Effects of organic additives on the microstructural, rheological and electrical properties of silver paste for LTCC applications

Silver powders were prepared by a chemical reduction method. Effects of different surfactants during this preparation process were examined. Using these silver powders, silver pastes were prepared for low temperature co-fired ceramics (LTCC) applications. Another batch of different surfactants were added into the silver paste and evaluated. Moreover, various organic thixotropic agents were used to modify the thixotropy behavior of the silver paste. It was found that the surfactants of PVP (Polyvinyl pyrrolidone, for preparation of the silver powder) and Span-85 (for the preparation of silver paste) exhibited the best performance, with an electrical resistivity of 0.11 mΩ mm. The silver paste with the thixotropic agent of hydrogenated castor oil shows the largest thixotropic index of 1.56. Using different organic additives, the microstructural, rheological and electrical properties of the silver paste can be considerably improved, and our results shed light on the optimization of the silver paste for LTCC applications.


Introduction
In order to meet the demand of high transmission speed, high density, high reliability and low cost of wireless microwave communication systems, LTCC technology has attracted extensive attentions. Conductive paste for preparation of inner electrodes, usually silver, is one of the most important ingredients for LTCC process [1,2]. Generally, the conductive paste is composed of three parts. (1) metal powders, (2) glass additives as inorganic binder, and (3) organic mixtures (also called organic vehicle) for dispersing metal powders, which also determine the rheological behavior of the conductive paste [3,4]. The rheological properties, i.e., viscosity, viscoelasticity and thixotropy of the paste, will affect greatly the preparation processing and thus the final electrical properties of the electrodes [5][6][7]. The rheological properties depend on many factors such as the dispersion of the powders, the particle size, surfactants and/or polymeric binders. The surfactant is one of the key components for determining the rheological property of the paste [4]. Organic surfactants are usually adsorbed on the surface of metal particles, act like a bridge between the metal particles and the organic mixture to facilitate particles' surface wetting and dispersion in the system [3]. To obtain low electrical resistivity and proper rheological behavior, the surfactants have been extensively studied for the conductive paste [8,9].
For the preparation of silver paste, the first step is to prepare the silver powder. And there are many different methods for this purpose and the surfactants always play an important role in the preparation process. Guo et al. prepared ultrafine silver powders for silicon photovoltaic devices using a chemical reduction method. They studied the effects of various surfactants, i.e.. gelatin, oleic acid, polyethylene glycol (PEG), polyvinyl alcohol (PVA), gum Arabic, sodium citrate, Tween-80, on the dispersion of the silver powders and found that PVA was the most effective surfactant [8]. Zhang et al. produced well-dispersed silver powders for electrodes in multilayer ceramic capacitors with superfine particle size using the surfactant of PVP [10]. Tian et al. obtained ultrafine silver powders for general silver pastes by a similar chemical reduction method using various surfactants, i.e., arabic gum, Tween-80, PEG, PVP and PVA, and found that arabic gum was the best surfactant [11].
The silver paste is usually prepared by mixing the silver powders with the organic vehicle. Different types of solvents normally work with different surfactants. Since the organic vehicles for the silver paste is different from the organic solutions for the preparation of the silver powders, the surfactants used in these two steps would be different. Chen et al. produced a kind of silver ink for inkjet printing using the surfactant of 910 (ICI Americas Inc. USA, phosphate ester/phosphoric acid), 9250 (ICI Americas Inc. USA, light aromatic solvent naphtha/propylene glycol monomethyl ether acetate) or KD-6 (ICI Americas Inc. USA, propylene glycol) and found that the surfactant of 9250 can effectively improve the stability and lower the viscosity of the final silver paste [12]. Jiang et al. used stearic acid, ethyl cellulose as the surfactant to prepare the silver paste for thick film devices, and found that the rheological and thixotropic properties were mainly determined by the polymer surfactants, which can be qualitatively described by the viscosity of the organic vehicle [13]. Moreover, using several different surfactants in one formula can effectively improve the dispersibility of the silver paste. Lin et al. prepared silver paste for silicon solar cells with various surfactants of caprylic acid, triethanolamine, and the combination of the both [3]. The latter resulted in a proper thixotropic property through the steric effects. In addition to using surfactants for the dispersion of inorganic powders in polymer matrix, many scholars have also studied the physical methods for dispersing silver powders in silver paste. Lin et al. obtained highly dispersed silver paste by mechanical mixing with three-roller milling and sonication treatment [14].
