IR-sintering of ink-jet printed metal-nanoparticles on paper

Thin Solid Films 520 (2012) 2949–2955

Contents lists available at SciVerse ScienceDirect

Thin Solid Films

j ourna l homepage: www.e lsev ie r .com/ locate / ts f

IR-sintering of ink-jet printed metal-nanoparticles on paper

Daniel Tobjörk a,f,g,⁎, Harri Aarnio a,g, Petri Pulkkinen d,g, Roger Bollström b,g, Anni Määttänen b,c,g,Petri Ihalainen b,c,g, Tapio Mäkelä e,g, Jouko Peltonen c,g, Martti Toivakka b,g,Heikki Tenhu d,g, Ronald Österbacka a,f,g

a Physics, Department of Natural Sciences, Åbo Akademi University, Porthansgatan 3, FI-20500 Turku, Finlandb Laboratory of Paper Coating and Converting, Department of Chemical Engineering, Åbo Akademi University, Porthansgatan 3, FI-20500 Turku, Finlandc Laboratory of Physical Chemistry, Department of Natural Sciences, Åbo Akademi University, Porthansgatan 3, FI-20500 Turku, Finlandd Laboratory of Polymer Chemistry, Department of Chemistry, University of Helsinki, P.O. Box 55, FI-00014 Helsinki, Finlande VTT Microsystems and Nanoelectronics, P.O. Box 1000, FI-02044 VTT, Finlandf Graduate School of Materials Research, Turku, Finlandg Center for Functional Materials, Finland

⁎ Corresponding author at: Physics, Department of NUniversity, Porthansgatan 3, FI-20500 Turku, Finland. Te

E-mail address: [email protected] (D. Tobjörk).

0040-6090/$ – see front matter © 2011 Elsevier B.V. Alldoi:10.1016/j.tsf.2011.10.017

a b s t r a c t

Article history:Received 29 June 2011Received in revised form 27 September 2011Accepted 10 October 2011Available online 18 October 2011

Keywords:Printed electronicsNanoparticle sinteringPhotothermal sinteringInfraredPaper substrateAbsorptanceInkjet printingResistivity

Sintering of printed metal nanoparticles can be made not only by conventional heating, but also by, e.g., elec-trical, microwave, plasma, laser and flash lamp annealing. We demonstrate sintering by using low-cost incan-descent lamps as an effective way of obtaining highly conductive contacts of two types of ink-jet printedmetal-nanoparticle inks on paper; both alkanethiol protected gold nanoparticles and a commercially avail-able silver nanoparticle ink. This low-cost roll-to-roll compatible sintering process is especially suitable onpaper substrates because of the high diffuse reflectance, relatively high thermal stability and low thermalconductivity of paper. A volume resistivity of around 10 μΩ cm was achieved of the inkjetted silver nanopar-ticles within 15 s of exposure to an IR lamp, which corresponds to a conductivity of 10–20% of that of bulksilver. Too long exposure time and too high intensity, however, lead to darkening of the paper fibers. Boththe crack formation and the coffee ring effect of the inkjet printed gold nanoparticles were, furthermore,found to be reduced on paper as compared to glass or plastic substrates.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Paper has recently gained attention as a potential substrate [1] forlow-cost flexible electronics, partly because of its wide usage, lowcost and recyclability. Highly conducting printable conductors are de-sired in most flexible electronic applications, e.g., for contacts, wiresand antennas. While conducting polymers can be used when a rela-tively high resistance of the conductor is acceptable, metals shouldbe used when high and stable conductivities (>104 S/cm) are re-quired. Metals can be printed by using inks based on metal particlesof different size or low-viscosity inks based on organometallic precur-sors. When conducting metal structures are to be printed onto paper,high-viscosity particle-based inks are advantageous due to the limit-ed penetration of metal particles into the porous substrate. Amongthe advantages of nanoparticle (NP) based inks are the relativelysmall feature size that can be printed and the high conductivity that

atural Sciences, Åbo Akademil.: +358 2 215 4630.

rights reserved.

can be achieved after a sintering process. Ink-jet printing is rathercommonly used for applying NP-inks, because of its low-cost, flexibil-ity and low ink consumption. Among the metals that have been de-veloped for NP-based inks are silver (Ag) [2-4], gold (Au) [4-8] andcopper (Cu) [9, 10], and stable inks are made by capping the metal-NPs with sterically stabilizing organic molecules.

Sintering of printed metal-NPs, to some extent, is required in orderto achieve highly conducting structures. The coalescence of NPs attemperatures far below the melting temperature of the bulk materialis often ascribed to the strong size dependence of the melting pointof NPs, which can be derived by relatively simple thermodynamic con-siderations [11]. Assigning bulk-like thermodynamic properties to NPsis, however, problematic, and there are other processes than meltingthat can result in the coalescence of NPs. If given enough time it is pos-sible for the NPs to reduce their surface energy by forming necks be-tween each other through grain boundary and surface diffusion atmuch lower temperatures than predicted by thermodynamic consid-erations. However, in order for any sintering to take place, contact be-tween the NPsmust be enabled, which requires the removal ofmost ofthe organic ligands. This is why the lowest annealing temperature re-quired for achieving conducing structures of a metal-NP ink typically

2950 D. Tobjörk et al. / Thin Solid Films 520 (2012) 2949–2955

is given by the temperature at which the organic ligands are oxidized,pyrolyzed or sublimated [4].

