Ss

CAa

b

a

ARRAA

KSAAF

1

eeoossfpdicpuadap[

T

0h

Electrochimica Acta 113 (2013) 755– 761

Contents lists available at ScienceDirect

Electrochimica Acta

jo u r n al hom ep age: www.elsev ier .com/ locate /e lec tac ta

urface patterned dielectrics by direct writing of anodic oxides usingcanning droplet cell microscopy

hristian M. Siketa, Andrei Ionut Mardareb, Martin Kaltenbrunnera, Siegfried Bauera,chim Walter Hasselb,∗

Soft Matter Physics, Johannes Kepler University Linz, AustriaInstitute for Chemical Technology of Inorganic Materials, Johannes Kepler University Linz, Austria

r t i c l e i n f o

rticle history:eceived 18 January 2013eceived in revised form 6 July 2013ccepted 13 July 2013vailable online 26 July 2013

a b s t r a c t

Scanning droplet cell microscopy was used for patterning of anodic oxide lines on the surface of Al thinfilms by direct writing. The structural modifications of the written oxide lines as a function of the writingspeed were studied by analyzing the relative error of the line widths. Sharper lines were obtained forwriting speeds faster than 1 mm min−1. An increase in sharpness was observed for higher writing speeds.

eywords:canning droplet cell microscopenodised aluminiumluminium oxidelexible electronics

A theoretical model based on the Faraday law is proposed to explain the constant anodisation currentmeasured during the writing process and yielded a charge per volume of 13.4 kC cm−3 for Al2O3. Fromcalculated oxide film thicknesses the high field constant was found to be 24 nm V−1. Electrochemicalimpedance spectroscopy revealed an increase of the electrical permittivity up to ε = 12 with the decreaseof the writing speed of the oxide line. Writing of anodic oxide lines was proven to be an important stepin preparing capacitors and gate dielectrics in plastic electronics.

. Introduction

Since its development in the 90s, the capillary-based micro-lectrochemical cell was continuously improved and extended.g. by implementation of scanning capabilities [1], by the usef integrated micro-reference electrodes [2] or by the devel-pment of the contact operation mode [3] in which very highurface reproducibility was recently proven [4]. In the last period,canning droplet cell microscopy became one of the main sur-ace analysis tools used for mapping the electrochemical surfaceroperties of thin film combinatorial libraries for new materialsevelopment due to its high-throughput-experimentation capabil-

ties [5–7]. Thus, it is appropriate to define the scanning dropletell as a new form of microscopy. Apart from imaging surfaceroperties, the scanning droplet cell microscope (SDCM) can besed for many other types of electrochemical investigations suchs grain boundary electrochemistry [8], corrosion studies with

ownstream analytics [9], localized photoelectrochemistry [10,11]nd recently, its low required volume of electrolyte made it aerfect candidate for electrochemistry using organic electrolytes12]. A special application of the SDCM was recently found in

∗ Corresponding author at: Altenberger Strasse 69, 4040 Linz, Austria.el.: +43 732 2468 8700; fax: +43 732 2468 8905.

E-mail addresses: achimwalter.hassel@jku.at, hassel@elchem.de (A.W. Hassel).

013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.electacta.2013.07.114

© 2013 Elsevier Ltd. All rights reserved.

microelectrochemical lithography which involved localized anodicoxide growth in a point-by-point approach for surface patterning[13].

Among all valve metals, aluminium plays a key role due tothe possibilities of using its protective oxide easily obtainablethrough anodisation [14]. The properties and behaviour of theanodic oxides can be tuned by alloying Al with other metals (e.g.Fe, Mg, W) before anodisation, leading to modified oxides [15–17].Enhanced corrosion protection can be obtained in this manner[18]. The customization of the oxide properties extends towardsthe electronic applications, nanocomposite oxide films obtainedby anodisation of Al alloys showing improved dielectric properties[19].

Anodised Al is an excellent dielectric for flexible electronicdevices [20–24]. Direct writing of anodised oxide lines has potentialfor the preparation of patterned oxide structures, useful as capac-itors and gate dielectrics in flexible electronic circuits. Therebyadditional process steps such as optical lithography or shadowmasking are avoided. Furthermore, writing potentially allows forimplementation of the SDCM in roll-to-roll processing of macro-electronic items.

