Review—Electrophoretic Deposition of Phosphors for Solid-State Lighting

Electrophoretic deposition (EPD) is a facile method to produce phosphor particulate ﬁlms for solid-state lighting applications. EPD has produced ﬁlms in the conformal (deposited directly on the LED) or remote conﬁgurations (deposited on a substrate above the LED). Films of different blended phosphor compositions have been deposited to produce white emission, either by excitation with blue-emitting or near UV-emitting LEDs. Layered ﬁlms of sequentially deposited phosphors have also been shown to produce white light. The key results from both experiments and theory are described and summarized, which show the utility of EPD as a phosphor particle coating method.

Commercial white light from GaInN light emitting diodes (LEDs) arises from the combination of the blue light from the LED (450 nm), which is used to activate a broadband yellow-emitting phosphor (Y 3 Al 5 O 12 :Ce 3+ , YAG:Ce), and a small amount of red-emitting phosphor (e.g. Ca-αSiAlON:Eu 2+1 ) for color correction. The placement and arrangement of phosphors are crucial to the extraction efficiency of white-emitting LEDs. Phosphor arrangements in white-emitting LEDs are illustrated in Figure 1. Figure 1a shows the phosphor particles embedded in an encapsulant surrounding the LED and housed in a reflector cup (phosphor-in-cup), which is the configuration used for most commercial devices. Figure 1b shows a cross-sectional view of the dispersed phosphor particles surrounding a chip. Figure 1c shows a conformal phosphor layer, where the phosphor particles are coated directly onto the LED. A top view and cross-sectional view of the conformal layers produced by electrophoretic deposition (EPD) is shown in Figure 1d. A uniform, highly packed, conformal phosphor layer controls color and efficiency and improves the spatial color distribution of LEDs. 2 For near-UV emitting diodes (370-410 nm), thicker phosphor layers must be used for the conformal or remote phosphor distribution so that no UV light escapes the device.
The conformal and phosphor-in-cup distribution limits the light extraction efficiency of the device. Because the phosphor particles emit light in all directions, a large portion directly impinges on the LED chip where it can be re-absorbed (Figures 1a, 1c). This issue is critical in the conformal phosphor configuration due to the close proximity of the phosphor and the LED chip. If the phosphor is placed at a sufficiently large distance from the LED chip (remote phosphor configuration) the probability of light rays emanating from the phosphor and directly hitting the low reflectivity LED chip is small, improving the light extraction efficiency. Another advantage of the remote phosphor configuration is that it can reduce the operating temperature of the phosphor. Figure 1e shows a remote phosphor configuration, 3 in which a phosphor layer of uniform thickness is distributed over the reflector cup on a transparent substrate. However, there is still a probability of light rays being reflected by the reflector cup and being re-absorbed by the LED chip. Finally, Osram Optoelectronics has developed a new GaInN 3D structure, consisting of 2 μm diameter micro-rods that are 4 μm in height. Phosphor particles were deposited by EPD to completely fill in the regions around the micro-rods  ( Figure 1f), which increases the light scattering in the forward direction. 4 The micro-rod 3D configuration increases the brightness up to 20% and decreases the droop by 15% at 350 mA. 5 There are several parameters used to characterize the whitelight emission. One parameter is the chromaticity coordinates (x, y), which quantify the color on the CIE (Commission internationale de l'éclairage) diagram. For example, a daylight (noon) has x, y values of 0.314, 0.331. Closely related is the correlated color temperature (CCT), defined by the closeness of the coordinates to the blackbody locus. The CCT of most lamps range from 2700 K (warm light) to 6500 K (cool light). Finally, the color rendering index (CRI) is a measure of how close the light source resembles daylight (noon), which has a maximum CRI set at 100. Incandescent lamps have a CRI of 100 and fluorescent lamps range from 51-90. Target values set by the US Dept. of Energy for white-emitting LEDs are CRI > 95 and CCT between 3000-4500 K. 6 An advantage of the remote phosphor configuration is that heat from the LED is dissipated, resulting in a more uniform thermal distribution at the system level. Although most phosphors are refractory ceramics with high melting temperatures, the photoluminescence emission intensity is significantly affected in the operating temperature of a LED (∼150 • C), 1 as shown in the thermal quenching curves in Figure 2 for several solid-state lighting phosphors. 7 The reduced temperature of the phosphors, combined with a decrease in re-absorption and scattering of emitted light into the LED can lead up to a 20% improvement for the device. 8 The particle size also has a significant effect on the light scattering. Figure 3 shows the light scattering coefficient as a function of particle size. For very small particles (<∼50 nm) or large (>∼30 μm), the scattering coefficient is low. For particles in the range of commercial phosphors (5-10 μm), the scattering coefficient is large. 9 Therefore, there is great interest in producing phosphor particles in the nano-sized range, but there has been little work performed on EPD of nanophosphors.
Particulate phosphors layers (in either conformal or remote configurations) can be integrated with the LED by several methods. Slurry coating, particle settling from a solution and EPD are the most widely used processes. The slurry coating method uses a dipping, spinning or flowing process of the phosphors suspended in a photoresist. 10 After a sufficient thickness is achieved, the photoresist is cured to form a strongly adherent particulate film to the substrate. Settling of phosphor powders from a solution involves mixing the powders in a solution with potassium silicate and barium acetate electrolyte which forms SiO 2 microclusters that acts as a binder. 10 The powders settle, following Stokes' law (which is a slow process), and after the desired thickness is achieved, the solution is decanted and the particulate film is dried.
