Fabrication of microcraters on silicon substrate by UV nanosecond photonic nanojets from microspheres

Long range nanostructuring of silicon surfaces by photonic nanojets from microsphere

Langmuir films

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2013 J. Phys. D: Appl. Phys. 46 145102

(http://iopscience.iop.org/0022-3727/46/14/145102)

Download details:

IP Address: 139.124.58.70

The article was downloaded on 12/03/2013 at 13:34

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 46 (2013) 145102 (9pp) doi:10.1088/0022-3727/46/14/145102

Long range nanostructuring of siliconsurfaces by photonic nanojets frommicrosphere Langmuir filmsL N Deepak Kallepalli1,3, D Grojo1, L Charmasson1, P Delaporte1,O Uteza1, A Merlen2, A Sangar2 and P Torchio2

1 Aix Marseille Universite, CNRS, LP3 UMR 7341, 13288, Marseille, France2 Aix Marseille Universite et Sud Toulon Var, CNRS, IM2NP UMR 7334, 83957, Toulon, France

E-mail: [email protected]

Received 20 December 2012, in final form 5 February 2013Published 11 March 2013Online at stacks.iop.org/JPhysD/46/145102

AbstractLarge arrays of sub-micrometre blind holes and with a filling ratio up to 60% on areas ofmillimetre square are realized on silicon. The structuration ensues from combining bothLangmuir–Blodgett deposition technique and ultraviolet nanosecond laser-assisted photonicnanojet ablation through C18 functionalized silica microspheres. Different laser fluenceranges and numbers of laser shots are studied to understand the tradeoff between size, qualityof the craters and surface morphology after laser irradiation. In particular, tuning theirradiation fluence yields selectivity of the characteristic lateral dimension of the imprintedcraters on the substrate and laser operation in multishot mode allows obtaining high qualityand regularity of the surface morphology of the resulting millimetre square arrays of holes.This simple, fast, long-range and low-cost near-field nanolithography technique is of interestfor fabricating devices with new functionalities and finds applications in many fields innanoscience and nanoengineering.

(Some figures may appear in colour only in the online journal)

1. Introduction

Developing techniques able to fabricate micro- and nanos-tructures at the surface of a material and on a large scalearea is extremely valuable for fostering emerging applicationsin microelectronics, optics, biology and in nanoscience andnanoengineering as a general view. In addition to its tech-nological pertinency, the economic viability of an emergingtechnique will be strongly favoured if it is shown to be in-expensive and to have extensive manufacturing capabilities.The synthesis and processing of materials on large dimensionsand with highly ordered structures at the nano- and microscalehave been first performed using chemical routes and evapo-ration masks [1–4]. However, such techniques suffer limita-tions [3, 4] and often require numerous chemical steps makingthe overall process time consuming and not environmentallyfriendly. Ion beam lithography in particular, and any short

3 Author to whom any correspondence should be addressed.

wavelength based lithography techniques (electron, extremeultraviolet (UV) and x-rays) [5, 6], offer unbeatable nanoscaleresolution patterning with a high process quality (low affectedzone) but these techniques yield slow process and imply com-plex installations.

Direct laser machining has been proved to be one ofthe best technologies for material surface patterning with anacceptable quality for industrial applications [7, 8]. However,parallel (rapid) direct laser writing of sub-µm features (orsub-micro scale patterning) at the surface of a material andon a large scale area remains a challenging task and ofhigh interest for a large panel of applications in science andtechnology [9, 10]. One solution was promoted in [11] thatshows the ability of interference techniques combined withlaser ablation to fabricate periodic nanostructures on a largearea (>mm2). However, it requires a high level of controlof the coherence of the interfering beams making criticalthe alignment, and overall stability of the structuration set-up restricts the domain of applicability of this technique.

0022-3727/13/145102+09$33.00 1 © 2013 IOP Publishing Ltd Printed in the UK & the USA

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

Another technical route combines near-field optics and laserablation [12, 13]. In this context, photonic nanojets haveemerged as a promising and simple solution to address theproblem. This technology involves near-field focusing of lightwaves underneath the surface of a monolayer of particles tofabricate structures of micro- and nanometre size by reachingthe fluence above the damage threshold of the material ofinterest. When modest energy laser pulses are used, thesubstrate is modified only locally at the tip of the photonicnanojets [12, 14–17]. By utilizing the benefits of the self-assembly properties of microspheres, a single laser beam canproduce a well-ordered array of features with size on the orderof a few hundred nanometres or even below, depending onthe material response over relatively large surfaces [12, 18].Versatility of the technique and enlargement of its domain ofapplicability can be further extended by varying the natureand size of the spheres, and to a lesser extent the laserparameters [19], and also by combining with other techniquelike laser-induced forward transfer (LIFT) for instance tofabricate nanodot arrays with controlled dot size and dot-to-dot spacing [20, 21]. The outcome of the process primarilydepends on the ability to synthetize a perfect monolayer ofparticles onto the substrate to pattern. In the context oflaser studies, self-assembly of particles is widely obtainedby spin coating and dip-coating [12, 17, 19] but they areintrinsically limited in size of the monolayer grown (100 µm2

