APPLIED SCIENCES AND ENGINEERING Polymeric ......The “nanosponge” hydrogel probe “erases” patterns by hydrating and absorbing molecules into a porous hydrogel matrix via diffusion

SC I ENCE ADVANCES | R E S EARCH ART I C L E

APPL I ED SC I ENCES AND ENG INEER ING

1Department of Chemistry and Biochemistry, Southern Illinois University, Carbon-dale, IL 62901, USA. 2GSI Helmholtzzentrum, 64291 Darmstadt, Germany. 3Tech-nische Universität, 64289 Darmstadt, Germany.*Present address: Institute for Systems Research, University of Maryland, CollegePark, MD 20740, USA.†Present address: Vuzix Corporation, 2166 Brighton Henrietta Town Line Road,Rochester, NY 14623, USA.‡Corresponding author. Email: [email protected]

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

2017 © The Authors,

some rights reserved;

exclusive licensee

American Association

for the Advancement

of Science. Distributed

under a Creative

Commons Attribution

NonCommercial

License 4.0 (CC BY-NC).

http://advancesD

ownloaded from

Polymeric lithography editor: Editing lithographicerrors with nanoporous polymeric probesPradeep Ramiah Rajasekaran,1* Chuanhong Zhou,1† Mallika Dasari,1 Kay-Obbe Voss,2

Christina Trautmann,2,3 Punit Kohli1‡

A new lithographic editing system with an ability to erase and rectify errors in microscale with real-time opticalfeedback is demonstrated. The erasing probe is a conically shaped hydrogel (tip size, ca. 500 nm) template-synthesizedfrom track-etched conical glass wafers. The “nanosponge” hydrogel probe “erases” patterns by hydrating andabsorbing molecules into a porous hydrogel matrix via diffusion analogous to a wet sponge. The presence of aninterfacial 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 highbecause of the large free volume of the hydrogel matrix. The fast frictionless translocation and interfacial hy-dration resulted in an extremely high erasing rate (~785 mm2/s), which is two to three orders of magnitudehigher in comparison with the atomic force microscopy–based erasing (~0.1 mm2/s) experiments. The highprecision and accuracy of the polymeric lithography editor (PLE) system stemmed from coupling piezoelectricactuators to an inverted optical microscope. Subsequently after erasing the patterns using agarose erasers, apolydimethylsiloxane probe fabricated from the same conical track-etched template was used to precisely re-deposit molecules of interest at the erased spots. PLE also provides a continuous optical feedback throughoutthe entire molecular editing process—writing, erasing, and rewriting. To demonstrate its potential in devicefabrication, we used PLE to electrochemically erase metallic copper thin film, forming an interdigitated arrayof microelectrodes for the fabrication of a functional microphotodetector device. High-throughput dot and lineerasing, writing with the conical “wet nanosponge,” and continuous optical feedback make PLE complementaryto the existing catalog of nanolithographic/microlithographic and three-dimensional printing techniques. Thisnew PLE technique will potentially open up many new and exciting avenues in lithography, which remain un-explored due to the inherent limitations in error rectification capabilities of the existing lithographic techniques.

.scie

ugust 28, 2020ncem

ag.org/

INTRODUCTIONErrors are omnipresent in nature (1). Some naturally occurring errorsrectified and/or amplified over time positively influence the system, asinmutation-led evolution (2). On the contrary, the errors produced in asynthetic fabrication process can propagate and proliferate with timerather than self-rectify, devouring valuable time and resources (3).Therefore, error rectification is of utmost importance for fabricationprocesses.

Among the existing fabrication techniques, lithography (4) is a ne-oteric fabrication technology that forms the cornerstone of biotechnology(5), electronics (6), and the semiconductor industry (7). Commonly en-countered errors in standard lithographic techniques include maskcritical dimension error (8) and exposure system aberration (9) in photo-lithography, and stitching (10) and overlay errors (11) in e-beam (elec-tron beam) lithography; e-jet (electrohydrodynamic jet) (12), dip-pennanolithography (DPN) (13), and DPN variants can suffer from errorscommon to any conventional printing process.

Whereas a number of commercially successful techniques are avail-able for lithography and fabrication, only a handful of techniques areavailable for error correction. A few notable error-correcting tech-niques 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 abra-sion methods can incur considerable friction and may also lack contin-uous feedback for error detection and rectification (14–18). With theexisting business model and paucity of editing techniques, discardingdefective products is preferred to rectifying the errors in the fabricationprocess.

The global annual revenue for semiconductor industries reached$260 billion (19). A large number of these industries still use theconventional mask-based wet etching methods for fabrication with aprocess yield of ~80% (20). This relatively modest device fabricationyield can generate significant material waste and contribute to loss ofproductivity. With rapidly growing lithography-based industries, thereis a huge market to be conquered in post-fabrication error correction.The error-rectifying polymeric lithography editor (PLE) system intro-duced here demonstrates the feasibility of erasing and rectifying post-fabrication errors in lithography. This technique will be potentially ofgreat interest to the burgeoning semiconductor, biotechnology, mi-croelectromechanical 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 nov-elty of this newPLE systemare the “real erasing” of the deposited patternsfollowed by deposition at the erased spots. The erasing mechanism inPLE is analogous to cleaning a surface with a “wet sponge” but at the“micrometer” scale. PLE also boasts continuous optical feedback for highprecision and accuracy during erasing and deposition.

1 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

We have also used PLE to demonstrate fabrication of a working de-vice. PLE was used to fabricate a pair of interdigitated copper electrodesfor a functional microphotodetector device. This was achieved byelectrochemically erasing/etching a 200-nm copper film by translocat-ing an Fe3+-encapsulated PLE over the copper film that yielded an in-terdigitated electrode pair. In principle, PLE has the potential to evolveinto a “three-dimensional (3D) etcher” analogous to a 3Dprinter. This canbe accomplished by attaching a viscoelastic conical hydrogel or a porouspolymeric matrix probe to a commercial 3D printer’s tool head and repla-cing the printing ink reservoir with a reservoir containing an etching agent(25). Thus, PLE can potentially open upmany new and exciting avenues inlithography, which are currently unexplored due to the inherent limita-tions in erasing capabilities of the current lithographic techniques.

http://advanD

ownloaded from

RESULTS AND DISCUSSIONSynthesis and characterization of hydrogel PLE probesfor erasingPLE agarose erasers were template-synthesized from track-etched con-ical pores present in glasswafers (Fig. 1A).Agarose is a thermoreversiblehydrogel that is widely used in life sciences and biotechnology (26, 27)because of its biocompatibility (28) and highly porous nature (29). Po-rous agarose (5%w/w) canhold >95%of itsweight inwater (30) and hasan extremely large hydrated surface area of ~271m2 g−1, with an averagepore size of 55 nm (31). The aqueous agarose PLE in the hydrated formwas not imagedwith a scanning electronmicroscope (SEM) at high res-olution, but a polydimethylsiloxane (PDMS) PLE synthesized from the

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

on August 28, 2020

ces.sciencemag.org/

same template was imaged at high resolution using the SEM (Fig. 1B).The PDMS PLE consisted of dt = 500 nm, db = 22 mm, and h = 33 mm,where dt, db, and h are the average tip diameter, base diameter, andheight, respectively (Fig. 1B). The agarose PLE is expected to have thesame geometry as that of the PDMS PLE. The cross section of the lyo-philized agarose PLE revealed a dense population of nanopores in thesub–100 nm range (Fig. 1, C and D). The hydrated nanoporous natureof agarose imparted a “wet nanosponge”–like characteristic to the agar-ose PLE eraser. This facilitated wetting of the patterned molecules onthe surface and their transport into nanoporous PLE tip during erasing.

