 Use the plasma treated slides immediately in the next step.
5. Carefully place the square glass slide (from step 1.4) over the prepolymer on the master (from step 1.3) with the plasma-cleaned side facing down. Gently press down the glass slide onto the silicon master to remove any trapped air and to ensure a uniform film of PDMS is sandwiched between the master and the slide. 6. Place the sandwiched PDMS array from above step in a petri dish with the silicon master at the bottom (i.e., with the back of the glass slide facing upwards) and place the dish in an oven at 70−80 °C for 24−48 h to thermally cure the PDMS. 7. Remove the cured array from the oven and allow to cool for 15 min, then with a razor blade carefully remove any excess PDMS from the back and sides of the glass slide and use a stream of dry nitrogen to blow away any loose PDMS debris. Note: Take care not to scratch the silicon master with the razor blade, as this may damage the non-stick coating. 8. Wedge a razor blade into the corner of the array at a depth of 1 mm and carefully pry the array apart from the master. Perform this action in a single continuous lifting action; do not allow the arrays to fall back onto the master. 9. Carefully cut and scrape away 0.5 mm of the PDMS at the edges of the array with a razor blade. Use a stream of dry nitrogen to blow away any loose PDMS debris. Note: It may be easier to perform this trimming step under stereoscope or a magnifying glass. Take care not to scratch the silicon master with the razor blade, as this may damage the non-stick coating.

Array preparation and substrate mounting
1. Generate a hydrophilic surface on the probe array by O 2 plasma treatment: 1. Place the PPL pen array in a petri dish into plasma chamber then apply vacuum to 600 mTorr. Switch on the plasma generator (maximum setting) for 30 s. 2. Release the vacuum, remove the array and check its hydrophilicity by dropping 20 µL of deionized water onto the array and observing whether there is even spreading of the water across the surface. If this does not occur, subject the array to a second round of plasma treatment. Afterwards, dry the array thoroughly with a stream of dry nitrogen gas.
2. Using double-sided carbon tape, attach the array onto the middle of the probe holder. Mount the probe holder onto the AFM kinematic holder (Figure 2). 3. To load the PPL array with 16-mercaptohexadecanoic acid (MHA) ('inking'): 1. Prepare 1 mM 16-mercaptohexadecanoic acid (MHA) solution by dissolving 8.6 mg in 30 mL ethanol in a tube and placing it in an ultrasonic bath for 10 min to fully dissolve the compound. CAUTION: 16-mercaptohexadecanoic acid is toxic. Please read MSDS before working with this solution. Safety equipment must be worn while handling the chemical. 2. Using a micropipette, deposit 20 µL drop of the MHA solution on the array. Avoid contact of the pipette tips with the arrays. Allow it to spread throughout the array, then allow the ethanol to evaporate under ambient conditions. NOTE: The PPL array can alternatively be inked after the alignment has taken place. 10 4. Once the MHA solution has dried, mount the probe holder with the PPL array onto the AFM.

Preparation of gold substrates for PPL.
1. Gold substrates can either be purchased, or made in-house by thermal or electron beam deposition, and are constructed of a 2 nm titanium adhesion layer followed by 20 nm of gold on a glass or silicon wafer. 18 2. Where necessary, clean the substrates by oxygen plasma treatment using the parameters described in step 1.4. 3. Place the gold substrate in the middle of the AFM sample stage and secure with adhesive tape around the borders of the substrate ( Figure  2). Adjust the stage to the correct height as indicated in the manufacturer's operating instructions using the z-axis controller.

Automatic alignment of pen array
1. Open and run the stage controller setup program (SetupNSF.exe) on the computer to reset ('zero') all axes and angles to a pre-calibrated zero point, then use the stage x/y-axis controller console to move the substrate to the desired alignment/printing location. For optimal results, the substrate should be placed near the center of the stage, between the stage's force sensors. NOTE: In some models of computer, the x/y-axis controller USB signal may interfere with that from the z-axis controller. If this occurs, disconnect the x/y-axis controller USB cable after this step. It should then be reconnected after the alignment procedure (step 4.7). 2. Switch the stage release lever to release the sample stage and activate the triad of force sensors as indicated by the AFM manufacturer's instructions. Allow the force sensors to equilibrate for at least 15 min. For optimal results, allow 30-50 min. 3. Increase the z-axis height to bring the array into close proximity with the substrate by visually observing the probe array and surface.
NOTE: The closer the array is to the surface, the fewer iterations are required for the alignment process, thus saving time. 4. Open/run the Automatic Alignment program (Auto Alignment v16.exe) and enter relevant alignment parameters into the program.
1. Enter the desired 'Angle Step' parameter value, typically 0.15°. This parameter is the offset angle from the 'optimum' angle for each axis that is determined by the program. Set this parameter between 0.1 and 0.2°, as angles lower than this range do not result in a clearly detectable force difference upon approach of the probes to the surface. NOTE: Software accepts values in millidegrees (i.e., 1 x 10 -3 °). For example, for 0.15°, users should input '150.'