typical) and thus not favourable for large area (�mm2) surfacestructuration. In parallel, self-assembly techniques have beenwidely discussed in the last decade because of their use innumerous nanofabrication cycles [22, 23]. Among them, theLangmuir–Blodgett (LB) technique has been demonstratedto be a promising solution for providing controlled self-assembly on a large surface area but the approach still requirescomplex preparation protocols depending on the nature of themicrospheres that are used [24–26].

In this work, we thus report and measure the capability ofthe LB technique to prepare self-assembly of SiO2 particleas a monolayer on the Si substrate (see section 2) on avery large surface area (>cm2). After this simple step ofpreparation and initial characterization of the LB process,the synthetized monolayers are exposed to laser irradiationto determine the capability of microsphere-assisted laserablation to realize ordered structuring at a scale of a fewhundred nanometres of large areas of silicon surface (seesections 2 and 3). In section 3, we further explore theinfluence of laser irradiation parameters (fluence and numberof shots) on the quality and geometrical characteristics of theresulting material structuration. Our results evidence that thecombination of three simple techniques, which are LB filmsynthesis, near-field optics and laser ablation, allow accessingparallel long range (�mm2) structuration at the mesoscopicscale (100 nm–1 µm) of a material. This simple, fast, long-range and low-cost near-field nanolithography technique willbe particularly attractive for the realization of devices withnew functionalities in many fields including micro/nano andoptoelectronics, photonics, photovoltaics, energy harvesting,biosensing, imaging, etc [27–29].

2. Method for large-scale parallel nanostructuringof silicon surfaces

2.1. General approach

Our method relies on two main fabrication steps which aresummarized in figure 1. For a proper demonstration of itsinterest and applicability, the synthesis of model samples isreported.

Basically, the first step uses the LB deposition techniqueto grow a large area monolayer composed of microspheres ona substrate. Secondly, we irradiate the prepared sample witha laser appropriately chosen to realize its structuration at thesub-micrometre scale. To obtain an optimized structuration,each step of fabrication has to be carefully controlled, as isdescribed hereafter.

2.2. Synthesis of particle monolayer on Si substrate

In the experiments, commercial coated SiO2 spheres of1 µm diameter and functionalized with C18 carbon chains(Micromod Company) are mixed with ethanol solvent in a40 mg ml−1 concentration solution which is sonicated for 10 hfor complete miscibility. Silicon substrates (Siltronix, France;Intrinsic orientation: 100) are cut into ∼=2 × 2 cm2 area andsonicated in distilled water and ethanol each for 30 min. Theyare further treated with atmospheric plasma torch (Acxystechnologies) in order to remove the surface contamination andto increase their surface wettability. An LB film depositionmachine (KSV-Nima, model Mini) equipped with a surfacetension balance is then used to grow a sphere monolayer on theSi substrates. The LB machine is put on a table isolated frommechanical vibrations to preserve the quality of monolayerduring its synthesis.

We are not the first group to apply the LB techniqueto colloidal particles for their assembly on surfaces. Itwas demonstrated that the approach could already lead tomonolayers with coverage of the sample up to 80% [24, 30].However, most of the works concentrate on the transfer ofpolymer spheres which are more appropriate for the techniquebecause they are light and their natural hydrophobicity can beused to maintain them at air/water interfaces. Nevertheless,such spheres are not compatible with deep UV opticalapplications due to the strong absorption of polymer in the UVpart of the spectrum [31]. For UV laser applications, we needto use glass beads which usually make a poor combination witha water sub-phase in Langmuir troughs due to their miscibility.To the best of our knowledge, we report on the first successfuldirect transfer of silica spheres by this technique without usingchemicals to modify the sub-phase. We chose to work withcommercial SiO2 spheres functionalized with carbon chains(C18). The presence of C18 carbon chains renders themstrongly hydrophobic. This makes possible their assembly ata water surface. The microsphere solution is carefully spreadby small droplets (µl) at the surface of water before a classicalLB transfer.