Agarose PLE was synthesized using gelation of agarose followedby its aging. This is because the aged PLE yielded high quality andlarge capacity molecular erasing over multiple erasing cycles with-out any significant decrease in erasing performance. Therefore, ge-lation and aging of agarose warrant a brief discussion here. Thegelation of agarose is a series of phase transitions that starts outkinetically controlled and culminates in a thermodynamically stablelow-energy state (32). Dynamic light scattering has been often usedto investigate the gelation mechanism (33). Gelation is widely acceptedto proceed via liquid-liquid phase separation (34). As the temperatureof the solution approaches the gelation point, the Brownian motion ofthe polymer chains slows down, causing them to entangle. These en-tanglement points act as “nuclei” or “seeds” to recruit more fibers fromthe sol state to the gel state (32). As gelation progresses, the polymericchains eventually segregate into polymer-rich gel phase and polymer-depleted sol phase, yielding water-filled pores in agarose hydrogel. Dur-ing gelation, the correlation length between the polymer chainwas foundto increase from 40 to 400 nm [0.05 weight % (wt %) agarose] (32). Thepore size is directly related to the correlation length of the polymerchains. With gelation time, the local coagulation of the polymer-richareas progresses until they reach an equilibrium, thereby forming stablepores in the matrix. Thus, a larger number of polymer chains in highagarose wt% solution densely pack into smaller pores (1.5% agarose hasa mean pore size of 129 nm), whereas a smaller number of polymerchains in low agarosewt% solution sparsely arrange themselves to formlarge pores (0.5% agarose has a mean pore size of 226 nm) (32). Duringthe overnight pseudoequilibration stage or “aging,” the mean correla-tion length does not undergo any significant change; however, the re-maining free-floating chains pack themselves to their lowest local energyconfigurations (32). The organization and packing of the polymer chainsduring the aging step stabilize the porous network. In our experiments,the PLE erasers were allowed to equilibrate or “age” at 4°C in waterovernight, thereby stabilizing the porous network in the PLE tips. Webelieve that the aging of the agarose is the key for highly efficient mo-lecular erasing and editing using agarose PLE (fig. S1).

Mechanism of erasingThe agarose hydrogel PLE was mounted to an XYZ stage coupled to aninverted optical microscope (Fig. 2A). The humidity was maintained at55 ± 3% RH (relative humidity) using a humidifier and a hygrometer.The RH near the substrate-PLE interface increased exponentially asPLE agarose approached the substrate (fig. S2). High humidity at thePLE tip–substrate interface resulted in the formation of a water menis-cus upon contact, and the molecules present on the substrate werehydrated within the water meniscus volume. The solvated moleculeswere transported into the nanoporous agarose matrix via the water me-niscus at the PLE-surface interface.Humidity appears to have a complexrelationship with “erasing contact time” and erased pattern diameter.This is because of the presence of two independent sources of humidity

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

2 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from

in the erasing experiments: (i) ambient humidity and (ii) humidity fromthe hydrogel agarose PLE itself. This complex interplay between inter-facial humidity and agarose drying and shrinkage is likely to affect thepattern diameter and contact time when erasing experiments are per-formed for long durations. Overall, we believe that to fully understandthe effect of ambient humidity on the PLE-based editing, more thor-ough studies in precisely controlled environments are required (seethe Supplementary Materials for more details).

The diffusion mechanism of PLE molecular erasing was confirmedby the linear de-

ffiffiffiffite

pdependence (Fig. 2B), where de and te represent the

pattern diameter and erasing time, respectively. This erasing mecha-nism is analogous to the diffusion-based deposition mechanism ofthe DPN via the water meniscus (13). Lyophilized (dehydrated) agaroseshowed negligible or no erasing, suggesting that water plays crucial rolesboth in the dissolution of the adsorbed ink molecules in the watermeniscus and molecular diffusion into the porous agarose matrix ofthe PLE eraser. Erasing efficiency depends both on the aging of theagarose (fig. S1) and on the hydrodynamic radius of the erased mole-cules that diffuse through the PLEmatrix (35). Figure 2C schematical-ly shows molecular erasing via the water meniscus formed at thesubstrate-hydrogel interface.

Figure 2D shows a 6 × 6 dot array erased by the agarose PLE,orchestrated by contacting the PLEwith a fluorescein-coated glasswaferat 36 points. The agarose PLE erasers exhibited an erasing efficiency (Er)

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

of ~0.85 over 36 erasing spots with de ~11.7 ± 0.7 mm (te = 1.8 s) (Fig.

2D, fig. S1, and movie S1). We defineEr ¼ 1� IfIi

� �, where Ii and If are

the initial and final integrated emission intensities of the erased spots,respectively. The spectroscopic analysis of fluorescein moleculesextracted from the agarose erasers suggested that ~100 pmol was re-moved after 36 sequential erasing cycles of de ~11 mmspots. This yieldedanaverage areal erasingof~36 fmol/mm2of fluorescein inour experiments.

The presence of an interfacial water layer between the PLE and thepattern played an important role by decreasing the friction between theagarose PLE and the substrate. The interfacial water layer also alloweddeposition of continuous lines (Fig. 2E). This slidingmechanism is anal-ogous to skiing on ice, where the interfacial liquid water lubricates themovement of a ski on the snow (36). The analogous sliding of the agar-ose PLE over a surface (Fig. 2F) allowed continuous molecular deposi-tion with speed in the range of 10 mm/s over >600-mm writing lengthscale (Fig. 2E). The writing speed >10 mm/s was not tested in ourexperiments but is plausible in our experiments. Such fast depositionwas not possible with the PDMS PLE because of the lack of lubricationat the PLE-substrate interface, which in turn generated a stick-slip fric-tion (37) at the interface during PLE translocation.

To further understand the erasing mechanism for the agarose PLE,we used amacroscalemodel (dt = 1mm, db = 7mm, and h = 20mm) ofthe agarose PLE eraser to optically investigate the diffusion mechanism

Fig. 2. (A) A schematic of the custom-built PLE system with an XYZ stage coupled to an inverted optical microscope. (B) Linear de-ffiffit

pe 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 withthe 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 erasedpatterns was 11.7 ± 0.7 mm. (E) Deposition with agarose PLE—a continuous line deposition, with agarose PLE equilibrated in a 2% fluorescein solution. Scale bars, 30 mm(D and E). (F) A schematic showing the line pattern deposition of fluorescein using agarose encapsulated with fluorescein. Both the deposition and erasing occurred viathe water meniscus formed upon contact between the conical hydrogel tip and the substrate.

3 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from

during the erasing process (figs. S3 and S4). By analyzing the time-lapsesnapshots of the erasing by the macroscale agarose PLE (movie S2), weconfirmed that the erasing proceeds through the following three keysteps: (i) rehydration of the deposited molecules upon tip approach(time of 1 to 4 s inmovie S2), (ii) diffusion of the solvated inkmoleculesfrom the wet surface to the agarose macro-PLE tip upon contact (timeof 5 to 8 s inmovie S2), and (iii) themigration of the absorbedmoleculesfrom the tip into the agarose matrix. This was observed in theexperiments through fading of the blue color at the contact spot (time9 to 16 s in movie S2). The rapid diffusion of the absorbed moleculesaway from the tip into the matrix maintained a considerably large con-centration gradient between the tip and the substrate, which supportedthe constant Er (Fig. 2D and fig. S1). The results of the macroscale PLEmodel (figs. S3 and S4) corroborate the erasing phenomenon in themi-croscale PLE probe (Fig. 2B). Although some forms of probe-basederasing were previously reported in the literature (14–18), we believethat this is the first demonstration of “real/conventional” erasing usinga submicrometer nanosponge probe at the micrometer scale withoututilization of an externally applied stimulus. The real erasing hereimplies that the ink molecules are erased or transported into the matrixof the eraser.

Parameters influencing capacity, efficiency, and speed ofPLE erasingThe large and undiminished erasing efficiency observed in our ex-periments becomes apparent after a simple numerical analysis. Wedefined Vr = Vf/Ve, where Vf is the total available free PLE volumeand Ve is the volume of the fluorescein molecules picked up by thePLE in each erasing cycle. Thus, Vf is composed of the sum of conicalPLE eraser volume (VPLE) and the supporting agarose base volume(Vbase), which is 5 mm × 5 mm × 5 mm in dimension. In our case,Vr ~10

10 suggested that Vf ≫ Ve (the fluorescein film thickness wasassumed as ~10 nm). Even without a supporting base, Vr ~67,000provided a constant Er ~0.85 for 36 erasing cycles. If we assume thatthe whole free volume of the agarose PLE and the supporting agarosebase can be completely used to pack the erasedmolecules in the matrix,we estimate that a single PLE eraser would erase a 100-nm-thick fluo-rescein (or similar molecules) film of an area >1 m2. However, theerasing of this large area would affect Er and would also require manydays to achieve this type of erasing using a single PLE eraser underpassive erasing conditions.