Entering the desired Coarse
Step' parameter value, typically 0.6 µm. This parameter is the z-axis step size used by the stage as it approaches the probes in the initial rough alignment. Set this parameter between 0.2 and 1 µm. Larger step sizes decrease the time taken for the alignment process but reduce the accuracy of the alignment, and increase the wear on the probes. NOTE: Software accepts values of coarse steps in micrometers. For example, for 0.6 µm users should input '0.6'.

Enter the desired 'Fine
Step' parameter value, typically 0.2 µm. This parameter is the z-axis step size used for fine adjustment of the optimum alignment. For most applications, set this parameter between 0.1 and 0.4 µm. Larger value step sizes will decrease the amount of time taken for the alignment process but reduce the quality of the alignment. NOTE: Software accepts values of fine steps in micrometers. For example, for 0.2 µm, users should input '0.2.' 4. Configure the 'Excel file path' and attach an unmodified copy of the provided spreadsheet template file by using the 'folder' icon, navigating to the file location, and pressing 'OK'. This file contains the raw and calculated data that is used to determine the optimum stage tilt angles of the stage.
5. Open/run the AFM control software. Navigate to the spectroscopy component of this program by clicking the 'spectroscopy' button (according to the manufacturer's instructions), and set the AFM scan head z-axis to oscillate by 10 µm over 100 ms, with a pause time of 250 ms, then to retract 10 µm over 100 ms with a pause time of 250 ms (Figure 3). 6. As the AFM head is oscillating, click the 'start' button of the alignment software to begin the automated alignment process. When the program is running, the software is writing and reading data in the file described in the probe array with a newly prepared array (step 2) and repeat the alignment (step 4.4). 9. Move the stage upwards in the z-axis using the stage controller console for that axis. The stage should be moved in 500 nm increments until contact can be observed from the top view camera of the AFM. Contact between the array and substrate can be observed as a 'white dot' of high contrast at the apex of the individual probe pyramids. 10. At this point, click the 'stop' button on the AFM control software to stop the spectroscopy program from step 4.5. This will retract the array by 10 µm, therefore leaving 10 µm of possible z-axis extension. Check the image from the top view camera of the AFM to ensure that the probes are not in contact with the substrate.