The processing parameters, namely the surface tensionand processing speed during the sample dipping process,are carefully adjusted to optimize the quality, uniformity

2

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

Figure 1. Fabrication steps of large scale structured samples at the sub-micrometre scale. (a) LB deposition technique of C18 silica spheresonto silicon substrates. (b) Illumination of the silica spheres by a UV nanosecond laser pulse and (c) subsequent local substrate ablation andsimultaneous cleaning of the silicon substrate.

and compactness of the monolayers. The preparation ofmonolayers at the surface pressure (SP) of 10 mN m−1 yieldsthe formation of a sub-monolayer with voids at several places(>10 µm2) which are observable through optical microscopy.This gives us the first indication that the formation ofmonolayer starts to be effective at 10 mN m−1, but this valueis not sufficient to obtain a close-packed assembly. Whenexperiments are performed at a SP superior to 15 mN m−1,a much better coverage is observed. But scanning electronmicroscope (SEM) characterization reveals the presence ofmultilayers in the assembly showing that a further increasewill not lead to any additional improvement. In order toobtain a good coverage and an optimized monolayer over alarge area, the SP of 15 mN m−1 is thus selected. For thesample dipping procedure, we test different speeds from 1to 10 mm min−1. Operation at high speed (10 mm min−1)

results in a larger number of voids because the maximumrate of compression is too low to maintain a close packedmonolayer at the water surface during the transfer to thesubstrate. Finally, the experiments are performed at a rate of5 mm min−1 corresponding to the best compromise betweenmonolayer quality and sample processing time.

To evaluate the quality of the grown monolayers, wefirst make a direct observation of the structure upon whitelight illumination. In figure 2(a), a photograph of one of thesynthesized samples is shown. We clearly see different coloursat the surface of our 2 × 2 cm2 sample due to the diffractionof light with different view angles. In fact, near visible light isdiffracted from the 2D periodic arrangement of microspheresin a similar manner to the diffraction of x-rays from ionic ormolecular crystals. The observation of this effect confirmsthat the ordered sphere assembly covers the entire sample.The sample has a black region on left which corresponds tothe silicon surface without any sphere. This region was at the

touch of the holder in the LB system during the depositionprocess and obviously did not receive any particles. In thephotograph, we also note white regions revealing multilayersor aggregates of SiO2 particles and, upon careful inspection,we find a few voids corresponding to grey dots. From this firstmacro-scale characterization, we conclude that we are still farfrom obtaining a defect free monolayer but the coverage witha monolayer is ensured over very long ranges and in all theregions of the sample.

For quantitative analyses of the quality of the monolayer,we determine the filling percentage of voids, monolayers andmultilayers in five different regions regularly distributed onthe processed zone. These regions are marked as R1–R5 infigure 2(a) and are further observed with SEM (see figure 2(b)as an example). The SEM images are converted into threegrey level pictures clearly indicating the zones of voids,close-packed monolayers and multilayers. The procedure isperformed manually by using the Igor Pro software (version6.1). Then, the area of voids, monolayers and multilayersare automatically calculated by dividing the measurementsobtained with the total image area, giving the percentage of areacovered by close-packed monolayers, voids and multilayers.The results of these analyses are shown in table 1. Similarresults are obtained in all of these five individual regionsshowing the uniformity of the deposition. We estimate that∼=40% of the region consists typically of either multilayersor voids. Thus, we conclude that ∼=60% of our samplesare covered by a close-packed monolayer, which determinesthe monolayer filling ratio FRmonolayer associated with ourdeposition method.

Here, it is important to emphasize that the maximum areaexhibiting a close-packed monolayer remains on the orderof 150 µm2 which is similar to other studies dealing withmicrosphere monolayer on surfaces. However, it is worth

3

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

Figure 2. Photograph of a 1 µm diameter C18-silica sphere monolayer on silicon upon illumination with white light (a). SEM image of oneof the five analysed regions (b).

Table 1. Results of software analyses on SEM imagescorresponding to five regions of a typical sample (see figure 2).

Sample Close packedregion Ri monolayer Void Multilayer

R1 63.5% 6% 30.5%R2 58.9% 2.3% 38.8%R3 60.0% 4.4% 35.6%R4 59.7% 8.6% 31.7%R5 58.1% 5.6% 36.3%

Average coverage 60.1% 5.3% 34.6%

noting that our filling percentage FRmonolayer is obtained ona long range evaluation since we report on measurementsover a very large area of the sample (∼4 cm2) which is notaccessible using other techniques like drop coating or spincoating [12, 17, 19].

