Laser-Based Deposition Technique: Patterning Nanoparticles into Microstructures

Transcription

1 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures Edward M. Nadgorny Jaroslaw Drelich Michigan Technological University, Houghton, Michigan, U.S.A. INTRODUCTION The laser-based particle deposition (BPD) technique is one of direct-write techniques that can be used for patterning nanoparticles into a variety of microstructures. If properly developed, BPD may have wide application in manufacturing novel electronic, sensing, catalytic, and other devices that require specific structures, combinations of unlike materials, and unconventional substrates. The technique makes use of laser-induced optical forces to guide particles by a narrow laser beam from an atomizer source until depositing them on a substrate. One of the BPD versions based on aperture guidance is reviewed in this paper. The discussion covers both the fundamentals of both laser guidance and solid liquid interactions during patterning, including recommendations for the best setups, as well as a more detailed description of optical and other forces exerted on particles inside the BPD apparatus. Some examples of nanoparticle deposition and patterning are also presented. ASER-BASED PARTICE DEPOSITION TECHNOOGY A wide range of techniques generally referred to as direct-write technologies have recently emerged in response to the needs for rapid prototyping of micropatterns made of continuous phase or fine particles. [1] Technologies such as inkjet printing, micropen writing, plasma spray, laser particle guidance, and matrix-assisted pulsed-laser evaporation are capable of direct fabrication of microstructures on a variety of substrates using a wide range of materials. All of these approaches have the advantage of avoiding the need for masks and complicated lithographic technologies associated with microscale electronic device fabrication. These rapid prototyping techniques are often able to consolidate powder materials into three-dimensional structures. aser-based particle deposition technique is one of many such techniques under recent development that has a promising potential for the fabrication of electronics and sensors based on microvolume materials. The BDP technique makes use of laser-induced optical forces to guide particles by a narrow laser beam until depositing them on a substrate. As described in Ref. [2], BPD has many advantages over existing patterning techniques....in contrast to photolithography, the process adds material to the surface (as opposed to etching material) and does not require harsh or corrosive chemicals. In contrast to robotic microspotting, deposition, inkjetting, and screenprinting, particles are strongly localized within the laser beam and the deposition accuracy can be below one micrometer. Most importantly, nearly any material in either liquid or aerosol suspension can be captured and deposited as long as convection and gravity are weaker than the guidance forces (typically in the nanonewton range). Additionally, BPD allows one the fabrication of various porous structures of nanoparticles, which can serve, for example, as microsensors. [3] Although perfection in nanoparticle arrays is not required in such applications, achieving sufficient control over deposited micropatterns to manufacture porous microstructure is not a trivial task. There are only two promising techniques for such applications at present, inkjet writing and BPD. Each of them has different merits and demerits, so that they rather complement each other being suitable for a different range of applications. Although both techniques can produce similar structures, the resolution of inkjet writing at present reaches only mm; BPD is not restricted either by solvent viscosity to the degree required by inkjet techniques. More importantly, BPD is also temperature-flexible: When the transporting particles absorb laser light, the resulting heat, if required, can activate chemical reactions and/or phase transformations during and after deposition. At the same time, the laserinduced heat can also be rather low: Because of high transporting velocities and such option as cooling carrier gas, both the droplet and substrate temperatures can be controlled allowing only moderate short-duration temperature increases. Because the BPD-transported material is in the form of either liquid precursor or colloidal suspension atomized to micrometer-size droplets, the deposited particles can be of any size between several nanometers to about a micron. Finally, BPD is capable Dekker Encyclopedia of Nanoscience and Nanotechnology 1565 DOI: /E-ENN Copyright D 2004 by Marcel Dekker, Inc. All rights reserved.

2 1566 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures of codepositing dissimilar materials offering a potential to fabricate both heterostructured and porous patterns that combine distinct electronic, organic, and biological materials using the same basic fabrication technology. The BPD technique, although still in its early stages of development, has been used to fabricate micron-scale structures from a wide variety of materials, such as metals (Au, Ag, In, Cu, Pt), semiconductors (Si), oxides (CuO, RuO 2, Al 2 O 3 ), ferroelectrics (BaTiO 3 ), ionic crystals (NaCl, KI), biological cells (neurons, others) on such various substrates as glass, alumina, silicon, sapphire, and various polymers. [2 8] In principle, BPD allows depositing virtually any material that can be suspended in a liquid or formed by decomposition of a precursor. The primary restrictions on the BPD technique are that the substrate is transparent to light at the laser frequency (to avoid substrate heating) and that the particles being deposited should not undergo unwanted reactions (e.g., decomposition) during transport in the laser beam. Historically, it was Ashkin [9] who was the first to demonstrate the optical levitation that moved and suspended micrometer-sized particles against gravity in a focused laser beam. Stable three-dimensional trapping of such particles in the minimum-waist region of a highly focused laser beam was demonstrated later. [10] These and other findings formed the basis of the so-called optical tweezers used in many applications. [11] Renn and Pastel [4,5] were the first to demonstrate laser guidance and trapping of mesoscale particles from suspensions and liquid droplets in hollow-core optical fibers. The obtained data gave rise to industrial applications of a laser-guided direct write of electronic and biological components. [2] Finally, BPD technique with aperture guidance was demonstrated recently. [12] Unlike optical trapping techniques, [9 11] BPD utilizes optical transverse gradient forces to confine particles inside weakly focused laser beams and take advantage of the radiation pressure to move particles axially along through an aperture onto a substrate within the near-field region. This version of the BPD technique as developed at Michigan Technological University is described in this review. The principal BPD setup is shown schematically in Fig. 1. It includes a mist chamber (MC), source chamber (SC), process chamber (PC), and a laser system. The mist chamber includes an atomizer and delivery system to make a mist of micron-sized liquid droplets (or particles) that are precursors for the deposited material. The droplets are delivered from the atomizer to the source chamber by a carrier gas, and a mass flow meter measures and controls the gas flow. The source chamber is connected to a process chamber by a micron-sized aperture, with a typical size in the range from 15 to 35 mm. The droplets are usually much smaller, of the order of a micron or less. The laser beam is coupled into the aperture by an optical lens of low numerical aperture to allow particle guidance to a substrate that is mounted on a translational stage. The supply chamber both provides an environment suitable for the deposition process and shields the particles from Fig. 1 Schematic of the MTU BPD system: source of deposited material (carrier gas supply and atomizer inside mist chamber MC); supply (SC) and process (PC) chambers linked by a high-power laser aperture (details shown in the insets); laser system (laser and focusing lenses); movable substrate on a translational stage. (View this art in color at

3 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures 1567 Fig. 2 Examples of materials deposited by the BPD apparatus: (a) SEM micrograph of microspheres formed by 7-nm Au S(CH2)7CH3 particles deposited by a 200-mW laser; (b) AFM image of a cluster of 100-nm polystyrene particles; (c) SEM micrograph of an array of 400-nm polystyrene spheres partially coated with 100-nm polystyrene spheres; (d) optical image of the 7-nm Au S(CH2)7CH3 particles deposited on an array of gold microelectrodes by a 200-mW laser; (e) similar to (d), after an electrode was produced by in situ melting by a 500-mW laser beam. (f) Structure made of 20-nm Au S(CH2)7CH3 particles deposited after atomization of 2 4 g/ suspension of nanoparticles in toluene. (View this art in color at convection currents that can deflect particles out of a laser s confinement. As shown in Fig. 1, in contrast to optical tweezers,[9 11] BPD allows droplets to be captured continuously from the supply chamber and transport them onto the substrate to write on it. Guidance in the BPD aperture-based system is achieved by a weakly focused laser beam and diffracted laser radiation within near-field limits at the front and the rear of the aperture, respectively.[12] Also shown in Fig. 1 (inset b) are the main forces exerted on a particle in the apparatus. Three of them, the optical radiation force Fz, the optical transverse gradient force Fr, and optical axial gradient force Fga, are resulted from the interaction of the particle with incident laser beam, and the drag force FD is due to the interaction of a moving particle with the ambient. They and other forces will be discussed in detail below. Examples of different materials deposited by the BPD apparatus can be found elsewhere.[2,3,7,8,13] Several additional examples are shown in Fig. 2. BPD BASICS Optical Field Inside the BPD System It is important to discuss the spatial distribution of laser optical field near an aperture as the light intensity provides both axial and transverse guidance of particles. In particular, it is the optical axial force Fz and the optical transverse gradient force Fr that determine the guidance efficiency, the particle velocity, and the throughput rate of an BPD apparatus. We will consider the laser beam with a TEM00 (or fundamental) mode that can be represented by an ideal Gaussian intensity profile. Such a beam with the total power P propagates in free space along the z direction toward a centered circular aperture with radius a (Fig. 1b). The intensity (or irradiance) I(z, r) of such a beam is then.[14] Iðz; rþ ¼ I0 e 2r 2 =w2 ð1þ where I0(z) = 2P/pw2 is the axial (peak) intensity at r = 0, w(z) is theqspot radius diverging with the distance as ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi wðzþ w0 1 þ ðz=zr Þ2, and the smallest spot radius w0 (at the beam waist) is at z =0. For such a beam, the distance zr, the waist spot radius w0, and the beam fullangle divergence y are beam parameters defined asa zr ¼ pw20 =l; y ¼ 2lM 2 =pw0 a ð2þ Eq. 2 is analogous to the optical (the Smith Helmholtz) invariant in an optical system comprising only lenses. In practice, lasers as a rule are characterized not only by the beam-waist spot w0 but also by the beam divergence y.

4 1568 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures where M2 1 is a factor representing the beam quality (for an ideal Gaussian beam, M2 =1), and for a beam behind a circular aperture with radius a, zr = pa2/l. The parameter zr is referred to as the Rayleigh range indicating an approximate border between the near-field diffraction region (at z <zr) and the far-field diffraction region (at z> zr). The extent along the beam between ±zr points relative to the beam waist is commonly considered as the collimated waist region. The diffraction angle y of a beam spreads in the far field, where w(z) w0z/zr = lz/pw0 at z zr. The angle y is often considered as a laser primary parameter because divergence is of fundamental importance for many laser applications. However, any one of the beam parameters completely characterizes the optical field along the entire Gaussian beam because only one is independent at a particular wavelength l. Therefore the axial intensity I0 of a laser cannot be increased by reducing w0 without also increasing its divergence. As we will see, this is especially important for the BPD systems. The beam intensity falls off very fast with radius r beyond the spot size w: at r = w the intensity decreases to 0.135I0 (1/e2 criterion) and to about 0.01I0 (often called 99% criterion) at r = pw/2. The circular aperture, which links the supply and process chambers of the BPD apparatus (Fig. 1), changes the beam optical field by two different ways. First, due to aperture truncation, the power transmission of the passing beam can be cut off considerably depending on the ratio a/w0, especially at a/w0 1. Even an aperture with radius a = w0 transmits only 86% of the total beam power P. Fig. 3 The transverse diffracted intensity I calculated for different ratios a/w0 as a function of the transverse distance r from the beam axis normalized by the aperture radius a, and corresponding diffraction images. The arrow in the plot marks the first minimum for the ratio a/w0 = 1.4. The ratio a/w0 increases from (a) to (e); the latter corresponds to a typical working setup with a/ w In (a), the size of the first ring is about 70 mm. The images were obtained with a digital camera in the far-field diffraction region of the 35-mm aperture. (View this art in color at