The most common technique of sintering NPs is by conventionalheating in an oven or on a hot plate. However, this method usuallyrequires long annealing time (>30 min) or high temperatures(>200 °C), and is consequently incompatible with a fast roll-to-rollmanufacturing process or with low-cost flexible substrates. Thereare, nevertheless, several other sintering techniques [12] that couldbe used. The electrically induced insulating-to-conducting transitionof metal-NPs (by Joule heating) has not only been considered formemory applications [13], but also as an interesting technique forsintering metal-NPs [14]. Among the other techniques for sinteringprinted metal-NPs are microwave [15, 16] and argon plasma [17] sin-tering. It has even been demonstrated that sintering of NPs, to someextent, spontaneously can take place at room temperature throughsurface diffusion [18] when the organic dispersion agents are first re-moved by dipping in a solvent [19, 20], by tuning the Zeta-potential ofthe NPs [21], by oxidizing the protective organic ligand [22], by in-cluding a destabilizing electrolyte in the dispersion agent [23] or bydesorption into a photo paper substrate [24].

Another possibility is to induce the sintering by using various lightsources. A popular method is to use a laser, since this enables selectiveheating of the printed areas on the substrate, and even patterning ofconducting structures by sintering or ablation [25]. An additional ad-vantage is that the wavelength of the laser can be chosen to matchthe local light absorption maximum of the NPs, which improves theenergy transfer efficiency. However, using a laser is not only expen-sive, but also less suitable for larger areas, and best results are typicallyobtained when using a slow writing speed of around 0.2 mm/s. Adifferent type of light source that has been shown to be promisingfor sintering metal-NPs is the use of a xenon flash lamp [26, 27] emit-ting millisecond high-intensity pulses (e.g. 25 kW/cm2). The shortsintering time reduced the problems of oxidation of the Cu-NPs in airand of heating a plastic substrate (after filtering away the UV-light).

In this work, we used incandescent lamps for sintering ink-jetprinted metal-NPs on a paper substrate. Two different NP-inks wereused; one based on alkanethiol protected small Au-NPs, and one com-mercial Ag-ink consisting of relatively large NPs with a polymeric dis-persion agent. Both the resistivity and the optical appearance of thesintered ink-jet printed NP-inks were studied.

2. Material and methods

The recyclable multilayer-coated paper substrates [28] that wereused in this work were developed for having good barrier propertiesand a low surface roughness and desired porosity. The barrier layer con-sisted of kaolin (Barrisurf HX, ImerysMinerals Ltd.) blendedwith 30 pphethylene acrylic acid copolymer latex (Tecseal, Trüb Emulsions ChemieAG), while the calendered topmineral pigment layer was based on kao-lin (Barrisurf FX, Imerys Minerals Ltd.) with 12 pph of latex binder (DL920, Styron Europe GmbH). The root mean square surface roughnessof the paper substrates was less than 100 nm as measured by atomicforce microscope (AFM) on a 100 μm×100 μm area, and the pore sizein the top coating was measured to 0.04 μm by mercury porosimetry.

Microscopy glass slides and polyethylene therephtalate (PET) films(Mylar A, DuPont Teijin Films) were used as reference substrates. Inorder to reduce the surface energy of the substrate and the spreadingof the NP inks, the glass substrates were treated with a self assembledmonolayer of hexamethyldisilazane in a petridish on a hot plate at100 °C for 15 min. The cross section area of the ink-jet printed lineson the glass substrate was measured by AFM.

A commercial Ag-NP ink (SunTronic Jettable Silver U5603, SunChemicals Corp.) was used. The ink consisted of 20 wt.% Ag in the sol-vents ethanol and ethylene glycol, and had a viscosity of 10–13 mPasand a surface tension of 27–31 mN/m. The diameters of the Ag-NPs inthis type of ink are around 20–50 nm [29].

The synthesis of dodecanethiol and butanethiol protected Au-NPswas made in a similar way to what has been reported in literature[6, 7]. The dodecanethiol-Au-NPs were synthesized by first adding1 mmol of both dodecanethiol and chloroauric acid (HAuCl4) to10 ml of tetrahydrofuran and by stirring for 20 min. 3 mmol of the re-ducing agent superhydride (1 M in tetrahydrofuran) was added drop-wise to the solution and was stirred for 30 min. The resulting mixturewas precipitated in three repeated cycles; by first adding 40 ml etha-nol and centrifuging, and finally removing the supernatant and redis-solving the NPs in tetrahydrofuran.