Afterwards, we also verify the microscopic characteristicsof the assembled particles. Using the same SEM pictures fromthe different regions, the mean size and associated standarddeviation of the SiO2 particles deposited on the substrateare calculated using 250 individual measurements. Thisprocedure yields the following particle diameter: daverage =927 ± 67 nm, in relatively good accordance with the dataprovided by the microsphere supplier (diameter d = 1 µm;polydispersity index PDI<0.2). The different processing stepsthus are not associated with the degradation or separation ofa specific size distribution. It is obvious that the quality ofthe microsphere monolayers is highly dependent on their sizedistributions. Non-commercial particles featuring narrowersize distributions could be employed to improve the qualityof the deposition [19]. Indeed, narrower size distributions ofparticles (either commercial or noncommercial) would leadto more homogeneous monolayers and the quality of surfacestructuration resulting from laser irradiation would be furtherimproved also. The use of the LB technique in tandem withhydrophobic SiO2 particles thus yields a large monolayerarea of good optical quality for UV applications but alsolong range homogeneity which is not accessible with othertechniques. In the past, we used spin-coating or drop-coatingas deposition techniques for the self-assembly of spheres whichallows us to produce self-assembled monolayers of particleswith areas exceeding 100 µm2. This was enough for proof-of-concept experiments [12, 18] but it could not be envisioned for

applications because some regions of centimetre square scaleof the samples did not contain any microspheres. Even if agood hexagonal close packed structure is not obtained overlonger ranges, the total coverage of the substrate is ensuredby the nature of LB technique. This is an improvement fortechnological considerations when large structured surfacesare required.

2.3. UV laser irradiation and silicon substratenanostructuring

An ArF excimer laser system (Lambda Physik, LP220i)delivering light pulses of 25 ns duration at 193 nm wavelengthis used for surface structuring experiments (see figure 1, secondstep). While surface structuring by microsphere near-fieldassisted ablation is applied today using a wide variety oflasers with wavelengths from UV to infrared (IR) and withpulse durations from femtosecond (fs) to nanosecond (ns)[14, 20], we choose to work in the UV nanosecond regimebecause it combines several advantages for the processing ofa large surface of silicon. Firstly, silicon strongly absorbsthe UV part of the spectrum. At 193 nm wavelength, theoptical penetration depth falls down to ≈5.6 nm making UVlasers good candidates for controlled processing applicationsconfined at the surface. Secondly, the lateral size of near-field photonic nanojets from microspheres scales with the laserwavelength [32]. Then, deep UV illumination of microspheresprovides a way to perform very high resolution interactions.Thirdly, the ‘flat top’ nature of beams from excimer lasers is amajor asset for uniform irradiation of targets. It avoids the useof complex optical elements for beam shaping.

For enhanced control of the process, a mask projectiontechnique is implemented to have a nearly top-hat beam(uniform illumination) onto the sample. First, a rectangularaperture (mask) closed to a size of 3.7 × 18 mm2 selects theregion of uniform intensity at the direct output of the laser.Then, an image relay technique with a lens (focal lengthf = 175 mm) for beam size reduction on the target is used.The typical laser spot on the target is a rectangle with sizeS = 0.6 × 2.9 mm2 allowing a maximum laser fluence onthe target of ∼=5 J cm−2 . The nearly top-hat beam qualityis confirmed by cross examining the impact uniformity onphoto and thermo-sensitive papers placed on the sample plane.

4

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

Figure 3. SEM images of the silicon surfaces obtained with single UV pulse illumination of the Langmuir films with gradually increasinglaser fluences. Each image shows a representative region of 5 × 14 µm2 in the irradiated zone.

Then, the reported laser fluences can be simply expressedas F = E/S, where E is the laser pulse energy and S isthe laser beam area at the target. The laser energy E ismeasured after the lens to account for transmission lossesfrom optical elements. A manually operated beam attenuator(Optec, AT4030) is used to vary the laser fluence onto thesubstrate in the range 0.15–5 J cm−2 during the experiments.The sample is placed on motorized XY Newport stages toperform each irradiation experiment in distinct sample regionsto avoid overlapping among the irradiated zones. The effect ofillumination conditions is investigated by repeating the laserirradiation experiments with gradually increasing laser fluenceand number of pulses. Figure 3 presents representative SEMimages of the target surface after single pulse illumination withincreasing laser fluences. Microscale structuration is obtainedon the whole irradiated area (≈1.7 mm2 ).