In the present work, the scanning capabilities of the SDCMare exploited for direct writing of continuous anodic oxide lines

on flexible substrates in the sub-mm range. The lines are easilyprepared under ambient laboratory conditions, avoiding addi-tional processing such as optical lithography or shadow maskingprocess.

756 C.M. Siket et al. / Electrochimica

2

2

otttG(wattsmciAtrTb

ffgmpcoopoetwwmoe

Fig. 1. Scheme of the preparation of an Al layer with a thickness gradient.

. Experimental details

.1. Substrate preparation

In order to investigate the possibilities of direct writing of anodicxides, thin Al films were prepared under various conditions. Ahermal evaporator with a base pressure of 10−4 Pa was used forhe deposition of Al thin films on 50 �m thick polyethylene naph-halate (PEN) plastic substrates (Teonex Q51, obtained from PützMBH + Co. Folien KG). As Al source, ultra-pure (99.9995%) Al wires

Alfa Aesar) cut in small pieces were positioned in direct contactith the surface of a W filament. The substrates were cut into rect-

ngular stripes (10 mm × 60 mm) and were placed directly abovehe evaporation source at a distance of 170 mm. In order to reachhe evaporation temperature of Al, a Sorenson XG 84-20 powerource capable of delivering up to 1.6 kW was employed. Self-ade LabView software including a PID controller was used for

ontrolling the power delivered to the evaporation source, hold-ng the evaporation rate at 2 nm s−1 during the deposition. All thel depositions were done at room temperature and the deposi-

ion geometry allowed a maximum substrate temperature of 60 ◦C,eliably avoiding any thermal degradation of the PEN substrates.he evaporation rates were monitored in situ using crystal quartzalances (Testborne).

Two different series of Al thin films were evaporated. First, a uni-orm Al thin film with a thickness of 25 nm was evaporated, usedor the oxide writing process. In a second deposition, a thicknessradient of the film was obtained in the deposited Al by using aobile shutter placed in close vicinity to the substrate holder. The

rinciple of the thickness wedge formation is described schemati-ally in Fig. 1. For this special approach the sample was positionedff axis and the moving shutter ensured a prolonged exposure ofnly one side of the substrate to the Al vapour phase. Due to theoor step coverage dictated by the directional behaviour of Al evap-ration, the shutter was actively shadowing the sample. Beforevaporation, the shutter completely covered the substrate. Afterhe desired evaporation rate was reached, the shutter was openedith a constant velocity (10 mm s−1). As soon as the entire substrate

as exposed to Al evaporation, the direction of the shutter move-ent was reversed. This resulted in an accentuated accumulation

f Al at one side of the substrate, producing an Al thickness gradi-nt. This type of sample was used for the analysis of the thickness

Acta 113 (2013) 755– 761

influence on potentiostatic anodisation during the oxide writingprocess.

2.2. Oxide patterning process using scanning droplet cellmicroscope for oxide writing

Patterning of oxide on the surface of Al was achieved by employ-ing a scanning droplet cell microscope (SDCM), where only a smallelectrolyte droplet comes in contact with the Al thin film. Surfaceoxide lines can be drawn by moving the electrolyte droplet duringanodisation. This was done by using the scanning function of theSDCM. The details about the microelectrochemical cell constructionwere presented elsewhere [13]. For the present work two differentcells were prepared. One cell with a tip diameter of 550 �m wasoperated in the free droplet mode [25] for the purpose of writinganodic oxide lines on the surface of Al/PEN substrate. The secondcell had a tip diameter of 300 �m and was prepared for operation inthe contact mode. A soft silicone sealing was formed on the rim ofthe capillary tip by dipping it in liquid silicone, followed by drying inN2 flow [3]. The tip diameter of the second cell was smaller in orderto allow for a complete addressing of the oxide lines written usingthe first cell. For both cells the counter electrode (CE) consisted ofa Au band wrapped around a �-AuHg/Hg2(CH3COO)2/NaCH3COOreference electrode (RE) capillary. Details about the fabrication ofthe RE were presented elsewhere [26]. The position of the SDCM tipwas controlled by a high precision X–Y–Z translation stage. Actua-tion in the X–Y plane was achieved for patterning purposes usingin house developed LabView controlling software. A tilting stageensured a planar positioning of the Al/PEN substrates. Using thevertical (Z) positioning, the tip was brought in contact or in closeproximity to the surface of the Al/PEN substrate. A force sensor wasemployed to control the force applied to the Al/PEN surface duringthe contact mode operation [4]. In the case of the free droplet modeoperation, a distance between the tip and the Al surface of approxi-mately 20 �m was achieved corresponding to an aspect ratio of tipdiameter tip distance of 1:25. For the purpose of growing a com-pact anodic oxide on Al, a citric acid/citrate buffer (0.265 g citricacid, 2.57 g anhydrous sodium citrate in 100 ml DI water producinga pH 6.0) was used.

