low-temperature silver pastes for HJ

Rheological properties and related screen-printing performance of low-temperature silver pastes for a-Si:H/c-Si heterojunction solar cellsDongpo Chen •Lei Zhao •Hongwei Diao •Wenbin Zhang •Ge Wang •Wenjing WangReceived:15July 2014/Accepted:1September 2014/Published online:14September 2014ÓSpringer Science+Business Media New York 2014Abstract Four different low-temperature silver pastes were utilized to make metal grids by screen printing for silicon heterojunction solar cells.The rheological behav-iors of the low-temperature silver pastes were characterized by viscosity test,thixotropy test,oscillatory stress sweep test and creep-recovery test.The correlationship between the screen-printing performance and the rheological prop-erties was investigated.It was found that the shear thinning behavior and the thixotropy behavior of the silver pastes were desirable for the screen-printing process.An obvious viscoelastic behavior of the silver paste was also helpful for improving its printability.Further,good recovery and low creep and recovery compliances could minimize the printing defects and the tendencies to bleed out during the screen-printing,and thus increase the aspect ratio of the printed grids.1IntroductionAmorphous/crystalline silicon (a-Si:H/c-Si)heterojunction (SHJ)solar cell,which combines c-Si wafer and a-Si:H thin film technologies,has been attracting more and more attention because of the several following advantages:low temperature process,compatibility with industrialrequirements,excellent surface passivation,and very high efficiency [1,2].Panasonic has recently announced a record conversion efficiency of 24.7%at research level for such solar cell using its heterojunction with instrinsic thin layer (HIT)structure [3].For this solar cell,the front silver grid is usually fabri-cated on transparent conductive oxide (TCO)electrode by screen-printing.However,due to the SHJ solar cell struc-ture is fabricated at low temperature usually lower than 250°C,the low-temperature silver paste is thus necessary for the grid fabrication.Up to now,some studies on low-temperature silver pastes utilized in SHJ solar cells can be found.Jiang et al.[4]pointed out that the proper choice of the paste curing mechanism could significantly boost the paste electrical conductivity,and thermoset system was better than thermoplastic system.Veschetti et al.demon-strated the use of low temperature resin silver paste for the front contact on TCO layer.However,in order to improve the fill factor of SHJ solar cells,the resistivity of such silver paste needed to be lowered further [5].Recently,we also did some primary studies on the electrical performance of the screen-printed silver grid fabricated on ITO layer via low-temperature sintering [6].Besides the electrical performance,in order to obtain high conversion efficiency for the solar cell,the shape of the silver grid should be also controlled.It should have a narrow width and a high height–width aspect ratio.The narrow width can reduce the shadowing effect.And the high height–width aspect ratio can improve the conductance.So the screen printability of the low-temperature silver paste becomes quite important.This feature is determined by the rheological behaviors of the low-temperature silver paste.Some researchers have studied the rheological proper-ties of the usually used high-temperature silver pastes for c-Si solar cells.For examples,Hoornstra et al.pointed outD.Chen ÁL.Zhao (&)ÁH.Diao ÁG.Wang ÁW.Wang Key Laboratory of Solar Thermal Energy and PV Systems of Chinese Academy of Sciences,Institute of Electrical Engineering,Chinese Academy of Sciences,Beijing 100190,Chinae-mail:zhaolei@W.ZhangChaori Solar Energy Science and Technology Co.Ltd.,Shanghai 201406,ChinaJ Mater Sci:Mater Electron (2014)25:5322–5330DOI 10.1007/s10854-014-2308-1that the rotation viscometry could be used as afirst order predictor for the print behavior of the paste,but did not duplicate the shear history of the paste during printing. They utilized the oscillation rheometry to simulate the recovery behavior of some commercially available silver pastes during the screen-printing[7].And Neidert et al.