5 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures 1569 However, this effect can be avoided by increasing the ratio: An aperture transmits as much as 99% after the ratio increase to a/w 0 p/2 = 1.6. Secondly, diffraction of the beam on an aperture produces more complex intensity patterns in both the near-field and far-field regions. [14,15]a In the far field, the axial intensity decreases due to standard Gaussian beam divergence: According to Eq. 1, the axial intensity changes with distance as I 0 (z)=(z R / z) 2 I 00, where I 00 is the axial intensity at the beam waist. Additionally, a diffracting aperture perturbs the beam creating well-known diffraction effects that significantly change the optical field distribution including the central far-field lobe intensity. In the near field, the changes in the characteristics of the diffracted beam are the most essential. The intensity Fresnel ripples appear and become more and more distinguished as the ratio a/w 0 decreases. The intensity variation DI/I can be as large as ± 20% even for a 99% aperture. (The far-field central intensity will also be decreased by 20% by such an aperture.) As a result, rings of a higher and lower intensity are typical in the near-field region, their location, amplitude, and number changing with distance and with the ratio a/w 0 (Fig. 3). On the other hand, the average intensity is practically a constant or even increasing with distance in the near-field region at z<z R. Finally, there is a peak at or near the beam axis that might be rather spiky closer to the aperture. For a typical double-frequency YAG-laser (l = 532 nm, y = 0.5 mrad and M 2 = 1), Eq. 2 then gives w 0 0.7mm and a long collimated waist region of 2z R 5.4 m. Focusing such a collimated beam with a lens of the focal length f= 35mm decreases the original waist spot diameter 2w 0 to 2w 0 =2yf =35 mm. At the same time, the divergence increases to y = w 0 y/w 0 20 mrad and the beam Rayleigh range z R diminishes to 1.8 mm. Obviously, the same z R =1.8mm is for the beam after passing an aperture with the diameter 2a =2w 0 =35mm. The intensity of a real laser beam can be different than that of an ideal Gaussian beam because of both a possible mixture with higher-order modes and effects of nonideal optics. Although the Rayleigh range is still the same in such cases, the waist spot size and divergence are larger, as can be seen from Eq. 2 for M 2 >1. The most distinguished feature, however, is the more complex intensity profile of such laser beams: Instead of Eq. 1, the intensity distribution exhibits local maxima and minima inside the collimated region or focus. Setup Optimization of the BPD System The discussion above and computational results based on the paraxial approximation of a Gaussian beam diffracted by a circular aperture [15 17] suggest several factors important for the optimal setup of BPD. The calculations and experimental comparison of the diffraction patterns with the best particle flux through the 25- and 35-mm apertures have shown that one of the most important factors is the ratio a/w 0. The ratio affects the far-field diffraction pattern as can be seen in Fig. 3 for seven ratios a/w 0. The calculations are performed for conditions close to real experimental setups: Gaussian beams with w 0 =3 20mm and circular apertures with diameters from 35 to 5 mm. Only three pairs of maxima and minima without a much higher central maximum are displayed. As seen, both the positions of the first minima and the diffraction contrast (determined as the intensity difference between the first maximum and minimum) depend on the ratio a/w 0. The contrast decreases as the ratio increases, and the first minimum completely vanishes at a/w Experimentally, the ratio a/w 0 can be changed by either displacing different focusing lenses relative to the laser (changing the waist radius w 0 only), or changing the aperture size, or both. The experimental data confirm the calculation results (Fig. 3). The images are obtained by axially displacing a 35-mm aperture closer and closer to the beam waist, until the first minimum cannot be seen anymore in Fig. 3(e). Using such images, one can find the best alignment of the laser lens aperture system. Numerous deposition experiments with apertures and laser beam waists of various sizes confirmed that deposition is impossible when the contrast between the first minimum and the first maximum is high, such as in Fig. 3(a) through (d). As the contrast decreases, deposition is finally made possible near a/w This number is larger than the ratio of required for a maximum axial intensity in the far-field range but closer to p/2 that corresponds to 99% power transmission. [14] The other important factor is the divergence of the beam ahead of the aperture. A large divergence in the region near the beam waist (as utilized in optical tweezers) [11] can lead to reverse radiation pressure, especially for small w 0 l. [18] Our experiments with 100-nm polystyrene spheres, 25-mm circular aperture, and 30-mm focusing lens have shown that the setup can be optimized to provide guidance at as low laser power as P100 mw. [17] High accuracy is essential for direct-write applications. Because it is determined by many factors, systematic studies are required to estimate their contributions. Preliminary experiments with five different materials [19] have shown that the deposition accuracy is inversely proportional to the transport distance. The accuracy also strongly depends on the material properties of the transported particles, such as the refractive index and absorption, and relatively weakly on the laser power when measured inside a relatively low-power range. To increase deposition precision and utilize the features of the nearfield intensity pattern, the substrates during BPD

6 1570 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures operations are usually located at z<z R, i.e., closer to the aperture. Particle Dynamics and Forces in the BPD System Two special features critically influencing the forces exerted on droplets inside a BPD apparatus are a nonuniform optical field with on-axis maximum intensity and a droplet small size of Rl. There are three types of forces important for BPD: optical forces, radiometric forces, and the drag force. The optical forces arise from scattering and absorption of light by the particle; the radiometric forces result from the temperature gradients either inside the particle or in the medium; and the drag force develops due to the resistance of the medium to a moving particle. Additionally, the force of gravity and hydrodynamic interaction forces between particles are also present. However, a characteristic feature of the dynamics of droplets inside the BPD system is that the interaction forces between particles and the force of gravity are, in most cases, considerably less than the other three. For submicron particles typical of the BPD technique, the force of gravity is below the N range and decreases significantly with the particle size, so that we can safely ignore the gravity force. The droplets can also be considered independent of one another inside the supply chamber because of their relatively low concentration so that we can neglect hydrodynamic forces as well. On the other hand, the particle interaction might be important during delivery from the atomizer to the supply chamber, as the clouds of particles not only move faster than the individual particles [26] but also the particles can coagulate forming larger droplets or clustered particles. When a particle