A schematic drawing of the SDCM describing the anodic oxidewriting process is presented in Fig. 2. An EG&G Instruments Poten-tiostat/Galvanostat Model 283 was used for the potentiostaticanodisation of Al thin films, which represented the working elec-trode (WE). Electrical contact between the potentiostat and the WEwas established by pressing a W needle against the Al thin film sur-face. During anodisation the current was continuously monitored.In part (a) of Fig. 2 the electrical connections of the potentiostat canbe seen together with the principle of oxide writing. Using the freedroplet mode, a constant tip velocity combined with a constantapplied anodic potential will ensure the writing of a continuousoxide line. Due to the capillary forces present between the SDCMtip and the Al surface, the electrolyte droplet will move togetherwith the tip during the scanning/writing process, this is because itis held by capillary forces. The water evaporation due to the contactto the surrounding atmosphere was minimal because of the smallproximity between the SDCM tip and the substrate surface. Thisfact combined with using a buffered electrolyte allowed entirelyneglecting the water evaporation process over the experimentationtime. A high detail photograph of the SDCM tip during the writingprocess of oxide lines on Al thin films is shown in Fig. 2(b). The tipof the cell is visible and no indications of gas evolution (e.g. bubbleformation) could be observed. A secondary image of the cell is seen

due to mirroring on the Al thin film surface. The Au counter elec-trode is visible inside the body of the cell giving it the golden colour.Several lines already written in previous scans are noticeable with agood contrast due to the different refractive indices of Al and Al2O3.

C.M. Siket et al. / Electrochimica Acta 113 (2013) 755– 761 757

F(

cpat3fscnonwopiwsbit

ig. 2. Schematic construction of the SDCM for operation in the free droplet modea) and optical image of its tip during the oxide writing process (b).

For the structuring of well defined oxide lines a certain pro-ess flow was developed. A schematic of the steps involved in theatterning of a dielectric layer is shown in Fig. 3. Starting withn Al/PEN substrate prepared by thermal evaporation (Fig. 3(a)),wo parallel oxide lines were written at a constant speed of.4 mm min−1. For both lines, the entire Al thin film was locallyully converted into anodic oxide. This was ensured by applying aufficiently high constant potential of 28 V vs. SHE which wouldompletely convert into oxide an Al film with a maximum thick-ess of 35 nm. This value of the potential was calculated using anxide formation factor for Al of 1.6 nm V−1, a natural oxide thick-ess of 2.6 nm, and a Pilling–Bedworth ratio of 1.28 [1,27]. Theidth of the fully anodized Al lines was defined by the diameter

f the SDCM tip of 550 �m. The procedure of writing these twoarallel lines (Fig. 3(b)) allowed a precise definition of an Al stripe

n between. In a final processing step, an anodic oxide layer wasritten on the Al stripe using a potential of 6 V vs. SHE and a con-

tant speed (Fig. 3(c)). This produced 12 nm thick Al2O3 layers, toe used as dielectric layers in electronic components such as capac-

tors or field effect transistors. Due to the design of such devices,he Al stripe has to be electrically insulated from the remaining Al

Fig. 3. Process steps for writing anodic oxide lines. (a) Evaporation of Al, (b) writingof fully anodized oxide lines, and (c) writing of the metal-dielectric.

thin film, becoming an electrode for addressing the active dielec-tric layer. Therefore, the second processing step of writing fullyanodized parallel lines is crucial for the device fabrication. Furtherprocessing steps can be designed, e.g. it is possible to remove all ofthe remaining not anodized areas of the Al layer via a single stepwet etching process. The use of various Al selective etchants is pos-sible. A mixture of 19 parts H3PO4 with one part CH3COOH, onepart HNO3 and two parts H2O was already successfully used in fab-rication of Al-based thin film capacitors [22], a typical Al etchingrate of 4 nm s−1 and approximately 100 times smaller Al2O3 etch-ing rate being quite convenient for this purpose. Since the etchingrate of Al2O3 is almost negligible, no additional masking processis necessary for removing the unwanted Al. This then allows forthe deposition of a top electrode or a semiconducting film abovethe patterned active dielectric layer, rendering the fabricated struc-tures ready for direct use in flexible electronic applications.