While the particle removal (or laser cleaning) threshold[33] is exceeded at a laser fluence of 0.15 J cm−2 , we note thatlocal silicon ablation is observable by SEM at a typical laserfluence of about 0.2 J cm−2 . In the range 0.2–0.75 J cm−2 ,the single pulse structuration consists of an array of sub-µmdiameter features composed of a blind hole surrounded by anaffected zone and separated by a pitch of 1 µm determined bythe SiO2 particle diameter. From the images, we note that theother size parameters (crater diameter and extension of affectedzone) can be adjusted according to the applied laser fluence butalso the applied number of pulses as will be discussed later.

Above 0.75 J cm−2 , the quality of structuration is severelyaffected with more irregular patterns due to coalescence

(and/or overlapping) of the imprinted features (see figure 3).This is not surprising as the fluence for melting bare silicon ismeasured at ∼=1 J cm−2 [31] fixing the upper limit for periodiclocal structuration of silicon in a controlled way.

Using an image processing approach similar to theprocedure described before, we further determine themicrostructuration filling ratio FRstructured as the percentageof area covered by sub-µm features with respect to the totalconsidered area. Considering a single pulse irradiation ata fluence of 0.55 J cm−2 and performing statistical analysesover five regions distributed in the irradiated zone (thatis over 130 × 100 µm2 in total), the measurement yields:FRstructured = 63.4 ± 5% which is in good accordance withthe previously reported monolayer filling ratio FRmonolayer forthe same sample before irradiation. This connects directlythe surface microstructuration by the laser interaction with themonolayer partially covering our silicon substrates.

From the same images, we can also extract the character-istic size of the imprinted features dfeature corresponding to thediameter of the blind holes surrounded by affected zones. Themeasurement is performed using the SEM software (JEOL,JSM 6390) over a statistically significant number of craters(n > 50). The procedure leads to dfeature = 822 nm witha standard deviation of 142 nm. As will be discussed later,this corresponds to relatively large features and a broad sizedistribution associated with single shot large energy laser pro-cessing.

These results confirm the capability of microsphere-assisted UV laser ablation to realize ordered structuration

5

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

Figure 4. Measurements of feature sizes (diameters) on silicon forsingle laser shot irradiation of the Langmuir film with different laserfluences. The data are presented for single and double pulseillumination as a function of laser fluence. The error bars stand forthe standard deviation of 50 individual measurements for eachexperimental condition.

on the silicon surface. The applications of this technologyare today limited by the extent to which ordered near-fieldmasks can be prepared over large areas. Here, we demonstratethe successful transfer of centimetre-scale microsphere filmsleading to a microstructured surface coverage exceeding 60%after a single pulse irradiation. Such results rely on the use ofthe LB deposition technique with strongly hydrophobic silicaspheres. Figure 3 also shows the wide variety of featuresthat can be obtained while varying the laser fluence. A morequantitative and complete analysis requires statistical studies toevaluate the size and reproducibility of features with respect tothe laser parameters. The assessment of these aspects in detailis of major importance to develop efficient and suitable siliconnano/micro-fabrication technologies based on this photonicmethod.

3. Material structuring characteristics and controlparameters

To study the laser capability to structure the samples in aversatile way, we measure the evolution of feature size as afunction of fluence and for single and two pulse irradiations.Figure 4 shows the measurement of diameters of the producedfeatures and their associated standard deviations (error bars)for each irradiation condition using an approach similar to thatdescribed above (nfeature > 50 for all cases). The featuresize strongly increases above the threshold fluence (Fth ∼0.2 J cm−2 for 1 shot), but above the fluence of ≈0.4 J cm−2,there is no further significant increment in the crater sizethough fluence is increased. A very similar behaviour isobserved for two shot irradiations but saturation is observedat a smaller feature size level. To determine the origin ofthis observation, we show in figure 5 the measurement offeature sizes for a gradually increasing number of pulses at

Figure 5. Evolution of the feature size as a function of number oflaser pulses and with different laser fluence conditions. The first andsecond sequences correspond to the irradiation of the substrate withthe same laser fluence, set at 0.55 J cm−2 (F1, red) and0.36 J cm−2(F2, blue) respectively, and varying the number of shots.The third experiment is a mixed sequence where the first laser pulse(ejecting the spheres) is set at 0.36 J cm−2 and the following laserpulses (shots #2 to N) are set at a fluence of 0.55 J cm−2 (F3, black).

a fluence of 0.55 J cm−2 (red circles, F1) that is above thesaturation level. We note the size decrease associated withthe second laser shot. However, all subsequent laser shots areassociated with very modest change (if any) of the apparentsize of the features. In figure 6, we show SEM and atomicforce microscope (AFM) images of the structures showingunambiguously that the decrease in feature size is associatedwith the disappearance of large affected (or contaminated)zones surrounding the craters after the first laser shot.