Through a thorough survey of the published literature, we find most of the reported studies have been limited to either the surfactants for the preparation of silver powders, or the surfactants for the preparation of silver paste. And there is no systematic study on the synergistic effect of different surfactants on the preparation of both the silver powder and the silver paste. In current work, we systematically studied the effects of the surfactants used in different stages, i.e., firstly during the silver powder preparation and then during the silver paste preparation process, on the rheological and electrical properties of the final silver paste.
Another important property for the silver paste is thixotropy, especially for the via-fill silver paste in the LTCC process and the silver paste for the front side metallization of solar cells [2,[15][16][17][18]. The thixotropic behavior is described as the decreasing of the viscosity under constant high shear stress (or shear rate) and the recovery of viscosity with time after the shear stress is decreased or removed [19]. This property is very important for the screen printing process. During the printing step, the silver paste is pushed through the openings of the screen by a squeegee with a high shear stress, and the viscosity of the silver paste is low due to the thixotropic property. Therefore, the silver paste flows easily onto the substrate. After the printing, the shear stress will be gone (snap-off), and the silver paste on the substrate needs to return to its original high viscosity state quickly in order to keep the well-defined shape with as less spreading as possible. Thixotropic property of the system can be modified with inorganic or organic additives, i.e., the so-called thixotropes [20][21][22]. Commonly used thixotropes include organometallic clays, fumed silica, hydrogenated castor oil, aminebased wax, etc. [23,24]. For silver paste, the organic thixostrope are preferred because they can be fired off during the sintering process. The mechanism of thixotropy is due to the destruction of a network structure inside the paste under the external shear stress (viscosity decreases), and reconstruction after removal of the external shear stress (viscosity increases) [25]. This network structure is formed by multiple interactions, i.e., intra-silver particles, silver particle-polymer, and polymer-polymer [3]. Therefore, the optimal thixotropes varies from case to case, due to differences in the formula of the organic vehicles and the surface modifications of the silver particles. In current study, we will also investigate the effects of different organic thixotropes on the thixotropic properties of our silver pastes.
The aim of our research is to optimize the surfactant combinations for the preparation of silver powders as well as silver pastes, and then to identify the effective thixotropes for the silver paste. We firstly prepared micron-sized silver powders by the chemical reduction method using six different surfactants, i.e., gelatin, oleic acid, PVA, sodium citrate, Tween-80 and PVP. Then, for the preparation of the silver paste, four organic vehicles with different surfactants, i.e., tributyl phosphate, Span-85, castor oil and ammonium polymethacrylate (APMAA), were used. Finally, we added eight different thixotropes (Glycerol trioleate, Acrylic resin, Corn oil, 2-Furoic acid, Octadecylamine, hydrogenated castor oil, Palmitic acid, Polyamide) into the silver paste formula with surfactants combination of PVP and Span 85 to study their effects on the thixotropic properties of the final silver paste.

Synthesis of silver powders
There are many reports on synthesizing silver powders with various sizes and/or shapes using physical or chemical methods [26][27][28][29][30]. In current case, micronsized silver powders were prepared by a chemical reduction method. The raw materials were analytical grade AgNO 3 and used as received without further purification. The reductants used by this method are usually ascorbic acid [31,32], zinc [33], glycerol [34], hydrazine hydrate [35][36][37], formaldehyde, glucose [35] and H 2 O 2 [38]. Ascorbic acid was chosen as the reducing agent in this work [31,39]. The chemical reaction of this process is shown as below [32].
We added 5% more ascorbic acid to ensure the complete reduction of AgNO 3 .

Preparation of silver ammonia solution
3.5 g ammonia water (22-28 wt%, Damao Chemical Reagent Factory, Tianjin, China) was added into 294 mM AgNO 3 aqueous solution. Pellucid-turbidpellucid appearance of the mixture was then observed.

Preparation of reduction solution
1 g surfactant (gelatin, oleic acid, PVA, trisodium citrate, Tween-80, PVP) was added into 150 mM ascorbic acid aqueous solution and then stirred for 10 min until the solution was completely pellucid.