[8] further found a fast recovery could help the silver pastes to achieve grids with good aspect ratio.However,they used few characterization methods.And at the same time,the high-temperature silver pastes they characterized are usu-ally made of micro-sized silver particles with glass frit,which are obviously different from the components of the low-temperature silver pastes we concern here.The low-temperature silver pastes are mainly made of nano-sized particles and without glass frit.So to investigate the rhe-ological properties of the low-temperature silver pastes completely by utilizing various characterization methods will be very helpful for the fabrication of SHJ solar cells.Here,we studied systematically the screen-printing performance of four different kinds of low-temperature silver pastes and their corresponding rheological behaviors. As a result,the correlationship between the printing per-formance and the rheological properties was found.2ExperimentalFour different low-temperature silver pastes(namely paste SM1,SM2,SM3and SM4)were investigated.These pastes are uniformly dispersed mixtures of silver particles and some organic components such as binders and solvents. Table1provides the details of these paste components. Although all the pastes adopt bisphenol A epoxy resin as the binder,its fraction is different.Different organics are utilized as the solvents.And the metal particle size is also different.Such differences are expected to make the pastes show different rheology behaviors[9].The printing trials of the four pastes were carried out with the screen printing parameters shown in Table2.The surface and cross-sectional morphologies of the printed grids were observed by scanning electron microscopy (SEM)(FESEM S4800,Hitachi,Japan).In order to correlate the printing performance of the silver pastes with their rheological properties,several rheological measurements were carried out on Rheometer MCR302(Anton Par Ltd,Germany),including viscosity test,thixotropic test,oscillatory amplitude sweep test,and creep-recovery test.As shown in Fig.1,the parallel plate measuring geometry consists of a25mm diameter stain-less steel upper plate and a40-mm diameter stationary lower plate.The gap width between the parallel plates should be set about ten times the largest particle diameter in order to avoid any anomalous effect due to particle jamming and deformation[10].Taking this into consider-ation,a gap width of0.5mm was chosen for all the rhe-ological measurements here.For the viscosity test,a shear rate ranged from0.1to 1,000.0s-1was used.And for understanding the structural breakdown and recovery of the silver pastes during the screen-printing process[11,12],the thixotropy measure-ment was carried out by a control shear rate(CSR)step test.Three steps were created to simulate the behavior of the silver pastes during the screen-printing process:(1)pre-printing,(2)printing,and(3)post-printing,as shown in Fig.2.During the pre-printing step,the paste was in the rest on screen,but slightly pre-stressed by pre-stirring and flooding.So for this step,a low shear rate of0.1s-1forTable1Particle size and components of the four kinds of low-temperature silver pastesPaste Particle size Main compositionsMetal Binder Solvents Curing agentSM1*100–150nm*87.3%Ag*7.2%Bisphenol A epoxy resin*4.1%C10H20O4*0.4%ModifiedimidazoleSM2*2–8l m*56.6%Ag,*25.5%Cu *3.1%Bisphenol A epoxy resin*7.2%C8H18O3,*6.4%C7H8O*0.3%ModifiedimidazoleSM3*100–200nm*83.4%Ag*2.7%Bisphenol A epoxy resin*4.1%C6H10O4,*9.0%C7H12O4*0.2%ModifiedimidazoleSM4*150–250nm*91.7%Ag*2.9%Bisphenol A epoxy resin*4.9%C8H10O2*0.3%ModifiedimidazoleTable2Screen-printing parametersSnap-off distance(mm) 1.2Squeegee length(mm)180.0Squeegee contact angle(°)70Squeegee pressure(MPa)0.30Scraper pressure(MPa)0.28Counter pressure(MPa)0.10Squeegee velocity(mm s-1)80Scraper velocity(mm s-1)12020s was used.For the printing step,a high shear rate of 1,000.0s -1for 9s was used to simulate the printing pro-cess.For the post-printing step,the shear rate returned back to the initial of 0.1s -1for a long period so that the pastes could recover and regain the structures [13].Further,the oscillatory amplitude sweep test was carried out to identify the viscous and elastic properties of the silver pastes [14].Such test could be also designed to determine the linear viscoelastic region (LVR)of silver pastes where the sample structure remains undamaged,irrespective of the applied shear stress [15].The test was performed by applying sinusoidal varying stress,ranged from 0.1to 500.0Pa,with a constant frequency of 1Hz to the sample,as shown in Fig.3,and measured the induced