Polymer pen lithography (PPL)
1. Navigate to the lithography component of the control software by clicking the 'lithography' button on control software. Choose the zmodulation operating mode and import a raster (bitmap) or vector image that will be used as the lithography pattern. In order to generate the features shown in the representative results, use a bitmap consisting 20 x 20 black pixels (see supplemental material), corresponding to the lithography of a grid of 20 x 20 dots per probe on the PPL array. 2. Enter the lithography parameters into the 'Pixel Graphic Import' window of the AFM controller software.
1. Configure the 'Size' of the pattern to be generated, e.g., 40 µm in length and width. These parameters indicate the width and length over which the image in the bitmap will be scaled. To generate features shown in the representative results, use a width and length of 40 µm in both dimensions. 2. Set the 'Origin' of the pattern to be generated at 25 µm on both x and y axis. These parameters determine the center of the image relative to the center of the AFM x/y-axes. Set these parameters to avoid the region of the surface where the probes were in contact during the alignment process. 3. Set the printing 'Parameters'. These values determine how the probes are to be extended (i.e. brought into contact with the surface) in response to each pixel in the bitmap image. 1. Select from the drop-down menu 'Modulation Abs Z Pos' and 'Simplify to' two layers. This mode instructs the AFM to extend the probes by an absolute distance determined by only two results, either 'Black (0)' or 'White (1)' fields. 2. Set the values in the 'Black (0)' and 'White (1)' fields to 5 and -5 µm, respectively. These values determine the distance the probes should be moved in response to a black or white pixel on the bitmap image and are typically set between 3 and 5 µm for 'Black' (i.e., extend probes downwards by that distance relative to the zero point of that axis) and -3 to -5 µm for 'White' (i.e., withdraw the probes upwards by 3 to 5 µm relative to the zero point). NOTE: These representative distances assume that a 5 µm extension results in the probes coming into contact with the surface and hence the generation of a feature, while a 5 µm withdrawal lifts the probes away from the surface resulting in no contact. Zextension affects feature size by determining the extent of probe contact with the surface, greater extensions result in the probes being pressed further into the surface, resulting in larger features. 10 3. Click the 'OK' button to implement these settings and close the window.
3. Enter the 'pause time' in the lithography window of the AFM control software, typically 1 s. This setting determines the length of time the probes remain in the extended 'Black' position, which is typically set between 0.1 and 10 s. NOTE: Longer pause times result in larger feature sizes due to the larger amount of MHA transported to the gold surface. Further details on controlling the size of features generated can be found in other reports. 20 4. Prepare the atmospheric control enclosure. 1. Lower the atmospheric isolation chamber onto the AFM and open/run the manufacturer-supplied atmospheric control software (MHG_control.exe). 2. Set the atmospheric control software to maintain a relative humidity of 45%, a temperature of 25 °C, and an atmosphere exchange 'Flow rate' of 500 mL by entering these values into the software. Click 'Use' to implement the settings. The atmospheric control module will then begin to pump humidified air into the chamber. NOTE: Higher humidity levels result in larger feature sizes due to the formation of a larger water meniscus generated between pen arrays and surface. 21 This value is typically set between 40 and 60%. The flow rate is typically set between 300 and 500 mL. Larger flows allow the desired humidity level to be reached more rapidly but is less accurate. The representative results use a flow rate of 500 mL for initial generation of humidity and is decreased to 300 mL upon reaching the desired humidity, to maintain an accurate and stable level during lithography.

5.
Once desired humidity is obtained, start the lithographic sequence by pressing the 'start' button on the software interface. 6. Upon completion of the lithography, use the z-axis stage controller console to move the substrate away from the array by retracting the stage by 500 µm. Then remove the atmospheric isolation chamber from its mount. 7. Switch the stage release lever to lock the sample stage and deactivate the force sensors, as indicated by the AFM manufacturer's instructions, then remove the substrate from the stage.

Pattern visualization
1. Patterns can be visualized using one of the following methods, lateral force scanning probe microscopy or chemical etching. 2. Scan the patterned surface on AFM with lateral force mode using contact mode cantilever to examine the features nondestructively. NOTE: Lateral force microscopy can be used as a nondestructive method of viewing the features produced by polymer pen lithography; however, using this method, only a relatively small area can be visualized (typically 100 x 100 µm).
4. Subsequently submerge the substrate in 2 mL of a solution of fluorescently conjugated anti-rabbit secondary antibody (diluted with 5% (w/v) BSA in PBS at the manufacturer's specified dilution, 2 drops/mL), cover in tin foil and incubate at room temperature for 1 h, then wash three times with 0.1% PBST. 5. To label actin filaments, submerge 2 mL of fluorescently conjugated phalloidin at a dilution of 1:250 in PBS, cover in tin foil and incubate at 4 °C for 30 min then wash three times with PBS. 6. Simultaneously stain cell nuclei and mount the substrate by applying a drop of mounting medium containing DAPI and cover with a coverslip.
4. Visualize cells using a fluorescence microscope according to the manufacturer's instructions, with excitation filters of 365 nm for nuclei (DAPI), 488 nm for F-actin and 594 nm for fibronectin.