3. Results and discussion

In order to analyze the anodic oxide writing process necessaryfor structures patterning on Al thin films, the relationship betweenthe writing speed and the properties of the written oxide lines mustbe understood. For this purpose it is relevant to develop the rela-tionship between the anodisation current and the writing speed.The use of Faraday’s law allows a direct correlation between theoxide volume converted per time unit and the anodisation current:

dVox

dt= M

�zrFI (1)

where Vox represents the oxide volume, t is the time, M is the oxidemolar mass, � is the oxide density, r is the roughness factor, F theFaraday constant, and I is the absolute anodisation current. Theoxide volume can be expressed as a product between the line width,thickness, writing velocity, and time which transforms Eq. (1) into:

I = q̃whv (2)

where q̃ is a material constant with units of charge per volume, wis the width of the oxide line, h is the effective oxide thickness, and

v is the tip velocity. Here, wh can be interpreted as the cross sectionarea of the written oxide line normal to the writing direction.

For studying the speed effect on the edge sharpness of the writ-ten oxide lines, the relative error of the line width was analyzed

7 imica Acta 113 (2013) 755– 761

fStwwtlsBteo

Acwcceistlibrtwttt

ovagorrfcw

twwotrtfotbtttogEssyiTf

58 C.M. Siket et al. / Electroch

or various writing speeds, during full Al anodisation at 28 V vs.HE. Eq. (2) can be directly used to calculate the relative errors ofhe line widths from the recorded anodisation currents, during theriting process. These results are summarized in Fig. 4 togetherith optical imaging of the written oxide lines at three represen-

ative velocity values. In part (d) of Fig. 4 the relative error of theine width is presented as a function of the anodic oxide writingpeed. A clear decay of the error for larger velocities is observed.etween 1 and 2 mm min−1 writing speed, a change in the slope ofhe relative error plot can be identified. Above this threshold therrors decrease at a slower rate until the highest investigated speedf 34 mm min−1 is reached.

At speeds below 1 mm min−1 a fuzzy boundary between thel2O3 and the Al parent metal can be found in part (a) of Fig. 4. Thisan be explained by considering a discrete wetting mechanism inhich new Al zones are wetted in discrete jumps rather than in a

ontinuous manner. During the anodisation process, the electro-apillarity plays an important role for reaching surface tensionquilibrium. A slow movement of the SDCM tip will not result in anmmediate movement of the electrolyte droplet on the addressedurface. The sequential minimizing of the surface tension duringhe quasi equilibrium processes involved in the oxide writing atow speeds may be responsible for the droplet jump to new pos-tions. This happens as soon as the force exerted on the dropletecomes sufficiently high for bringing the droplet to a new equilib-ium position. This discrete type of electrolyte movement duringhe anodic oxide growth can be observed in the part (a) of Fig. 4here equidistant concentric arcs are identified on the surface of

he written oxide line. The equal distance between the arcs is dueo the constant tip velocity and can be directly related to the surfaceension of the electrolyte in contact with the surface.

Increasing the writing speed leads to a continuous movementf the electrolyte droplet on the Al surface. Above the thresholdelocity value, a dynamic equilibrium between electrolyte dropletnd Al surface is reached. This leads to more stable and homo-eneous anodisation conditions, resulting in sharper edges at thexide/metal boundary as can be seen in part (b) and (c) of Fig. 4 cor-esponding to writing speeds of 3.4 mm min−1 and 34 mm min−1,espectively. The smoothness of the oxide/metal boundary is higheror the highest oxide writing speed. In both cases no equidistantoncentric circles are visible in the optical images of the oxide linesritten at speeds above the dynamic equilibrium threshold.