Considering the single pulse regime, the saturationreported in figure 4 could be attributed to the self-limitingcharacter of ablation experiments using particles as focusingelements, as already reported in [34]. Indeed, since theparticles are ejected by momentum transfer from the ablatedspecies, the higher the laser energy, the sooner the particledetachment and the disappearance of the focusing power atthe target surface. Then, increasing the laser fluence does notlead necessarily to an increase in the deposited energy and, asa consequence, of the feature size on the target.

When increasing the number of shots, the scenario issomewhat different since surface laser ablation by photonicnanojets is a one-shot process by nature because the ablatedproducts eject the spheres [33]. After the first shot, anysubsequent pulse interacts with a structured substrate butfree of particles (neglecting particle redeposition effects). Infigures 5 and 6, as the impinging fluence is relatively low,

6

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

Figure 6. Microstructured silicon surfaces at the fluence of 0.55 J cm−2 with (a) single shot (b) two shots and (c) ten shots.

at least much below the ablation threshold of virgin (notstructured) Si substrate (Fth,Si = 1 J cm−2 ), we do notexpect the second pulse or any subsequent pulse to affect theunmodified regions of the substrate. To confirm this point,we performed near-field calculations to evaluate the potentialfield enhancement by interaction with craters with size similarto those observed by SEM for single shot irradiations (seefigure 6). We found very little enhancement (<1.2, notshown here) suggesting the change in feature size with thenumber of shots is obtained below the laser fluence ablationthreshold for all shots (N > 2). The observations of figure 6suggest that the improvement of the surface regularity with thenumber of shots can be a result of laser annealing or cleaningof the modified materials surrounding the spheres. Thus,the multishot experiments are sensitive to two laser energythresholds: the threshold for ablation and the threshold forcleaning/annealing resulting in the decrease in the producedfeature size.

In order to reveal these aspects, we studied the influenceof the number of applied laser pulses while varying thelaser fluence. In figure 5, we show the evolution of thefeature size as a function of the number of shots N for threedifferent sequences. As already described, the first set ofexperiments is conducted with a fluence F1 = 0.55 J cm−2

above the saturation fluence. The second sequence is similarbut performed at the laser fluence F2 = 0.36 J cm−2 just belowthe saturation fluence. The third set mixes both F1 and F2

fluences, consisting of one shot at 0.36 J cm−2(F2) followedby (N − 1) shots at 0.55 J cm−2.

Whatever the fluence, a rapid decrease in the feature size isobserved considering a few shots (typically 2) before saturationoccurs for N > 5. Figure 6 shows the fluence F1 case. Infact, it reveals that the first shot imprints the feature whilethe following shots remove the debris and the affected zoneat the edges of the feature, both having an ablation thresholdsmaller than unmodified Si. As a result, cleaning of the surface,apparently without degrading it, is obtained thus revealing theinner structuration (crater) left by the first shot (see figure 6).Moreover, the depths of these features range in between 250and 300 nm whatever the number of shots and the effectof cleaning and smoothening of the surface using multishotirradiation is particularly evident as shown in the AFM profiles(see figure 6, middle and right). A very good regularity of theprinted pattern (here, an array of sub-micrometre blind holesof 250 nm depth and separated of 1 µm pitch) is thus obtainedin the multishot irradiation regime.

The differences between the two fluences F1 and F2considering the feature size obtained are not significant (seefigures 4 and 5), presumably due to small differences ofresidual absorption when using fluences slightly below orclose to the ablation threshold of the Si structured surface.We recall that we do not know with precision this parameterbut it is expected to be somewhat smaller than 1 J cm−2

(Fth,Si). However, we note that the mixed sequence (F3)

7

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

allows reducing significantly the feature size level. The onlydifference between the sequences F1 and F3 relies on thefirst ‘structuring’ shot and the only difference between thesequences F2 and F3 relies on the ‘processing’ shots (N > 2).Also we can interpret these measurements in the way that0.36 J cm−2 corresponds to a sufficient fluence for structuring(figure 5) but inducing craters with size smaller than thoseproduced at 0.55 J cm−2. This is revealed by an efficientcleaning sequence accessed only at 0.55 J cm−2.

A first conclusion from this measurement is that a smallnumber of shots (<5) appear to be sufficient for an optimizedstructuration. However, while the fluence of the first shot canbe adjusted for controlling the size of craters, the fluence ofthe subsequent pulses must exceed a threshold (between 0.36and 0.55 J cm−2) to reveal ‘clean’ ablation features. Furtherexperiments are needed to determine the optimized sequenceof combined fluences and number of shots. This work alreadydemonstrates the interest and versatility of such an approach.