Preparation of silver powders
Silver ammonia solution was poured gradually into the vigorously stirred reduction solution in 1 min. The colorless solution immediately became brown due to the formation of silver powders, and the solution was further stirred for another 1 h to fully complete the reaction. The silver powders were collected by centrifuging at 4000 rpm for 10 min and then washed three times with deionized (DI) water and ethanol. The obtained silver powders were dispersed by sonification for 4 h with Ultrasonic cell pulverizer (OM-650Y, Ou Meng, Shanghai, China) with a power of 1250 W. Finally, the powders were freeze-dried at -41°C for 24 h. We found that such treatment (freeze-drying) can further improve the dispersion of the silver powders. Solvents, plasticizer and surfactants were firstly weighed and mixed, then the ethyl cellulose and 1-hexadecanol was added to the mixture according the formula shown in Table 1. The mixture was stirred for 2 h until the solid ethyl cellulose and 1-hexadecanol were fully dissolved.

Preparation of organic vehicles
The amount of solvent terpineol was decreased from 20 to 19 wt% when the thixotrope (1 wt%) was added to the organic vehicle.  Table 2 and then mixed by a roller mill for 30 min. The powder mixture was placed in an Al 2 O 3 crucible and heated to 1400°C at a rate of 10°C/min in a muffle furnace. The mixture was kept at 1400°C for 30 min and then poured into DI water directly. The obtained glass powders were ball-milled (MSK-SFM-1, KJMTI, Hefei, China) at 400 rpm for 20 h with zirconia balls in ethanol medium. Finally, the mixture was dried in air at 60°C for 5 h.

Preparation of silver paste
The organic vehicle/silver powders/glass frits were mixed in a weight ratio of 19/78/3 and then the mixture was stirred in a blender for 5 h at a temperature of 70°C. The mixture was then ball-milled in Nylon jar using zirconia balls of 2-12 mm diameters at a speed of 500 rpm for 15 h. The schematic diagram of the preparation processes of silver powders and silver pastes is shown in Fig. 1.
We used screen printing method to prepare silver electrode lines. The silver paste was printed on alumina ceramic substrate. During the sintering of the silver paste, the temperature was increased from 0 to 100°C, kept at 100°C for 2 h, and then raised to 870°C with a heating rate of 10°C/min, dwelled for 10 min, and finally cooled to room temperature naturally.

Characterizations
The microstructures of the silver powders and the fired silver pastes were studied by a field-emission scanning electron microscopy (SU8010, HITACHI, Tokyo, Japan). Thermal gravity analysis (TGA, STA449F5, NETZSCH, Bavarian, Germany) was used to quantitatively determine the content of the surfactant and the organic vehicle in the silver paste, and also their thermal decomposition behavior with temperatures. The rheological properties of the silver paste were measured with a rotational viscosimeter (MCR301, Anton Paar, Graz, Austria) using a parallel plate geometry and the temperature was maintained at 25°C. The shear rate was changed from 0 to 600 s -1 during the measurements. The thixotropic index was defined as the ratio of viscosity measured at the shear rate of 6 s -1 to that measured at 60 s -1 as the shear rate is increasing during the measurements. The specific surface area was measured by the Brunner-Emmet-Teller (BET, V-Sorb 2008B, Gold APP Instrument Corporation, Beijing, China). The resistance of the sintered silver electrode was measured by four-point method (Keithley 2635B, Tektronics, Shanghai, China).

Results and discussions
The effects of surfactants (gelatin, oleic acid, PVA, trisodium citrate, Tween-80 and PVP) as well as the sonification treatments on the morphology of silver powders are shown in Fig. 2. In Fig. 2a, c, e, g, i and k are shown the microstructures of the silver powders prepared with the surfactant of gelatin, trisodium citrate, PVA, PVP, Tween-80 and oleic acid, respectively. All of these powders underwent sonification treatments for 4 h. The silver particles are quasispherical and well-dispersed. One can find that the surface of the silver particles prepared with gelatin, PVA, and PVP surfactants are smooth and the particles are also well-dispersed, compared to those prepared with trisodium citrate, Tween-80 and oleic acid surfactants. On the other hand, in Fig. 2b, d, f, h, j and l, are shown the microstructures of those silver powders without sonification treatments. These silver powders are poorly dispersed, especially for the samples prepared with the surfactant of oleic acid and gelatin, as shown in Fig. 2l and b, respectively. Figure 3 shows the particle size distributions of the silver powders synthesized with different surfactants. It is obviously observed that the silver powders with gelatin as a surfactant have the narrowest size distribution and also the smallest averaged particle size. The averaged particle size is 0.46 lm, 0.70 lm, 0.86 lm, 0.61 lm, 0.58 lm and 0.93 lm, for the silver powders prepared with the surfactant of gelatin,  were chosen for our silver paste. For these three kinds of silver powders, their specific surface areas were measured to be 1.86, 0.24, and 0.43 m 2 /g, respectively.