In addition to the volume factor involved in erasing, the time lag(t) between each erasing cycle also contributed to the observed higherasing 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 esti-mated distance (d) travelled by the “picked-up” ink molecules from the tipof the PLE eraser into the matrix in t = 10 s was ~70 mm (d ≈

ffiffiffiffiffiffiffiffi2Dt

p).

That is, the absorbed molecules will travel a distance of ~70 mm fromthe tip into the matrix between consecutive erasing cycles (t ~10 s).This simple analysis suggested that a high concentration gradient be-tween the PLE tip (low ink concentration) and the substrate (high inkconcentration) was maintained during molecular ink erasing. Evenwith a modest t of 100 ms between erasing steps, the molecules willtravel >7 mm, resulting in a large concentration gradient between thesubstrate and the PLE favoring a constant Er.

Er was also found to depend on the agarose concentration and gelsetting time—both of these parameters are related to cross-linking den-sity of the agarose matrix. A freshly prepared agarose gel PLE erasershowed a significant decline in the Er after six erasing cycles (that is,

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

Er ~0; fig. S1). However, a porous hydrated agarose PLE that was agedovernight exhibited a constantly high erasing efficiency (Er ~ 0.85) for≥36 erasing cycles (fig. S1). This decline in erasing efficiency for a fresh-ly prepared agarose PLE is attributed to incomplete gelation and poorlyformed porous hydrogel network. Gelation of agarose is both a time-and temperature-dependent process, where a complete gelation usuallytakes place over several hours at low temperature (33). Agarose PLEerasers gelled and aged overnight are believed to attain a pseudoequilib-rium, where a highly cross-linked density network with stable poresyielded a constant high erasing capacity for the PLE (see text abovefor more information on agarose gelation and its aging).

On the basis of the above-mentioned observations, the erasing effi-ciency for the extremely large area of the porous PLE can be furtherimproved by the following steps: (i) increasing the lag time betweenthe erasing cycles, which will generate a high concentration gradientbetween PLE and the substrate, and (ii) actively driving the erasedmolecules away from the tip through an applied external stimulus.Electric field-induced electrophoresis and/or electroosmotic flowmay allow selective control over molecular transport rate duringthe erasing process through controlled application of electric field.Recently, electric field–induced enhanced transport of the proteinmolecules with nanofountain probes was demonstrated, where themolecular transport rate of the protein molecules was enhanced byup to three orders of magnitude over diffusion-based transport (39).(iii) Last, soaking the erasers in a suitable solvent with a high partitioncoefficient for erased molecules will create “free space” for the incom-ing erased molecules.

Comparing PLE erasing with existing technologiesIt is important to compare agarose PLE with the other erasingexperiments reported in the literature. The areal erasing speed (∂Ae

∂t ) ofthe PLE is extremely large (~785 mm2/s), where Ae = pde

2 is the erasedarea and de is the diameter of the erased spot. That is, a surface area of1mm2 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, inan AFM nanoshaving experiment, with de ~100 nm and a scan rateof 1 mm/s (40), it would take ~115 days to erase or shave an area of1 mm2. The molecular erasing using PLE is real; that is, the PLE ac-tually soaks up molecules off the patterns, allowing erasing of thepatterns. Furthermore, the PLE erasing capacity is many orders ofmagnitude larger than that for the AFM tip–based erasing ex-periments. This is attributed to porosity and size differences betweenagarose and SiOx/SiNx-based probes. Although it may be practicallyunlikely to use an AFM tip to erase an area of 1mm2, it must be takeninto consideration that there is no current technique that can per-form ultrafast and high-quality real erasing at the micrometer scalesimilar to PLE.

Because of the abrasive nature of erasing involved in nanoshavingexperiments, where mechanical friction is needed to remove the mole-cules absorbed or bound to the surface, irreversible damage to the tipand the substrate is a concern. In contrast, the agarose hydrogel PLE cansmoothly translate across the substrate without damage to the surface.The PLE erasing picks up erasedmolecules into itsmatrix, but the AFMshaving may leave behind debris of erased molecules on the substrate,whichmay require post-erasing processes for further substrate cleaning.Another difference between the AFM-based and the PLE-based erasingis the spatial resolution of the removal of the patterned molecules. Thespatial resolution of the AFM-based erasing is two to three orders ofmagnitude better than the PLE erasing, which may be highly beneficial

4 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from

and desirable in many nanotechnological and life and materials scienceapplications.

Spatial resolution of PLE easing: Experiments and modelingWe were intrigued by the fact that even though agarose PLE had a tipdiameter of ~500 nm, the smallest de (referred to as de,smallest) obtainedin our experiments was ~10 mm.Under identical conditions, the PDMSPLE tip (dt ~500 nm) generated dot patterns measuring ~600 nm indiameter (fig. S5). That is, the smallest patterns produced by the PDMSPLE tip was only ~20% larger than dt, whereas the agarose tip con-sistently yielded patterns de,smallest≈ 20 × dt. The diameter of the erasedarea depends on three key parameters: the deformation of the tip due toapplied force upon contact with the surface, water meniscus at the PLEtip–surface contact point, andmechanical properties of the tip and surface.Even under very careful maneuvering, de,smallest ~10 mm was the smallestfeature size obtained by our experiments. The de,smallest ≈ 20dt was un-likely due to swelling of agarose because agarose does not swell signif-icantly in water (41).

We argue that significantly larger than expected erasing spotdimensions in our experiments using agarose PLE were due to large di-mensional changes in the agarose PLE tip upon contactwith the surface.The interaction forces involved between the PLE tip and the surface arevan derWaals (vdW), capillary, and electrostatic in nature (see the Sup-plementary Materials). The electrostatic interactions are considered tobe minor because of the lack of charged species in our system. Further-more, the presence of a water layer on the PLE and the substrate willdepolarize any electrical charge present on the surfaces, reducing theelectrostatic contributions to the overall PLE-substrate interaction energy.

The two major interactive forces considered in our system are vdWand capillary forces at the interface (Eqs. 1 and 2). The vdW forces areattractive forces between materials that can arise through dispersionand dipolar-dipolar interactions between atoms or molecules (42). Al-though these forces are known to be weaker than covalent and ionicforces, they are responsible for one of the strongest reversible adhesiveforces observed in nature [for example, the adhesion of gecko setae tovarious surfaces is attributed to vdW interactions (43)]. The vdW forcesbetween the PLE tip and the substrate were estimated using Eq. 1, bymodeling the PLE tip and the substrate as two flat surfaces with wateradsorbed on their surfaces (Fig. 3A). The maximum capillary force wasestimated using Eq. 2, where R, g, q, D, and f are the radius of the PLEtip, 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 Supple-mentary Materials.

F Dð Þ ¼ 16p

A232′

D3�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA121A32′3

p

ðDþ TÞ3 �"

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffip ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffip #

ðDþ T′Þ3 þ A1′2′1′A121

ðDþ T þ T′Þ3 L2 ð1Þ

Fcap ¼ 4pRgD cosq ð2Þ

Here, F(D), D, Aijk, T′, and T represent the force between the PLE tipand the substrate, the PLE-substrate distance, the Hamaker constant,the thickness of the water layer adsorbed on the PLE tip, and the thick-ness of the water layer on the glass substrate, respectively (44). i, j, and k

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

represent differentmaterials in our system (Fig. 3A). For example,A1,2,3

is the nonretarded Hamaker constant for the heterointeractions be-tween 1 and 3 across 2. The subscripts 1, 2, 3, 2′, and 1′ in the Eq. 1 cor-respond to agarose (or PDMS), water, air, water, and fluorescein,respectively. The value of the Hamaker constant for agarose used inour calculationswas assumed to be the same as that for cellulose becauseboth the agarose and the cellulose are chemically very similar to oneanother (45). TheHamaker constants for the PDMS (46) and fluoresceinwere 4.4 × 10−20 and 5 × 10−20 J, respectively. The estimated values ofAijk for the agarosePLE systemwere 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, re-spectively (see the Supplementary Materials). Similarly, the estimatedvalues for A232′, A121, A32′3, A1′2′1′, and A323 for the PDMS PLE systemwere 3.7 × 10−20, 0.031 × 10−20, 3.7 × 10−21, 0.10 × 10−20, and 3.7 × 10−20 J,respectively.