Alternatively, EPD has emerged as a useful method to fabricate particulate phosphor films. 11 The process is simple, scalable and costeffective. EPD has been used for the preparation of specialized CRTs since the 1950s [12][13][14] and to prepare flat panel emissive displays. 10,15,16 More recently, EPD has been found to be a viable method for depositing phosphors for white LED lighting, both in the conformal and remote configurations. Various phosphors compositions in these configurations are shown to emit white light located on the black bodyline on the CIE diagram. The application of EPD to deposit phosphors for white solid-state lighting is promising and offers facile tuning of emissive color by varying the phosphor blend compositions. This review discusses recent developments in EPD in the solid-state lighting industry.

The Electrophoretic Deposition Process
Experimental considerations.-Electrophoresis refers to the movement of charged particles in a suspension and the deposition process is the formation of a layer of the particles on a substrate. The three major steps are 1) charging the particles in suspension, 2) transport of the particles to a substrate under an electric field and 3) deposition and adherence of the particles onto the substrate. Figure 4a shows a schematic diagram of the process where an anode and cathode are placed in a solution of positively charged particles that migrate to the cathode under the application of an electric field. Although EPD may be done in an aqueous or non-aqueous medium, EPD of phosphors has been performed in non-aqueous baths, primarily alcohols, which avoids the gas evolution (hydrogen at the cathode) due to electrolysis and gives better control over thickness and uniformity of the deposit. Also, EPD of phosphors has mostly been conducted under constant applied voltage, as opposed to a constant current condition, with deposition occurring on the cathode, being either ITO-coated glass or the LED itself. There have been several reviews of the EPD process in general. 17,18 The important suspension properties are the solid loading, viscosity and dielectric constant. Also, the suspension may benefit from the addition of a dispersant and binder. The zeta potential (ζ) of the particles is measured to indicate the charge on the particles in the suspension and the ability of the suspension to be electrostatically dispersed and to calculate the electrophoretic mobility. The zeta  potential is defined as the electrical potential at the shear plane with respect to the bulk liquid and can be measured by several methods. 19 The important deposition properties are the voltage and deposition time. Table I summarizes these EPD parameters and how they influence the final deposit characteristics. The EPD of phosphors was studied thoroughly in our previous work in a bath of isopropyl alcohol (IPA) with nitrate salts. [20][21][22][23][24][25][26] First, the effect so the suspension medium was investigated by determining the dissociation of Mg(NO 3 ) 2 in IPA by measuring molar conductivity. 20 Using the limiting conductivity and the concentration of ions calculated from the dissociation constant, the mobility of the ions was determined. Dissociation is very small in liquids of low dielectric constant such as IPA. In the concentration range of 10 −4 to 10 −3 M Mg(NO 3 ) 2 typically used for EPD, Mg(NO 3 ) + is the main cation available to charge the particles. Other salts have been used. For example, Grosso et al. 27 investigated EPD of a phosphor in IPA and mixtures of acetone and IPA with the addition of a number of nitrate, chloride, and bromide salts and a small amount of water. However, only nitrates and chlorides of magnesium, aluminum, and lithium produced adherent deposits. Secondly, zeta potentials of several oxide, sulfide, and oxysulfide phosphors were measured in IPA and in IPA containing nitrate salts and water. 20,21 In pure IPA, the zeta potentials of phosphors were negative, but with the addition of 5 × 10 −4 M nitrate salt, the zeta potentials became positive. Both nitrate salt concentrations and pH were shown to affect the zeta potentials of Zn 2 SiO 4 :Mn (P-1) and ZnS:Ag (P-11) phosphor particles, in particular. 21 During EPD, electrolysis of the water present in the IPA produces OH − at the cathode that reacts with Mg(NO 3 ) + to produce Mg(OH) 2 at high water content (>5 vol%) 21,22 and Mg(C 3 H 7 O) 2 at low water concentrations (<1 vol%). At intermediate water concentrations, a mixture of the two materials forms. With increased amounts of water, the dissociation of nitrate salt and thus, conductivity of the solution also increases, as shown in Figure 5. 26 Therefore, the Mg(NO 3 ) 2 in the EPD bath both charges the particles positively, maintains a positive zeta potential at the high pH at the cathode, and forms the binder. Note, however, that hydrogen evolution can also occur with a high concentration of water. The amounts of deposited particles and magnesium hydroxide binder were simply modeled by integrating the flux of material over time, 21,23 which agreed well with experimental results provided that the Mg(NO 3 ) 2 concentration was greater than ∼10 −4 M to form the binder to adhere the particles. By baking the deposits at 425 • C for 1 h, Mg(OH) 2 and alkoxide binder convert to MgO, which is a stronger binder. [23][24][25] The adhesion strength was strongest with 2 vol.% glycerin added to the IPA bath.
Generally, in commercial EPD processes, water absorption into the bath is not controlled. However, a small amount of water is needed for the precipitation of Mg(OH) 2 at the cathode to adhere the phosphor particles to the substrate. But too much water can cause undesired hydrogen gas formation, which can damage the film or even prevent deposition, The effects of up to 10 vol.% water in the bath on the deposition and adhesion strength of EPD phosphor screens were studied. 23 There was dramatic increase in the adhesion strength as the added water content in the bath was increased from 1 to 5 vol.%. However, with water concentrations higher than 5%, the screens became irregular and have poor quality, due to hydrogen evolution.
