Polymeric lithography editor: Editing lithographic errors with nanoporous polymeric probes

See allHide authors and affiliations

Science Advances  09 Jun 2017:
Vol. 3, no. 6, e1602071
DOI: 10.1126/sciadv.1602071


A new lithographic editing system with an ability to erase and rectify errors in microscale with real-time optical feedback is demonstrated. The erasing probe is a conically shaped hydrogel (tip size, ca. 500 nm) template-synthesized from track-etched conical glass wafers. The “nanosponge” hydrogel probe “erases” patterns by hydrating and absorbing molecules into a porous hydrogel matrix via diffusion analogous to a wet sponge. The presence of an interfacial liquid water layer between the hydrogel tip and the substrate during erasing enables frictionless, uninterrupted translation of the eraser on the substrate. The erasing capacity of the hydrogel is extremely high because of the large free volume of the hydrogel matrix. The fast frictionless translocation and interfacial hydration resulted in an extremely high erasing rate (~785 μm2/s), which is two to three orders of magnitude higher in comparison with the atomic force microscopy–based erasing (~0.1 μm2/s) experiments. The high precision and accuracy of the polymeric lithography editor (PLE) system stemmed from coupling piezoelectric actuators to an inverted optical microscope. Subsequently after erasing the patterns using agarose erasers, a polydimethylsiloxane probe fabricated from the same conical track-etched template was used to precisely redeposit molecules of interest at the erased spots. PLE also provides a continuous optical feedback throughout the entire molecular editing process—writing, erasing, and rewriting. To demonstrate its potential in device fabrication, we used PLE to electrochemically erase metallic copper thin film, forming an interdigitated array of microelectrodes for the fabrication of a functional microphotodetector device. High-throughput dot and line erasing, writing with the conical “wet nanosponge,” and continuous optical feedback make PLE complementary to the existing catalog of nanolithographic/microlithographic and three-dimensional printing techniques. This new PLE technique will potentially open up many new and exciting avenues in lithography, which remain unexplored due to the inherent limitations in error rectification capabilities of the existing lithographic techniques.



Errors are omnipresent in nature (1). Some naturally occurring errors rectified and/or amplified over time positively influence the system, as in mutation-led evolution (2). On the contrary, the errors produced in a synthetic fabrication process can propagate and proliferate with time rather than self-rectify, devouring valuable time and resources (3). Therefore, error rectification is of utmost importance for fabrication processes.

Among the existing fabrication techniques, lithography (4) is a neoteric fabrication technology that forms the cornerstone of biotechnology (5), electronics (6), and the semiconductor industry (7). Commonly encountered errors in standard lithographic techniques include mask critical dimension error (8) and exposure system aberration (9) in photolithography, and stitching (10) and overlay errors (11) in e-beam (electron beam) lithography; e-jet (electrohydrodynamic jet) (12), dip-pen nanolithography (DPN) (13), and DPN variants can suffer from errors common to any conventional printing process.

Whereas a number of commercially successful techniques are available for lithography and fabrication, only a handful of techniques are available for error correction. A few notable error-correcting techniques discussed in literature are nanoshaving (14), nanografting (15), electron-induced diffusion and evaporation using a tip (16), and electrochemical removal using a conductive atomic force microscope (AFM) tip (17). These probe-based error correction techniques (18) can be labor-intensive, expensive, and time-consuming. The dry abrasion methods can incur considerable friction and may also lack continuous feedback for error detection and rectification (1418). With the existing business model and paucity of editing techniques, discarding defective products is preferred to rectifying the errors in the fabrication process.

The global annual revenue for semiconductor industries reached $260 billion (19). A large number of these industries still use the conventional mask-based wet etching methods for fabrication with a process yield of ~80% (20). This relatively modest device fabrication yield can generate significant material waste and contribute to loss of productivity. With rapidly growing lithography-based industries, there is a huge market to be conquered in post-fabrication error correction. The error-rectifying polymeric lithography editor (PLE) system introduced here demonstrates the feasibility of erasing and rectifying post-fabrication errors in lithography. This technique will be potentially of great interest to the burgeoning semiconductor, biotechnology, microelectromechanical system (MEMS), and Bio-MEMS industries (21), as well as other industries relying on traditional lithographic patterning.

PLE is a probe-based technique conceptually related to DPN (13), polymer pen lithography (PPL) (22), beam pen lithography (BPL) (23), and hard-tip, soft-spring lithography (HSL) (24). The highlight and novelty of this new PLE system are the “real erasing” of the deposited patterns followed by deposition at the erased spots. The erasing mechanism in PLE is analogous to cleaning a surface with a “wet sponge” but at the “micrometer” scale. PLE also boasts continuous optical feedback for high precision and accuracy during erasing and deposition.

We have also used PLE to demonstrate fabrication of a working device. PLE was used to fabricate a pair of interdigitated copper electrodes for a functional microphotodetector device. This was achieved by electrochemically erasing/etching a 200-nm copper film by translocating an Fe3+-encapsulated PLE over the copper film that yielded an interdigitated electrode pair. In principle, PLE has the potential to evolve into a “three-dimensional (3D) etcher” analogous to a 3D printer. This can be accomplished by attaching a viscoelastic conical hydrogel or a porous polymeric matrix probe to a commercial 3D printer’s tool head and replacing the printing ink reservoir with a reservoir containing an etching agent (25). Thus, PLE can potentially open up many new and exciting avenues in lithography, which are currently unexplored due to the inherent limitations in erasing capabilities of the current lithographic techniques.


Synthesis and characterization of hydrogel PLE probes for erasing

PLE agarose erasers were template-synthesized from track-etched conical pores present in glass wafers (Fig. 1A). Agarose is a thermoreversible hydrogel that is widely used in life sciences and biotechnology (26, 27) because of its biocompatibility (28) and highly porous nature (29). Porous agarose (5% w/w) can hold >95% of its weight in water (30) and has an extremely large hydrated surface area of ~271 m2 g−1, with an average pore size of 55 nm (31). The aqueous agarose PLE in the hydrated form was not imaged with a scanning electron microscope (SEM) at high resolution, but a polydimethylsiloxane (PDMS) PLE synthesized from the same template was imaged at high resolution using the SEM (Fig. 1B). The PDMS PLE consisted of dt = 500 nm, db = 22 μm, and h = 33 μm, where dt, db, and h are the average tip diameter, base diameter, and height, respectively (Fig. 1B). The agarose PLE is expected to have the same geometry as that of the PDMS PLE. The cross section of the lyophilized agarose PLE revealed a dense population of nanopores in the sub–100 nm range (Fig. 1, C and D). The hydrated nanoporous nature of agarose imparted a “wet nanosponge”–like characteristic to the agarose PLE eraser. This facilitated wetting of the patterned molecules on the surface and their transport into nanoporous PLE tip during erasing.