Figure 3B provides some interesting observations for the inter-actions between the PLE and the substrate and also on the interactionenergy dependence on the material properties of PLE and substrate.The PLE composed of agarose appeared to have energetically higherinteractions [more negative E(D) values] with the substrate at all Dthan the PDMS PLE–substrate interactions with the same T andT ′ values. For small D (<0.5 nm), the vdW force interactions dominatedtotal E(D) values, whereas the contribution of the capillary interactionforces to total E(D) was minor (Fig. 3B). However, for D > 5 nm, theinteractions due to the capillary forces were larger than the vdW inter-actions between the agarose PLE tip and the surface. This is evident inan increase in the E(D) at large D (>2 nm) for the agarose PLE (asshown in Fig. 3B). The capillary interactions for the PDMS-basedPLE were insignificant compared to the vdW interactions for D < 2 nmbecause of its hydrophobic nature. Finally, the effect of magnitude ofthe T and T′ values on the overall E(D) was minor as long as there wasa thin layer of water (that is, both T and T′ ≠ 0) present on both theagarose PLE and the fluorescein-coated glass substrate. Because of thehydrophobic nature of PDMS (T′ was assumed as either 0 or 0.2 nm),the capillary force interactions were also negligible for all values ofT ′ and D.

Our analysis also suggested that the PLE and substrate surfacematerial properties are important for the probe-basedmolecular erasingand patterning. For example, the PLE-substrate interactions aredependent on Hamaker constants (for vdW interactions) and watercontact angle (for capillary interactions). Depending on the Hamakerconstants, the vdW forces can be negative, which would affect theE(D)-D dependence (42). The electrostatic interactions for the electri-cally charged surfaces should also be considered in the analysis, whichmay dominate at small D when the ionic strength of the solution at thePLE-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 andthe substrate. With an elastic compression modulus of ~930 kPa for5% (w/w) of agarose (47), we estimate that a relatively large strain (e)of ~17 will be exerted on the agarose PLE tip when it is in close contactwith the surface. The consequence of this large stress at the agarose PLEtip can be a large deformation of the agarose PLE tip—consistent withour experimental results that yielded de,smallest ≈ 20dt.

If we assume that the total interaction energy at the PLE-surfaceinterface is converted into kinetic energy, then the PLE may attain avelocity (v) ≈

ffiffiffiffiffiffiffiffiffiffiE

2m=p

as it approaches the surface. Here,m is the total

5 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from

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 wasalso observed under a microscope during visual examination but wasnot captured by our camera because of the low acquisition rate (5 framesper second) of the CCD (charge-coupled device) camera used in theexperiments. The interplay of the attractive forces between the agarosePLE and the substrate at short distances can satisfactorily explain signif-icantly larger thanexpectedpatterns thatwereobtained inour experiments.

The estimated strain exerted at the PDMS PLE tip because of theattractive forces was ~7.2 at D = 0.5 nm. However, the experimentalobserved minimum deposition spot diameter was comparable to dt.The discrepancy between the experimental de and expected de fromthe estimated strain values on the PDMS PLE tip is believed to bedue to the viscoelastic nature of the PDMS material used for the PLEfabrication (48). The stress generated at the PLE tip because of the con-tact with the substrate appeared to be quickly relieved without affectingthe pattern size dimensions.

Additionally, the largewater content in the agarose PLE and the highhumidity at the interface (fig. S2) create a large water meniscus uponcontact with the substrate. It has been shown by Rozhok et al. (49) thathumidity plays a key role in the size of the patterns deposited. They havedemonstrated that with the same AFM tip dimensions under same ex-perimental conditions, higher-humidity experiments yielded larger pat-tern size (13). Thus, high humidity at the PLE-substrate contact interface(fig. S2) can also contribute to larger than expected erasing patterns inour erasing experiments.

Finally, both the experimental results and calculations further revealthat to improve the pattern size and resolution of the erasing, thematerials and surface properties of the PLE and ink deserve special at-tention. For example, the Young’s modulus, Poisson ratio, and stressrelaxation characteristics of PLE will affect the pattern size and resolu-tion. Probe composed of hard materials (such as SiOx/SiNx with largeYoung’s modulus) and fast stress relaxation properties may providehigher spatial resolution and smaller de. The PLE composed ofmaterialswith a negative Poisson ratio (50) may show a decrease in erasing (anddeposition) pattern size with an increase in the stress on it. There areseveral opportunities where the agarose properties can be tunedthrough simple chemical means (51–53). It is possible to achievesmaller-sized patterns with modified hydrogels, for example, hydrogelPLE with enhanced mechanical modulus (54) and reduced water con-

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

tent along with precisely controlled humidity at the PLE tip–surfaceinterface may allow molecular editing at a higher resolution than dem-onstrated in the present studies.

Editing with PLE probes: “Writing-erasing-rewritingmolecular patterns”To demonstrate molecular editing, erasing must be complementedwith an appropriate deposition technique. PPL-inspired depositionfollowed by “nano–wet sponge erasing” and redeposition of mole-cules at the erased spots was sequentially orchestrated to demon-strate the “molecular editing process” using PLE. Mirkin’s group haspioneered probe-based deposition techniques, such as DPN (13), PPL(22), BPL (23), and HSL (24), with a variety of molecules on manydifferent surfaces at the sub–100 nm scale. The complete editing pro-cess using the PDMS and agarose polymer probes are shown in theform of a fluorescence intensity surface plot in Fig. 4. Initially, a PDMSpen (dt ~500 nm and tc ~15s) was used to deposit a 3 × 3 array offluorescein dots measuring ~15 mm (Fig. 4A). The deposition pattern sizewas monitored and tuned in real time by adjusting the z-piezoelectricstage while the deposition was in progress. The reversibly compress-ible PDMS can be flattened when pressed on the substrate to fabricatethe patterns with a premeditated dimension (55). Although the PDMSpens can be used to deposit submicrometer features (fig. S5), a largepattern size of ~15 mm was chosen to demonstrate pattern editing be-cause of the constraints imposed by the limiting achievable eraser spotdimension with an agarose PLE of ~10 mm. Optically transparent PDMSpens and agarose erasers were attached to the z-piezo stage using a trans-parent glass holder. The transparent window on the z-piezo stage allowedmonitoring of the PLE probe during the editing process. The ability tolock a desired location with high precision and accuracy was importantfor the editing process (registration), because the same location mustbe accessed consecutively for writing, erasing, and rewriting. Theoptical transparency of the agarose PLE, PDMS PLE, and the PLE holdercoupled to an inverted optical microscope made locations of the regis-tration possible. PLE editing can also be accomplished with opaquesubstrates using upright microscopes.

Although both the PDMS and agarose PLE are optically transparent,they appeared opaque under an optical microscope (fig. S7). Thisintriguing phenomenon was analyzed using a custom finite-differencetime-domain (FDTD) code (56, 57). Our simulations indicated that,analogous to a prism, the angled cone walls refracted the incoming light

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 coveredwith 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 ofthe 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.

6 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from

away from the straight path that yielded the transparent PLE micro-structures appearing as dark circles in the optical micrographs (fig.S7A). ThePLE lost the prismeffect at the PLE-substrate interface, whichallowed optical flux through the flattened part of the PLE tip generatinga dark “doughnut”-shaped feature surrounded by a bright central part(fig. S7B). This clear doughnut-shaped feature was used to locate thecontact point (registration) and to control the contact area of the PLEwith the substrate during the editing process.

After depositing a 3 × 3 array with a conical PDMS pen, an agarosePLE eraser attached to the z-piezo stage was translated and positionedsuch that the eraser was partially above the first deposition in the array.After locking the xy coordinates of the patterns, the PLE eraser wastranslated vertically down, thereby making contact with the molecularpattern for te = 3 ± 1 s. A partial erasing of the patterned moleculesalong a diagonal direction was demonstrated (Fig. 4, B to D). The con-tact point and contact area for PLE were tuned and adjusted by manip-ulating the position and dimension of the optical doughnut-shapedPLE. The patterns along the diagonal were chosen for partial erasingto demonstrate the “precision and confined registration” nature ofthe agarose PLE system. The precision and accuracy of the registrationfor writing and erasing depend only on the precision and accuracy ofthe piezoelectric stage used for the translocation of the PLE. In ourexperiments, the accuracy (repeatability) of our piezo stage was 150nm. This allows accurate placement of the PLE eraser (and pen)within ±150 nm of a given registration spot. After demonstrating acontrolled partial erasing along the diagonal, complete removal of thediagonal patterns was demonstrated by moving the eraser along the di-agonal twice in contact mode (te = 15 s) (Fig. 4E).