In figure 6(c), we show that a surface morphology ofhigh regularity at the sub-micrometre scale can be obtainedusing a few numbers of shots. However, we need toverify the integrity of the nanostructured silicon to envisionthis photonic method to fabricate nanostructure templatesfor applications. Micro-Raman spectroscopy is recognizedas a very sensitive optical method to detect alterations incrystalline silicon through Raman spectrum analysis [35].Hence, we perform a comparison between the pristine and theirradiated zone (craters) of our samples with laser irradiationconditions corresponding to figure 6(c). During the diagnostic,a continuous wave laser at 488 nm is focused at a controlledposition on the surface and we measure the Raman responseof the materials using a Horiba Labram HR800 spectrometer.Details of the experiment can be found in [36]. The results areshown in figure 7. Both spectra show a narrow and symmetricRaman peak at ∼=521 cm−1. It was demonstrated that a phasetransition (amorphous to crystalline) or mechanical stresses[36–39] are revealed by a peak shift and/or broadening ofthe Raman peak. Here, none of these effects are observedin the structured zone confirming most likely that the laserablation process and supplementary energy deposition shotsafter shots do not lead to a structural transformation of thematerial while structuring and cleaning the surface. Fromthese measurements, we also note that the Raman signalassociated with the structured silicon is enhanced. Whilewe concentrate this paper on fundamental and technologicalaspects associated with nanosphere mediated structuringof silicon, we will focus future developments on thisobservation representing an interesting effect for siliconnanophotonics/sensor applications.

4. Conclusions

We have successfully shown the structuration with micro/nanofeatures on silicon by photonic nanojets with UV lasercompatible Langmuir films. The monolayer films weremade of C18 functionalized silica spheres of 1 µm diameter.With our method, we have demonstrated more than 60%of microstructured coverage of the silicon surface, with

Figure 7. Confocal micro-Raman analysis showing pristine (Red)and irradiated zones (Blue). Laser exposure: 10 shots andirradiation dose per shot equal to 0.55 J cm−2. The inset shows abright field optical image of the substrate and the locations ofmicro-regions from where the Raman spectra are obtained.

periodic features over surfaces exceeding 1 mm2 which iscoincident with the coverage of spheres by the depositedfilm. The coverage homogeneity over such large samplesis a significant improvement for microsphere-assisted lasermicro-structuring applications brought by the LB technique.For silicon structuration, we conclude that working with thefluence range below 1 J cm−2 with few shots allows obtaininggood quality craters with diameter around 800 nm and depthof 250 nm that can be controlled by the laser fluence. Theperiodicity of the structure is controlled by the diameterof spheres which are used. It can be easily tuned by theprocessing step of sphere deposition. In this regime, theobserved structures rely on the combination of two laser–matter interaction mechanisms. The first laser shot ablates andstructures the substrate while the subsequent irradiation doseslead to surface cleaning phenomena yielding an improvementof the apparent morphology confirmed through SEM, AFMand confocal micro-Raman spectroscopy techniques. Laserfluences above 1 J cm−2 exceed the melting threshold for baresilicon. While the intensity contrast provided by the nanojetsstill allows us to microstructure silicon with the first shot,there is an overlap between the modified zones underneath thespheres that degrade the apparent quality of the surface. Thestructure is further damaged if more laser shots are applied.It is always a tradeoff between crater size, quality of cratersand invasiveness of the surface materials that decide the fluencerange of interest. The assessment of these aspects in detail is ofmajor importance for applications. This study opens a route todevelop efficient and suitable silicon nano/micro-fabricationtechnologies based on this laser method. In addition, thistechnique appears to be particularly simple and convenient

8

J. Phys. D: Appl. Phys. 46 (2013) 145102 L N Deepak Kallepalli et al

to implement for the patterning of very large areas whenmicro- or nanoscale amounts of material must be processedin a controlled way.

Acknowledgments

The authors thank French ANR agency for financial supportthrough the FELINS-ANR-10-BLAN-946 and CARIOCA(number 2010-JCJC-918-01) projects.