The packing fraction of particles, , in an EPD film can be calculated using the following equation: 11 = m hρ p [1] where m is the weight of the deposited film per area of the sample, h is the thickness of the deposit, and ρ p is the material density. The maximum (ideal) packing fraction is 0.74 for uniform particles, however most EPD deposits range from 0.25 to 0.56, depending on the deposition parameters that influence the deposited mass. 28 If the deposition time is short and the suspension is relatively well dispersed and stable, EPD can be performed in a quiescent bath. This is often the case if depositing thin layers of phosphors. However, if there is settling, bath agitation by stirring or flow can be used. This must be done is a way to not disturb the morphology and uniformity of the deposit, which can be a challenge. For nano-sized particles, it was found that placing the EPD cell into a sonication bath and applying sonication during deposition prevented sedimentation of the particles and yielded a more uniform deposit from the removal of bubbles generated by electrolysis of water. 29 A typical laboratory set-up is shown in Figure 4b, where the anode is an aluminum sheet separated 1-2 cm from the cathode, an ITOcoated glass slide. A general procedure is: the phosphors (∼5-10 μm diameter, ∼1-5 g/L concentration) are suspended in a bath of a polar solvent (e.g. IPA) along with dissolved metal salts (e.g. Mg(NO 3 ) 2 ), a dispersant that helps to keep the particles from agglomerating in solution (e.g. glycerin) and occasionally an adhesive agent. The beaker housing the solution, electrodes and a magnetic stirrer is placed on a magnetic plate and stirred for ∼30 min. Under the application of an electric field gradient (typically between 30-80 V/cm), a particulate film forms on the ITO-coated glass (deposition times ∼5-3600 s). After the film is dried (∼100-500 • C), Mg(OH) 2 , converts to the MgO at the higher temperature; either Mg(OH) 2 or MgO can act as a binder for particles to the substrate and to each other. Other salts can be used, such as Al(NO 3 ) 3 , Y(NO 3 ) 3 , and La(NO 3 ) 3, which may improve the adhesion strength compared to Mg(NO 3 ) 3 . 23 When a phosphor is coated on a LED (conformal configuration), the thickness of the layer is a critical issue. For blue-emitting LEDs, the phosphor layer must be thin enough for the light from the LED to emerge from the phosphor layer to combine with the light from the yellow-emitting phosphor layer. However, if the layer is too thin, the chromaticity coordinates will not be optimal for white light and the spectrum will have too much blue-emission. If the layer is too thick, too much yellow-emission from the phosphor is observed. Thus, an optimal balance between phosphor thickness and chromaticity must be determined. For near-UV LEDs, the deposit thickness must be optimized so that it is thick enough to ensure full conversion (no unwanted near-UV light escapes), but not so thick as to reduce light output; these films must be between 10-40 μm thickness for full conversion. 28 The creation of thick films creates processing problems with EPD: thick films typically require longer deposition times. At long times if the solution is not stirred, settling of the powders has to be taken into account as the concentration of particles in the bath decreases with time. Stirring may cause non-uniform deposition. In addition, as the thickness increases, the effective electric field decreases due to the resistance of the insulating film. For our EPD process for thicker layers needed for near-UV LEDs, lower voltages (∼80 V) and longer deposition times (up to 30 min.) 30,31 were used than previous EPD processes to attain uniform deposits. 16,20,21,23 For both blue-emitting and near-UV diode, other requirements are that the deposit must be of uniform thickness to give consistent optical performance, the packing density should be optimized for the best light output, and the amount of non-luminescent materials (binders) should be minimized. Finally, the deposit must have sufficient adhesion strength to withstand handling during manufacturing, as well as during use.
There are three phosphor arrangements in a particulate film that have been investigated for white-emission from excitation by blue or near-UV light. For blue emitting LEDs, one method is to mix the blue light from the chip and yellow light from a phosphor layer above the chip (conformal or remote) (Figure 6a), the second is to create a sequentially deposited, layered structure of single-colored phosphors (Figure 6b), and the third is to disperse red-, green-and yellowemitting phosphors in the same bath and deposit them in a random arrangement ( Figure 6c). The impetus for creating a three-layered over the random arrangement of particles is to potentially reduce readsorption of emission wavelengths of the phosphors. 32,33 It has been shown that the luminous efficiency is improved by 8% by having a separate green-emitting layer on top of a red-emitting layer on a blue LED chip, due to a decrease of reabsorption of green light by the red-emitting phosphor. 34 In other work, 35 it was found that a separate yellow-emitting phosphor on top of a red-emitting layer, with respect to the blue LED, led to an 18% increase in luminous flux compared to the blended yellow and red phosphor film. The enhanced efficiency was attributed to the reduced reabsorption of yellow light by the redemitting phosphor. For near UV-LEDs, red-, green-and blue-emitting phosphors are used, which can be deposited sequentially (Figure 6d) or as a single layer with multi-colored phosphors ( Figure 6e) to a thickness where the chip light cannot escape from the device.