Fig. 1

(A) An SEM image showing a cross section of a single track-etched pore in a glass substrate. (B) A typical SEM image of a single PDMS PLE fabricated using a conical pore as a template. The tip diameter (dt), base diameter (db), and height (h) of the PLE are 500 ± 100 nm, 22 ± 1 μm, and 38 ± 1 μm, respectively. This was an exact replica of the agarose PLE because both PLEs are synthesized from the same template. (C) A typical SEM image of the cross section of the agarose PLE shows nanochannels in the agarose matrix, which allowed adsorption and transport of the molecules from the patterns during the erasing process. (D) An SEM image of the agarose PLE showing nanochannels at the surface. Inset shows corresponding AFM image of the agarose surface.

Agarose PLE was synthesized using gelation of agarose followed by its aging. This is because the aged PLE yielded high quality and large capacity molecular erasing over multiple erasing cycles without any significant decrease in erasing performance. Therefore, gelation and aging of agarose warrant a brief discussion here. The gelation of agarose is a series of phase transitions that starts out kinetically controlled and culminates in a thermodynamically stable low-energy state (32). Dynamic light scattering has been often used to investigate the gelation mechanism (33). Gelation is widely accepted to proceed via liquid-liquid phase separation (34). As the temperature of the solution approaches the gelation point, the Brownian motion of the polymer chains slows down, causing them to entangle. These entanglement points act as “nuclei” or “seeds” to recruit more fibers from the sol state to the gel state (32). As gelation progresses, the polymeric chains eventually segregate into polymer-rich gel phase and polymer-depleted sol phase, yielding water-filled pores in agarose hydrogel. During gelation, the correlation length between the polymer chain was found to increase from 40 to 400 nm [0.05 weight % (wt %) agarose] (32). The pore size is directly related to the correlation length of the polymer chains. With gelation time, the local coagulation of the polymer-rich areas progresses until they reach an equilibrium, thereby forming stable pores in the matrix. Thus, a larger number of polymer chains in high agarose wt % solution densely pack into smaller pores (1.5% agarose has a mean pore size of 129 nm), whereas a smaller number of polymer chains in low agarose wt % solution sparsely arrange themselves to form large pores (0.5% agarose has a mean pore size of 226 nm) (32). During the overnight pseudoequilibration stage or “aging,” the mean correlation length does not undergo any significant change; however, the remaining free-floating chains pack themselves to their lowest local energy configurations (32). The organization and packing of the polymer chains during the aging step stabilize the porous network. In our experiments, the PLE erasers were allowed to equilibrate or “age” at 4°C in water overnight, thereby stabilizing the porous network in the PLE tips. We believe that the aging of the agarose is the key for highly efficient molecular erasing and editing using agarose PLE (fig. S1).

Mechanism of erasing

The agarose hydrogel PLE was mounted to an XYZ stage coupled to an inverted optical microscope (Fig. 2A). The humidity was maintained at 55 ± 3% RH (relative humidity) using a humidifier and a hygrometer. The RH near the substrate-PLE interface increased exponentially as PLE agarose approached the substrate (fig. S2). High humidity at the PLE tip–substrate interface resulted in the formation of a water meniscus upon contact, and the molecules present on the substrate were hydrated within the water meniscus volume. The solvated molecules were transported into the nanoporous agarose matrix via the water meniscus at the PLE-surface interface. Humidity appears to have a complex relationship with “erasing contact time” and erased pattern diameter. This is because of the presence of two independent sources of humidity in the erasing experiments: (i) ambient humidity and (ii) humidity from the hydrogel agarose PLE itself. This complex interplay between interfacial humidity and agarose drying and shrinkage is likely to affect the pattern diameter and contact time when erasing experiments are performed for long durations. Overall, we believe that to fully understand the effect of ambient humidity on the PLE-based editing, more thorough studies in precisely controlled environments are required (see the Supplementary Materials for more details).

Fig. 2

(A) A schematic of the custom-built PLE system with an XYZ stage coupled to an inverted optical microscope. (B) Linear de-Embedded Image dependence suggested diffusion-based molecular erasing by the agarose PLE. The error bars in (B) is SD in de (n = 3). (C) A schematic representation of the absorption of molecules upon contact with the PLE tip with a contact time (te) of ~ 1.8 s. (D) A fluorescence micrograph of the PLE erasing with a 5% (w/w) of agarose after gelation. The diameter of the erased patterns was 11.7 ± 0.7 μm. (E) Deposition with agarose PLE—a continuous line deposition, with agarose PLE equilibrated in a 2% fluorescein solution. Scale bars, 30 μm (D and E). (F) A schematic showing the line pattern deposition of fluorescein using agarose encapsulated with fluorescein. Both the deposition and erasing occurred via the water meniscus formed upon contact between the conical hydrogel tip and the substrate.

The diffusion mechanism of PLE molecular erasing was confirmed by the linear de- Embedded Image dependence (Fig. 2B), where de and te represent the pattern diameter and erasing time, respectively. This erasing mechanism is analogous to the diffusion-based deposition mechanism of the DPN via the water meniscus (13). Lyophilized (dehydrated) agarose showed negligible or no erasing, suggesting that water plays crucial roles both in the dissolution of the adsorbed ink molecules in the water meniscus and molecular diffusion into the porous agarose matrix of the PLE eraser. Erasing efficiency depends both on the aging of the agarose (fig. S1) and on the hydrodynamic radius of the erased molecules that diffuse through the PLE matrix (35). Figure 2C schematically shows molecular erasing via the water meniscus formed at the substrate-hydrogel interface.

Figure 2D shows a 6 × 6 dot array erased by the agarose PLE, orchestrated by contacting the PLE with a fluorescein-coated glass wafer at 36 points. The agarose PLE erasers exhibited an erasing efficiency (Er) of ~0.85 over 36 erasing spots with de ~11.7 ± 0.7 μm (τe = 1.8 s) (Fig. 2D, fig. S1, and movie S1). We define Embedded Image, where Ii and If are the initial and final integrated emission intensities of the erased spots, respectively. The spectroscopic analysis of fluorescein molecules extracted from the agarose erasers suggested that ~100 pmol was removed after 36 sequential erasing cycles of de ~11 μm spots. This yielded an average areal erasing of ~36 fmol/μm2 of fluorescein in our experiments.