To complete the molecular editing process, fluorescein moleculeswere precisely redeposited at the erased spots having similar dimensions

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

to those of the erased patterns. To accomplish redeposition at the erasedspots, a PDMS PLE pen immersed in a 10 mM fluorescein solution for60 s and dried underN2 gas streamwas used. The optical doughnut wasused to manipulate the size of the deposited pattern. The emission in-tensity of newly deposited molecules along the diagonal in Fig. 4F wascomparable to the original unexposed dot patterns in Fig. 4A, demon-strating reproducibility of the PLE system. Figure 4G shows a schematicof the partial erasing using an agarose PLE.

The fluorescence emission intensity of the nonerased patterns wasfound to reduce by ~50% in comparison to their original emission in-tensity (Fig. 4, A and F). This decrease in the emission intensity was dueto photobleaching of the fluorescein molecules during registration,PLE-surface contact alignment, and image acquisition processes. Thephotobleaching damage to nonerased patterns can be reduced byminimizing the photoexposure of the dye molecules during the erasingprocess. This can be achieved by reducing the photoexcitation apertureand intensity through automation of the erasing process. The above-mentioned three series of experiments demonstrated the editingprocess—“writing-erasing-rewriting” at themicrometer scale with highprecision and accuracy using our PLE system. Another important fea-ture of the PDMS PLE is that the same PDMS pen was also usedmultiple times (>100 times) for writing experiments without any signif-icant observed deterioration in its performance. With automation andhigh precision tools, this simple proof-of-concept demonstration couldbe translated into a lithographic system for editing and rectification oferrors at the submicrometer scale.

Device fabrication using PLE and potentialapplications of PLEIt is also possible to rectify photolithographic errors during the devel-oping and etchingprocesseswithPLE.Many commonphotolithographicdevelopers used in semiconductor processing [for example, RD6 (58),Microposit 351 (59)] are aqueous in nature. An error that occurred duringa development process in photolithography can be rectified by (i) raster-ing a developer-loaded agarose PLE over an erroneous spot and (ii) afollow-up photoresist deposition over the erased spot by a photoresist-loaded hydrophobic PLE. Fabrication of porous hydrophobic nano-structures (for example, composed of nanoporous PDMS) has beendemonstrated in literature (60). This process using PLE may allowrectification of photolithographic errors at the micrometer scale.

Furthermore, many common etchants used in the semiconductorindustry for etching Si, SiNx, and SiO2 are aqueous in nature (for ex-ample, KOH and HF). Therefore, etchant-loaded PLE can be used forlocalized etching of semiconductors. Agarose encapsulated with ap-propriate molecular etchants are shown to etch metal oxides andsemiconductors (61). Thus, tailoring the PLE polymeric materialsand loading it with appropriate molecules may allow PLE to play a cru-cial role in the semiconductor industry.

We have demonstrated editing of the dot pattern array and the dep-osition of line patterns using the agarose PLE.We now demonstrate aninteresting application of PLE as a “microelectrochemical translating re-actor” toward device fabrication. A porous agarose hydrogel matrix en-capsulating a 2% FeCl3 was translocated on a copper-coated glasssubstrate to electrochemically etch metallic copper (~200 nm thick) forfabricating an array of interdigitated electrodes. Briefly, the Fe3+(aq)–encapsulated agarose PLE tip was rastered at a speed of ~10 mm/s whilein contact with the copper surface in a zigzag pattern. The encapsulatedFe3+(aq) ions diffused out of the PLE and electrochemically oxidized themetallic copper Cu0(s) to Cu2+(aq) (Eq. 3). The aqueous solvated

Fig. 4. “Editing”—Write-erase-rewrite using PLE. (A) A 3 × 3 fluorescent arraydeposited using a PDMS PLE pen (dt = 500 nm, db = 22 mm, and h = 38 mm). (B toD) Partial consecutive erasing of patterns along the diagonal. (E) Completeerasing of the fluorescent dots along the diagonal. (F) Demonstration of the rec-tification process of the patterns. AU, arbitrary units. (G) A schematic of the agar-ose PLE erasing mechanism showing partial erasing. The optical micrographs ofthe PDMS PLE pen (H) and agarose PLE eraser (I) in contact with the surface, asseen through the microscope. Scale bars, 25 mm (H and I).

7 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on August 2

http://advances.sciencemag.org/

Dow

nloaded from

products, Fe2+ and Cu2+, were absorbed back into the porous agarosematrix or were washed away with water.

Fe3þ þ Cu0 → Fe2þ þ Cu2þ ðEocell ¼ 0:43VÞ ð3Þ

Figure 5A shows an optical micrograph of the resultant interdigitatedcopper electrode fabricated using the “microelectrochemical translator”PLE probe. The average width of the finger electrodes was 60 ± 9 mm(dark area) with an interelectrode separation of 25 ± 9 mm (nonconduc-tive, light area in Fig. 5A). The electrical resistance of >1 gigaohm acrossthe fabricated interdigitated electrode confirmed successful erasing of theCu0 by the PLE.We estimated combined erasing capacity of an agarosePLE (dt = 500 nm, db = 22 mm, and h = 38 mm) and the supportingagarose base (5mm×5mm×5mm) loadedwith 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 erasingcapacity will be significantly lower than the theoretical estimated ca-pacity because of kinetics andmass transport processes involved in theetching process.

It is also possible to electrochemically redeposit or rewrite metalsusing hydrogels. The encapsulation and deposition of silver (62) viaelectrochemical methods using highly porous and aqueous agaroseare demonstrated in the literature. An AFM-DPN tool was previouslyused to electrochemically deposit metallic platinum (63). Further-more, agarose can be made ionically conductive hydrogel which willallow metallic deposition or erasing using PLE by modulating thedirectional molecular transport through electrophoresis (64). More-over, more opportunities are also possible wherein PLE can be custo-mized to deposit organic and inorganic materials (65) and to erasemetal 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 (totalarea, ~15 mm2). The photoinduced current-time measurements wereacquired by switching the light “ON” and “OFF” with a time interval(ts) of 1 s (Fig. 5B). The excitation was achieved with a broadband light

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

source with an intensity of ~130 mW/m2 at a distance of 0.6 m. Thephotoinduced current (Ip) closely followed the light ON and OFF pat-tern. The dark current (Ip,dark) was <0.1 nA, which is below the detec-tion limit of our instrument under experimental conditions. For ts =1 s, Ip was ~3.3 and 2.7 mA with the lights ON and OFF, respectively, atan applied potential of 5 V (Fig. 5B). The device remained functionalfor >1 week under wet laboratory conditions without significant loss ofits functionality. The photoactive device fabrication using the PLE sys-tem demonstrates the utilization of PLE in the device fabrication,which complements many currently available fabrication tools.

CONCLUSIONIn conclusion, we have developed a new probe-based erasing systemthat can perform “erasing” with high precision and accuracy. The“nanosponge”-like hydrogel probe hydrates the molecules depositedon the surface and erases them via diffusion. The erasing efficiency ishigh because the free volume available in the agarose eraser is abundantin comparison to the volume of the erased molecules. The interfacialliquid layer between the tip and the substrate circumvents the stick-slipfriction and enables frictionless continuous deposition and erasing. Bymounting PLE probes on piezoelectricmotors coupled to an opticalmi-croscope, the entire process was monitored and manipulated continu-ously. Combining erasing with the deposition, we demonstrated theediting process: writing-erasing-rewriting at the micrometer scale byusing our PLE system. To further demonstrate its versatility, we usedPLE as a “microelectrochemical reactor and translocator” device by en-capsulating redox species in the PLE hydrogel matrix. An interdigitatedarray of microelectrodes for the fabrication of microphotodetector de-vices was accomplished using the agarose PLE. The soft polymericprobes can aid with the deposition and etching at the submicrometerand micrometer scales with high accuracy and precision withoutdamaging the substrate. The porous matrix can also be used to encap-sulate, transport, and deliver awide variety of chemical and biochemicalmolecules for the fabrication of functional devices.With an appropriateadditive to the hydrogel or by using a porous hydrophobic polymer, thePLE can even be coupled to the lithographic process in the semi-conductor industrywhere organic solvents are used. Thus, the PLE system

8, 2020

Fig. 5. (A) An interdigitated copper electrode fabricated by redox reaction between the Cu0- and Fe3+-containing agarose PLE erasers. Scale bar, 150 mm. (B) A typicalphotoinduced current-time response for a device shown in (A). Black and yellow stars represent light turned OFF and ON, respectively.