The thermogravimetric (TG) and differential scanning calorimetry (DSC) profiles for the silver powders prepared with various surfactants are shown in Fig. 4. Figure 4a shows the TG-DSC curves for the silver powders prepared with the surfactant of gelatin. It shows a weight loss of 0.3% as the temperature was increased to 280°C, without obvious change of the heat flow. Then, with the temperature further increased, a weight loss of * 2% was observed, which was accompanied by an exothermic peak around 300°C. This exothermic peak was due to the reaction between the gelatin surfactant and oxygen, i.e., the firing of the gelatin. There was no further decrease in the weight above 350°C, which indicates that gelatin has been completely fired.
In Fig. 4b-e, similar phenomenon was observed for the silver powders prepared with trisodium citrate, PVA, PVP, and Tween-80, respectively, except the variations in the exothermic peak temperatures, which were 198°C, 230°C, 212°C and 211°C, respectively. These exothermic peaks were due to the decomposition of trisodium citrate, PVA, PVP, and Tween-80, respectively [40][41][42][43][44]. Figure 4f presents the TG-DSC curves for the silver powders prepared with oleic acid as the surfactant. The TG curve indicates two stages of weight loss. The first stage of weight loss occurred between 160 and 260°C with a weight loss of 3%. The second stage occurred at 274-302°C with a weight loss of 1.1%. There was no change in weight above 350°C. This two-stage weight loss may be due to the partial oxidation of the oleic acid on the surface of the silver powders prior to this measurement and thus formation of some unknown species which have different temperature behavior in the TG measurements [45,46]. Our results indicate that all these organic surfactants will be completely burned out during the sintering process of the silver paste, which usually occurs at a temperature of 800-900°C.
The TG and DSC profiles for the silver paste prepared with various surfactant combinations are shown in Fig. 5. Three different surfactants were used during the preparation of the silver powders, i.e., gelatin, PVA and PVP, while four different surfactants were added during the preparation of the   Table 3. The weight loss in this stage was attributed to the decomposition of other residual organics, i.e., ethyl cellulose and the surfactants, and was accompanied by distinct exothermic  [43,44], as indicated in Fig. 5. Generally speaking, the silver pastes with the surfactant of gelatin (for the preparation of silver powders) exhibits larger weight loss (averaged over the four silver pastes with different surfactants-APMAA, tributyl phosphate, Span-85 and castor oil), i.e., 18.19% versus 17.26% for the silver pastes with PVA, and 16.00% for the silver pastes with PVP. This is consistent with the TG results of the silver powders as shown in Fig. 4.
Through the analysis of these TG-DSC curves, one can obtain the precise temperature range when the organic vehicle will evaporate and/or decompose. Another important aspect worthy of notice for these curves is the temperature-dependent evaporation/ decomposition rate of the organic vehicles. This is important, due to the fact that it will affect the formation of microstructural defects during the sintering process of silver pastes. For example, if the organic vehicle is burned out too fast at some specific temperature, then there will be an abrupt change in the weight loss as well as in the volume, which will probably result in local microcracks or microcavities, etc. These defects will greatly deteriorate the performance of the final products. So, we analyzed the evaporation/decomposition rate of the organic vehicles by linearly fitting the TG curves in the temperature range covering stage I, II, and III, where the major weight loss occurred. It was found that the slopes of the fitting lines were -0.  Fig. 6. According to the fitting results, the silver paste with surfactant combination of PVA and Span-85 shows a relatively small slope (-0.051) and high coefficient of determination (0.984), which implies that the evaporation/decomposition of this organic vehicle is relatively slow and uniform throughout the temperature range. Electrical resistivity of the sintered silver pastes (i.e. silver conductor lines) is defined as.