The butanethiol-Au-NPs were synthesized by first dissolving2.7 mmol of tetraoctylammonium bromide in 80 ml of toluene,whichwas followed by adding 0.787 mmol of HAuCl4 in 25 ml of deio-nized water. The aqueous phase was removed after the transfer to theorganic phase had taken place, and 0.93 mmol of butanethiol wasadded to the mixture. 4 mmol of sodium borohydride in 25 ml ofwater was then added, whichwas followed by stirring for 3 h. Toluenewas removed by using a rotary evaporator, and the Au-NPs were pre-cipitated in ethanol, filtered and washed with ethanol and acetone,and finally dispersed in toluene.

After evaporating the organic solvents, both types of Au-NPs wereredispersed in p-xylene to a concentration of 10 wt.%, which was pos-sible to use as an ink for ink-jet printing. The size of the Au-NPs wasestimated to around 2–5 nm by wide angle X-ray scattering measure-ments. Thermogravimetric analysis was carried out at a heating rateof 10 °C/min in nitrogen atmosphere, and showed that the pyrolyzationof butanethiols took place at 100–200 °C resulting in a weight loss ofaround 10 wt.% and at around 250 °C of the dodecanethiols resultingin a weight loss of around 20 wt.%. By thermal annealing on a hot platea conducting gold-colored film was achieved already at 120 and 170 °Cfor butanethiol and dodecanethiol protected Au-NPs, respectively.

Ink-jet printing of the metal NP-inks was performed on a drop-on-demand Dimatix Materials Printer (DMP-2831, Dimatix-Fujiifilm Inc.)using a cartridge (DMC-11610) that consists of 16 piezoelectric noz-zles with a diameter of around 20 μm, and eject droplets with a nom-inal volume of 10 pL. The inkjetting frequency was 15 kHz and thecustomized voltage waveform had a maximum of around 20 V and apulse width of 10 μs, resulting in a drop velocity of up to 10 m/s.2–10 layers of the Au-NP ink were printed with a drop spacing of 20,25 and 50 μm, while only one layer with a drop spacing of 20 μm(25 μm on glass) was used for the Ag-NP ink. The line-edge roughnessof ink-jet printed interdigitated Ag-contacts on plastics and paper wasmeasured using the software SuMMIT (EUV Technology) from opticalmicroscope images that were taken with light coming from below thesubstrate. The dimensions of the inkjetted structures used formeasur-ing the resistance were 1 cm long and 0.25 mm (Au) and 1 mm (Ag)wide. In order to avoid contributions from contact resistance to theresistance measurements, 4-point resistance measurements wereperformed on a digital multimeter (Keithley 2100 USB, KeithleyInstruments Inc.).

Two incandescent lamp setups were used. The larger setup(A) consisted of three 30 cm long 2 kW strip light bulbs (IRT System,Hedson Technologies AB) with an intensity that was measured to2–4 W/cm2 at a distance of 15 cm by using a power meter (FieldMas-ter 33–0506, Coherent Inc.). The smaller setup (B) consisted of striplight bulbs (Ceramicx HQE) with a FeCrAl-resistance wire and atotal power of 500 W. By assuming that the 500 W IR-lamp worksas a black-body radiator and reaches the tabulated temperature of772 °C, the maximum intensity should be around 4 μm. A simpleway of controlling the intensity is to change the distance to the lamp.The intensity of the IR-lamp B was measured with the power meter asa function of the distance (d), and is shown in Fig. 1. The intensity fol-lows the theoretical 1/d2 dependence rather well at d>10 cm, butshould decay as 1/d at short distances. The intensity at very short dis-tances, however, showed quite large lateral variation and was hard tomeasure accurately. For sintering the printed NPs with this setup, a

Fig. 1. The measured light intensity is shown as a function of the distance to the 500 WIR-lamp.

2951D. Tobjörk et al. / Thin Solid Films 520 (2012) 2949–2955

distance of 3–15 cm was used. Another factor affecting the sinteringprocess was the substrate on which the sample was placed. A setupwas used where the paper samples only were attached at two ends,andwith the backside of the paper not being in contactwith a substrate.

The absorption spectra of spin coated metal-NP inks on glass weremeasured on an absorption spectrophotometer (U-3200, HitachiHigh-Technologies Corp.). Measurements of the absorptance ofpaper substrates (4.9 cm2) with and without ink-jet printed Ag-NPswere made by using an integrating sphere with a highly reflectingbarium sulfate surface and by illuminating with a xenon lamp(140 W) both at diffuse and at (mainly) direct illumination of thesamples. Samples consisting of a 100 nm thick thermally evaporated(5 Å/s at 1 mPa) Ag-film on PET and paper were also used as a refer-ence. The experimental setup is schematically shown in Fig. 2. A490 nm cut-off filter was applied in front of the incoming light tothe integrating sphere, in order to prevent any contributing effectfrom the photoluminescence of the paper surface. The light intensityfrom the integrating sphere was measured using a fiber optics spec-trometer (USB2000, Ocean Optics Inc.) for obtaining the whole spec-trum (300–900 nm). The power of the incoming light to theintegrating sphere from the xenon lamp, with no monochromatoror cut-off filter applied, was measured with the power meter toaround 50 mW (assuming a wavelength of 820 nm).