For studying the speed influence on the anodisation duringhe direct writing of Al2O3 lines on Al, the anodisation currentas recorded. For this particular experiment the Al/PEN substrateas patterned as shown in Fig. 3 with a defined Al stripe width

f 200 �m. The Al stripe was scanned by the SDCM for writinghe dielectric layer. Generally, the circular geometry of the tipesults in an inhomogeneous anodisation time distribution acrosshe addressed area. For precise time investigations a uniform sur-ace anodisation time distribution is desired. Since the cell diameterf 550 �m was significantly larger than the line width of 200 �m,he anodisation time for each location can be considered constantecause the addressed area on the Al stripe can be approximatedo a rectangle, which shares the symmetry centre with the SDCMip. In Fig. 5 the dependence of the cross section current density onhe writing speed is shown for a wide range of velocities. Resultsbtained from fully anodized lines (28 V vs. SHE) and dielectric linesrown at 6 V vs. SHE are presented together in Fig. 5. According toq. (2) a linear dependence between the cross section current den-ity and the writing velocity is to be expected. A linear fit of the lowpeed range of the cross section current density for dielectric lines

ields a slope of 13.4 kC cm−3, while the theoretical proportional-ty constant appearing in Eq. (2) leads to a value of 19.9 kC cm−3.his value is smaller than the theoretically predicted value by aactor of 0.68. This suggests that Eq. (2) has to be improved. Due

Fig. 4. The relative errors of the anodic oxide line widths (d) written on Al thin filmsat various speeds (a–c).

C.M. Siket et al. / Electrochimica Acta 113 (2013) 755– 761 759

Fv

tdtittAtitbc

tovtaal

Fv

ig. 5. Dependence of the anodisation cross section current density on the writingelocity for 6 V anodisation and full anodisation at 28 V.

o the geometry of the oxide line formation, where dragging of theroplet with a circular geometry results in an overlapping betweenwo consecutively addressed regions, a geometric factor must bentroduced multiplicatively in Eq. (2). For the given dimensions ofhe SDCM, the geometric factor must account for the difference inhe slope measured in Fig. 5 and the theoretically predicted value.t high writing velocities the fully anodized lines show little devia-

ion from the predicted linear behaviour, described by the linear fitn Fig. 5. However, writing of dielectric lines shows a strong devia-ion from the linear fit for velocities above 15 mm min−1. This maye due to insufficient time for the anodisation to reach stationaryonditions.

In Fig. 6 the inverse of the effective oxide thickness vs. anodisa-ion time and writing velocity is presented. The effective thicknessf the oxide lines was calculated by using Eq. (2) in which theelocity is described by the ratio between the tip diameter andhe anodisation time. Therefore, a direct relationship between bothbscissas can be easily derived. Aluminium is a valve metal and its

nodic oxide grows according to the high field regime [28]. Fol-owing this model, the time dependence of the thickness cannot

1 10 1000.6

0.8

1.0

1.2

1.4

1.6

1/h/108·m

-1

t / s

= 24 nm·V-1

10 1v / mm·min-1

ig. 6. Inverse of the effective oxide thickness vs. anodisation time and writingelocity.

Fig. 7. Representative EIS spectra recorded on an Al2O3 line after being written at aspeed of 1 mm min−1.

be modelled analytically, but a good approximation is given by aninverse logarithmic growth law of the form:

1h

= A − B ln(t) (3)

The factor B is given by

B = 1ˇ(E − E0)

(4)

where ̌ is the high field constant and E − E0 is the potential drop onthe oxide layer [28]. From a plot of inverse thickness vs. logarithmicanodisation time, as shown in Fig. 6, the value of B can be exper-imentally obtained by linearly fitting the data. The measurementpoints show small deviations from the obtained linear fit presentedas a solid line. A single stray point measured at high velocity wasexcluded from the fitting process. Using the so obtained B, thevalue of the high field constant ̌ is calculated as 24 nm V−1. Thisis in good agreement with previously reported values [28]. At highanodisation velocities (corresponding to short anodisation times)the deviation from the assumed uniform surface anodisation timedistribution becomes more prominent due to the inverse logarith-mic growth law (Eq. (3)). This is most likely the reason for the straypoint in Fig. 6 at around 1 s anodisation time. Decreasing the writingvelocity over two time decades resulted in a roughly 20% thickeranodic oxide film. In principle this may suggest that denser anodicoxides are to be expected at low writing speeds. This expectation isderived from approaching the time for establishing the stationarycorrosion conditions on Al2O3 which is about 1000 s. Also, higheranodisation times result in a reduced amount of defects/traps inthe oxide film [1].