where q (mX mm) is the electrical resistivity, R (mX) is the electrical resistance defined by Ohm's law, A (mm 2 ) is the cross-sectional area, L (mm) is the line length, w (mm) is the width, and th (mm) is the thickness. Table 4 shows the electrical resistivities of the sintered silver pastes. It was found that silver pastes with Span-85 as the surfactant (during preparation of silver paste) result in a relatively small electrical resistivity. On the other hand, the silver pastes with PVP as the surfactant (during preparation of silver powders) also exhibit a relatively low electrical resistivity. These results lead to the conclusion that the silver paste prepared with surfactant combination of PVP and Span-85 exhibits the best performance, with an electrical resistivity of as low as 0.11 mX mm.
The electrical resistivity of our silver paste is several times larger than that of the electrical resistivity of bulk silver, which is 0.0161 mX mm [47]. However, our result is comparable with the published results of other researchers, as shown in the Table 5. One should be aware that the electrical resistivity values listed in the table are highly dependent on the sintering temperature, which is related to different applications of the silver pastes. In current study, the sintering temperature was not optimized yet. The larger resistivity of our samples can be attributed to several factors. The first one is the size and the shape of the silver powders. It is generally accepted that by mixing silver powders with different shapes or size distributions, the resistivity can be reduced greatly [5]. Yet, in current case, we did not optimize the shape and size parameters. The second one is the mixing methods. We used ball-milling to mix the silver powders and the organic vehicles, which is not the best technique for the preparation of silver paste, especially for high-loading silver paste. And the more delicate mixing method is through three-roller milling [3]. The third one is the inorganic additive (glass component), which will affect the resistivity of the silver paste. We have not yet optimized the size and the filling ratio of the glass component, on which we will focus in the future. Moreover, the screen printing process for the formation of the green silver pastes was not optimized, which may result in poor uniformity of the final fired silver pastes. And this also will lead to a higher value of the resistivity observed in this study.  The microstructures of the sintered silver pastes prepared with various surfactant combinations are shown in Fig. 7. There is no obvious trends regarding the effects of various surfactants on the microstructures of the sintered silver pastes. However, for the surfactants during the powder preparation, gelatin and PVP seem to be better than PVA since the sintered silver pastes exhibit a denser microstructure, as shown in Fig. 7a-c and j-l. Regarding the surfactant during the paste preparation, the APMAA seems to be more appropriate for producing a dense microstructure of the sintered silver paste, as shown in Fig. 7b, f, and j. It is also worthy of notice that the silver paste prepared with surfactant combination of PVP and Span-85 also exhibits a dense microstructure, with the averaged grain size of 5.88 lm. This sample also shows the lowest resistivity. The distributions of the grain size for all the sintered silver pastes are summarized in Fig. 8. The samples with lower electrical resistivity usually exhibit larger averaged grain size. For example, the sintered silver paste with gelatin-castor oil surfactants and PVA-  castor oil surfactants showed the electrical resistivity of 0.18 and 0.14 mX mm, respectively, and their averaged grain sizes were found to be 4.9 and 5.26 lm, respectively. Larger grain size means less grain boundary, which usually possess higher electrical resistance. However, other factors, such as voids, cracks, and impurities also play important role in determining the electrical resistivity of the sintered silver paste. Finally, we studied the effects of different organic additives (thixotropic agent) on the thixotropic behavior of the silver paste. Thixotropic behavior refers to a reduction in magnitude of rheological properties of a system, such as elastic modulus, yield stress, and viscosity, on application of shear strain, which can reversibly and isothermally recover to its origin state after removal of the shear strain [48]. It is also called shear-thinning behavior. Therefore, the thixotropic behavior is shear-strain dependent and also time dependent. The thixotropic behavior is related to an internal structure of the dispersion system. For a particle dispersion system, the internal structure is formed from a network-like arrangements of anisometric dispersed particles, in a liquid medium, having areas or points at which adhesive forces may arise. Such a network would provide a structure enabling the dispersion to appear to react