3. Calculation

The absorptance (A), reflectance (R) and transmittance (T) can bederived by using the measured intensities (L) of the outgoing light

Fig. 2. A schematic image of the measurement setup is shown. The intensity of the lightcoming out from the integrating sphere was measured with a spectrometer. The differentmeasured light intensities for the situations without (Ldark) or with (Lplug and Lsample)an inserted plug and/or sample were used for deriving A, R and T of the sample.

from the integrating sphere e.g. when placing the sample on thereflecting plug area. The light absorptance could be calculated byusing a similar approach as was used by Pålsson et al. [30] and deMello et al. [31] for deriving the photoluminescence. In their approach,the light absorptance is given by A=(Ldiffuse−Ldirect) /Ldiffuse. Thiswayof calculating the absorptance, however, can only be used when theincoming light in the diffuse light setup has the same intensity asthat in the direct light setup. In addition to this, all of the incominglight in the direct light setup should first be reflected directly ontothe sample (i.e. no diffuse light first reflected on the walls). Whenusing the xenon lamp as the light source for the integrating spheresetup (Fig. 2), neither of these two criteriawas fulfilled. In ourmeasure-ment setup the incoming light was lower in the diffuse light setup thanby using direct light and a part of the incoming direct light was first dif-fusely reflected on the walls before reaching the sample. This, conse-quently, resulted in too low values of the calculated absorptance.

A different method of calculating A, T and R of a sample can be de-rived by using the following relations of the measured light intensi-ties from the integrating sphere when using a constant partlydiffuse incoming light: Ldark=Lplug (1−α−β), Lsample (no plug behind)=Lplug (1−α (A+T)−β (A+T)), Lsample (plug behind)=Lplug (1−α (A+AT)−β (A+AT)), where Lplug is the intensity of the outgoing lightwhen the reflecting plug is in place, Lsample and Ldark are the measuredintensities when the plug area was replaced with the sample andwhen itwas removed,α is the fraction of the light that is directly incom-ing to the plug area and β is the fraction (coefficient) of diffuse light thatilluminates the plug area after being reflected on the walls of the inte-grating sphere (also including multiple reflections). The assumption ofa constant value of the coefficient β is, however, not exactly correct,since less secondary reflections are taking place when the plug is re-moved than with a reflecting sample. While corrections to these effectscould be applied, they were for simplicity not used in this work. Thelight absorptance was also assumed to be independent of the angle ofthe incoming light.

From the previous relations it turns out that A+AT can be simplycalculated by using that A+AT≈(Lplug−Lsample (plug behind))/(Lplug−Ldark). When the transmittance of the sample is small, this could givean approximate but slightly overestimated value of the absorptance(A≈A+AT). In a similar way the reflectance can be calculated byusing that R=1−A−T≈(Lsample (no plug behind)−Ldark)/(Lplug−Ldark).By inserting the previous expression for A+AT and that A+T≈(Lplug−Lsample (no plug behind))/(Lplug−Ldark), the absorptance can, e.g.,be calculated by using that A=(1+(A+T)) /2−√((1+(A+T))2/4−(A+AT)). Similarly, the transmittance can be calculated by usingthat T=−(1−(A+T))/2+√((1+(A+T))2/4−(A+AT)).

4. Results and discussion

4.1. Light absorption of metal-NPs and paper

The measured light transmittance of spin coated Ag-NPs anddodecanethiol-Au-NPs on a glass substrate, before and after thermalannealing on a hot plate, are shown in Fig. 3 as a function of wave-length. The local absorption maximum due to plasmons is around420 nm of the Ag-NPs and around 515 nm of the Au-NPs. Another im-portant feature is the increase in the transmittance of the unsinteredNPs with an increasing wavelength (λ>600 nm). The large decreasein the light transmission at longer wavelengths after thermal sinteringis mainly due to the increase in the reflectance of the metallic films.

In Table 1 the average values of the measured reflectance, absorp-tance and transmittance over the wavelengths 520–820 nm areshown for the paper substrate, with and without ink-jet printed andevaporated Ag. The largest wavelength dependence was measuredfor the transmittance of the clean paper substrates, which increasedwith almost a factor of two within this range. A small decrease inthe absorptance with increasing wavelength was seen for all samples.

Fig. 3. The measured transmittance spectra of (a) Ag-NPs and (b) Au-NPs on a glass substrate are shown before and after thermal annealing on a hot plate at 120 °C for 30 min andat 200 °C for 1 h, respectively.

2952 D. Tobjörk et al. / Thin Solid Films 520 (2012) 2949–2955

An interesting difference in Table 1 is that the reflectance of the Ag-film is much higher on PET than on paper. The enhanced light absorp-tion of Ag on paper is an obvious advantage for the light induced sinter-ing, and could be explained bymultiple reflections (light trapping) andsurface plasmons. An even larger increase in the absorptance has previ-ously been reported for laser ablated structures on a metal surface [32].

The relatively low light absorptance of the clean paper substratecan be explained by the high reflectance of the calcium carbonateand kaolin pigment particles that were used as fillers and in the topcoating. The reflectance of calcium carbonate is even higher thanthat of kaolin, but the reflectance of kaolin increases with the wave-length, which also is the case for bleached uncoated papers [33].