Experimental results.-Some of the first reports of using EPD of phosphors for solid-state lighting are in the patent literature. They describe methods to produce conformal YAG:Ce coatings on a blueemitting LED to produce white light. [36][37][38] One method involves applying a voltage between the submount and the charged phosphor particles in a solution with binders above the LED, resulting in the phosphor layer (<30 μm thick) being deposited directly on the LED surface, either singly ( Figure 7a) or in an array. 36 A photoresist and mask is applied to the submount so that the phosphor only deposits on the LED. The patent also describes a method to deposit a conducting surface on the LED if the LED substrate is non-conducting. In another patent, 38 an anode is placed in the bath with the LED and a voltage is applied between the anode and the p-side of the LED to be positive with respect to the p-side and a second voltage is applied between the p-side and n-side of the LED, causing deposition of the phosphor onto the LED. Deposition of the phosphors can occur on three sides of the LED exposed in the solution, by changing the bias voltages. This allows fine control over material deposition on a semiconductor device, improving manufacturing efficiency by reducing material waste, and providing coating only where required. This method is better for color-critical applications when compared to the current phosphor deposition system. A further refinement has been reported, 37 where a method that prevents ion deposition from the LED and supports during the EPD process. An auxiliary meshed cathode was inserted between the anode and LED to capture ions such as ions of Ag, Pb, Cu, Sn and Ni that prevent their deposition onto the LED. Figure 7b is a schematic illustration of the cross-section of the resultant device, showing the conformal phosphor layer on the chip.
Most work on EPD for solid-state lighting application has been on depositing phosphors on ITO-coated glass and optimizing the deposition parameters so that the desired packing density and thickness were achieved. Single compositions, phosphor blends and layered deposits have been explored, and these layers were excited by blue or near UV light that ultimately produced white light. A summary of recent procedures and results follows.
Yum et al. 39  270 V/cm). The optimal conditions for achieving a good white light (x = 0.290 and y = 0.320) when the deposit was excited by 450 nm were 3 g/L at 230 V/cm for 240 s, which resulted in a deposit weight of 9.88 mg/cm 2 . The optimal thickness was 30 μm, which had a packing density of 51% (ideal = 74%), higher than those deposited by settling or slurry methods (<37%). However the adhesive strength, determined by impinging a jet of N 2 at 4 atm on the phosphor layer, was lower than those formed from the slurry or settling methods. To improve the adhesive strength, a modified EPD process was carried out, where a mixture of polyvinyl alcohol (PVA), ammonium dichromate (ADC), and distilled (DI) water was spray coated on the deposit and cross-linked under UV light. This process significantly improved the adhesive strength, which matched that from the slurry and settling methods. The morphologies of the EPD and modified EPD films had no distinguishable differences and the surfaces were more uniform than the slurry or settling methods, as shown in Figures 8a, 8b. 40 Furthermore, EPD could more easily control the thickness and uniformity than the slurry and settling methods and the high packing density from EPD allowed a thinner layer to be fabricated (optimal thickness of 30 μm, compared to ∼90 μm for the slurry and settling methods).
The water content was varied to increase the conductivity of the bath. Additions of up to 5 vol.% water increased the zeta potential from 30 mV to 50 mV, which increased the deposit amount from ∼2 mg/cm 2 to 4 mg/cm 2 . The adhesion strength was further investigated by varying the water concentration and with and without post-deposition annealing at 425 • C. 41 Increasing the water concentration in the bath with annealing of the deposit increased the adhesion strength, however the best results were found using the UV-curable mixture. In follow up work, YAG:Ce was deposited (with PVA, ADC and DI water) on a flexible ITO-coated polyethylene terephthalate substrate and the concentration of Mg(NO 3 ) 2 was investigated. 42 In pure IPA, the zeta potential was negative and reached a maximum of 16 mV at 10 −5 M. The addition of 1 vol% DI water increased the conductivity from 2 μScm −1 to 8 μScm −1 . A thickness of 27 μm and packing density of 53% was achieved under 400 V, 2 g/L phosphor concentration and 180 s deposition times. To increase the packing density, the phosphor on the polymer substrate was compressed under 20 MPa, which resulted in a packing density of 62%. However, the phosphor packing density as deposited on a glass substrate was higher (72%), even without post-deposition compression. The phosphor strongly adhered to the substrate, determined by the N 2 jet impingement method.
In other work by Chen et al. 43 EPD of YAG:Ce powder (5-10 μm diameter) was performed on ITO-coated glass from a bath of IPA, 0.75 g/L powder, 5 × 10 −4 M Mg(NO 3 ) 2 and 1-4 vol.% DI water under an electric field of 200 V/cm. When the solution was stirred during the deposition, the density of the phosphor layer increased. The same deposition method was applied to two different surface mounted blue-emitting LED devices: one had the die surrounded by epoxy in a bowl reflector, the other had normal die-bonding. These LEDs were compared to the conventional method of dispersing the phosphor in the epoxy. With the thin epoxy layer, the light color was more controlled; however, the luminous efficiency was lower than that of without the thin epoxy layer. It was found that both EPD LEDs had a lower efficiency than the dispersed phosphor in an epoxy. It was suggested that this was due to a larger difference in refractive index between the phosphor and the InGaN compared to the difference between epoxy and InGaN, decreasing the blue light extraction. This work was followed by examining the luminescence properties of an EPD phosphor plate of YAG:Ce suspended above a blue-emitting diode. 44 By altering the solid loading in the bath, the thickness of the deposit could be controlled between 16-40 μm with CCT values between 3800-6000 K. The deposits had a packing density of ∼30%, which allows the blue light to be transmitted through the phosphor layer. The light flux was increased using the phosphor plate that was placed 500 μm above the LED (72 lm) compared to the dispersed phosphor in epoxy (69 lm).