The presence of an interfacial water layer between the PLE and the pattern played an important role by decreasing the friction between the agarose PLE and the substrate. The interfacial water layer also allowed deposition of continuous lines (Fig. 2E). This sliding mechanism is analogous to skiing on ice, where the interfacial liquid water lubricates the movement of a ski on the snow (36). The analogous sliding of the agarose PLE over a surface (Fig. 2F) allowed continuous molecular deposition with speed in the range of 10 μm/s over >600-μm writing length scale (Fig. 2E). The writing speed >10 μm/s was not tested in our experiments but is plausible in our experiments. Such fast deposition was not possible with the PDMS PLE because of the lack of lubrication at the PLE-substrate interface, which in turn generated a stick-slip friction (37) at the interface during PLE translocation.

To further understand the erasing mechanism for the agarose PLE, we used a macroscale model (dt = 1 mm, db = 7 mm, and h = 20 mm) of the agarose PLE eraser to optically investigate the diffusion mechanism during the erasing process (figs. S3 and S4). By analyzing the time-lapse snapshots of the erasing by the macroscale agarose PLE (movie S2), we confirmed that the erasing proceeds through the following three key steps: (i) rehydration of the deposited molecules upon tip approach (time of 1 to 4 s in movie S2), (ii) diffusion of the solvated ink molecules from the wet surface to the agarose macro-PLE tip upon contact (time of 5 to 8 s in movie S2), and (iii) the migration of the absorbed molecules from the tip into the agarose matrix. This was observed in the experiments through fading of the blue color at the contact spot (time 9 to 16 s in movie S2). The rapid diffusion of the absorbed molecules away from the tip into the matrix maintained a considerably large concentration gradient between the tip and the substrate, which supported the constant Er (Fig. 2D and fig. S1). The results of the macroscale PLE model (figs. S3 and S4) corroborate the erasing phenomenon in the microscale PLE probe (Fig. 2B). Although some forms of probe-based erasing were previously reported in the literature (1418), we believe that this is the first demonstration of “real/conventional” erasing using a submicrometer nanosponge probe at the micrometer scale without utilization of an externally applied stimulus. The real erasing here implies that the ink molecules are erased or transported into the matrix of the eraser.

Parameters influencing capacity, efficiency, and speed of PLE erasing

The large and undiminished erasing efficiency observed in our experiments becomes apparent after a simple numerical analysis. We defined Vr = Vf/Ve, where Vf is the total available free PLE volume and Ve is the volume of the fluorescein molecules picked up by the PLE in each erasing cycle. Thus, Vf is composed of the sum of conical PLE eraser volume (VPLE) and the supporting agarose base volume (Vbase), which is 5 mm × 5 mm × 5 mm in dimension. In our case, Vr ~1010 suggested that VfVe (the fluorescein film thickness was assumed as ~10 nm). Even without a supporting base, Vr ~67,000 provided a constant Er ~0.85 for 36 erasing cycles. If we assume that the whole free volume of the agarose PLE and the supporting agarose base can be completely used to pack the erased molecules in the matrix, we estimate that a single PLE eraser would erase a 100-nm-thick fluorescein (or similar molecules) film of an area >1 m2. However, the erasing of this large area would affect Er and would also require many days to achieve this type of erasing using a single PLE eraser under passive erasing conditions.

In addition to the volume factor involved in erasing, the time lag (τ) between each erasing cycle also contributed to the observed high erasing efficiency. With the diffusion coefficient (D) of ~2.5 ± 2 × 10−10 m2 s−1 for a small molecule in the agarose gel (38), the estimated distance (d) travelled by the “picked-up” ink molecules from the tip of the PLE eraser into the matrix in τ = 10 s was ~70 μm (Embedded Image). That is, the absorbed molecules will travel a distance of ~70 μm from the tip into the matrix between consecutive erasing cycles (τ ~10 s). This simple analysis suggested that a high concentration gradient between the PLE tip (low ink concentration) and the substrate (high ink concentration) was maintained during molecular ink erasing. Even with a modest τ of 100 ms between erasing steps, the molecules will travel >7 μm, resulting in a large concentration gradient between the substrate and the PLE favoring a constant Er.

Er was also found to depend on the agarose concentration and gel setting time—both of these parameters are related to cross-linking density of the agarose matrix. A freshly prepared agarose gel PLE eraser showed a significant decline in the Er after six erasing cycles (that is, Er ~0; fig. S1). However, a porous hydrated agarose PLE that was aged overnight exhibited a constantly high erasing efficiency (Er ~ 0.85) for ≥36 erasing cycles (fig. S1). This decline in erasing efficiency for a freshly prepared agarose PLE is attributed to incomplete gelation and poorly formed porous hydrogel network. Gelation of agarose is both a time- and temperature-dependent process, where a complete gelation usually takes place over several hours at low temperature (33). Agarose PLE erasers gelled and aged overnight are believed to attain a pseudoequilibrium, where a highly cross-linked density network with stable pores yielded a constant high erasing capacity for the PLE (see text above for more information on agarose gelation and its aging).

On the basis of the above-mentioned observations, the erasing efficiency for the extremely large area of the porous PLE can be further improved by the following steps: (i) increasing the lag time between the erasing cycles, which will generate a high concentration gradient between PLE and the substrate, and (ii) actively driving the erased molecules away from the tip through an applied external stimulus. Electric field-induced electrophoresis and/or electroosmotic flow may allow selective control over molecular transport rate during the erasing process through controlled application of electric field. Recently, electric field–induced enhanced transport of the protein molecules with nanofountain probes was demonstrated, where the molecular transport rate of the protein molecules was enhanced by up to three orders of magnitude over diffusion-based transport (39). (iii) Last, soaking the erasers in a suitable solvent with a high partition coefficient for erased molecules will create “free space” for the incoming erased molecules.