In the optimal situation only the NPs should be heated, while thepaper should remain unaffected. However, unfortunately there is atrade-off between the absorptance of the NPs and of the paper sub-strate, since the reflectance of the kaolin particles that covered thepaper surface increases with the wavelength, while the metal-NPs ab-sorb most light at shorter wavelengths. Unfortunately, we were notable to measure the absorptance further into the IR-part of the spec-trum. However, if the absorptance of paper increases at higher wave-lengths (>3 μm), the heating of the paper could be reduced by usingan incandescent lamp with a high temperature.

The around 10 times lower light absorptance of the paper sub-strate than of the Ag-NPs on paper obviously results in a lower tem-perature increase of the surrounding unprinted paper area uponillumination. The increase in temperature of the paper substrate andthe NP-layer, furthermore, depends on the values of the specificheat, density and thickness. Considering that the specific heat isaround 5 times larger, the thickness around 500 times larger andthe density around 10 times lower of paper than of Ag, this would re-sult in a roughly 250 times larger temperature increase of the Ag-NPsthan of the paper substrate for the same amount of absorbed light en-ergy. Some energy is, however, consumed by the decomposition ofthe organic protecting agents in the NP-film, and some energy is dis-sipated to the surrounding. Another important factor is the thermal

Table 1The reflectance, absorptance and transmittance of the paper substrate with and with-out evaporated Ag and IR-sintered inkjetted Ag-NPs are shown. The values of a PETsubstrate with evaporated Ag are shown for comparison. The values were calculatedfrom the measured intensities from the integrating sphere setup (in Fig. 2) by usingthe equations derived in Section 3, and the average values and standard deviationswere taken over the wavelength range of 520–820 nm, where the values were relative-ly constant.

R (%) A (%) T (%)

Paper 79±4 5±2 16±4Ag-NPs on paper 28±1 55±2 17±2Ev. Ag on paper 57±2 38±2 4±1Ev. Ag on PET 85±3 9±3 5±1

conductivity, which is more than around 1000 times larger of Ag thanof paper. Because of the low thermal conductivity of paper (k≈0.01–1W/Km) [34, 35], the hot metal film heats the paper substrate mainlyin the interface region. This was confirmed by the observation after arelatively long IR-irradiation that the backside of the paper substrateonly turned dark on the areas where the metal-NPs were printed.

The heating of the paper substrate by thermal conduction from themetal-NPs could be reduced by using a higher intensity and therebyreducing the exposure time. While very high light intensities can beachieved with a flash lamp or a laser, they are, however, more expen-sive than incandescent lamps, and there might be problems withorganic residues in some NP-inks if the annealing is too fast. Further-more, it is easier to control the exposure by using a light source with alower intensity. The IR-sintering was, however, not compatible withlow-cost plastic substrates, such as PET, that quickly were deformedbecause of the large thermal expansion coefficients and the lowglass transition and melting temperatures.

4.2. Resistivity

The measured volume resistivity (ρ) of inkjetted dodecanethiolprotected Au-NPs (5 layers of a 0.25 mm wide and 10 mm long line)on paper after IR-sintering was down to 25 μΩ cm, which correspondsto around ten times the bulk resistivity of Au. While lower resistivityvalues of printed Au-NPs of even down to 3–14 μΩ cm have beenreported, e.g. by laser sintering [7, 36-38] and thermal annealing(>200 °C) [39, 40] on glass or polyimide, in this work the Au-NPswere printed directly onto paper. Although Ag and Cu, typically, arepreferred over Au because of thematerial cost, theremay still be appli-cations where Au is more suitable, e.g., because of its high work func-tion. As a result of the relatively large surface roughness and porosityof paper substrates, the resistivity of thin conductors on paper sub-strates is generally higher than on plastics or glass [1, 41]. However,it is not possible to reach the bulk resistivity level in thin metal filmseven by thermal evaporation on very smooth substrates [42], due tosurface and grain boundary scattering.

The measured volume resistivity after IR-sintering of the ink-jetprinted commercial Ag-NP-ink on paper was lower than that of theAu-NP-ink. In Fig. 4 the measured volume resistivity of inkjettedAg-NPs on paper is shown as a function of the exposure time to an in-candescent lamp (A). While a volume resistivity of less than 8 μΩ cmwas reached within 20 s, the paper surface usually turned yellow-brown upon this long IR-exposure. Consequently, with the usedpaper substrate and IR-sintering setup the highest acceptable volumeresistivity limit for the application in mind of the inkjetted Ag-NPsshould be above 10 μΩ cm. Considering the relatively small thickness(300–500 nm) of the Ag-layer on paper, this is a really low resistivitylevel, and less than 10 times the bulk resistivity of Ag. This is also sim-ilar to the volume resistivity values that were reported by Yung et al.

Fig. 4. The measured volume resistivity of inkjetted Ag-NP-structures (10 mm long and1 mmwide) on paper is shown after different exposure times to the IR-lamp A at an in-tensity of 2–4 W/cm2. The dashed line in the graph indicates the bulk resistivity of Ag.