8 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

presented here complements the existing catalog of nanolithographic/microlithographic and 3D printing techniques for the fabrication ofelectronic, biological, optical, and micro- and nanodevices. PLE willoffer exciting opportunities for new ventures in lithography that arecurrently unexplored because of the inherent limitations in the error-rectifying capabilities of the existing lithographic techniques.

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from

MATERIALS AND METHODSChemicalsThe glass substrates were purchased from SCHOTT AG. PDMSSYLGARD 184 elastomer kit was purchased from Dow Corning.P3HT and phenyl-C61-butyric acid methyl ester (PCBM) were pur-chased fromSigma-Aldrich and Fisher Scientific, respectively. All oth-er reagents (agarose, fluorescein, hydrofluoric acid, calcium nitrate,H2SO4, H2O2, FeCl3, and 1,2-dichlorobenzene) and supplies were pur-chased from Fisher Scientific and were used as received unless specifiedin the text. Tri-decafluoro-1,1,2,2-tetrahydrooctyl)triethoxy-silane(DTTS) was obtained from Gelest Inc.

Characterization and instrumentationElectron micrographs were taken with FEI Quanta FEG 450. AFMmeasurements were carried out in tapping mode on a TopoMetrixExplorer using silicon tips. NTS10 piezoelectric stages from DiscoveryTechnology International mounted on an inverted microscope wereused to manipulate and translate PLE in the z axis. Optical images (bothfluorescent and bright-field) were obtained using an inverted DM IRBLeica microscope. The fluorescence spectroscopy measurements wereperformed on a PerkinElmer LS 55 spectrometer. The photovoltaic de-vice characterization was carried out by using an AM1.5 solar simulator(Newport Inc.) with a 150-W xenon arc lamp. A Melles Griot photo-detector was used to measure the photoinduced response of the PLE-fabricated devices. Photoinduced current-time curves were measuredusing Keithley 6487 at an external bias of 5 V.

Ion tracking of glass substratesThe PLE pens and erasers used in our studies were fabricated throughtemplate-assisted synthesis (66, 67). The pores in the glass substrateswere fabricated through a track-etched process in which 2.2-GeV Auions were irradiated onto the glass slides at the linear accelerator inGSIHelmholtzzentrum,Darmstadt, Germany. The tracking in the glasssubstrates was performed in a 70-mm-thick SCHOTT D 263T boro-silicate glass substrate. The kinetic energy of a given irradiating ionwas 2.2 GeV. With averaged electronic energy loss of ~25 keV/nm,the ions fully penetrated 70-mm-thick glass samples (68).

Etching of patterned and randomly tracked substratesOur previous experiments showed that the average etching rate of thetracked 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 re-quired for a complete etching) to avoid a complete etching of the sub-strates. A partial etching of the tracked substrates provided sharp(pointed) pores for PLE pens/erasers fabrication.

Fabrication of PLE pens and erasersThe etched glass chips were first thoroughly cleaned with piranha solu-tion (3:1 H2SO4/H2O2) and were then fluorinated with a monolayer ofDTTS for easy lift-off. DTTS functionalization of the TGwas performedby immersing the chips in a solution of 4% (v/v) DTTS, 1% (v/v) acetic

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

acid, and 95% (v/v) ethanol for 1 hour. The TGswere rinsed thoroughlyin ethanol and deionized water and were baked at 90°C for 5 hours.Following the silane treatment, PDMS (or agarose) was poured overTG, filling the conical pores with polymer. The PLE-containing poly-mer layer was peeled off the TG after it was cured.

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

Encapsulation of Fe(III) in PLEA 2% solution of FeCl3 with 0.1 M HCl was made in deionized water.Preformed agarose PLE erasers equilibrated in deionized water over-night were immersed into a petri dish containing the FeCl3 solution for30 min. FeCl3 diffused into water-filled pores within the agarose matrix,changing its color from whitish to yellowish. The Fe(III)-impregnatedPLE agarose eraser was carefully and gently rinsed with water. It wasair-dried for ~10 min to remove excess water before erasing.

LyophilizationPrefabricated agarose gel PLE eraser cut into pieces measuring 1 cm2

was immersed in 20 ml of water in a 50-ml corning centrifuge tubesand frozen in liquid nitrogen. Upon freezing, the lids of the tubes werepunctured, and the samples were placed in a lyophilizer until all the wa-ter molecules were completely evaporated, leaving behind a free-standing solid agarose framework. The SEM of the cross section ofthe sample was analyzed with ImageJ.

FDTD simulationsTo gain a better understanding of optically opaque to optically cleartransition upon contact with the surface, FDTD simulations involvinglight propagation through PLE pens were performed. Briefly, a home-written code based on solving Maxwell’s electrodynamic equations wasused for FDTD simulations. More details on FDTD simulations can beobtained from our previous publication (69).

Fabrication of the photoinduced detector deviceA mixture of P3HT and PCBM (20 mg each) was dissolved in 1 ml of1,2-dichlorobenzene. The solution was heated up to 60°C to speed upthe dissolution ofmaterials in the solvent. Themixturewas cooled downto room temperature and was centrifuged at 2000 rpm for 5 min to re-move 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 excitationwas a broadband sourcewith an intensity of ~130mW/m2

at a distance of 0.6m. The photoinduced current (Ip) wasmeasured usinga Keithley 6487 picoammeter with photoexcitation ON and OFF.

Erasing capacity of a PLE microelectrochemicaltranslating reactorAgarose PLEwith a db = 22 mmand h=38 mmcanhold ~4.8 × 10−6ml ofwater (~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 ofcopper can be erased using all the Fe3+ present in the PLE pen. With a

9 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

coppermetal density of 8.96 g/cm3, this corresponds to a copper volumeof 6.8 × 10−10 cm3 or 42 × 1011 nm3. Therefore, this PLE microelectro-chemical translating reactor can erase a copper film of 100 nm of an areaof ~6.8 × 109 nm2. Total area would be 0.0068mm2. However, with a baseof volume (5 × 5 × 5mm3) of 125mm3, ~0.25 g of Fe or 3.5 × 10−3mol ofcopper 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 of100 nm of an area of ~2.5 × 1017 nm2 = 1016 nm2 = 2500 cm2.

http://advances.sD

ownloaded from

SUPPLEMENTARY MATERIALSSupplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/3/6/e1602071/DC1fig. 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 thesurface.fig. S7. FDTD simulation and microscopic visualization of a PLE conical pen before and aftercontact.RH–versus–PLE substrate gap dependenceRole of humidity in writing and erasingMechanism of erasingNature of force between the PLE and the substrateCapillary interactions for the PLE-surface pairFDTD analysismovie S1. This video was captured through the eyepiece of an inverted microscope whileerasing the 6 × 6 array (Fig. 2D).movie S2. This video was captured through the eyepiece of an optical microscope underbright-field conditions.

on August 28, 2020

ciencemag.org/

REFERENCES AND NOTES1. D. Allchin, “To err is science,” paper presented at the American Association for the

Advancement of Science Meetings, Washington, DC, 20 February 2000).2. A. Stoltzfus, L. Y. Yampolsky, Climbing mount probable: Mutation as a cause of

nonrandomness in evolution. J. Hered. 100, 637–647 (2009).3. H. Akhavan, M. El-Beheiry, O. Levi, Nano-fabrication dependent quality factor in photonic

crystal slab biosensors, in 2010 International Symposium on OptomechatronicTechnologies (ISOT’10), Toronto, Ontario, 25 to 27 October 2010.