Comparing PLE erasing with existing technologies

It is important to compare agarose PLE with the other erasing experiments reported in the literature. The areal erasing speed (Embedded Image) of the PLE is extremely large (~785 μm2/s), where Ae = πde2 is the erased area and de is the diameter of the erased spot. That is, a surface area of 1 mm2 coated with a small molecule (~1 to 3 nm thick) can be erased in ~20 min using an agarose PLE probe from our studies. In contrast, in an AFM nanoshaving experiment, with de ~100 nm and a scan rate of 1 μm/s (40), it would take ~115 days to erase or shave an area of 1 mm2. The molecular erasing using PLE is real; that is, the PLE actually soaks up molecules off the patterns, allowing erasing of the patterns. Furthermore, the PLE erasing capacity is many orders of magnitude larger than that for the AFM tip–based erasing experiments. This is attributed to porosity and size differences between agarose and SiOx/SiNx-based probes. Although it may be practically unlikely to use an AFM tip to erase an area of 1 mm2, it must be taken into consideration that there is no current technique that can perform ultrafast and high-quality real erasing at the micrometer scale similar to PLE.

Because of the abrasive nature of erasing involved in nanoshaving experiments, where mechanical friction is needed to remove the molecules absorbed or bound to the surface, irreversible damage to the tip and the substrate is a concern. In contrast, the agarose hydrogel PLE can smoothly translate across the substrate without damage to the surface. The PLE erasing picks up erased molecules into its matrix, but the AFM shaving may leave behind debris of erased molecules on the substrate, which may require post-erasing processes for further substrate cleaning. Another difference between the AFM-based and the PLE-based erasing is the spatial resolution of the removal of the patterned molecules. The spatial resolution of the AFM-based erasing is two to three orders of magnitude better than the PLE erasing, which may be highly beneficial and desirable in many nanotechnological and life and materials science applications.

Spatial resolution of PLE easing: Experiments and modeling

We were intrigued by the fact that even though agarose PLE had a tip diameter of ~500 nm, the smallest de (referred to as de,smallest) obtained in our experiments was ~10 μm. Under identical conditions, the PDMS PLE tip (dt ~500 nm) generated dot patterns measuring ~600 nm in diameter (fig. S5). That is, the smallest patterns produced by the PDMS PLE tip was only ~20% larger than dt, whereas the agarose tip consistently yielded patterns de,smallest ≈ 20 × dt. The diameter of the erased area depends on three key parameters: the deformation of the tip due to applied force upon contact with the surface, water meniscus at the PLE tip–surface contact point, and mechanical properties of the tip and surface. Even under very careful maneuvering, de,smallest ~10 μm was the smallest feature size obtained by our experiments. The de,smallest ≈ 20dt was unlikely due to swelling of agarose because agarose does not swell significantly in water (41).

We argue that significantly larger than expected erasing spot dimensions in our experiments using agarose PLE were due to large dimensional changes in the agarose PLE tip upon contact with the surface. The interaction forces involved between the PLE tip and the surface are van der Waals (vdW), capillary, and electrostatic in nature (see the Supplementary Materials). The electrostatic interactions are considered to be minor because of the lack of charged species in our system. Furthermore, the presence of a water layer on the PLE and the substrate will depolarize any electrical charge present on the surfaces, reducing the electrostatic contributions to the overall PLE-substrate interaction energy.

The two major interactive forces considered in our system are vdW and capillary forces at the interface (Eqs. 1 and 2). The vdW forces are attractive forces between materials that can arise through dispersion and dipolar-dipolar interactions between atoms or molecules (42). Although these forces are known to be weaker than covalent and ionic forces, they are responsible for one of the strongest reversible adhesive forces observed in nature [for example, the adhesion of gecko setae to various surfaces is attributed to vdW interactions (43)]. The vdW forces between the PLE tip and the substrate were estimated using Eq. 1, by modeling the PLE tip and the substrate as two flat surfaces with water adsorbed on their surfaces (Fig. 3A). The maximum capillary force was estimated using Eq. 2, where R, γ, θ, D, and φ are the radius of the PLE tip, the surface tension of water, the PLE-water contact angle, the PLE-surface distance, and the PLE-water rise angle, respectively (see also fig. S6). Details on the capillary force calculations are given in the Supplementary Materials.Embedded Image(1)Embedded Image(2)Here, F(D), D, Aijk, T′, and T represent the force between the PLE tip and the substrate, the PLE-substrate distance, the Hamaker constant, the thickness of the water layer adsorbed on the PLE tip, and the thickness of the water layer on the glass substrate, respectively (44). i, j, and k represent different materials in our system (Fig. 3A). For example, A1,2,3 is the nonretarded Hamaker constant for the heterointeractions between 1 and 3 across 2. The subscripts 1, 2, 3, 2′, and 1′ in the Eq. 1 correspond to agarose (or PDMS), water, air, water, and fluorescein, respectively. The value of the Hamaker constant for agarose used in our calculations was assumed to be the same as that for cellulose because both the agarose and the cellulose are chemically very similar to one another (45). The Hamaker constants for the PDMS (46) and fluorescein were 4.4 × 10−20 and 5 × 10−20 J, respectively. The estimated values of Aijk for the agarose PLE system were 3.7 × 10−20, 0.23 × 10−20, 3.7 × 10−20, 0.1 × 10−20, and 3.7 × 10−20 J for A232′, A121, A32′3, A1′2′1′, and A323, respectively (see the Supplementary Materials). Similarly, the estimated values for A232′, A121, A32′3, A1′2′1′, and A323 for the PDMS PLE system were 3.7 × 10−20, 0.031 × 10−20, 3.7 × 10−21, 0.10 × 10−20, and 3.7 × 10−20 J, respectively.

Fig. 3

(A) A schematic of the vdW interactions between the PLE and fluorescein-coated surfaces. Both the PLE and the glass were modeled as flat surfaces covered with a layer of water at its interface with air. The thicknesses of the water layers (T and T′) were varied in our calculations to accommodate for the changes in the RH of the atmosphere. (B) E-D dependence for the PLE glass interface. Here, E is the interaction energy due to vdW and capillary forces between the PLE tip and the substrate, and D is the PLE-surface distance.