2953D. Tobjörk et al. / Thin Solid Films 520 (2012) 2949–2955

[27] when sintering Ag-NPs on photopaper by using a flash lamp(>10.5 μΩ cm). However, by electrical sintering for 1 min of aninkjetted Ag-NP-structure between two contacts on photopaperAllen et al. [14] were able to reduce the volume resistivity down to al-most 2.7 μΩ cm. Rather low resistivity values of sintered Ag-NPs havealso been reported on plastic substrates by using microwave(5–30 μΩ cm) [15, 16] and plasma sintering (10 μΩ cm) [17] whenpreheating the structures at 110–150 °C. Even by using various chem-ical sintering techniques at room temperature, volume resistivityvalues of down to 6–90 μΩ cm have been demonstrated of Ag-NPson photopaper and plastic substrates after relatively long exposures[19-21, 23, 24].

Using a distance of only 3 cm to the IR-lamp B, a volume resistivityof less than 6 μΩ cm (and a surface resistivity of around 0.1 Ω/Sq) wasachieved of inkjetted Ag-NPs on a paper substrate within 10 s. How-ever, in order to avoid degradation of the paper substrate, lower in-tensity or shorter exposure time (≈3 s) should be used, resulting ina larger volume resistivity (>10 μΩ cm). When using the same dis-tance for sintering inkjetted Ag-NPs on a glass substrate, longer expo-sures were required in order to achieve conducting structures. Avolume resistivity of less than 9 μΩ cm was measured after 1 min,and around 3 μΩ cm after 5 min of exposure to the IR-lamp. The lon-ger sintering time on glass than on paper can be explained by thelarger thermal conductivity and thickness of glass, the large transmit-tance of glass and the larger reflectance of the Ag-film on glass thanon paper.

An advantage with light induced sintering of metal-NPs over othersintering techniques is the negative feedback of the process. The heat-ing process actually slows down as the NPs are sintered due to the

Fig. 5. Optical microscope images are shown of IR-sintered ink-jet printed Au-NPs on glass (Au-NPs.

decrease in the light absorptance. This prevents overheating of theNP-structures, which easily can take place in e.g. microwave sinteringor electrical sintering (Joule heating) processes that have a positivefeedback [43]. A disadvantage with paper substrates is, however,that the polymeric binder in the coating and the cellulose fibersturn darker after too large light exposure and heating, which resultsin a positive feedback to the heating process of the paper.

4.3. Coffee stain effect and cracks of ink-jet printed metal-NPs on glassand paper

Opticalmicroscope images of ink-jet printed lines of dodecanethiolprotected Au-NPs in xylene on glass slides and on paper are shown inFig. 5 after exposure to the IR-lamp A for 2 min and 1 min, respective-ly. One of the differences between the printed structures on paper andon glass is the clearly visible coffee stain effect on the glass substrate.The coffee stain effect is a common phenomenon in ink-jet printingand is caused by the faster evaporation at the edges of the depositedlow viscosity ink (where the contact line is pinned) due to the largerair interface and by the compensating capillary flow within the ink.The reduced coffee stain effect on paper, as compared to glass or plas-tics, could be explained by the liquid absorption into the substrate thatcompetes with the initial spreading and the final evaporation.

Another issue with the printed Au-structures, clearly visible inFig. 5, is the formation of cracks that can be detrimental to the conduc-tivity. An important observation is that the crack formation of NP-inksseems to be less distinct on paper substrates than on plastics. The visu-al difference was also confirmed by electrical measurements showingthat most of the Au structures on glass were non-conducting, while avolume resistivity of down to 25 μΩ cm was measured on the papersubstrate after exposure to IR-lamp B.

Crack formation has previously been reported in literature [10, 13,44-47], and is caused by the large volume reduction during the ther-mal annealing when both the organic protecting agents are removedand the structure is getting denser due to the sintering process. Thiseffect is especially seen when the thickness of the NP-film is large,the volume reduction is large, there are pre-existing cracks and an in-homogeneous stress distribution [10, 45], the adhesion to the sub-strate is poor [13, 48] and the thermal expansion of the substrate isdifferent from that of the NPs [47]. The weight concentration of thealkanethiols in the dried Au-NP ink was around 20 wt.%, which be-cause of the large difference in density corresponds to a volume con-centration of around 75 vol.%. It is not hard to imagine that this largevolume reduction upon thermal annealing could lead to large stressand even cracks in the NP-film. While the final cracks form during

top) and on paper (bottom), where the brighter (yellow colored) areas are the sintered

Fig. 6. Optical microscope images of inkjetted interdigitated Ag-contacts are shown on a PET (to the left) and on a paper (to the right) substrate after thermal annealing at 120 °C for30 min and exposure to the IR-lamp A for 15 s, respectively. The black color of the printed lines is due to the illumination from below the substrates that was used in order to en-hance the contrast.