elastically to a low stress. And under a high stress, catastrophic breaking of the network linages occurs and the anisometric particles assume random orientation with respect to the active regions for adhesion, and continuous flow begins to take place in the dispersion. When the external stress is stopped, random Brownian movement will bring areas or points of adhesion of the anisometric particles into the volume of space in which linkage could again be formed and the dispersion will enter a state of mechanical rest again [49]. Thixotropic behavior is important for the silver paste. During screen printing, the silver paste must stay rest on the screen before the squeegee stroke. And when the squeegee moves across the paste, the paste viscosity should decrease in order to allow itself flowing through the screen openings. After snap-off, the silver paste should recover its initial viscosity in order to avoid spreading and keep good shape definition [5]. In current report, we will try to find the effects of different organic additives on the thixotropic behavior of our silver paste. We firstly studied the thixotropic behavior of the organic vehicles with various thixotropic agents, then that of the silver pastes, and finally try to find out the relationship between their thixotropic behaviors. Figure 9 shows the viscosity vs. shear rate curves of organic vehicles using different thixotropic agents. Thixotropic index is often used to characterize the thixotropic property of the silver pastes [3,21,50]. Yield stress can also be used to quantify the thixotropic property [6]. Larger thixotropic index or yield stress indicate a more prominent thixotropic behavior. Lin et al. found that the optimum thixotropic index (defined as the ratio of viscosity measured at 3.84 s -1 to that measured at 38.4 s -1 ) for their silver paste was in the range of 2.5-3.0 [3]. Zhang et al. claimed a thixotropic index (defined as the ratio of viscosity measured at 5 rpm to that measured at 50 rpm) of 9.35 to be a proper value for the organic vehicle of their silver paste [50]. Table 6 summarized the thixotropic indices of various organic vehicles with different thixotropic agents, based on the data shown in Fig. 9. It was found that the organic vehicle with hydrogenated castor oil as the thixotropic agent exhibited the largest thixotropic index (1.56), while the organic vehicle with octadecylamine as the thixotropic agent showed the smallest thixotropic index (1.11).
Substances with plastic or pseudo-plastic flow properties exhibit a yield stress. This yield stress is related to the internal structures of the system, where randomly connected particles or long molecular chains form a stable network at rest [51]. In order to obtain the yield stress (s 0 , Pa) of our organic vehicles, we used Bingham (as shown in Eq. 3) and Casson (as shown in Eq. 4) models to extrapolate the yield stress values through fitting the shear stress vs. shear rate curves of the organic vehicles. A Bingham model (Eq. (3)) is characterized by a linear relationship between the shear stress and the shear rate, above a threshold. For a Casson model (Eq. (4)), there is also a relationship between the shear stress and shear rate above the yield stress, but it is non-linear [52]. This method was also adopted by other researchers [53].
where s is the shear stress (Pa), s 0 is the yield stress (Pa) and g 1 is the viscosity (Pa s) when the shear rate _ c (s -1 ) reaches an infinite value. Figure 10a shows the fitting curves of shear stress vs. shear rate for the organic vehicle with hydrogenated castor oil as the thixotropic agent, based on the Bingham and Casson model. The shear rate range for the fittings was chosen to be 100-600 s -1 . The extrapolated yield stress values were 138.8 Pa and 25.5 Pa based on the Bingham and Casson fittings, respectively. Both fittings seem fairly good with the coefficient of determination of 0.989 and 0.994, respectively. It is not unusual for a yield stress obtained by one model to be very different from that by another model [52]. Indeed, there are many models suitable for different dispersion systems and one master model for all situations does not exist [54]. On the other hand, it was shown that the Casson model is a semi-empirical model that has been applied to fit the flow curves of many paints and printing ink formulations [51]. Figure 10b plots the yield stress values for our organic vehicles with 2 Pa, respectively. It was found that both models give higher yield stress values for the organic vehicles using the thixotropic agents of hydrogenated castor oil, Corn oil and Acrylic resin. These results are coincident with the thixotropic indices as shown in Table 6, where these three samples exhibit larger thixotropic index values as 1.56, 1.37, and 1.32, respectively. Our results confirm that both the thixotropic index and the yield stress can be used to quantify the thixotropic behavior of the organic vehicles.