Figure 3B provides some interesting observations for the interactions between the PLE and the substrate and also on the interaction energy dependence on the material properties of PLE and substrate. The PLE composed of agarose appeared to have energetically higher interactions [more negative E(D) values] with the substrate at all D than the PDMS PLE–substrate interactions with the same T and T′ values. For small D (<0.5 nm), the vdW force interactions dominated total E(D) values, whereas the contribution of the capillary interaction forces to total E(D) was minor (Fig. 3B). However, for D > 5 nm, the interactions due to the capillary forces were larger than the vdW interactions between the agarose PLE tip and the surface. This is evident in an increase in the E(D) at large D (>2 nm) for the agarose PLE (as shown in Fig. 3B). The capillary interactions for the PDMS-based PLE were insignificant compared to the vdW interactions for D < 2 nm because of its hydrophobic nature. Finally, the effect of magnitude of the T and T′ values on the overall E(D) was minor as long as there was a thin layer of water (that is, both T and T′ ≠ 0) present on both the agarose PLE and the fluorescein-coated glass substrate. Because of the hydrophobic nature of PDMS (T′ was assumed as either 0 or 0.2 nm), the capillary force interactions were also negligible for all values of T′ and D.

Our analysis also suggested that the PLE and substrate surface material properties are important for the probe-based molecular erasing and patterning. For example, the PLE-substrate interactions are dependent on Hamaker constants (for vdW interactions) and water contact angle (for capillary interactions). Depending on the Hamaker constants, the vdW forces can be negative, which would affect the E(D)-D dependence (42). The electrostatic interactions for the electrically charged surfaces should also be considered in the analysis, which may dominate at small D when the ionic strength of the solution at the PLE-substrate interface is low.

On the basis of the vdW and capillary interactions on the PLE tip, the stress [= F(D)/A] generated by these forces at the agarose PLE tip–surface interface was estimated to be ~15.5 MPa (T = 1 nm, T′ = 1 nm, and D = 0.5 nm; Fig. 3B). A is the area of contact between the PLE and the substrate. With an elastic compression modulus of ~930 kPa for 5% (w/w) of agarose (47), we estimate that a relatively large strain (ε) of ~17 will be exerted on the agarose PLE tip when it is in close contact with the surface. The consequence of this large stress at the agarose PLE tip can be a large deformation of the agarose PLE tip—consistent with our experimental results that yielded de,smallest ≈ 20dt.

If we assume that the total interaction energy at the PLE-surface interface is converted into kinetic energy, then the PLE may attain a velocity (v) Embedded Image as it approaches the surface. Here, m is the total mass of the PLE and cantilever. For both the agarose and PDMS PLEs, with m ~0.5 g yielding v ~ 5.5 cm/s (D = 0.5 nm, T = 1 nm, and T′ = 1 nm). The collision between the agarose PLE tip and the substrate was also observed under a microscope during visual examination but was not captured by our camera because of the low acquisition rate (5 frames per second) of the CCD (charge-coupled device) camera used in the experiments. The interplay of the attractive forces between the agarose PLE and the substrate at short distances can satisfactorily explain significantly larger than expected patterns that were obtained in our experiments.

The estimated strain exerted at the PDMS PLE tip because of the attractive forces was ~7.2 at D = 0.5 nm. However, the experimental observed minimum deposition spot diameter was comparable to dt. The discrepancy between the experimental de and expected de from the estimated strain values on the PDMS PLE tip is believed to be due to the viscoelastic nature of the PDMS material used for the PLE fabrication (48). The stress generated at the PLE tip because of the contact with the substrate appeared to be quickly relieved without affecting the pattern size dimensions.

Additionally, the large water content in the agarose PLE and the high humidity at the interface (fig. S2) create a large water meniscus upon contact with the substrate. It has been shown by Rozhok et al. (49) that humidity plays a key role in the size of the patterns deposited. They have demonstrated that with the same AFM tip dimensions under same experimental conditions, higher-humidity experiments yielded larger pattern size (13). Thus, high humidity at the PLE-substrate contact interface (fig. S2) can also contribute to larger than expected erasing patterns in our erasing experiments.

Finally, both the experimental results and calculations further reveal that to improve the pattern size and resolution of the erasing, the materials and surface properties of the PLE and ink deserve special attention. For example, the Young’s modulus, Poisson ratio, and stress relaxation characteristics of PLE will affect the pattern size and resolution. Probe composed of hard materials (such as SiOx/SiNx with large Young’s modulus) and fast stress relaxation properties may provide higher spatial resolution and smaller de. The PLE composed of materials with a negative Poisson ratio (50) may show a decrease in erasing (and deposition) pattern size with an increase in the stress on it. There are several opportunities where the agarose properties can be tuned through simple chemical means (5153). It is possible to achieve smaller-sized patterns with modified hydrogels, for example, hydrogel PLE with enhanced mechanical modulus (54) and reduced water content along with precisely controlled humidity at the PLE tip–surface interface may allow molecular editing at a higher resolution than demonstrated in the present studies.

Editing with PLE probes: “Writing-erasing-rewriting molecular patterns”

To demonstrate molecular editing, erasing must be complemented with an appropriate deposition technique. PPL-inspired deposition followed by “nano–wet sponge erasing” and redeposition of molecules at the erased spots was sequentially orchestrated to demonstrate the “molecular editing process” using PLE. Mirkin’s group has pioneered probe-based deposition techniques, such as DPN (13), PPL (22), BPL (23), and HSL (24), with a variety of molecules on many different surfaces at the sub–100 nm scale. The complete editing process using the PDMS and agarose polymer probes are shown in the form of a fluorescence intensity surface plot in Fig. 4. Initially, a PDMS pen (dt ~500 nm and τc ~15s) was used to deposit a 3 × 3 array of fluorescein dots measuring ~15 μm (Fig. 4A). The deposition pattern size was monitored and tuned in real time by adjusting the z-piezoelectric stage while the deposition was in progress. The reversibly compressible PDMS can be flattened when pressed on the substrate to fabricate the patterns with a premeditated dimension (55). Although the PDMS pens can be used to deposit submicrometer features (fig. S5), a large pattern size of ~15 μm was chosen to demonstrate pattern editing because of the constraints imposed by the limiting achievable eraser spot dimension with an agarose PLE of ~10 μm. Optically transparent PDMS pens and agarose erasers were attached to the z-piezo stage using a transparent glass holder. The transparent window on the z-piezo stage allowed monitoring of the PLE probe during the editing process. The ability to lock a desired location with high precision and accuracy was important for the editing process (registration), because the same location must be accessed consecutively for writing, erasing, and rewriting. The optical transparency of the agarose PLE, PDMS PLE, and the PLE holder coupled to an inverted optical microscope made locations of the registration possible. PLE editing can also be accomplished with opaque substrates using upright microscopes.

Fig. 4 “Editing”—Write-erase-rewrite using PLE.