4. A. Piqué, D. Chrisey, Direct-Write Technologies for Rapid Prototyping Applications(Academic Press, 2002).

5. K. Lee, H. Jung, Drawing lithography for microneedles: A review of fundamentals andbiomedical applications. Biomaterials 33, 7309 (2012).

6. M. A. Reed, C. Zhou, C. J. Muller, T. P. Burgin, J. M. Tour, Conductance of a molecularjunction. Science 278, 252–254 (1997).

7. R. Held, T. Heinzel, P. Studerus, K. Ensslin, M. Holland, Semiconductor quantum pointcontact fabricated by lithography with an atomic force microscope. Appl. Phys. Lett.71, 2689 (1997).

8. A. K. Wong, Lithographic effects of mask critical dimension error, in 23rd AnnualInternational Symposium on Microlithography (International Society for Optics andPhotonics, 1998).

9. C. J. Progler, D. C. Wheeler, Optical lens specifications from the user’s perspective.Proc. SPIE 3334, 256–268 (1998).

10. D. J. Dougherty, R. E. Muller, P. D. Maker, S. Forouhar, Stitching-error reduction in gratings byshot-shifted electron-beam lithography. J. Lightwave Technol. 19, 1527–1531 (2001).

11. P. B. Fischer, S. Y. Chou, 10 nm electron beam lithography and sub‐50 nm overlay using amodified scanning electron microscope. Appl. Phys. Lett. 62, 2989–2991 (1993).

12. B. H. Kim, M. S. Onses, J. B. Lim, S. Nam, N. Oh, H. Kim, K. J. Yu, J. W. Lee, J.-H. Kim,S.-K. Kang, C. H. Lee, J. Lee, J. H. Shin, N. H. Kim, C. Leal, M. Shim, J. A. Rogers, High-resolution patterns of quantum dots formed by electrohydrodynamic jet printingfor light-emitting diodes. Nano Lett. 15, 969–973 (2015).

13. R. D. Piner, J. Zhu, F. Xu, S. Hong, A. Mirkin, “Dip-pen” nanolithography. Science 283,661–663 (1999).

14. N. A. Amro, S. Xu, G.-y. Liu, Patterning surfaces using tip-directed displacement and self-assembly. Langmuir 16, 3006–3009 (2000).

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

15. M. Liu, N. A. Amro, G.–y. Liu, Nanografting for surface physical chemistry. Annu. Rev.Phys. Chem. 59, 367–386 (2008).

16. L. G. Rosa, J. Liang, Atomic force microscope nanolithography: Dip-pen, nanoshaving,nanografting, tapping mode, electrochemical and thermal nanolithography.J. Phys. Condens. Matter 21, 483001 (2009).

17. J.-W. Jang, D. Maspoch, T. Fujigaya, C. A. Mirkin, A “molecular eraser” for dip‐pennanolithography. Small 3, 600–605 (2007).

18. A. Tinazli, J. Piehler, M. Beuttler, R. Guckenberger, B. Tampé, Native proteinnanolithography that can write, read and erase. Nat. Nanotechnol. 2, 220–225 (2007).

19. S. B. Boyd, in Life-Cycle Assessment of Semiconductors (Springer, 2011), pp. 28.20. M. Henry, J. Wendland, P. M. Harrison, D. Hand, “Rapid laser patterning versus

wet-etch lithography for flat panel display manufacture: A technical & commercialcomparison (Technical Report M505, Laser Institute of America, 2007);http://www.powerlase-photonics.com/wp-content/uploads/2011/data-sheets/ICALEO07%20Oct07%20Rapid%20Laser%20Patterning.pdf.

21. A. C. R. Grayson, R. S. Shawgo, A. M. Johnson, N. T. Flynn, Y. Li, M. J. Cima, R. Langer,A BioMEMS review: MEMS technology for physiologically integrated devices.Proc. IEEE 92, 6–21 (2004).

22. F. Huo, Z. Zheng, G. Zheng, L. R. Giam, H. Zhang, C. A. Mirkin, Polymer pen lithography.Science 321, 1658–1660 (2008).

23. F. Huo, G. Zheng, X. Liao, L. R. Giam, J. Chai, X. Chen, W. Shim, C. A. Mirkin, Beam penlithography. Nat. Nanotechnol. 5, 637–640 (2010).

24. W. Shim, A. B. Braunschweig, X. Liao, J. Chai, J. K. Lim, G. Zheng, C. A. Mirkin, Hard-tip, soft-spring lithography. Nature 469, 516–520 (2011).

25. Aleph Objects, “LulzBot TAZ Dual Extruder Tool Head V2” (Aleph Objects, 2016);www.lulzbot.com/store/tool-heads/lulzbot-taz-dual-extruder-tool-head-v2.

26. Y. Luo, M. S. Shoichet, A photolabile hydrogel for guided three-dimensional cell growthand migration. Nat. Mater. 3, 249–253 (2004).

27. A. J. Millán, R. Moreno, M. I. Nieto, Thermogelling polysaccharides for aqueousgelcasting—Part I: A comparative study of gelling additives. J. Eur. Ceram. Soc. 22,2209–2215 (2002).

28. A. J. Senesi, D. I. Rozkiewicz, D. N. Reinhoudt, C. A. Mirkin, Agarose-assisted dip-pennanolithography of oligonucleotides and proteins. ACS Nano 3, 2394–2402 (2009).

29. H. Li, H. L. Xin, D. A. Muller, L. A. Estroff, Visualizing the 3D internal structure of calcitesingle crystals grown in agarose hydrogels. Science 326, 1244–1247 (2009).

30. M. Djabourov, A. H. Clark, D. W. Rowlands, S. B. Ross-Murphy, Small-angle x-ray scatteringcharacterization of agarose sols and gels. Macromolecules 22, 180–188 (1989).

31. J. Zhou, M. Zhou, R. A. Caruso, Agarose template for the fabrication of macroporousmetal oxide structures. Langmuir 22, 3332–3336 (2006).

32. J.-Y. Xiong, J. Narayanan, X.-Y. Liu, T. K. Chong, S. B. Chen, T.-S. Chung, Topology evolutionand gelation mechanism of agarose gel. J. Phys. Chem. B 109, 5638–5643 (2005).

33. J. Narayanan, V. W. Deotare, R. Bandyopadhyay, A. K. Sood, Gelation of aqueous pectinsolutions: A dynamic light scattering study. J. Colloid Interface Sci. 245, 267–273(2002).

34. D. Bulone, D. Giacomazza, V. Martorana, J. Newman, P. L. San Biagio, Ordering of agarosenear the macroscopic gelation point. Phys. Rev. E 69, 041401 (2004).

35. S. V. Lee, A. R. Bahaman, Discriminatory power of agarose gel electrophoresis in DNAfragments analysis. InTech, 10.5772/36891 (2012).

36. B. Glenne, Sliding friction and boundary lubrication of snow. J. Tribol. 109, 614–617(1987).

37. F. Wu-Bavouzet, J. Clain-Burckbuchler, A. Buguin, P.-G. De Gennes, F. Brochard-Wyart,Stick-slip: Wet versus dry. J. Adhes. 83, 761–784 (2007).

38. T. Støren, A. Simonsen, O. J. Løkberg, T. Lindmo, L. O. Svaasand, A. Røyset, Measurementof dye diffusion in agar gel by use of low-coherence interferometry. Opt. Lett. 28,1215–1217 (2003).

39. O. Y. Loh, A. M. Ho, J. E. Rim, P. Kohli, N. A. Patankar, H. D. Espinosa, Electric-field induceddirect delivery of proteins by a nanofountain probe. Proc. Natl. Acad. Sci. U.S.A. 105,16438–16443 (2008).

40. O. El Zubir, I. Barlow, G. J. Leggett, N. H. Williams, Fabrication of molecular nanopatternsat aluminium oxide surfaces by nanoshaving of self-assembled monolayers ofalkylphosphonates. Nanoscale 5, 11125–11131 (2013).

41. A. Hayashi, T. Kanzaki, Swelling of agarose gel and its related changes. Food Hydrocoll. 1,317–325 (1987).

42. V. A. Parsegian, Van der Waals Forces: A Handbook for Biologists, Chemists, Engineers, andPhysicists (Cambridge Univ. Press, 2005).