Next we used the organic vehicles with the most prominent thixotropic behavior, i.e., organic vehicles with thixotropic agent of hydrogenated castor oil, Corn oil and Acrylic resin, to prepare silver pastes and studied their rheological properties. The viscosity curves of these silver pastes are shown in Fig. 11a. All the three silver pastes exhibit obvious shearthinning behavior. At a very low shear rate, for example 0.1 s -1 , the viscosity is 998.0, 465.0, and 240.0 Pa s for the silver paste with hydrogenated castor oil, Corn oil and Acrylic resin as the thixotropic agent, respectively. When the shear rate was increased to 100 s -1 , the viscosity was decreased to 12.5, 15.0, and 4.5 Pa s, respectively. And after the shear rate returned back to 0.1 s -1 , the viscosity values were 74.7, 164.0, and 16.2 Pa s, respectively. Thus a hysteresis loop was clearly observed for all these samples. It has been shown that the gap between the upper and lower part of the hysteresis curve can be also used to measure the degree of thixotropy of the system [55]. In that sense, one can easily identify that the silver paste with the hydrogenated castor oil as the thixotropic agent exhibits the highest degree of thixotropy and that with Acrylic resin as the thixotropic agent possesses the lowest degree of thixotropy, as shown in Fig. 11a. This is consistent with the thixotropic indices of these silver pastes, which were 3.95, 3.17, and 2.77 for the silver pastes with hydrogenated castor oil, Corn oil and Acrylic resin as the thixotropic agent, respectively. In addition, we extrapolated the yield stress values of these three silver pastes by fitting their shear stress vs. shear rate curves, as we did for the organic vehicles in Fig. 10. The yield stress values are shown in Fig. 11b, which are 677.6, 578.6, and 168.7 Pa based on the Bingham 3 Pa, respectively. As discussed above, larger yield stress value indicates a more prominent thixotropic behavior. Therefore, the silver paste using the thixotropic agent of hydrogenated castor oil possesses the highest degree of thixotropy. This is fairly in line with the results of the organic vehicles, where the organic vehicle with the hydrogenated castor oil also possesses the highest degree of thixotropic index, as shown in Table 6. Therefore, we can conclude that through studying the thixotropic behavior of the organic vehicles, one can roughly predict the thixotropic behavior of their related silver pastes. This will be helpful in determining the optimal formulation of the custom-oriented silver paste. It can save time and reduce cost for the development of silver paste or other similar systems. The viscosity recovery profiles of the silver pastes after removal of external shear forces are shown in Fig. 12. The at-rest viscosity of the silver pastes prepared with the thixotropic agents of Hydrogenated Castor Oil, Corn oil, Acrylic resin is 1117 Pa s, 541 Pa s, and 255 Pa s, respectively. After the application of a constant high shear rate, i.e., 100 s -1 for 5 min, the viscosity was decreased to 5.59 Pa s, 7.01 Pa s, and 4.98 Pa s, respectively. These values are the starting points of those curves shown in Fig. 12. One can find that, after the application of a constant high shear rate, the silver paste prepared with the thixotropic agent of Hydrogenated Castor Oil exhibits the largest amount of viscosity decrease, i.e., from 1117 to 5.59 Pa s. Moreover, it can be seen from the Fig. 12 that the silver paste with the hydrogenated castor oil as the thixotropic agent recovers faster than the other two silver pastes. However, after 10 min, the viscosity of the silver pastes with Corn oil and Acrylic resin as the thixotropic agent gradually saturated while that of the silver paste with hydrogenated castor oil as the thixotropic agent was still increasing. This means that the silver pastes with hydrogenated castor oil as the

Conclusions
In this report, we systematically studied the effects of the surfactants used in different stages, i.e., firstly during the silver powder preparation and then during the silver paste preparation process, on the rheological and electrical properties of the final silver pastes. The silver paste prepared with surfactants combination of PVP and Span-85, exhibited best performance with the electrical resistivity of 0.11 mX mm. The microstructure of the sintered silver paste is very dense. We show that both the thixotropic index and the yield stress can be used to quantify the thixotropy degree of the silver paste. Organic vehicles and their silver pastes exhibit similar thixotropy behaviors. Hydrogenated castor oil was shown to be an effective thixotropic agent for silver paste. Our results are beneficial for developing silver pastes for LTCC applications in a fast and economic way.