sinusoidal varying strain.The applied stress and the resultant strain are expressed as:s ¼s 0sin ðx Át Þð1Þc ¼c 0sin ðx Át þd Þð2Þx ¼2p fð3Þwhere,d is the phase shift,f is the frequency,x is the angular velocity,s 0is the stress amplitude,c 0is the max-imum strain and t is the time.The ratio of the applied shearstress to the maximum strain is called the complex modulus (G *),which is an indicator of the resistance to the defor-mation for material,and is defined as:G üs 0c 0ð4ÞThe complex modulus can be divided into elastic and viscous components.The elastic component is calledstorage modulus (G 0),which expresses the ability of the material to store energy.It shows the material stiffness and structuring degree and is defined as:G 0¼s 0c 0 cos ðd Þ¼G Ãcos ðd Þð5ÞThe viscous component is called loss modulus (G 00),which describes the viscous dissipation of energy in the form of heat.It is related to the fluidity of the material and is defined as:G 00¼s 0c 0 sin ðd Þ¼G Ãsin ðd Þð6ÞThe loss factor tan(d )can be expressed as a function of the storage modulus and the loss modulus,which is given by Eq.(7).It gives an indication on material internal structure strength and helps to define viscoelasticity.The larger the loss factor is,the stronger the energy dissipation willbe.Fig.1The rheometer and the schematic diagram of the parallel-plate measurementgeometryFig.2CSR step test with three intervals:constant low/constant high/constant low shear rate,to simulate the pre-printing,printing and post-printingstepsFig.3The applied oscillatory stress profiletan ðd Þ¼G 00Gð7ÞFurther,in order to evaluate the relaxation time of the silver pastes,the creep-recovery test was also carried.Such test is quite useful for monitoring low-shear phenomena such as settling in suspensions,structure build-up,break-down over time and slump under the influence of gravity [16].The creep-recovery test was performed by applying a constant stress (s 0,within the LVR)for a 300s creep period and then no stress for a recovery period of 600s.The induced strain (c )in the sample was recorded,the result of the test was obtained as the compliance (J )of the pastes,which was the ratio of the strain developed in the sample to the applied stress and given by the equation below:J ðt Þ¼c ðt Þ=s 0ð8Þ3Results and discussion3.1Morphologies of the printed gridsFigure 4presents the printed grids after drying at 130°C for 25min for all the four silver pastes.The grids were obtained by using the same screen-printing pattern.Theline width,the line height and the aspect ratio are shown in Fig.5.It can be observed that paste SM2can produce grid with relatively flat surface.However,it has a much lower thickness,a larger line width,and a significantly lower aspect ratio than the other three pastes.Paste SM3is only a little better.These make pastes SM2and SM3not acceptable for use.What is more,it was found that these two pastes could clog the screen mesh easily during the printing process.Paste SM1can produce the grid with obvious rough surface and edges,and the mist formation is also observed,while the aspect ratio of the printed line is larger.The grid printed with paste SM4has little spreading along the edges,and both of the height and the aspect ratio are the largest.So,paste SM4may be the most suitable one for using in SHJ solar cells if only considering the print-ability,which is thought to be correlated with the corre-sponding rheological properties of the paste.3.2Viscosity test of the pastesFigure 6presents the viscosity versus shear rate curves.When the shear rate increases from 0.1to 50.0s -1,all the four pastes display dramatic decrease in the viscosity.AfterFig.4SEM surface micrographs of the different grids printed on ITO substrates from the four kinds of silver pastes:a SM1,b SM2,c SM3,d SM4the region,the viscosity of the pastes decreases gently as the shear rate increases further.The decrease in viscosity with the increasing shear rate is known as shear thinning behavior [17–19].It is a desirable property to enable the silver pastes to flow through the screen mesh easily under the pressing of the squeegee stroke during the screen-printing process.However,the critical shear rate viscosity,which is defined as the minimum shear rate to initiate paste flow,is not observed in our experiment.The reason may be that the critical shear rate of all the pastes is \0.1s -1.Among the four silver pastes,pastes SM3and