Representative Results
To check whether the automated alignment had been successful, the graphs plotted from the alignment data (in the spreadsheet from step 4.8) were examined. Where the alignment process had been successful the two plots, corresponding to the angle by which the sample stage has been tilted along the θ and φ axes, showed a series of rising and descending data points. In each of the plots, two linear fits of the data points showed a well-defined intersect "peak" indicating the maximum z-extension and the corresponding angle at which alignment was achieved (Figure 4A and 4B). This process is repeated four times (i.e., twice for each axis) and plotted as a set of four coordinates. The intersection of each pair of coordinates thus shows the overall optimum angles (Figure 4C). 13 In cases where the alignment was not successful, it can be observed that their corresponding θ and φ angle plots do not give good quality linear fits, or do not intersect (Figure 5). Such failed alignments are typically as a result of the arrays being improperly trimmed or mounted to the probe holder (steps 1.7, 1.8, and 2.2). In these cases, the arrays were discarded and a new one prepared and mounted (steps 1 and 2), and the alignment process repeated (step 4).
Upon successful alignment and lithography with MHA by PPL, patterned gold substrates were then imaged using lateral force microscopy to examine whether deposition had taken place. A larger area examination of the printed surfaces was also conducted by optical microscopy of the substrates after etching of the gold not protected by the deposited thiol (Figure 6 and Figure 7). However, the etched patterns cannot be used for further functionalization and should only be used to confirm patterning on representative samples of a batch of printed surface substrates. If the etched patterns show blank areas corresponding to individual pens (Figure 8), this result indicates that the production of probe arrays has not been done successfully, and that some probes are damaged or missing. This inhomogeneity of the probes may be due to the use of an old master where the perfluorinated coating has worn away (step 1.3), resulting in some probes being torn away when the array is separated from the master. In these cases, a new master should be used. The result may also be due to the presence of air bubbles trapped between the glass backing and the master (step 1.5), or if the probe array was not cleanly separated from the master after curing (step 1.8).
Florescent microscopy images of the fibronectin functionalized surfaces incubated hMSCs were also collected (Figure 9). In general, all substrates were stable within the in vitro culture environment and the cells adhered and adapted their morphology to the printed patterns in case of smaller isolated 20 x 20 array of features.    . It can be seen that in some areas no patterns are generated, due to missing probes in those locations. In the areas where only two lines of dots are produced, this result indicates that a probe is present but it is not of the same height as the fully functioning probes, so only deposit features when the array is extended to the full z-axis distance. In this image, the contrast has been deliberately altered to enable observation of the partially printed areas. Please click here to view a larger version of this figure.

Discussion
This protocol serves to provide users with a convenient methodology to rapidly carry out nanolithographic patterning with high uniformity and controllable feature size over large (cm The key enabler of this protocol is the automation of the alignment procedure (step 4) that allows highly uniform and high-throughput production of features on surfaces, down to nanoscale resolution, which enables the rapid turnover of cell culture experiments. The polymer pen lithography carried out using this alignment algorithm is able to generate nanoscale features within approximately 30 min. The reproducibility and accuracy of automated alignment, and thus the uniformity of the patterned features, is however critically dependent on the quality of the probe arrays that are produced (step 1 and 2). Any flaws in their preparation that result in blunt, broken or missing probes; such as trapped air bubbles (step 1.5) or improper separation of the probes from the master (step 1.8) can result in inaccurate alignment and poor quality lithography.
This reported method shares a limitation in common with other alignment methods that rely on force feedback. The accurate determination of when the probes are in contact with the surface is constrained by the need to account for background vibrations caused by the ambient environment and the movement of the sample stage. In general, the sensors have a force sensitivity in the µN regime (2 µN in this case), but the alignment algorithm is designed to only register a force of at least 490 µN as definitive contact between the probes and the surface, in order to avoid any 'false positives' resulting from background noise. 13 Thus, this method tends to produce large features (1-2 µm) since the probes must extended a large distance on the z-axis (with a consequent higher force) in order to register contact. In order to compensate, smaller features can be generated by reducing the z-axis distance travelled during the lithography step (e.g., entering the 'Black' setting in step 5.2.3.2 as 3 µm instead of 5 µm).
Nevertheless, even with this limitation, the automation algorithm is able to address a critical aspect in the application of parallelized scanning probe lithography methods, as alignment was previously the most time demanding and imprecise step in the implementation of these techniques. This automation now shifts the rate-limiting step of the fabrication process from the alignment to the lithographic writing itself. While this protocol demonstrates the application of this alignment procedure to PPL, the framework could be applied to a number of SPL techniques such as lipid-DPN 26 and matrix-assisted lithography 27 as well as potential future catalytic probe systems. 28

Disclosures
The alignment algorithm and software were developed by, and are proprietary to, the University of Manchester. It is available for download at http://www.click2go.umip.com/.