Electrochemical impedance spectroscopy was used for checkingthe quality of the dielectric oxide films grown at 6 V vs. SHE usinga wide range of writing velocities starting from 0.3 mm min−1 upto 3.4 mm min−1. For this purpose, the SDCM operating in the con-tact mode was used. Between all the impedance curves recordedfor oxides obtained by using different writing velocities no signif-icant difference is directly observable. As example, a typical Bodeplot is shown in Fig. 7 for a writing speed of 1 mm min−1. Start-ing from high frequencies, the impedance shows a typical plateaudescribing the series resistance between electrolyte and oxide.Decreasing the frequency resulted in the appearance of a −1 slopein the impedance plot, suggesting mainly capacitive behaviour. Thephase also changes with frequency. Starting from a maximum value

of approximately −25 ◦ at high frequencies, the phase is smoothlydecreasing to the value corresponding to ideal capacitive behaviourof −90 ◦. These observations are supporting the idea of using a sim-ple equivalent circuit for fitting all of the measured impedance

760 C.M. Siket et al. / Electrochimica Acta 113 (2013) 755– 761

Fw

scdfi

nsrafottTiioeboeadctqtdfi

swioFttasfTbipc

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

4

5

6

7

8

9

I/µA

d / cm

10

12

14

16

18

20

22

linear fit for d < 1 cm

linear fit for d > 1 cm

h/nm

theory for 9 V vs. SHE

Fig. 9. Recorded anodisation current and corresponding thickness for the anodis-ation of an Al graded sample. Blue and black line represents linear fits of the datatill and above d = 1 cm, respectively. The red line represents current and thickness

ig. 8. Writing speed dependence of capacitance and electric permittivity of theritten anodic oxide lines.

pectra. The equivalent circuit used, contained a series resistanceharacteristic to the electrolyte and a parallel R–C combinationescribing the electrical conduction and capacitance of the Al2O3lm.

Upon fitting each impedance spectra (ZView Software – Scrib-er Associated Inc.) the values of anodic oxide capacitance can betudied as a function of the writing speed. These results are summa-ized in Fig. 8(a). The capacitance values are normalized to the areaddressed by the SDCM. Values for the capacitance per area rangingrom 700 to 950 nF cm−2 were found, showing no clear dependencyn the anodic oxide writing velocity used. The obtained effec-ive oxide thickness calculated as previously described was usedo estimate the electrical permittivity of the anodic oxide films.hese results are presented in part (b) of Fig. 8. Unlike the capac-tance behaviour, the electrical permittivity slightly decays withncreasing writing speed. This may suggest that the quality of thexide is improving with lowering the writing speed. For the high-st writing speed values strong fluctuations are visible due to theefore discussed thickness inaccuracies. Even though an increasef the normalized capacitance for faster writing speeds would bexpected due to the thinner oxide film thickness, a rather invari-ble behaviour is observed. This can be understood in terms of aecreasing electrical permittivity with the writing speed, thereforeompensating the decreasing thickness. The behaviour of the elec-rical permittivity reinforces the idea of increasing the anodic oxideuality for lower writing velocities. Nevertheless, various applica-ions of the written anodic oxide lines may require a certain speedue to technological processing requests which would lead towardsnding an optimized writing velocity.

The specially designed thickness gradient sample was used fortudying the effect of variable Al thickness on the anodic oxideriting process. The value of the absolute current recorded dur-

ng the oxide writing along the thickness gradient with a speedf 3.4 mm min−1 and at a potential of 9 V vs. SHE is presented inig. 9 as a function of the oxide line length. Additionally, the oxidehickness calculated using Eq. (2) is presented as a secondary ver-ical axis. Each of the points presented in Fig. 9 was obtained byveraging 50 experimental data recordings measured at the corre-ponding distance d and the error bars were statistically obtainedrom the standard deviation. Two distinct regions are identified.he first region, roughly 1 cm long, has a positive slope which can

e observed in the fitting line of the respective points. This slope

s characterizing the Al thickness gradient since the anodisationotential of 9 V vs. SHE was sufficiently high to cause a completeonversion of the Al thin film into Al2O3. The thickness gradient

as predicted by theory for anodisation at 9 V vs. SHE. (For interpretation of the ref-erences to color in this artwork, the reader is referred to the web version of thearticle.)