SM2present larger viscosity than SM 1and SM4in the higher shear rate,which makes them more difficult to transfer through the printing mesh during the printing process.The mesh is especially easy to be logged,even adjusting the printing parameters.This means that obvious shear thin-ning behavior is beneficial for the screen-printing.3.3Thixotropic test of the pastesThixotropy refers to the time-dependent behavior of the paste,in which the paste’s viscosity decreases with thetime of shearing and gradually recovers when the shear is removed.We adopted the shearing rate of 0.1and 1,000.0s -1to simulate the squeegee off and on steps during the printing process respectively.The results are as shown in Fig.7.In general,all of the four silver pastes display the thixotropy behavior.However,the extracted data given in Table 3presents obvious difference.Paste SM2and SM3have a higher viscosity than paste SM1and SM4at the shear rate of 1,000.0s -1.When the shear rate goes back to the initial of 0.1s -1,paste SM2can sub-stantially recover its initial state in a relatively long time of about 134s,and paste SM3can recover about 92%in a relatively short time of about 20s.For SM1,the recovery level is about 41%,and the recovery time is about 51s.While paste SM4can recover very quickly,although its recovery level is only about 13%.The rapid recovery can prevent the grid from deforming,and thus obtain narrower lines and higher aspect ratio.3.4Oscillatory amplitude sweep test of the pastes The measured results of G 0and G 00for the four different silver pastes are presented in Fig.8.All silver pastes present two different regions.At low shear stress,an initial flat region with constant G 0,which is known as LVR,is found.In LVR,the induced strain is proportional to the applied shear stress and the internal structure remains undestroyed,showing the elastic behavior of the silver pastes.It is also observed that the storage modulus G 0is dominant over the loss modulus G 00for all the silver pastes in LVR,indicating that the silver pastes are solid-like structured fluid.Beyond LVR,it is the nonlinear viscoelastic region.In this region,the value of G 0decreases gradually with the increasing applied shear stress.This decrease in the G 0value indicates the break-down of the structure of silver paste,which illustratesthatFig.5Geometrical data of the printed grids achieved with the four different kinds ofpastesFig.6Viscosity as a function of the shear rate for the four different low-temperature silverpastesFig.7Time-dependent viscosity function and structural regeneration of the four kinds of silver pastes during the low (0.1s -1)/high (1,000.0s -1)/low (0.1s -1)shearing rate processesthe paste structure begins to weaken and the paste grad-ually breaks down,resulting in a more liquid-like behavior.When the stress increased up to about180Pa, the G0and G00value of paste SM1,SM3and SM4 decrease rapidly,illustrating that these three pastes are more vulnerable to the structural breakdown,which can be attributed to the separation between the silver particles and the organic components.While there is an increase in the G0and G00value for paste SM2beyond72Pa,which indicates a shear thickening behaviour.Table4shows the range of LVR,G0,G00,G0/G00,the applied shear stress at G0=G00and phase angles d for all the four silver pastes.Paste SM4with an end value of 16Pa shows the largest LVR,followed by SM1,SM3and SM2.The highest value of G0is also produced by SM4, followed by SM1,SM3and SM2.The wider LVR and the higher G0of SM4indicate that the paste has a higher structural resistance towards to the applied stress and it can withstand a broader range of deformation before its struc-ture breaks down.Further analysis of the elastic behavior of the paste was carried out by calculating the ratio of solid-like to liquid-like behavior(G0/G00)within the LVR.The ratio of G0/G00 gives an indication of the interaction strength within the internal structure of the silver pastes.A higher G0/G00ratio indicates that the material is elastic in nature[20].Paste SM4has the highest G0/G00of3.7,followed by SM1with a value of3.3,SM3with a value of1.6and SM2with a value of1.5.The higher ratio indicates that the paste SM4is more elastic and cohesive compared to the pastes SM1, SM2and SM3,which can prevent the paste from bleeding between the screen and the ITO substrate and thus provide a higher resistance to slump.We also