is illuminated by an incident beam, its electromagnetic energy and momentum are lost because of absorption and scattering processes. The use of conservation of energy and momentum allows calculating the beam-induced stress tensor on the surface of the particle and the corresponding optical force field; such forces are collectively called the radiation forces. They depend on the size, shape and material of the particle, the beam polarization and intensity distribution, and the position of the particle inside the beam. [20 22] Then the forward force exerted on a spherical particle of radius R inside a gaseous medium with the refractive index n b 1 by a beam with the axial intensity I 0 and the waist spot w 0 R is given by [20] where Q pr, Q ext, and Q sca are the so-called efficiencies for radiation pressure, extinction, and scattering, respectively, Q ext =Q sca +Q abs (where Q abs is the efficiency for absorption), c is the velocity of light in free space, and the average cosine of the scattering angle cos y is the scattering asymmetry parameter. Because the refractive index of gases is very close to 1, we can use c instead of the velocity of light in the medium in Eq. 3. For lowabsorbing particles (such as liquid droplets of water-based dilute solutions), Q pr ð1 cos yþq sca ; in the opposite limit of particles with no scattering (ideal black particles), Q pr Q abs. Eq. 3 allows calculating the forward axial force F z based on various approximations of light scattering models. [20 22] To include both absorption and scattering processes into the calculations, the refractive index m of the particle is represented by a complex number, m = n+ik, where the real part n is responsible for scattering and the imaginary part k (or the extinction coefficient) is responsible for absorption. The extinction coefficient k is related to the absorption coefficient a (sometimes also called the turbidity or attenuation coefficient) of the Beer ambert law, I t = I i exp( al), as k = la/4p, where the incident intensity I i decreases to I t over the path length l. The refractive index m can vary over a wide range of 1m<1, where the lower limit corresponds to translucent liquids and solids, and the upper limit, typical of metals in the infrared region, corresponds to total reflection. The other way to characterize particles is through the relative dielectric constant e r = e 1 +ie 2 = m 2, so that e 1 = n 2 k 2 and e 2 =2nk. In the calculation, the particle size is represented by the normalized size parameter x = 2pR/l, which for particle sizes typical for BPD, 50R500 nm, is approximately within a range of 0.6x6. b Such a range requires a rigorous calculation to find F z, and it can be performed using the generalized orenz Mie theory (GMT). [20] If the particle is spherical, scattering can be simulated with the GMT independently of the spheres diameter and refractive index. An example of such a calculation is shown in Fig. 4; [19] similar calculations were performed for weakly absorbing spheres. [20 22] As seen, the average envelope of the Q pr x curve shows a relatively small change as x decreases from 20 to 3 remaining at Q pr 0.5, with a flat maximum near x = 6. Taking BPD typical conditions for a collimated beam of w 0 10 mm and P = 1 W yields F z and N for the particles of R = 500 and 200 nm, respectively. c The other feature is a superposition of ripple structures with maxima and sharp spikes that originate in resonant electromagnetic normal modes of a sphere. [22] Such structures are typical of the morphology-dependent resonances (MDRs); they F z ¼ pr2 I 0 c Q pr ¼ pr2 I 0 c ðq ext Q sca cos yþ ð3þ b Note that although we have propelled and deposited as small solid particles as 7 nm, [3,7,8] they were in suspensions. c Note that the force on particles with stronger absorption might be significantly higher; we deposited metallic particles in suspension at as low laser power as less than 100 mw.

7 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures 1571 magnitude larger than the thermal energy of the ambient molecules, which ensure particle guidance under ambient conditions. The models also revealed that an additional, axial gradient force F ga can exist in tight-focused systems that can produce trapping in the diverging part of the beam (the inverse radiation pressure). To avoid this, a better optical design with focusing lenses of low numerical aperture is required. The drag force F D is given by the well-known Stokes formula F D ¼ 6pZRv ð5þ Fig. 4 Typical size dependence of the efficiency factor Q pr for a polystyrene sphere (the refractive index m=1.431) obtained by numerical calculation based on a generalized orentz Mie theory with no absorption. The arrows show the corresponding radii R for l=532 nm. (From Ref. [19].) were both predicted by the Mie theory and observed in experiments involving several different techniques. [23] During the deposition process, laser-induced evaporation of the moving droplets changes the particle size so that the forward force experiences the corresponding fluctuations of MDRs. However, the droplet motion in gaseous medium is normally overdamped because of the drag force (see below) so that such short peaks should not affect the particle motion. The classical Mie orentz theory assumes only planar waves in the incident optical field, [29] but it can also be applied to a real laser beam provided that the waist spot w 0 R. To include into consideration also gradient forces, important for both BPD and optical trapping, other approach is required. A gradient force arising from electrical forces on a polarizable particle can be calculated directly from a standard expression, F g ¼ ða=2þre 2 ð4þ where re is the gradient of laser electric field E and a =4pR 3 (e r 1)/(e r +2) is the particle polarizability for small particles when x1/k. [22] For larger particles, the expression for a can be corrected. [24] Alternative models have also been proposed to calculate both gradient and forward forces for small particles of 2Rl. The models use either an extended GMT [18] or a two-component approach, [25] and in all cases the transverse gradient force F r is larger than the axial force F z. Such gradient force F g is large enough to confine the particles to the laser beam: The force is attracted toward high intensity for positive a and thus provides the necessary guidance along the laserbeam axis. Estimates also demonstrate [4] that the potential created by such gradient forces is several orders of where Z is the ambient shear viscosity (R/l1, l is the mean free path of the ambient molecules), and v is the particle velocity. Depending on the particle radius R, the force F D diminishes either linearly or steeper with R. As l10 nm in air at atmospheric pressure, F D decreases linearly as far as R is less than several hundreds of nanometers, and approximately quadratically at R 20 nm (molecular kinetics regime). A transition region exists at intermediate particle sizes where one of the two dependencies trends to the other. The drag force F D can be reduced substantially, practically almost to zero, by decreasing the ambient pressure. Absorption