Kitabatake et al. 45 reported on EPD of Eu 2+ -activated Ca-α-SiAlON (emission wavelength, λ em = 585 nm, yellow) with a particle size of ∼1 μm in diameter and containing agglomerates of up to ∼20 μm on ITO-coated glass. 45 Two g of powder was added to 100 ml of ethanol with phosphate ester (PE) and polyethylenimine (PEI) as dispersants and polyvinyl butyral (PVB) as the binder. After the stirring was stopped, the deposition was either (1) started immediately or (2) started after one minute so the larger agglomerates would settle out. The deposition rate was slightly higher for condition (2). Under 30 V/cm, a packing density of 55-60% was achieved for deposition times up to 300 s, with resultant film thicknesses between 2-6 μm, depending on the deposition time. Figure 9a shows the deposition yield and film thickness as a function of time for condition (2). As shown, the deposition rate (∼0.11 mg/cm 2 •s) was constant for up to ∼100 s and then saturated at a thickness of ∼5.5 μm. Excitation by 450 nm resulted in blue-white to yellow-white light, depending on the thickness of the film, as shown in Figure 9b. The optimal chromaticity coordinates for condition (2) were x, y = 0.374, 0.260. The chromaticity coordinates and light output were greater for condition (2) compared to condition (1). These results indicate that the removal of the large, agglomerated particles give a more uniform film.
The adhesion strength of Ca-α-SiALON:Eu 2+ deposits was improved by adding tetraethyl orthosilicate (TEOS, 5 mM) to the ethanol bath. 46 After a 60 s deposition on conducting polypyrrole coated glass slides (cathode), the films were dried and calcined at 500 • C to burn off the polymer and convert the silicon precursor to SiO 2 . Subsequently, the dried film was impregnated with a TEOS/HCl/ethanol solution and calcined again at 500 • C to form a composite film of the phosphor surrounded by SiO 2 . In parallel, EPD of the powders was carried out with PE, PEI and PVB in the ethanol bath and then infiltrated with the TEOS/HCl/ethanol solution and calcined at 500 • C. The adhesive strength (by a tape test) was significantly higher for the composite film compared to the films deposited with only the binders. The external quantum efficiency was also higher, suggesting that light propagation is improved by filling the voids in the film with SiO 2 . These results demonstrate that the adhesive strength and PL properties can be significantly improved by infiltrating the deposit with an inert material.
Zhang et al. 47,48 followed the above work by coating SiO 2 on Ca-α-SiALON:Eu 2+ , with the aim to improve the luminescence properties by mitigating surface defects. The SiO 2 coating reduced the zeta potential significantly, but the addition of 10 −3 M of Mg(NO 3 ) 2 equalized the zeta potentials between the uncoated and coated particles, as shown in Figure 10. 48 The coated and uncoated particles were separately suspended in a bath of IPA and glycerol and deposited for 300 s under an electric field of 100 V/cm. The coated particles had slightly higher PL excitation and emission intensities. This result indicates that phosphor particles with highly different zeta potentials can be rendered similar if an inert shell is coated on one or both of the particles.  (a) Examples of the multicolor light emitting panels prepared by sequential electrophoretic deposition (EPD) of yellow-, green-and redemitting phosphors on ITO-coated glass substrates. The laminated films were composed of red-yellow (red deposited first), yellow-green, green-yellow, green-red, red,-green, yellow-red, green-yellow-red and red-yellow-green. Photoluminescence emission spectra (excitation wavelength = 450 nm) and photographs of laminated films of (b) yellow/green (50 μm/50 μm) and (c) green/yellow (15 μm/15 μm). Photoluminescence emission spectra of EPD films of mixed powders of (d) of yellow-and green-emitting phosphors of various ratios, deposited for 15 s (30-45 μm thick) and (e) red-, green-and yellow-emitting phosphors of various ratios (45-67 μm thick). Adapted from Ref. 32. Blended phosphor mixtures of Eu 2+ -activated Ca-α-SiAlON, β-SiALON (λ em = 540 nm, green) and CaAlSiN 3 (λ em = 620 nm, red) were deposited onto an ITO-coated substrate with particles sizes ranging from 4-20 μm in diameter. 32 Three baths were prepared from one gram of each powder, which was added to 100 ml of ethanol with phosphate ester and polyethylenimine as dispersants and polyvinyl butyral as the binder. In another bath, 1g of blended powders were added to the ethanol solution. Laminated films were also prepared by sequential deposition of the red-, yellow-and/or greenemitting phosphors. To tune emission colors, the thickness and the order of the sequentially deposited layers of phosphors was altered, as shown in Figure 11a. 32 The light emission from the deposited surfaces was characterized by comparing the intensities of the excited yellow light and the transmitted blue light, both of which decreased as the film thickness increased. For green (β-SiALON:Eu 2+ ) and yellow (Ca-α-SiAlON:Eu 2+ ) phosphor layers, the individual phosphor deposition thickness (15-50 μm) and order of the deposition greatly R3114 ECS Journal of Solid State Science and Technology, 5 (1) R3107-R3120 (2016) affected the color, tunable from yellow-white (3000 K) to blue-white (5000 K) (Figures 11b,11c). If the green-emitting phosphor was deposited first and the yellow-emitting phosphor second, the emission color was 5000 K; if the yellow was deposited first and the green second, the emission was 3000 K. It was concluded that the first deposited layer mostly absorbed the blue light and the second layer was not excited. The white light emitting devices prepared by this method showed a broad spectrum, comparable to those of fluorescent lamps. The blended films were prepared by a single deposition using suspensions that contained two (yellow and green) or three (yellow, green and red) different phosphors. For yellow and green mixtures, the chromaticity coordinates were optimized for a ∼50/50 mixture (Figure 11d). For the three-phosphor mixture, the addition of the red component improved the color point (Figure 11e), however only a small amount (<10%) could be added else the red-emission swamped the emission spectrum.