(A) A 3 × 3 fluorescent array deposited using a PDMS PLE pen (dt = 500 nm, db = 22 μm, and h = 38 μm). (B to D) Partial consecutive erasing of patterns along the diagonal. (E) Complete erasing of the fluorescent dots along the diagonal. (F) Demonstration of the rectification process of the patterns. AU, arbitrary units. (G) A schematic of the agarose PLE erasing mechanism showing partial erasing. The optical micrographs of the PDMS PLE pen (H) and agarose PLE eraser (I) in contact with the surface, as seen through the microscope. Scale bars, 25 μm (H and I).

Although both the PDMS and agarose PLE are optically transparent, they appeared opaque under an optical microscope (fig. S7). This intriguing phenomenon was analyzed using a custom finite-difference time-domain (FDTD) code (56, 57). Our simulations indicated that, analogous to a prism, the angled cone walls refracted the incoming light away from the straight path that yielded the transparent PLE microstructures appearing as dark circles in the optical micrographs (fig. S7A). The PLE lost the prism effect at the PLE-substrate interface, which allowed optical flux through the flattened part of the PLE tip generating a dark “doughnut”-shaped feature surrounded by a bright central part (fig. S7B). This clear doughnut-shaped feature was used to locate the contact point (registration) and to control the contact area of the PLE with the substrate during the editing process.

After depositing a 3 × 3 array with a conical PDMS pen, an agarose PLE eraser attached to the z-piezo stage was translated and positioned such that the eraser was partially above the first deposition in the array. After locking the xy coordinates of the patterns, the PLE eraser was translated vertically down, thereby making contact with the molecular pattern for τe = 3 ± 1 s. A partial erasing of the patterned molecules along a diagonal direction was demonstrated (Fig. 4, B to D). The contact point and contact area for PLE were tuned and adjusted by manipulating the position and dimension of the optical doughnut-shaped PLE. The patterns along the diagonal were chosen for partial erasing to demonstrate the “precision and confined registration” nature of the agarose PLE system. The precision and accuracy of the registration for writing and erasing depend only on the precision and accuracy of the piezoelectric stage used for the translocation of the PLE. In our experiments, the accuracy (repeatability) of our piezo stage was 150 nm. This allows accurate placement of the PLE eraser (and pen) within ±150 nm of a given registration spot. After demonstrating a controlled partial erasing along the diagonal, complete removal of the diagonal patterns was demonstrated by moving the eraser along the diagonal twice in contact mode (τe = 15 s) (Fig. 4E).

To complete the molecular editing process, fluorescein molecules were precisely redeposited at the erased spots having similar dimensions to those of the erased patterns. To accomplish redeposition at the erased spots, a PDMS PLE pen immersed in a 10 mM fluorescein solution for 60 s and dried under N2 gas stream was used. The optical doughnut was used to manipulate the size of the deposited pattern. The emission intensity of newly deposited molecules along the diagonal in Fig. 4F was comparable to the original unexposed dot patterns in Fig. 4A, demonstrating reproducibility of the PLE system. Figure 4G shows a schematic of the partial erasing using an agarose PLE.

The fluorescence emission intensity of the nonerased patterns was found to reduce by ~50% in comparison to their original emission intensity (Fig. 4, A and F). This decrease in the emission intensity was due to photobleaching of the fluorescein molecules during registration, PLE-surface contact alignment, and image acquisition processes. The photobleaching damage to nonerased patterns can be reduced by minimizing the photoexposure of the dye molecules during the erasing process. This can be achieved by reducing the photoexcitation aperture and intensity through automation of the erasing process. The above-mentioned three series of experiments demonstrated the editing process—“writing-erasing-rewriting” at the micrometer scale with high precision and accuracy using our PLE system. Another important feature of the PDMS PLE is that the same PDMS pen was also used multiple times (>100 times) for writing experiments without any significant observed deterioration in its performance. With automation and high precision tools, this simple proof-of-concept demonstration could be translated into a lithographic system for editing and rectification of errors at the submicrometer scale.

Device fabrication using PLE and potential applications of PLE

It is also possible to rectify photolithographic errors during the developing and etching processes with PLE. Many common photolithographic developers used in semiconductor processing [for example, RD6 (58), Microposit 351 (59)] are aqueous in nature. An error that occurred during a development process in photolithography can be rectified by (i) rastering a developer-loaded agarose PLE over an erroneous spot and (ii) a follow-up photoresist deposition over the erased spot by a photoresist-loaded hydrophobic PLE. Fabrication of porous hydrophobic nanostructures (for example, composed of nanoporous PDMS) has been demonstrated in literature (60). This process using PLE may allow rectification of photolithographic errors at the micrometer scale.

Furthermore, many common etchants used in the semiconductor industry for etching Si, SiNx, and SiO2 are aqueous in nature (for example, KOH and HF). Therefore, etchant-loaded PLE can be used for localized etching of semiconductors. Agarose encapsulated with appropriate molecular etchants are shown to etch metal oxides and semiconductors (61). Thus, tailoring the PLE polymeric materials and loading it with appropriate molecules may allow PLE to play a crucial role in the semiconductor industry.

We have demonstrated editing of the dot pattern array and the deposition of line patterns using the agarose PLE. We now demonstrate an interesting application of PLE as a “microelectrochemical translating reactor” toward device fabrication. A porous agarose hydrogel matrix encapsulating a 2% FeCl3 was translocated on a copper-coated glass substrate to electrochemically etch metallic copper (~200 nm thick) for fabricating an array of interdigitated electrodes. Briefly, the Fe3+(aq)–encapsulated agarose PLE tip was rastered at a speed of ~10 μm/s while in contact with the copper surface in a zigzag pattern. The encapsulated Fe3+(aq) ions diffused out of the PLE and electrochemically oxidized the metallic copper Cu0(s) to Cu2+(aq) (Eq. 3). The aqueous solvated products, Fe2+ and Cu2+, were absorbed back into the porous agarose matrix or were washed away with water.Embedded Image(3)

Figure 5A shows an optical micrograph of the resultant interdigitated copper electrode fabricated using the “microelectrochemical translator” PLE probe. The average width of the finger electrodes was 60 ± 9 μm (dark area) with an interelectrode separation of 25 ± 9 μm (nonconductive, light area in Fig. 5A). The electrical resistance of >1 gigaohm across the fabricated interdigitated electrode confirmed successful erasing of the Cu0 by the PLE. We estimated combined erasing capacity of an agarose PLE (dt = 500 nm, db = 22 μm, and h = 38 μm) and the supporting agarose base (5 mm × 5 mm × 5 mm) loaded with 2% Fe3+ of ~100 cm2 of Cu0, assuming that the thickness of the deposited copper is 10 nm (see the Supplementary Materials). However, in practice, the erasing capacity will be significantly lower than the theoretical estimated capacity because of kinetics and mass transport processes involved in the etching process.