not only directly increases the forward force but also determines the guided particle temperature and therefore creates two additional effects: the appearance of the radiometric force F T and initiation of such thermal processes as evaporation, in-flight chemical reactions inside transported droplets, and even melting of strongly absorbing particles. Estimates show that the absorption becomes especially important for the particles with R 50 nm as the scattering contribution into the forward force decreases as R 6 in this size range, while the absorption contribution only as R 3. The radiometric force F T arises from a temperature gradient DT inside the particle illuminated from one side and points against the gradient, i.e., in the forward axial direction. Because in BPD systems R/l1, F T provides an additional axial forward force due to a rather complicated and different than normal viscous flow process, often called radiometric flow or thermal creep. Under a simplified assumption that DT is constant over a spherical particle (which is rarely the case for dielectric particles [6] but provides a reasonable estimate) with the thermal conductivity w, the radiometric force is given as. [26] F T ¼ 3pZ2 R g pm RDT 3pZ2 R g a b 2pMw RI 0 ð6þ where R g is the molar gas constant, p the pressure, M the molecular weight of the gas, and a b the absorption factor (for the absolutely black particle, a b = 1). Estimates show that proper illumination can create F T that is comparable

8 1572 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures with F z in strongly absorbing particles; an effect of such a radiometric force was also observed experimentally. [27] SOID IQUID INTERACTIONS DURING PATTERNING The BPD technique relies on precise and controlled deposition of liquid droplets, suspensions, and wet formulations. The capillary forces operating in these multiphase systems govern the dimensions and morphology of the deposited clusters as well as the extent of fusion among deposited clusters. An understanding of the solid liquid interactions is crucial in achieving three major objectives in laser-deposition technology: 1) to prepare suspensions that remain stable during the time of deposition; 2) to control phase separation (promote or avoid, depending on the need) in solid particle-in-liquid droplets transported by a beam; and 3) to manipulate the spread of deposited droplets and formation of pattern over a substrate. Fulfillment of these tasks requires an understanding and the manipulation of the surface properties of solids, both dispersed in liquid and used as substrates for patterning. If surface properties of particles or substrates cannot be manipulated, selection of the appropriate solvent or solution is needed. The solid liquid interactions specific for the processes important for laser-deposition systems have not been considered in depth so far. The discussion presented below relies rather on fragmentary experimentation and observations during the laser deposition process combined with the knowledge of solid liquid interactions extracted from broad surface chemistry research activities. A unified (but simplified) theory on solid liquid interactions is reviewed in the first part of this section and then the applicability of this theory to laser-deposition systems is discussed in subsequent parts. Solid iquid Interfacial Energy The interfacial interactions are analyzed in this paper using the ifshitz van der Waals ewis acid base interaction theory developed by van Oss, Good, and Chaudhury at the end of 20th century. [28 30] According to this theory, the total free energy of interaction between two surfaces, e.g., liquid and solid, is given by DG S ¼ DG W S þ DGAB S where DG W S and DG AB S are the ifshitz van der Waals component and ewis acid base component, respectively, of the free energy of interaction between a solid (S) and a liquid (). The DG W S component refers to interfacial apolar interactions caused by orientation, induction, and ð7þ dispersion forces, whereas the DG AB S component reflects the contribution from the (ewis) acid base (polar) interfacial interactions. Both components, as well as their combination, are assumed in this theory to follow the Dupre equations, [29,30] DG W S DG AB S ¼ gw S ¼ gab S gw S g W gab S g AB DG S ¼ g S g S g ð10þ where g is the surface or interfacial free energy; the subscripts S,, and S refer to solid and liquid surfaces, and solid liquid interface, respectively; and the superscripts W and AB denote the ifshitz van der Waals and acid base interactions, respectively. Because of the symmetrical nature of the W interactions, both the total W interaction energy and the W surface free energy components follow a combining rule, [28 30] ffiffiffiffiffi S ¼ 2 g W S gw DG W g W ð8þ ð9þ ð11þ qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi 2 S ¼ g W g W ð12þ S The combining rule does not apply to asymmetrical AB interactions; instead, they have been defined (a priori) by the following two equations, S ¼ 2 g S gþ þ DG AB g AB i ¼ 2 g þ i g i g þ S g ð13þ ð14þ where the superscripts and + refer to the electron donor (ewis base) parameter and electron acceptor (ewis acid) parameter, respectively. Combining Eqs. 11 and 13 with Eq. 7 then yields DG S ffiffiffiffiffi ¼ 2 þ g W S gw g þ S g þ g S gþ ð15þ Another expression for the total free energy of interaction between two surfaces DG S can be obtained by substituting the Young s equation, g S g S ¼ g cos y ð16þ into Eq. 10, in the form DG S ¼ g ð1 þ cos yþ ð17þ

9 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures 1573 Table 1 Surface tension (surface free energy) components for selected liquids a iquid g [mj/m 2 ] g W [mj/m 2 ] g [mj/m 2 ] g + [mj/m 2 ] Benzaldehyde Benzene Chloroform Diiodomethane Dodecane Ethanol Ethyl acetate Ethylene glycol Formamide Glycerol Heptane Methanol N-Octanol Toluene Water b 25.5 b a See footnote h. b Assumed standard values. Source: From Ref. [33]. where g is the surface tension of liquid and y>0 is the contact angle for liquid on the solid surface. The two equations (15 and 17) are commonly used for the calculations of the surface free energy and surface free energy components of a solid surface and a liquid from contact angle measurements. Because Eq. 15 contains three unknowns (g S W, g S, g S + ), the contact angles of three liquids with known surface tension must first be measured, so that three simultaneous equations can then be solved. [28 30] The g W, g +, and g components of surface free energy were determined for a number of liquids and solids and several examples are shown in Tables 1 and 2; the values for g + and g are relative, as they refer to a standard value assumed for water, g + =g =25.5 mj/m 2. [29,30] Table 2 Surface free energy components for selected solids a Solid g S [mj/m 2 ] g S W [mj/m 2 ] g S [mj/m 2 ] g S + [mj/m 2 ] Alumina Apatite Calcite Cellulose Cellulose nitrate Dolomite Glucose Hematite Human fibronectin (dry) actose Maltose Polyethylene Polyisobutylene Polymethyl methacrylate Polystyrene Polyvinyl alcohol Polyvinyl chloride Rutile a See footnote h. Source: From Ref. [33].