References

[1] Martin C R 1994 Science 266 1961[2] Masuda H, Yasui K and Nishio K 2000 Adv. Mater. 12 1031[3] Nam W, Seo H, Park S C, Bae C H, Nam S H, Park S M and

Ha J S 2004 Japan. J. Appl. Phys. 43 7793[4] Bae C H, Park S M, Park S C and Ha J S 2006 Nanotechnology

17 381[5] Watt F, Bettiol A A, Van Kan J A, Teo E J and Breese M B H

2005 Int. J. Nanosci. 4 269[6] Feiertag G et al 1997Appl. Phys. Lett. 71 1441[7] Brown M S and Arnold C B 2010 Fundamentals of

laser–material interaction and application to multiscalesurface modification Laser Precision Microfabrication(Springer Series in Materials Science) ed K Sugioka et al(Berlin: Springer) chapter 4

[8] Schaeffer R 2012 Fundamentals of Laser Micromachining(Boca Raton, FL: CRC Press)

[9] Perriere J, Millon E and Fogarassy E 2006 Recent Advances inLaser Processing of Materials (Oxford: Elsevier)

[10] Prasad P N 2004 Nanophotonics (Hoboken, NJ: Wiley)[11] Bekesi J, Meinertz J, Ihlemann J and Simon P 2008 Appl.

Phys. A 93 27[12] Pereira A, Grojo D, Chaker M, Delaporte P, Guay D and

Sentis M 2008 Small 4 572[13] Chong T C, Hong M H and Shi L P 2010 Laser Photon. Rev.

4 123[14] Zhou Y, Hong M H, Fuh J Y H, Lu L, Luk’yanchuk B S,

Wang Z B, Shi L P and Chong T C 2006 Appl. Phys. Lett.88 023110

[15] Cai W and Piestun R 2006 Appl. Phys. Lett. 88 111112[16] McLeod E and Arnold C B 2008 Nature Nanotechnol.

3 413

[17] Langer G, Brodoceanu D and Bauerle D 2006 Appl. Phys. Lett.89 261104

[18] Grojo D, Charmasson L, Pereira A, Sentis M and Delaporte Ph2011 J. Nanosci. Nanotechnol. 11 9129

[19] Grojo D, Boarino L, De Leo N, Rocci R, Panzarasa G,Delaporte P, Laus M and Sparnacci K 2012 Nanotechnology23 485305

[20] Landstrom L, Klimstein J, Schrems G, Piglmayer K andBauerle D 2004 Appl. Phys. A 78 537

[21] Kabashin A V, Delaporte Ph, Pereira A, Grojo D, Torres R,Sarnet Th and Sentis M 2010 Nanoscale Res. Lett. 5 454

[22] Yang S M, Gyu Jang S, Choi D G, Kim S and Hyung K Y2006 Small 2 458

[23] Grzelczak M, Vermant J, Furst E M and Liz-Marzan L M 2010ACS Nano 4 3591

[24] Goldenberg L M, Wagner J, Stumpe J, Paulke B R andGornitz E 2002 Langmuir 18 5627

[25] Bardosova M, Pemble M E, Povey I M and Tredgold R H 2010Adv. Mater. 22 3104

[26] Vogel N, Goerres S, Landfester K and Weiss C K 2011Macromol. Chem. Phys. 212 1719

[27] Poole C P and Owens F J 2003 Introduction toNanotechnology (Hoboken, NJ: Wiley)

[28] Ebbesen T W, Lezec H J, Ghaemi H F, Thio T and Wolff P A1998 Nature 391 667

[29] Rigneault H, Capoulade J, Dintinger J, Wenger J, Bonod N,Popov E, Ebbesen T W and Lenne P F 2005 Phys. Rev. Lett.95 117401

[30] Fulda K U and Tieke B 1994 Adv. Mater. 6 288[31] Grojo D, Cros A, Delaporte P and Sentis M 2007 Appl. Surf.

Sci. 253 8309[32] Kim M S, Scharf T, Muhlig S, Rockstuhl C and Herzig H P

2011 Opt. Express 19 10206[33] Grojo D, Cros A, Delaporte P and Sentis M 2006 Appl. Phys.

B 84 517[34] Grojo D, Delaporte Ph and Sentis M 2008 J. Adhes. Sci.

Technol. 22 623[35] Bonse J, Brzezinka K W and Meixner A J 2004 Appl. Surf. Sci.

221 215[36] Merlen A, Sangar A, Torchio P, Deepak K L N, Grojo D,

Uteza O and Delaporte P 2013 private communication[37] Palani I A, Vasa N J, Singaperumal M and Okada T 2010

J. Laser Micro/Nanoeng. 5 150[38] Lucazeau G and Abello L 1997 J. Mater. Res. 12 2262[39] De Wolf I, Maes H E and Jones S K 1996J. Appl. Phys. 79

7148

9