2954 D. Tobjörk et al. / Thin Solid Films 520 (2012) 2949–2955

the sintering process, some aggregation takes place already duringthe drying process of the NP-ink. While the thermal expansion ofplastic substrate is larger than that of the NPs, the thermal expansionof glass and paper [34] is rather similar to that of metals. The advan-tage of better sintering and less crack formation on paper than onplastics and glass can also be explained by the better adhesion tothe paper substrate and by the porosity of paper that allows penetra-tion into the substrate of solvents, organic ligands and products fromthe decomposed organic ligands.

A higher print resolution was achieved when ink-jet printing thecommercial hydrophilic Ag-NP ink than with the Au-NP ink, andthere was no problem with crack formation or the coffee stain effectneither on plastics nor on paper, as shown in the optical microscopeimages in Fig. 6. This could be explained by the lower organic to NPvolume ratio, the larger size of the Ag-NPs than of the Au-NPs, thehigher ink viscosity and the slower evaporating solvents. Another ad-vantage of using larger NPs is that less NPs are absorbed into thepores of the paper substrate. The average distance between theprinted interdigitated Ag-contacts was around 33 μm both on plasticsand paper, while the line-edge roughness was around 1.6 μm on plas-tics and 4.6 μm on paper. The larger lateral variation on paper than onplastics can be explained by the larger topological and surface energyvariations on paper.

5. Conclusions

IR sintering is a fast, low-cost and roll-to-roll compatible methodof achieving highly conducting structures of printed metal-NP inkson a mineral pigment coated paper. The technique is especially suit-able on paper substrates because of the high diffuse reflectance, lowthermal conductivity and high thermal stability of paper. In additionto this, a larger light absorptance of the printed NPs on paper thanon plastics or glass was found, and was explained by multiscatteringon the paper. The lowest obtained volume resistivity values after IR-sintering were down to 25 μΩ cm of the inkjetted Au-NPs andbelow 10 μΩ cm of the inkjetted Ag-NPs on paper, which is lessthan ten times the bulk resistivity of pure Au and Ag, respectively.While coffee stain effect and crack formation were observed for theinkjetted Au-NP ink on glass, both of these problems were stronglyreduced when using paper substrates instead of plastics or glass,which was explained by the porosity of paper and better adhesion.

Acknowledgments

Financial support is acknowledged from the European RegionalDevelopment Fund in South Finland and the Academy of Finlandthrough the National Center of Excellence program. The authors alsoacknowledge EUV Technology for providing the SuMMIT softwarethat was used for calculating the line-edge roughness.

References

[1] D. Tobjörk, R. Österbacka, Adv. Mater. 23 (2011) 1935.[2] Y. Li, Y. Wu, B.S. Ong, J. Am. Chem. Soc. 127 (2005) 3266.[3] K.J. Lee, B.H. Jun, T.H. Kim, J. Joun, Nanotechnology 17 (2006) 2424.[4] B.T. Anto, S. Sivaramakrishnan, L.-L. Chua, P.K.H. Ho, Adv. Funct. Mater. 20 (2010)

296.[5] M.J. Hostetler, J.E. Wingate, C.-J. Zhong, J.E. Harris, R.W. Vachet, M.R. Clark, J.D.

Londono, S.J. Green, J.J. Stokes, G.D. Wignall, G.L. Glish, M.D. Porter, N.D. Evans,R.W. Murray, Langmuir 14 (1998) 17.

[6] M. Brust, J. Fink, D. Bethell, D.J. Schiffrin, C.J. Kiely, Chem. Soc. Chem. Commun.(1994) 801.

[7] D. Huang, F. Liao, S. Molesa, D. Redinger, V. Subramanian, J. Electrochem. Soc. 150(2003) G412.

[8] Y. Wu, Y. Li, P. Liu, S. Gardner, B.S. Ong, Chem. Mater. 18 (2006) 4627.[9] P. Pulkkinen, J. Shan, K. Leppänen, A. Känsäkoski, A. Laiho, M. Järn, H. Tenhu, ACS

Appl. Mater. Interfaces 1 (2009) 519.[10] J.S. Kang, H.S. Kim, J. Ryu, H.T. Hahn, S. Jang, J.W. Joung, J. Mater. Sci.: Mater. Elec-

tron. 21 (2010) 1213.[11] P. Buffat, J.-P. Borel, Phys. Rev. A 13 (1976) 2287.[12] R.M. German, Sintering Theory and Practice, John Wiley & Sons, New York, 1996.[13] S. Sivaramakrishnan, P.-J. Chia, Y.-C. Yeo, L.-L. Chua, P.K.-H. Ho, Nat. Mater. 6

(2007) 149.[14] M.L. Allen, M. Aronniemi, T. Mattila, A. Alastalo, K. Ojanperä, M. Suhonen, H.