10 1574 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures Preparation of Suspensions The stability and homogeneity of the suspension used for the generation of a mist in laser-based deposition technology is crucial for maintaining a constant concentration of solid particles carried by droplets. The quality and reproducibility of patterned structures strongly depend on the concentration of particles and reproducibility of this concentration in the droplets. The stability of suspensions is controlled by particle particle forces, which include the van der Waals (apolar) interactions, ewis acid base (polar) interactions, electrostatic interactions, and Brownian movement forces. [29 31] As intense mixing results in dynamic conditions of suspension, the effect of Brownian movement forces on the stability of suspensions used in the formulation of mist can be ignored. For the same reason, any flocculation effects caused by long-range forces and by the presence of a secondary minimum in attraction [29 31] can also be neglected in this simple analysis. A more accurate analysis of the interactions between surfaces as a function of distance is beyond the scope of this paper and therefore is excluded from further discussion. Finally, although the electrostatic forces originating from the Coulomb interaction between surface charges are important to many aqueous suspensions and some other organic solventbased suspensions (and can be used in stabilization of suspensions), they are also ignored to simplify our analysis; the electrostatic interactions in apolar and weakly polar solvents have often only negligible effect on the stability of suspensions. Keeping the abovementioned simplifications in mind, it can be said that suspensions of particles remain stable if solid liquid adhesion exceeds the cohesion forces for liquid molecules and particles. As a result, the interaction energy between two solid surfaces immersed in a liquid can be estimated from the ifshitz van der Waals ewis acid base interaction theory using the following equation: [29,30] DG SS ¼ 2 g W S þ g W 2 þ 4 g þ S g þ g þ S g S g S gþ g þ g ffiffiffiffiffi g W S gw ð18þ Spontaneous dispersion and complete stability of the suspension can only be accomplished when the DG SS value is positive. [29,30] However, such a strong stability of suspension is not always necessary in the formulation of the mist for laser deposition. In our research, different suspensions with small negative DG SS values were successfully used in making the mist and used for patterning. d For example, 8 20-nm gold nanoparticles modified with self-assembled monolayers of thiols terminated with either methyl or benzyl functionality have been used for patterning of porous chemiresistor sensors. [3,13] The gold-thiolate particles functionalized with CH 3 groups disperse well in many apolar or weakly polar solvents such as saturated hydrocarbons, chloroform, and toluene, but because of the hydrophobic nature of these particles they flocculate in polar solvents such as water and, to a lesser extent, in ethanol and methanol. The behavior of Au S(CH 2 ) n CH 3 particles, or other particles, in solvents can be predicted from the calculated DG SS value. e Examples of calculation of the DG SS value for Au S(CH 2 ) n CH 3 dispersed in different solvents are shown in Table 3. It can be seen that the DG SS value for Au S(CH 2 ) n CH 3 particles is close to zero for such good solvents as chloroform, toluene, or dodecane but increases for polar solvents from 5.4 mj/m 2 for ethanol to over 102 mj/m 2 for water. f The surface free energy characteristics for gold-thiolate particles change drastically if the self-assembled monolayer of a long-chain thiol with CH 3 functionality is replaced with a monolayer of a short-chain thiol ended with the benzyl group (Table 3). The Au S(CH 2 ) 2 C 6 H 5 particles disperse spontaneously in chloroform, ethylene glycol, and water but have a tendency to flocculate in toluene, dodecane, ethanol, and methanol. As water is a convenient liquid carrier in the laser-based deposition technique because of its high surface tension and limited evaporation, the formulation of stable suspensions of d Spherical and relatively large particles, 100 nm and larger, are easier to disperse in liquids than nanoparticles having a diameter of a few nanometers and crystalline shape due to increasing adhesion expressed per volume of the particle with decreasing particle size. e ong-range colloidal forces should be included in detailed analysis of the system; see Ref. [33] and footnote 8 for more details. f The drawback of the ifshitz van der Waals ewis acid base theory is that small DG SS values are not very reliable. The values of electron acceptor and electron donor parameters for both solids and liquids can change significantly with only a small variation in measured contact angles. For example, it should be recognized that the DG SS values in Table 3 were calculated based on not always precise surface free energy components of solids and liquids and therefore the values shown in Table 3 should be treated as estimates. This is particularly true for methanol and ethanol, for which electron donor and electron acceptor parameters need to be determined with better accuracy. Additionally, ethanol can carry a trace amount of water as a contaminant or pick it up from the laboratory humid environment. Such a contamination might affect the suspension stability if water separates or adsorbs on the particle surface. The other concern is that gold nanoparticles might carry less perfect self-assembled monolayers of thiols as those commonly formed on flat macroscopic surfaces of gold. Any uncoated areas of gold particles will enhance the attractive interactions between suspended nanoparticles.