Choi et al. 30,31 investigated EPD to deposit both blended and layered phosphor films that generated white light using Eu 2+activated Sr 2-x Ca x Si 5 N 8 , Ba 2 SiO 4 , LiCaPO 4 , (Sr 0.75 Ba 0.25 ) 2 SiO 4 and (Sr 0.5 Ba 0.5 ) 3 SiO 5 (red-, green-, blue-, yellow-and orange-emitting, respectively) for near-UV LEDs; some physical properties are listed in Table II. The depositions were carried out in a bath of 100 mL IPA, 5 g/L phosphor loading, ∼10 −5 M Mg(NO 3 ) 2 and 2 vol% glycerin to help disperse the particles in the bath and deposited on ITO-coated glass slides. The nitrate concentration was chosen so that the zeta potential of ∼36 mV was attained for each phosphor. Voltages between 50-100 V/cm for times ranging from 1-8 min. For the individual phosphor deposits, the deposition rate was similar, except for the nitride, which deposited at a much higher rate (∼13 μg/cm 2 s compared to the other compositions at ∼4 μg/cm 2 s), for reasons yet unknown. Assuming a packing density of 56%, the thicknesses of the films were ∼6 μm, except for the nitride film (22 μm). White-emitting films were prepared by blending the phosphors with two different blend ratios (in wt%) to attain a thickness of ∼20 μm (deposition time = 5 min). One was composed of three phosphors (63% blue, 15% yellow, 22% orange) and the other was composed of four phosphors (70% blue, 13% green, 5% orange, 12% red). Figure 12 shows the emission results on a CIE chromaticity diagram, calculated from the PL emission spectra. For the three-phosphor blend (Figure 12a), CCT = 3203 and CRI = 75 was obtained, with the color point falling on the black bodyline. For the four-phosphor blend (Figure 12b), a CCT = 3346 and CRI = 94 was obtained, with the color point falling on the black bodyline. A higher CRI was found with the four-phosphor blend because of the inclusion of the red-emitting phosphor.
To investigate the effect of re-adsorption of emitted light from other phosphors by the red-emitting phosphor, layered deposits were prepared. White-emitting laminate films were fabricated by the sequential depositions of mixtures of green:blue-emitting phosphors (2:3 wt ra- tio) and of red:orange-emitting phosphors (1:1 wt ratio), 31 from the compositions given in Table II. In one configuration, the red/orange mixture was deposited first (1 min deposition) on an ITO-coated glass substrate followed by the green/blue mixture (480 s deposition) (Figure 13). The deposition time was longer for the second layer because of the insulating effect of the first layer decreased the electric field. Near UV light illuminated the structure either on the deposit and measuring the emission from the substrate or vise versa, as shown in Figures 13a,13b. Both show similar CCT ∼3160 K and CRI = 90. In the second configuration, the green/blue mixture was deposited first (180 s) on the substrate followed by the red/orange mixture (300 s). The deposit was examined both by illuminating with near-UV light from below the substrate and above the deposit, as shown in   Figures 13c,13d. The color points are identical, with CCT = 2720 K and CRI = 90. These results demonstrate that the order of depositing the red/orange or green/blue layers does not affect the PL emission properties, and that the red/orange mixture does not re-absorb the emission from the green/blue phosphors. Compared to other work on sequential depositions of two phosphor layers, 32 in the above work the deposition rate was slower (0.02-0.83 μm/s compared to 0.9-2.7 μm/s), which could be due to the lower solid concentration (5 g/L) than the previous work (10 g/L). The first layer thickness, 5-15 μm, was thin enough for the near UV light to excite the second layer. In the previous work, 32 the first layer thickness was 40-50 μm, which was too thick for excitation of the second layer.
Comparison of the blended films with the layered films in the above works is difficult since the depositions were optimized so that the chromaticity coordinates were close to the blackbody line. The thicknesses of the blended and layered films are not the same, nor are the ratios of the various phosphors. The use of blended phosphors in one bath is easier to prepare, but because of the different deposition rates of the individual phosphors, it takes experimental optimization to determine the quantity of each phosphor that should be added to the bath. Additionally, the zeta potential of each phosphor must be the same to achieve a uniform deposition rate of each phosphor and the particle sizes should be roughly equivalent, or the larger particles will settle out of the bath. Layered phosphors take at least two separate depositions, but both steps are more easily controlled in terms of the mass deposited.
Color tunability was found for β-SiALON:Eu 2+ mixed with various amounts of 25 nm diameter SnO 2 powders. 49 It was proposed that scattering from SnO 2 nanoparticles, which have the same refrac- tive index as β-SiALON, could improve the overall uniformity of the irradiation of the phosphor particles. One gram of β-SiALON:Eu 2+ (rod-shaped, ∼4 μm × 23 μm) was dispersed in 100 mL of IPA with 2 mL of DI water, 2 mL of glycerol and 0.1 mole of Mg(NO 3 ) 2 and 0-1 g of SnO 2 . The mixture was deposited on ITO-coated glass with a stainless steel sheet as the anode. A voltage of 50 V/cm was applied for deposition times from 10-300 s. After deposition, the film was dried and calcined at 425 • C for 1 h. The films were irradiated with 405 nm light below the substrate. The quantum efficiency (QE) varied with SnO 2 concentration and was maximized at 0.05 g. Increasing the concentration changed the chromaticity coordinates from the yellow/green to the blue region in the CIE diagram. This shift was attributed to a lower packing density obtained by increasing the amount SnO 2 , which allowed more of the 405 nm light to be transmitted through the phosphor layer.