43. K. Autumn, M. Sitti, Y. A. Liang, A. M. Peattie, W. R. Hansen, S. Sponberg, T. W. Kenny,R. Fearing, J. N. Israelachvili, R. J. Full, Evidence for van der Waals adhesion in gecko setae.Proc. Natl. Acad. Sci. U.S.A. 99, 12252–12256 (2002).

44. J. N. Israelachvilli, Internmolecular and Surface Forces (Elsevier, ed. 3, 2011).

45. F. L. Leite, C. C. Bueno, A. L. Da Róz, E. C. Ziemath, O. N. Oliveira Jr., Theoretical modelsfor surface forces and adhesion and their measurement using atomic forcemicroscopy. Int. J. Mol. Sci. 13, 12773–12856 (2012).

10 of 11

SC I ENCE ADVANCES | R E S EARCH ART I C L E

http://advances.sciencemag.org

Dow

nloaded from

46. C. J. Drummond, D. Y. C. Chan, Van der Waals interaction, surface free energies, andcontact Angles: Dispersive polymers and liquids. Langmuir 13, 3890–3895 (1997).

47. V. Normand, D. L. Lootens, E. Amici, K. P. Plucknett, P. Aymard, New Insight into agarosegel mechanical properties. Biomacromolecules 1, 730–738 (2000).

48. J. Mark, Ed., Polymer Data Handbook (Oxford Univ. Press, 1998) pp. 411–435.49. S. Rozhok, R. Piner, C. A. Mirkin, Dip-pen nanolithography: What controls ink transport?

J. Phys. Chem. B 107, 751–757 (2003).50. R. S. Lakes, Foam structures with a negative Poisson’s ratio. Science 235, 1038–1040

(1987).51. M. Malkoch, R. Vestberg, N. Gupta, L. Mespouille, P. Dubois, A. F. Mason, J. L. Hedrick,

Q. Liao, C. W. Frank, K. Kingsbury, C. J. Hawker, Synthesis of well-defined hydrogelnetworks using click chemistry. Chem. Commun. 26, 2774–2776 (2006).

52. A. M. Kloxin, A. M. Kasko, C. N. Salinas, K. S. Anseth, Photodegradable hydrogels fordynamic tuning of physical and chemical properties. Science 324, 59–63 (2009).

53. J. R. Tse, A. J. Engler, Preparation of hydrogel substrates with tunable mechanicalproperties. Curr. Protoc. Cell Biol., 10.1002/0471143030.cb1016s47 (2010).

54. J. Li, Z. Suo, J. J. Vlassak, Stiff, strong, and tough hydrogels with good chemical stability.J. Mater. Chem. B 2, 6708–6713 (2014).

55. D. J. Eichelsdoerfer, X. Liao, M. D. Cabezas, W. Morris, B. Radha, K. A. Brown, L. R. Giam,A. B. Braunschweig, C. A. Mirkin, Large-area molecular patterning with polymer penlithography. Nat. Protoc. 8, 2548–2560 (2013).

56. A. Taflove, Computational Electrodynamics: The Finite-Difference Time-Domain Method(Artech House, 1995).

57. P. R. Rajasekaran, J. Wolff, C. Zhou, M. Kinsel, C. Trautmann, S. Aouadi, P. Kohli, Twodimensional anisotropic etching in tracked glass. J. Mater. Chem. 19, 8142–8149(2009).

58. Futurrex Inc., “Resist Developers” (Futurrex Inc., 2007); http://futurrex.com/en/products/resist-developers.html (accessed on May 8, 2017).

59. Shipley Company, “MICROPOSIT 351 Developer” (Shipley Company, 2001);http://docs.capitolscientific.com/manufacturer/dow_electronic_materials/datasheets/351_Developer.pdf (accessed on May 8, 2017).

60. K. Jiao, C. L. Graham, J. Wolff, R. G. Iyer, P. Kohli, Modulating molecular and nanoparticletransport in flexible polydimethylsiloxane membranes. J. Memb. Sci. 401–402, 25–32(2012).

61. S. K. Smoukov, B. A. Grzybowski, Maskless microetching of transparent conductiveoxides (ITO and ZnO) and semiconductors (GaAs) based on reaction-diffusion.Chem. Mater. 18, 4722–4723 (2006).

62. U. Hasse, F. Scholz, Tuning the size of silver deposits by templated electrodepositionusing agarose gels. J. Solid State Electrochem. 10, 380–382 (2006).

63. Y. Li, B. W. Maynor, J. Liu, Electrochemical AFM “dip-pen” nanolithography. J. Am.Chem. Soc. 123, 2105–2106 (2001).

Rajasekaran et al., Sci. Adv. 2017;3 : e1602071 9 June 2017

64. D. L. Smisek, D. A. Hoagland, Electrophoresis of flexible macromolecules: Evidence for anew mode of transport in gels. Science 248, 1221–1223 (1990).

65. J. Zhong, G. Sun, D. He, Classic, liquid, and matrix-assisted dip-pen nanolithography formaterials research. Nanoscale 6, 12217–12228 (2014).

66. C. R. Martin, Nanomaterials: A membrane-based synthetic approach. Science 266,1961–1966 (1994).

67. C. R. Martin, Z. Siwy, Molecular filters: Pores within pores. Nat. Mater. 3, 284–285 (2004).68. B. Afra, M. D. Rodriguez, C. Trautmann, O. H. Pakarinen, F. Djurabekova, K. Nordlund,

T. Bierschenk, R. Giulian, M. C. Ridgway, G. Rizza, N. Kirby, M. Toulemonde,P. Kluth, SAXS investigations of the morphology of swift heavy ion tracks in a-quartz.J. Phys. Condens. Matter 25, 045006 (2013).

69. C. Zhou, P. R. Rajasekaran, J. Wolff, X. Li, P. Kohli, Photo-pens: A simple and versatile toolfor maskless photolithography. Langmuir 26, 17726–17732 (2010).

Acknowledgments: We thank V. Kollipara for useful discussions and suggestions. Weacknowledge C. Watts for machining some parts for the fabrication and assembling of thePLE system. We also acknowledge J. Bozzola of the Integrated Microscopy and GraphicsExpertise Center at the Southern Illinois University, Carbondale (SIUC), for his help in theacquisition of the SEM images and their analysis. Funding: We acknowledge partial financialsupport of this work from the NSF (CHE-0748676 and CHE-0959568), the NIH (GM 106364and GM 080711), and the Office of Vice Chancellor of Research at the SIUC. The SEMused 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 thephotoresponse experiments of the working devices fabricated using PLE. C.Z. performedthe 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 theexperiments. P.K., P.R.R., K.-O.V., and C.T. critically read the paper. P.K. and P.R.R. wrote themanuscript. Competing interests: The authors declare that they have no competing interests.Data and materials availability: All data needed to evaluate the conclusions in thepaper are present in the paper and/or the Supplementary Materials. Additional data related tothis paper may be requested from the authors.

Submitted 30 August 2016Accepted 23 March 2017Published 9 June 201710.1126/sciadv.1602071

Citation: P. R. Rajasekaran, C. Zhou, M. Dasari, K.-O. Voss, C. Trautmann, P. Kohli, Polymericlithography editor: Editing lithographic errors with nanoporous polymeric probes. Sci. Adv. 3,e1602071 (2017).

o/

11 of 11

n August 28, 2020

Polymeric lithography editor: Editing lithographic errors with nanoporous polymeric probesPradeep Ramiah Rajasekaran, Chuanhong Zhou, Mallika Dasari, Kay-Obbe Voss, Christina Trautmann and Punit Kohli

DOI: 10.1126/sciadv.1602071 (6), e1602071.3Sci Adv 

ARTICLE TOOLS http://advances.sciencemag.org/content/3/6/e1602071

MATERIALSSUPPLEMENTARY http://advances.sciencemag.org/content/suppl/2017/06/05/3.6.e1602071.DC1

REFERENCES

http://advances.sciencemag.org/content/3/6/e1602071#BIBLThis article cites 54 articles, 10 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.Science AdvancesYork Avenue NW, Washington, DC 20005. The title (ISSN 2375-2548) is published by the American Association for the Advancement of Science, 1200 NewScience Advances

Copyright © 2017, The Authors

on August 28, 2020

http://advances.sciencemag.org/

Dow

nloaded from