Fig. 5

(A) An interdigitated copper electrode fabricated by redox reaction between the Cu0- and Fe3+-containing agarose PLE erasers. Scale bar, 150 μm. (B) A typical photoinduced current-time response for a device shown in (A). Black and yellow stars represent light turned OFF and ON, respectively.

It is also possible to electrochemically redeposit or rewrite metals using hydrogels. The encapsulation and deposition of silver (62) via electrochemical methods using highly porous and aqueous agarose are demonstrated in the literature. An AFM-DPN tool was previously used to electrochemically deposit metallic platinum (63). Furthermore, agarose can be made ionically conductive hydrogel which will allow metallic deposition or erasing using PLE by modulating the directional molecular transport through electrophoresis (64). Moreover, more opportunities are also possible wherein PLE can be customized to deposit organic and inorganic materials (65) and to erase metal oxides and semiconductors (13).

Using PLE, we fabricated a microphotodetector device by drop-casting a photoactive mixture composed of 20 mg each of poly(3-hexylthiophene) (P3HT) (electron donor) and C60 (electron acceptor) dissolved in 1,2-dichlorobenzene on the interdigitated electrodes (total area, ~15 mm2). The photoinduced current-time measurements were acquired by switching the light “ON” and “OFF” with a time interval (τs) of 1 s (Fig. 5B). The excitation was achieved with a broadband light source with an intensity of ~130 mW/m2 at a distance of 0.6 m. The photoinduced current (Ip) closely followed the light ON and OFF pattern. The dark current (Ip,dark) was <0.1 nA, which is below the detection limit of our instrument under experimental conditions. For τs = 1 s, Ip was ~3.3 and 2.7 μA with the lights ON and OFF, respectively, at an applied potential of 5 V (Fig. 5B). The device remained functional for >1 week under wet laboratory conditions without significant loss of its functionality. The photoactive device fabrication using the PLE system demonstrates the utilization of PLE in the device fabrication, which complements many currently available fabrication tools.


In conclusion, we have developed a new probe-based erasing system that can perform “erasing” with high precision and accuracy. The “nanosponge”-like hydrogel probe hydrates the molecules deposited on the surface and erases them via diffusion. The erasing efficiency is high because the free volume available in the agarose eraser is abundant in comparison to the volume of the erased molecules. The interfacial liquid layer between the tip and the substrate circumvents the stick-slip friction and enables frictionless continuous deposition and erasing. By mounting PLE probes on piezoelectric motors coupled to an optical microscope, the entire process was monitored and manipulated continuously. Combining erasing with the deposition, we demonstrated the editing process: writing-erasing-rewriting at the micrometer scale by using our PLE system. To further demonstrate its versatility, we used PLE as a “microelectrochemical reactor and translocator” device by encapsulating redox species in the PLE hydrogel matrix. An interdigitated array of microelectrodes for the fabrication of microphotodetector devices was accomplished using the agarose PLE. The soft polymeric probes can aid with the deposition and etching at the submicrometer and micrometer scales with high accuracy and precision without damaging the substrate. The porous matrix can also be used to encapsulate, transport, and deliver a wide variety of chemical and biochemical molecules for the fabrication of functional devices. With an appropriate additive to the hydrogel or by using a porous hydrophobic polymer, the PLE can even be coupled to the lithographic process in the semiconductor industry where organic solvents are used. Thus, the PLE system presented here complements the existing catalog of nanolithographic/microlithographic and 3D printing techniques for the fabrication of electronic, biological, optical, and micro- and nanodevices. PLE will offer exciting opportunities for new ventures in lithography that are currently unexplored because of the inherent limitations in the error-rectifying capabilities of the existing lithographic techniques.



The glass substrates were purchased from SCHOTT AG. PDMS SYLGARD 184 elastomer kit was purchased from Dow Corning. P3HT and phenyl-C61-butyric acid methyl ester (PCBM) were purchased from Sigma-Aldrich and Fisher Scientific, respectively. All other reagents (agarose, fluorescein, hydrofluoric acid, calcium nitrate, H2SO4, H2O2, FeCl3, and 1,2-dichlorobenzene) and supplies were purchased from Fisher Scientific and were used as received unless specified in the text. Tri-decafluoro-1,1,2,2-tetrahydrooctyl)triethoxy-silane (DTTS) was obtained from Gelest Inc.

Characterization and instrumentation

Electron micrographs were taken with FEI Quanta FEG 450. AFM measurements were carried out in tapping mode on a TopoMetrix Explorer using silicon tips. NTS10 piezoelectric stages from Discovery Technology International mounted on an inverted microscope were used to manipulate and translate PLE in the z axis. Optical images (both fluorescent and bright-field) were obtained using an inverted DM IRB Leica microscope. The fluorescence spectroscopy measurements were performed on a PerkinElmer LS 55 spectrometer. The photovoltaic device characterization was carried out by using an AM1.5 solar simulator (Newport Inc.) with a 150-W xenon arc lamp. A Melles Griot photodetector was used to measure the photoinduced response of the PLE-fabricated devices. Photoinduced current-time curves were measured using Keithley 6487 at an external bias of 5 V.

Ion tracking of glass substrates

The PLE pens and erasers used in our studies were fabricated through template-assisted synthesis (66, 67). The pores in the glass substrates were fabricated through a track-etched process in which 2.2-GeV Au ions were irradiated onto the glass slides at the linear accelerator in GSI Helmholtzzentrum, Darmstadt, Germany. The tracking in the glass substrates was performed in a 70-μm-thick SCHOTT D 263T borosilicate glass substrate. The kinetic energy of a given irradiating ion was 2.2 GeV. With averaged electronic energy loss of ~25 keV/nm, the ions fully penetrated 70-μm-thick glass samples (68).

Etching of patterned and randomly tracked substrates

Our previous experiments showed that the average etching rate of the tracked glass (TG) using 24.5% hydrofluoric acid is ~87.5 nm/s (57). Therefore, TG substrates were etched for 600 s (~75% of the time required for a complete etching) to avoid a complete etching of the substrates. A partial etching of the tracked substrates provided sharp (pointed) pores for PLE pens/erasers fabrication.