These researchers also applied a magnetic field (6 and 12 T) parallel to the electric field during the deposition of β-SiALON:Eu 2+ to align the crystals (hexagonal, a = 7.64 nm, c = 2.93 nm). 50 Both magnetic field strengths aligned the powders, which increased in alignment as the deposition time increased, along with increasing the deposit density. Figure 14 shows the color points for 80 μm and 100 μm thick films, with and without a 12 T field, with λ ex = 405 nm. As the film thickness with the magnetic field applied, the color point moves toward the green region, indicating more of the phosphor is excited. Additionally, the application of the magnetic field increased the external quantum efficiency increased by 10-20%, due to the deposit density increase.
Electrophoretic deposition of nanophosphors.-Typically, micron-sized phosphors are being used in solid-state lighting. However, the micron-sized phosphors have higher optical scattering losses due to an increased optical path length and increased number of reflections or transmissions at interfaces. 51 Nano-phosphors (<300 nm) with comparable quantum efficiency to typical micron-sized phosphors may be a means to reduce the losses and improve efficiency of the white UV-based LEDs. 52 But it is well known that the emission intensity of phosphor decreases as the crystallite size decreases. 53 This has been attributed to surface defect states that trap the emitted photons and thus quench the emission intensity. To alleviate the impact of surface defects in nanocrystalline powders, core/shell structured nanoparticles have been used to stabilize the surface of the nanoparticles. 54,55 It was found that the luminescence intensity, chemical stability and thermal stability were improved by the presence of inert shells. [56][57][58] It has been found that EPD from suspension of nano-sized particles of (Ba 0.97 Eu 0.03 ) 2 SiO 4 in an IPA bath yielded a non-uniform and porous film, 28 as shown in Figure 15, for both photographs (under excitation of 365 nm light) and SEM micrographs. The increased surface area of nano-sized particles adsorbs a larger quantity of water than for micron-sized particles. With that additional water in the bath, more hydrogen evolution occurred at the cathode, resulting in poor quality films.
To avoid this problem with the EPD of nano-sized phosphors, amyl alcohol, which has low water solubility, was used as the suspension medium instead of IPA. 28 Micron-sized (5 μm) and nano-sized (320 nm) (Ba 0.97 Eu 0.03 ) 2 SiO 4 was compared using bath a consisting of amyl alcohol, 5 g/L of the phosphor and Mg(NO 3 ) 2 (ζ = 22 mV). However, due to the limited availability of the nano-sized phosphors, a very small bath volume of 60 mL was used. Deposition times ranged from 300 -1800 s deposited sequentially from the same bath onto new substrates. For the micron-sized powders, the deposit weight increased linearly up to ∼600 s and then leveled off. If the solution was stirred, the deposition rate is lower and levels off around 1500 s for a deposit weight of ∼21 mg. For the nano-sized powders, leveling off also occurred at ∼1500 s at a weight of 20 mg, as shown in Figures 16a,16b (this figure will be discussed in more detail in the Theory section). Figure 17a shows SEM micrographs of cross-sections of the deposits at two different deposition times for the micron-and nano-sized powders. The films do not have a uniform thickness for t = 300 s, but become more uniform at increasing deposition time. For 30 μm thick films, the micron-sized powder films had a higher PL emission intensity than the nano-sized powders (Figure 17b). This is due to the inherent lower quantum efficiency of the nanosized powders compared to micron-sized ones. Table III summarizes experimental conditions that have been used to fabricate EPD films of phosphors for solid-state lighting. As shown, a variety of compositions (oxides, silicates, phosphates and nitrides) have been successfully deposited. Most researchers use a bath of IPA and Mg(NO 3 ) 2 , solid loadings between 1 to 20 g/L with electrodes a few centimeters apart to deposit on an ITO-coated glass substrate. Small amounts of dispersants are frequently used to keep the particles from agglomerating in the bath. Successful deposits have been fabricated from electric fields ranging from 50 to 230 V/cm, deposition times as short as 10 s to as long as 20 min, with deposit thicknesses ranging from 6 to 60 μm.