found in this manner was 3.6 nm cm−1. A second region is reachedas soon as the applied anodisation potential is not sufficient to fullyanodize the increasing thickness Al thin film. As a result only a con-stant thickness oxide line will be further written. Therefore, a zeroslope fitting line is expected for the remaining data points in Fig. 9.Additionally, the equivalent thickness line for an oxide grown at9 V vs. SHE with a writing speed of 3.4 mm min−1 theoretically cal-culated according to Eq. (2) is shown. A small deviation betweenthe fitted and theoretical lines is observed suggesting a good cor-relation between experiment and theory, a thickness accuracy of0.7 nm being found.

4. Conclusions

Scanning droplet cell microscopy was used for direct writingof anodic oxide lines on the surface of Al thin films. The struc-tural modifications of the written oxide lines as a function of thewriting speed were studied by analyzing the relative error of theline widths. High writing velocities had as a result a straight andsmooth Al/Al2O3 boundary of the written lines while low speedsproduced instabilities of the oxide line width. A speed threshold forthis behaviour was found around 1 mm min−1. A theoretical modelbased on the Faraday law was proposed to explain the constantanodisation current measured during the writing process. The the-ory explains the anodisation current for fully anodised oxide lines atall velocities, whereas for dielectric oxide films a deviation from thetheory was observed for higher writing speeds. From the recordedcurrent it was possible to calculate the effective oxide film thicknessdepending on writing speed. Through the use of electrochemicalimpedance spectroscopy the quality of the written oxide lines wasassessed. By fitting the recorded spectra the capacitance per areawas obtained and the electrical permittivity of the oxide was deter-mined. An increase of the electrical permittivity was found whiledecreasing the anodic oxide writing speed. A thickness graded sam-ple was used for studying the effect of Al layer thickness variation onthe anodic oxide writing process. The excellent dielectric propertiesof the oxide lines suggest applications in plastic electronics.

Acknowledgement

The financial support through the ERC Advanced InvestigatorsGrand “Soft-Map” is gratefully acknowledged.

imica

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

C.M. Siket et al. / Electroch

eferences

[1] A.W. Hassel, M.M. Lohrengel, Preparation and properties of ultra thin anodicvalve metal oxide films, Materials Science Forum 185 (1995) 581.

[2] A.W. Hassel, K. Fushimi, M. Seo, An agar-based silver|silver chloride referenceelectrode for use in micro-electrochemistry, Electrochemistry Communica-tions 1 (1999) 180.

[3] M.M. Lohrengel, A. Moehring, M. Pilaski, Electrochemical surface analysis withthe scanning droplet cell, Fresenius Journal of Analytical Chemistry 367 (2000)334.

[4] A.I. Mardare, A.W. Hassel, Quantitative optical recognition of highly repro-ducible ultrathin oxide films in microelectrochemical anodization, Review ofScientific Instruments 80 (2009) 046106.

[5] A.I. Mardare, A. Savan, A. Ludwig, A.D. Wieck, A.W. Hassel, A combinatorialpassivation study of Ta–Ti alloys, Corrosion Science 51 (2009) 1519.

[6] A.I. Mardare, A. Savan, A. Ludwig, A.D. Wieck, A.W. Hassel, High-throughputsynthesis and characterization of anodic oxides on Nb–Ti alloys, ElectrochimicaActa 54 (2009) 5973.

[7] A.I. Mardare, A.P. Yadav, A.D. Wieck, M. Stratmann, A.W. Hassel, Combinato-rial electrochemistry on Al–Fe alloys, Science and Technology of AdvancedMaterials 9 (2008) 035009.

[8] M.T. Woldemedhin, D. Raabe, A.W. Hassel, Grain boundary electrochemistry of�-type Nb–Ti alloy using a scanning droplet cell, Physica Status Solidi A 208(2011) 1246.

[9] S.O. Klemm, J.C. Schauer, B. Schuhmacher, A.W. Hassel, High throughput elec-trochemical screening and dissolution monitoring of Mg–Zn material libraries,Electrochimica Acta 56 (2011) 4315.