Fabrication of PLE pens and erasers

The etched glass chips were first thoroughly cleaned with piranha solution (3:1 H2SO4/H2O2) and were then fluorinated with a monolayer of DTTS for easy lift-off. DTTS functionalization of the TG was performed by immersing the chips in a solution of 4% (v/v) DTTS, 1% (v/v) acetic acid, and 95% (v/v) ethanol for 1 hour. The TGs were rinsed thoroughly in ethanol and deionized water and were baked at 90°C for 5 hours. Following the silane treatment, PDMS (or agarose) was poured over TG, filling the conical pores with polymer. The PLE-containing polymer layer was peeled off the TG after it was cured.

Agarose “erasers” were fabricated by casting a hot solution containing 5% agarose in water over the TG. The gel was allowed to set for 15 min before its removal from the TG. After removal from the TG, the erasers were placed overnight in water to stabilize the gelation process. To characterize the PLE erasers, the agarose gel structures were lyophilized (freeze-dried; see below). SEM characterization of the lyophilized PLE erasers was performed by sputtering a thin layer (~5 nm) of gold/palladium conductive layer on it.

Encapsulation of Fe(III) in PLE

A 2% solution of FeCl3 with 0.1 M HCl was made in deionized water. Preformed agarose PLE erasers equilibrated in deionized water overnight were immersed into a petri dish containing the FeCl3 solution for 30 min. FeCl3 diffused into water-filled pores within the agarose matrix, changing its color from whitish to yellowish. The Fe(III)-impregnated PLE agarose eraser was carefully and gently rinsed with water. It was air-dried for ~10 min to remove excess water before erasing.


Prefabricated agarose gel PLE eraser cut into pieces measuring 1 cm2 was immersed in 20 ml of water in a 50-ml corning centrifuge tubes and frozen in liquid nitrogen. Upon freezing, the lids of the tubes were punctured, and the samples were placed in a lyophilizer until all the water molecules were completely evaporated, leaving behind a free-standing solid agarose framework. The SEM of the cross section of the sample was analyzed with ImageJ.

FDTD simulations

To gain a better understanding of optically opaque to optically clear transition upon contact with the surface, FDTD simulations involving light propagation through PLE pens were performed. Briefly, a home-written code based on solving Maxwell’s electrodynamic equations was used for FDTD simulations. More details on FDTD simulations can be obtained from our previous publication (69).

Fabrication of the photoinduced detector device

A mixture of P3HT and PCBM (20 mg each) was dissolved in 1 ml of 1,2-dichlorobenzene. The solution was heated up to 60°C to speed up the dissolution of materials in the solvent. The mixture was cooled down to room temperature and was centrifuged at 2000 rpm for 5 min to remove any undissolved or undesirable large particles. The composite (1 ml) was drop-cast onto the PLE-fabricated interdigitated electrodes. A solar simulator with a 150-W xenon lamp was used as a light source. The excitation was a broadband source with an intensity of ~130 mW/m2 at a distance of 0.6 m. The photoinduced current (Ip) was measured using a Keithley 6487 picoammeter with photoexcitation ON and OFF.

Erasing capacity of a PLE microelectrochemical translating reactor

Agarose PLE with a db = 22 μm and h = 38 μm can hold ~4.8 × 10−6 μl of water (~4.8 × 10−6 mg of H2O). Two percent of the total volume of ~0.096 × 10−6 mg was occupied by Fe3+, which corresponds to 9.6 × 10−11 mol of Fe3+. This means that ~6 × 10−11 mol or ~6.1 × 10−9 g of copper can be erased using all the Fe3+ present in the PLE pen. With a copper metal density of 8.96 g/cm3, this corresponds to a copper volume of 6.8 × 10−10 cm3 or 42 × 1011 nm3. Therefore, this PLE microelectrochemical translating reactor can erase a copper film of 100 nm of an area of ~6.8 × 109 nm2. Total area would be 0.0068 mm2. However, with a base of volume (5 × 5 × 5 mm3) of 125 mm3, ~0.25 g of Fe or 3.5 × 10−3 mol of copper will be erased. This translates to 0.22 g of copper (volume, ~0.025 cm3). Thus, the PLE with a base can erase a copper film of 100 nm of an area of ~2.5 × 1017 nm2 = 1016 nm2 = 2500 cm2.


Supplementary material for this article is available at

fig. S1. Erasing efficiency.

fig. S2. Change in RH at the pattern-air interface.

fig. S3. A large-scale conical agarose eraser.

fig. S4. Optical micrograph of the PLE eraser.

fig. S5. Patterns deposited by PDMS PLE pens with a tip of 500 nm.

fig. S6. The schematic for the estimation of the capillary forces between the PLE and the surface.

fig. S7. FDTD simulation and microscopic visualization of a PLE conical pen before and after contact.

RH–versus–PLE substrate gap dependence

Role of humidity in writing and erasing

Mechanism of erasing

Nature of force between the PLE and the substrate

Capillary interactions for the PLE-surface pair

FDTD analysis

movie S1. This video was captured through the eyepiece of an inverted microscope while erasing the 6 × 6 array (Fig. 2D).

movie S2. This video was captured through the eyepiece of an optical microscope under bright-field conditions.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank V. Kollipara for useful discussions and suggestions. We acknowledge C. Watts for machining some parts for the fabrication and assembling of the PLE system. We also acknowledge J. Bozzola of the Integrated Microscopy and Graphics Expertise Center at the Southern Illinois University, Carbondale (SIUC), for his help in the acquisition of the SEM images and their analysis. Funding: We acknowledge partial financial support of this work from the NSF (CHE-0748676 and CHE-0959568), the NIH (GM 106364 and GM 080711), and the Office of Vice Chancellor of Research at the SIUC. The SEM used in this work was purchased through a grant from the NSF (CHE 0959568). Author contributions: P.R.R. performed all the experiments. M.D. performed the photoresponse experiments of the working devices fabricated using PLE. C.Z. performed the FDTD-based simulations of the interactions of electromagnetic waves with the PLE. P.K. performed the vdW and capillary calculations presented in the paper. C.T. and K.-O.V. provided the tracked substrates for the template synthesis. P.K. and P.R.R. designed the experiments. P.K., P.R.R., K.-O.V., and C.T. critically read the paper. P.K. and P.R.R. wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article