11 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures 1575 Table 3 The interaction energy for selected apolar and polar liquids with polystyrene and gold-hexadecanethiol Polystyrene (g S W =42, g S =1.1, g S + =0 mj/m 2 ) Au S(CH 2 ) n CH 3 ; n >7 (g S W =24.6, g S =gs + ffi0 mj/m 2 ) a Gold-thiolate Au S(CH 2 ) 2 C 6 H 5 (g S W =37.0, g S =52.7 g S + =0.46 mj/m 2 )*** iquid #G SS [mj/m 2 ] #G SV [mj/m 2 ] #G SS [mj/m 2 ] #G SV [mj/m 2 ] #G SS [mj/m 2 ] #G SV [mj/m 2 ] Chloroform +4.7 (dissolves) Toluene 2.6 (dissolves) Dodecane Ethanol ( 13.3) b (+4.9) b ( 5.4) b ( 2.0) b ( 4.0) b (+16.1) b Methanol ( 18.6) b (+1.4) b ( 9.6) b ( 5.6) b ( 4.0) b (+14.6) b Ethylene glycol Water a Arbitrarily chosen values based on the data from Ref. [49]. b Estimate due to uncertain values for the electron donor and electron acceptor parameters for ethanol and methanol h. Au S(CH 2 ) 2 C 6 H 5 nanoparticles in water without stabilizing chemicals is advantageous. The selection of a carrier liquid for polymeric particles is often more difficult. For example, Table 3 shows the calculated DG SS values for polystyrene dispersed in different apolar and polar liquids. Positive or close-to-zero negative value is predicted for chloroform and toluene. These two solvents, unfortunately, dissolve polystyrene. g In such cases more polar solvents are required for use in the formulation of suspension. Because of the poor compatibility between polystyrene and polar solvents such as ethanol, ethylene glycol, or water, the suspensions formulated without stabilizing agents will flocculate even under intense mixing conditions. A modification of the surface properties of polystyrene particles is required and it can be accomplished by using surfactants that reduce the hydrophobicity of the polymer and improve the polar interaction with liquid and enhance the repulsive interactions between surfactant-coated particles. Nanoparticles in Transported Droplets Suspensions with a reduced solid liquid interaction favor the formation of aggregates, as discussed in the previous section, and also often promote accumulation of particles at the liquid/droplet surface. A film of particles formed at a liquid surface increases the droplet s surface rigidity and influences the structure of deposited clusters. The particles carried on the surface of a droplet form more-orless regular monolayer structures that often promote the formation of spherical clusters with a well-organized assembly of particles. Also, due to reduced liquid air area, the rate of liquid evaporation is reduced during transportation by the laser beam. The tendency for particles to attach to the liquid droplet surface can be predicted by the ifshitz van der Waals ewis acid base interaction theory by calculating the interaction energy between a particle and a liquid gas interface: [29,30] ffiffiffiffiffi DG SV ¼ g W þ 2 þ q ffiffiffiffiffiffiffiffiffiffi g S gþ g W S gw þ g þ S g q 2 ffiffiffiffiffiffiffiffiffiffi g gþ ð19þ The positive DG SV value indicates that the particle will avoid the liquid gas interface, whereas the negative value indicates the preferential adhesion of particles to this interface. More negative DG SV values point toward stronger adhesion of the particle to the liquid air interface, which ultimately leads to a larger portion of the particle that is exposed to a gas phase after attachment. The position of the particle at the liquid gas interface is determined by the wetting characteristics of the solid expressed in contact angle values. [33]h Table 3 shows g Entropic term must be added to the equation on DG SS to calculate the interaction energy of dissolution and predict the behavior of the polymer in a solvent (Ref. [33].) h The contact angle measured for a liquid in contact with a nanoparticle is usually different from the contact angle measured on large (millimetersized) particles as a result of a contribution of the linear free energy at the highly curved three-phase line (Ref. [34].)

12 1576 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures examples of calculated DGSV values for polystyrene and gold-thiolate particles. Because of the positive DGSV values, polystyrene does not adhere to the surface of ethanol, methanol, and dodecane, although because of negative DGSS values the polystyrene particles will likely flocculate inside the bulk of these liquids. Therefore the structure and, to a lesser extent, shape of the deposited clusters might become unpredictable. It can be seen from the data in Table 3 that it requires the use of liquids with relatively high surface tension value such as ethylene glycol or water to generate droplets coated with films of polystyrene particles. Unfortunately, as also discussed in the previous section, the formulation of suspensions of polystyrene in water (or ethylene glycol) that are stable for at least a few seconds (time required to prepare a mist and transport the droplets to a substrate) is a challenge. In such systems the use of surface-active compounds or additional particulates that adsorb on the surface of polystyrene and strengthen the interactions of the particles with liquid is usually a need. The surface free energy of self-assembled monolayers of HS(CH2)nCH3 (n >7) on a gold surface is much smaller than the surface free energy of polystyrene, and, therefore, the gold-thiolate particles can coat almost any liquid listed in Table 3. However, small negative DGSV values for chloroform, dodecane, and toluene indicate that adhesion of Au S(CH2)nCH3 particles to these liquids is weak and the particles can easily detach from the liquid air interface. As a result, irregular and irreproducible clusters usually result from deposited suspensions formulated using one of these three solvents. The use of polar solvents such as ethanol, methanol, water, and ethylene glycol promotes the accumulation and alignment of Au S(CH2)nCH3 particles at the droplet surface. Again, due to strong particle particle interactions in these liquids, good stability of Au S(CH2)nCH3 suspension is difficult to maintain. The scenario is completely different for Au S(CH2)2C6H5 particles. As the DGSV values shown in Table 3 indicate these particles will avoid the liquid air interface when suspended in chloroform, toluene, dodecane, ethanol, and methanol. Although the Au S(CH2)2C6H5 particles disperse extremely well in chloroform, they tend to flocculate in toluene, dodecane, ethanol, and methanol. The DGSV values are also negative for the Au S(CH2)2C6H5 particles dispersed in ethylene glycol and water. These two systems, with positive DGSS values and negative DGSV values, seem to be exceptionally suitable for the deposition of uniformly shaped clusters of particles. Particles with negative DGSV values tend to form well-organized films on the surface of a liquid droplet during transportation and at the same