10] A.W. Hassel, M. Seo, Localised photoelectrochemical measurement with thescanning droplet cell, Electrochemical Society Proceedings PV 99-27 (1999)337.

11] J.P. Kollender, A.I. Mardare, A.W. Hassel, Photoelectrochemical scanning dropletcell microscopy (PE-SDCM), ChemPhysChem 14 (2013) 560–567.

12] J. Gasiorowski, A.I. Mardare, N.S. Sariciftci, A.W. Hassel, Scanning droplet cellmicroscopy for electrochemical characterization of organic semiconductors,Journal of Electroanalytical Chemistry 691 (2013) 77–82.

13] A.I. Mardare, A.D. Wieck, A.W. Hassel, Microelectrochemical lithography: amethod for direct writing of surface oxides, Electrochimica Acta 52 (2007) 7865.

14] Y. Li, H. Shimada, M. Sakairi, K. Shigyo, H. Takahashi, M. Seo, Formation andbreakdown of anodic oxide films on aluminum in boric acid borate solutions,Journal of The Electrochemical Society 144 (1997) 866.

[

[

Acta 113 (2013) 755– 761 761

15] Y. Liu, P. Skeldon, G.E. Thompson, H. Habazaki, K. Shimizu, Anodic film growthon an Al–21at.% Mg alloy, Corrosion Science 44 (2002) 1133.

16] H. Habazaki, K. Shimizu, P. Skeldon, G.E. Thompson, G.C. Wood, The behaviourof iron during anodic oxidation of sputtering-deposited Al–Fe alloys, CorrosionScience 43 (2001) 1393.

17] L. Iglesias-Rubianes, P. Skeldon, T.E. Thompson, H. Habazaki, K. Shimizu, Tran-sient anodic oxidation of an Al–W alloy: effects of current ratio, CorrosionScience 44 (2002) 751.

18] E. Akiyama, K. Asami, H. Habazaki, A. Kawashima, K. Hashimoto, Enhancementof corrosion resistance of amorphous aluminum alloys by alloying additions,in: Corrosion and Corrosion Protection, Electrochemical Society, 2001, pp. 1.

19] A. Mozalev, M. Sakairi, H. Takahashi, Structure, morphology, and dielectricproperties of nanocomposite oxide films formed by anodizing of sputter-deposited Ta–Al bilayers, Journal of The Electrochemical Society 152 (2004)F257.

20] A. Facchetti, M. Yoon, T. Marks, Gate dielectrics for organic field-effect transis-tors: new opportunities for organic electronics, Advanced Materials 17 (2005)1705.

21] J. Veres, S. Ogier, G. Lloyd, D. De Leeuw, Gate insulators in organic field-effecttransistors, Chemistry of Materials 16 (2004) 4543.

22] M. Kaltenbrunner, P. Stadler, R. Schwödiauer, A.W. Hassel, N. Sariciftci, S. Bauer,Anodized aluminum oxide thin films for room-temperature-processed, flexi-ble, low-voltage organic non-volatile memory elements with excellent chargeretention, Advanced Materials 23 (2011) 4892.

23] S. Goettling, B. Diehm, N. Fruehauf, Active matrix oTFT display with anodizedgate dielectric, Journal of Display Technology 4 (2008) 300.

24] A.I. Mardare, M. Kaltenbrunner, N.S. Sariciftci, S. Bauer, A.W. Hassel, Ultra-thinanodic alumina capacitor films for plastic electronics, Physica Status Solidi A209 (2012) 813.

25] A.W. Hassel, M.M. Lohrengel, The scanning droplet cell and its application tostructured nanometer oxide films on aluminium, Electrochimica Acta 42 (1997)3327.

26] K.A. Lill, A.W. Hassel, A combined �-mercury reference electrode/Au counter-electrode system for microelectrochemical applications, Journal of Solid StateElectrochemistry 10 (2006) 941.

27] A. Mozalev, V. Khatko, C. Bittencourt, A.W. Hassel, G. Gorokh, E. Llobet, X. Cor-reig, Nanostructured column-like tungsten oxide film by anodizing Al/W/Tilayers on Si, Chemistry of Materials 20 (2008) 6482.

28] M.M. Lohrengel, Thin anodic oxide layers on aluminium and other valve metals:high field regime, Materials Science and Engineering R 11 (1993) 243.