Seppä, Nanotechnology 19 (2008) 175201.[15] J. Perelaer, B.-J. de Gans, U.S. Schubert, Adv. Mater. 18 (2006) 2101.[16] J. Perelaer, M. Klokkenburg, C.E. Hendriks, U.S. Schubert, Adv. Mater. 21 (2009)

4830.[17] I. Reinhold, C.E. Hendriks, R. Eckardt, J.M. Kranenburg, J. Perelaer, R.R. Baumann,

U.S. Schubert, J. Mater. Chem. 19 (2009) 3384.[18] S. Iwama, K. Hayakawa, Jpn. J. Appl. Phys. 20 (1981) 335.[19] D. Wakuda, M. Hatamura, K. Suganuma, Chem. Phys. Lett. 441 (2007) 305.[20] D. Wakuda, K.-S. Kim, K. Suganuma, IEEE Trans. Comp. Packag. Technol. 32 (2009)

627.[21] S. Magdassi, M. Grouchko, O. Berezin, A. Kamyshny, ACS Nano 4 (2010) 1943.[22] M.J. Coutts, M.B. Cortie, M.J. Ford, A.M. McDonagh, J. Phys. Chem. C 113 (2009)

1325.[23] M. Grouchko, A. Kamyshny, C.F. Mihailescu, D.F. Anghel, S. Magdassi, ACS Nano 5

(2011) 3354.[24] M. Allen, J. Leppäniemi, M. Vilkman, A. Alastalo, T. Mattila, Nanotechnology 21

(2010) 475204.[25] S.-H. Ko, H. Pan, C.P. Grigoropoulos, C.K. Luscombe, J.M.J. Fréchet, D. Poulikakos,

Nanotechnology 18 (2007) 345202.[26] H.-S. Kim, S.R. Dhage, D.-E. Shim, H.T. Hahn, Appl. Phys. A 97 (2009) 791.[27] K.C. Yung, X. Gu, C.P. Lee, H.S. Choy, J. Mater. Proc. Tech. 210 (2010) 2268.[28] R. Bollström, A. Määttänen, D. Tobjörk, P. Ihalainen, N. Kaihovirta, R. Österbacka, J.

Peltonen, M. Toivakka, Org. Electron. 10 (2009) 1020.[29] H. Meier, U. Löffelmann, D. Mager, P.J. Smith, J.G. Korvink, Phys. Stat. Sol. A 206

(2009) 1626.[30] J.C. de Mello, H.F. Wittmann, R.H. Friend, Adv. Mater. 9 (1997) 230.[31] L.O. Pålsson, A.P. Monkman, Adv. Mater. 14 (2002) 757.[32] Y. Yang, J. Yang, C. Liang, H. Wang, Opt. Express 16 (2008) 11259.[33] T. Shakespeare, J. Shakespeare, Anal. Chim. Acta 380 (1999) 227.[34] E. Bøhmer, in: J. Borch, M.B. Lyne, R.E. Mark, C.C. Habeger Jr. (Eds.), second ed.,

Handbook of Physical Testing of Paper, Vol. 2, Marcel Dekker, New York, 2002.[35] P. Gerstner, Heat transfer through porous multiphase systems: measurement,

modelling and application in printing of coated papers, Doctoral Dissertation,TKK Reports in Forest Products Technology Series A13, Espoo, 2010.

[36] J. Chung, S. Ko, N.R. Bieri, C.P. Grigoropoulos, D. Poulikakos, Appl. Phys. Lett. 84(2004) 801.

[37] S.H. Ko, J. Chung, H. Pan, C.P. Grigoropoulos, D. Poulikakos, Sens. Actuators A 134(2007) 161.

[38] N.R. Bieri, J. Chung, S.E. Haferl, D. Poulikakos, C.P. Grigoropoulos, Appl. Phys. Lett.82 (2003) 3529.

2955D. Tobjörk et al. / Thin Solid Films 520 (2012) 2949–2955

[39] H.-H. Lee, K.-S. Chou, K.-C. Huang, Nanotechnology 16 (2005) 2436.[40] Y. Wu, Y. Li, B.S. Ong, P. Liu, S. Gardner, B. Chiang, Adv. Mater. 17 (2005) 184.[41] A.C. Siegel, S.T. Phillips, M.D. Dickey, N. Lu, Z. Suo, G.M. Whitesides, Adv. Funct.

Mater. 20 (2010) 28.[42] B. O'Connor, C. Haughn, K.-H. An, K.P. Pipe, M. Shtein, Appl. Phys. Lett. 93 (2008)

223304.[43] T. Öhlund, J. Örtegren, H. Andersson, H.-E. Nilsson, Sintering Methods for Metal

Nanoparticle Inks on Flexible Substrates, NIP 25 and Digital Fabrication Proceed-ings, Society for Imaging Science and Technology, 2009, p. 614.

[44] R.K. Bordia, A. Jagota, J. Am. Ceram. Soc. 76 (1993) 2475.[45] J.R. Greer, R.A. Street, J. Appl. Phys. 101 (2007) 103529.[46] J. Perelaer, A.W.M. de Laat, C.E. Hendriks, U.S. Schubert, J. Mater. Chem. 18 (2008)

3209.[47] D.J. Lee, J.H. Oh, H.S. Bae, Mater. Lett. 64 (2010) 1069.[48] J.-K. Jung, S.-H. Choi, I. Kim, H.C. Jung, J. Joung, Y.-C. Joo, Phil. Mag. 88 (2009) 339.