time particle particle repulsive forces (positive value of DGSS) are re- sponsible for a good dispersion of the remaining particles in the liquid bulk, protecting them against aggregation. As discussed above, the control of adhesion of particles to the droplet surface promotes the structuring of deposited clusters. Fig. 5 shows two extreme cases. Irregular clusters made of Au S(CH2)7CH3 particles that were deposited on a hydrophobized glass slide from particle-in-toluene suspensions are shown in Fig. 5a: DGSS = 0.3 mj/m2 and DGSV = 2.0 mj/m2. The negative value of DGSS is disadvantageous for uniform alignment and tight packing of the particles in a cluster. On the contrary, the negative value of DGSV is advantageous, although DGSV = 2.0 mj/m2 indicates a weak adhesion of particles to the interface and easy disturbance of particle-at-interface organization process. Spherical clusters made of regularly aligned polystyrene particles were deposited from suspension formulated in phosphate buffer solution with the addition of proteins and are shown in Fig. 5b; DGSS ffi 0 mj/m2 and DGSV 0 mj/m2 (exact values have not been determined). Packing density and alignment of nanoparticles in the film formed over the droplet surface are controlled by the size and shape of particles and interparticle interactions operating through both liquid and gas phases.[33,35] Capillary forces are negligible in these systems because of the small dimensions of particles and resulting flat meniscuses formed by the liquid between particles.[33] This is not necessarily true for the transported and deposited droplets if the laser beam strongly interacts with the droplet content. The resulting increase in temperature of the transported droplet causes the liquid to evaporate at a higher rate. Consequently, depletion of the liquid volume in the droplet that carries a high concentration of particles forces the liquid to shrink into interparticle space, and capillary forces start to operate Fig. 5 AFM phase images of clusters made of (a) 20-nm Au S(CH2)7CH3 particles and (b) 100-nm polystyrene particles. Gold nanoparticles were deposited from toluene suspensions whereas polystyrene particles were deposited from aqueous suspensions stabilized with surfactants and addition of avidin. (View this art in color at

13 aser-based Deposition Technique: Patterning Nanoparticles into Microstructures 1577 among newly formed meniscuses aligning particles into regular spherical clusters with particles tightly bound to each other.i Capillarity Phenomena for Droplet-on-Substrate Deposits The laser beam guides droplets or clusters (wet or dry) and deposits them on a substrate. Dry clusters and clusters carrying inside a small volume of liquid practically remain intact after deposition on a substrate. It is well documented that large droplets and wet clusters might deform during impact with the substrate, particularly if they travel at high velocity.[36,37] However, because of very small Reynolds number, Re= (1.5 to 6.0) 10 4, and Weber number, We = to 10 8 for the micron-sized droplets impacting substrates during the deposition with a laser beam,j droplet deformation is negligible and has no practical consequences in patterning process. The deposited droplets with no particles or particles weakly adhered to the liquid gas surface will spread over the substrate and the extent of spreading is governed by the substrate suspension interactions. Droplets and wet clusters will also fuse into each other and spread to form larger clusters if one is deposited on top or at very close proximity of another droplet or cluster. Fig. 6 shows two examples of gold-thiolate nanoparticle clusters deposited over a silanized glass slide in either dry (or almost dry) (a) or wet (b) form. Dry spherical clusters with a diameter of about 0.2 to 1 mm remain intact after deposition, although they have a tendency to form chains and aggregates of clusters likely as a result of attractive interactions among clusters stronger than interactions between clusters and substrate (Fig. 6a). Fig. 6b shows, on the other hand, that wet clusters formed plate-like structures as a result of liquid spreading and evaporation. Partially fused (disintegrated) clusters are also visible on top of the plate-like structure in Fig. 6b. Cracks between plates reflect boundaries between fused clusters and also some of them result from contraction of the volume of the deposited material during drying. i On the contrary, fast evaporation of a liquid carrying a small quantity of particles leads to droplets of such small sizes that the laser beam cannot guide them effectively so that the droplets spread in almost every direction instead of following the path of the beam. j Reynolds number Re describes a balance between inertial and viscous forces, Re = Dvr/Z, and Weber number We expresses a ratio of inertial to capillary forces, We = rv2d/g, where D is the diameter of the droplet [ 1 (m)], v the droplet velocity at impact [ (m/sec)], r the density of liquid or suspension [ 1000 (kg/m3)], and Z the liquid viscosity [ 10 3 (kg/msec)]. Fig. 6 Scanning electron micrographs of intact clusters made of (a) 7-nm Au S(CH2)7CH3 deposited from toluene suspension and (b) plate-like structure made of the same particles but formed from wet clusters that fused to each other. Spreading of deposited liquid drops (or wet clusters) over a substrate surface is controlled by the competition between the substrate liquid adhesion and cohesion forces of liquid. This can be analyzed through the spreading coefficient SSV,[38] SSV ¼ gs gs g ð20þ which for finite contact angles (y >0) becomes SSV ¼ g ðcos y 1Þ ð21þ where gs, g, and gs are the surface free energy of solid, liquid, and solid liquid interfacial free energy; and y is the contact angle. Using the ifshitz van der Waals ewis acid base model, the spreading coefficient can also be calculated from the following equation: SSV qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffi þ þ W W 2g ¼ 2 gs g þ gs g þ g S g ð22þ A spontaneous and complete spreading of liquid to form a thin film occurs when the SSV value is positive or zero, which corresponds to zero contact angle. This situation is very undesirable in patterning technology because of lack of control of the size and shape of the deposited droplets/ wet clusters, and the resulting structures. It is therefore necessary to select the substrate on deposition of liquid droplets and wet clusters with such a wetting characteristic that y 0. The contact angle from about y = 60 to 110 is usually preferable if control of the size of deposited individual clusters is desired. The base diameter of the deposited liquid droplet or wet cluster (db) is an important parameter if the size of the deposited clusters must be controlled. This parameter can be predicted from the volume of the droplet delivered by a