Theory
Knowledge of EPD kinetics is necessary in order to control the deposition rate and manipulate the microstructure of the deposit. EPD can be conducted under constant current or constant voltage. Constant voltage is mostly used as high voltage power supplies are, in general, more assessable; however, using a constant current condition gives a constant deposition rate as the voltage changes. The concentration of particles in the suspension can be considered either constant or decreasing with deposition time as deposition occurs. Modeling of the kinetics of EPD, in general, has been reviewed by others. 59,60 The deposited mass, m (kg), can be determined from the flux of particles toward the cathode by: [2] where C s is the suspension concentration (kg/m 3 ), v e is the electrophoretic velocity (m/s), A s is the deposition area (m 2 ), α is the fraction of particles adhering (usually assumed to be equal to 1) and t is the deposition time (s). This equation assumes C s and the electric field are constant and uniform in the bath, which is typically valid for short deposition times and stirred or flowing suspensions. If C s and v e are constant during the deposition and all particles that reach the substrate stick to it, the mass deposited can be expressed by the Hamaker equation: 61 [3] However, during EPD, particularly for longer deposition times, the concentration of particles decreases in the bath as deposited. Taking this into account, Sarkar and Nicholson 59,60 developed an equation for the deposited mass: where m 0 is the initial mass of particles (kg) and τ is the characteristic time (s) given by: where V is the volume of suspension (m 3 ). This equation indicates that the deposition rate is expected to decrease exponentially as a function of time due to depletion of the bath by particle deposition. For small values of t/τ, Eq. 4 reduces to Eq. 3. The electrophoretic velocity equals the product of the electrophoretic mobility, μ, (m 2 /Vs), and the electric field gradient, E (V/m). The electrophoretic mobility can be estimated from the Smoluchowski equation: 16 where ζ is the measured zeta potential (V), ε r is the relative permittivity of the suspension medium (found in reference tables), ε 0 is the vacuum permittivity (8.854 × 10 −12 J/V 2 m) and η is the viscosity (found in reference tables) of suspension medium (kg/ms). The settling velocity v s , (m/s) can be estimated by Stokes' law: v s = 2 9 ρ p − ρ f η gr 2 , [7] where ρ f is the density of fluid, g is the gravitational acceleration (9.81 m/s 2 ), and r is the radius of a spherical particle (m). However, the concentration in the suspension may change due to both deposition and settling. Taking both these effects into account the mass deposited can be expressed as: If the ratio v s A/v e A S 1, Eq. 8 reduces to Eq. 4 and the bath depletion only occurs from deposition. If the ratio is 1, then settling is the important factor, and Eq. 8 can be simplified to: 28 The experimental and theoretical values of deposited mass versus time from Eqs. 3, 4 and 8 are plotted in Figures 16a,16b for nano-sized particles and micron-sized particles, respectively, in an amyl alcohol bath. The experimental values for nano-sized particles in Figure 16a are in good agreement with the predictions from Eqs. 3, 4 and 8 until about 20 min of deposition time. This agreement is due to the suspension of nanoparticles being stable and that little material in the bath is being depleted for shorter deposition times. The deposition weight levels off after 20 min, which may be due to the affects of agglomeration of the particles and settling of these agglomerates since the bath was not stirred. It should be noted that in these experiments, v s A/v e A S = 2.4 × 10 −3 , thus Eqs. 4 and 8 yield the same results.
However, there is a large discrepancy between the measurements and theory for the micron-sized particles, as shown in Figure 16b. The experimental deposited mass is higher than the theoretical values. One possible explanation is the presence of vertical concentration gradients in the bath due to Rayleigh-Taylor (R-T) instability. 62 These R-T instabilities are unusual for EPD and developed in part due to the small Downloaded on 2015-10-23 to IP bath volume used in these experiments and because the bath was not stirred. 28 The R-T instabilities should not occur if the suspension is well mixed. Therefore, EPD was performed with gentle stirring of the bath by a small magnetic stirrer placed in the bottom of the bath under the same deposition conditions. With stirring, the experimental values showed better agreement with the theory except for the Hamaker equation (Eq. 3). The other model predictions are similar as depletion of the bath due to deposition is more significant at longer times and stirring mitigated settling of the powders. In the above experiments, the ratio v s A/v e A S = 5.7, thus Eq. 8 needs to be employed for the predicted mass.

Summary and Future Outlook
A wide variety of phosphors (oxides, nitrides, silicates, phosphates) have been deposited by electrophoretic deposition (EPD) for the application of solid-state lighting in several configurations. By understanding the fundamentals of the process, EPD has been used to coat uniformly thin, highly packed conformal (on blue-emitting LEDs) or a thick remote layer phosphor layer (both blue-emitting and near UV-emitting LEDs). It has been used to deposit micron, nanosize and core-shell phosphor particles for this application. EPD has advantages for controlling color and efficiency of solid-state lighting devices by coating single layers, multiple layers or blends of phosphor particles. The important points are summarized below: r EPD consists of three steps: charging of the particles, movement of the particles under an electric field (electrophoresis), and adherence of the particles to the substrate. These processes can be manipulated to fabricate films with controlled thickness, packing density and adhesive strength.
r The critical parameters in the bath are the suspension medium, water content, the solid and salt loading and additives to help dispersion and adhesion, all of which affect the zeta potential. The critical deposition parameters are the electric field and deposition time.
r Phosphors with different zeta potentials can be equalized by coating one or both with an inert shell.
r Magnetic fields applied during the deposition can be used to align anisotropic, magnetic powders to produce oriented films.
r The packing density of the as-deposited films typically range from 0.25 -0.50, but can be increased ∼15% by compression (20 MPa) of the film.
r The adhesive strength can be significantly improved by infiltrating the deposit with a silica precursor solution and subsequent heating to produce phosphor/SiO 2 composite films.
r Phosphor blends consisting of two or more compositions can be deposited, however the zeta potentials must be similar. If settling is an issue, and particle sizes and densities must also be similar. Experimental fine-tuning of the concentration of each phosphor in the bath must be done to achieve the desired chromaticity coordinates.
r Layered films can be produced by sequential deposition of different phosphor compositions from separate baths. If the first deposit is too thick, the second layer will not be excited. The desired thickness of each layer needs to be determined experimentally.
r Deposition of nanophosphors needs to be performed in a bath that has little water solubility, such as amyl alcohol.
r Theory to take account of both particle depletion and settling in the bath was developed, which may be important for longer deposition times needed for thicker films.
The future challenges of phosphor deposition for solid state lighting devices are in regards to the various chip configurations for optimization of light output and performance, heat transfer from the device and aging of the phosphors. EPD may be feasible for new device structure, such as 3D, and novel phosphor materials, such as nanosized and core-shell particles. EPD is being incorporated into additive manufacturing, 63 which will allow precise placement of materials in three dimensions and may benefit solid-state lighting applications in the future.