analyzed the applied shear stress at the point where the storage modulus was equivalent to the loss modulus(G0=G00)for the four pastes.In the nonlinear viscoelastic region,the values of G0and G00represent the solid and liquid characteristic.The crossover point at which G0=G00gives an indication of how solid or liquid-like the sample is.The lower the value of stress at the crossoverTable3Viscosity(unit:Pa s) and recovery level of the four silver pastes at different shear rates corresponding to the different steps in the printing process(25°C)Sample0.1s-1(beforesqueegee stroke)1,000.0s-1(squeegee stroke)0.1s-1(aftersqueegee stroke)Recovertime(s)Recoverylevel(%) SM198724005141SM26466646134100SM387961,6402092SM41,1962153413Fig.8Storage modulus G0and loss modulus G00as a function of shear stress for the four kinds of Ag pastes:a SM1,b SM2,c SM3,d SM4point is,the more liquid-like the sample should be[21].In addition,the stress at G0=G00can be used as an indicator for assessing the cohesiveness of the pastes.According to the result shown in Table4,paste SM2is the most liquid-like,followed by paste SM3,SM1and SM4,and paste SM2is less cohesive than others.The difference in rheo-logical behavior of the four pastes could be possible due to the formulation of the pastes itself[18].The interaction between the particles and the organic components has a direct effect on theflow behavior of the silver pastes.The phase angle(d)provides a quick way of assessing the viscoelastic behavior of silver pastes.A phase angle close to0°indicates an elastic(solid)behavior and an angle close to90°indicates a viscous(liquid)behavior. Viscoelastic behavior will result in the phase angle between0°and90°[22].For example,a low phase angle may indicate that the paste is very tacky.Alternatively,a high phase angle may indicate that a paste can slump easily.Table4summarizes the information of phase angle as a function of the applied shear stress.SM4and SM1 present the lower phase angle.All the above results show that SM4and SM1have obvious elastic(solid-like)bined with the printing performance shown in Fig.4,we can deduce that such elastic behavior of the silver paste is expected for the screen printing.3.5Creep and recovery test of the pastesThe results of creep compliance J c and recovery compli-ance J r for all the four pastes are shown in Fig.9and Table5.Paste SM2shows the highest J c and J r.And paste SM4presents the lowest ones.A high J c implies that the sample can be easily deformed and the extent of the deformation will depend on the strength of the micro-structure.Figure9and Table5also show that the values of recovery compliance J r are lower than the corresponding ones of J c,which indicates that the samples do not recover to their original structure completely after the removal of the applied stress.The poor recovery can be attributed to the structure breakdown and build-up of the samples.The recovery behaviours of the pastes can be used to study the slump characteristic,and it can be defined as the ratio of (J c-J r)/J c.The bar chart of recovery(%)for all the four pastes is presented in Fig.10.The highest relative recovery is shown by paste SM4with32.1%,followed by paste SM1with8.5%and paste SM3with6.1%.Paste SM2 exhibits the lowest recovery ratio of3.5%.The good recovery and the lowest compliances indicate that paste SM4has obvious elastic properties,which will be helpful to reduce the clogging and to prevent from oo-zing out through the valleys among pyramids on the sub-strate during the printing process.This should be one reason for the grid made from paste SM4to have the smallest width and the highest aspect ratio.On the con-trary,paste SM2with the poor recovery and the highest compliances gives the grid the largest width and the lowest aspect ratio.And at the same time,it was found that the screen mesh openings were clogged more easily by paste SM2,thus leading to the appearance of breakpoints and virtual prints.Such similar results have been found for solder pastes by others.Bao et al.[23]found that solder pastes with lower compliances(J c and J r)and highTable4Results from the oscillatory amplitude sweep testsSample G0(Pa)G00(Pa)G0/G00G0=G00(Pa)LVR range(Pa)dSM1404,142123,543 3.3245–95–69 SM229,77519,650 1.530–230–79 SM3125,25078,000 1.6185–827–78SM41,120,500303,000 3.7275–165–87Fig.9Measured creep compliance J c and recovery compliance J r for the four kinds of Ag pastesTable5Creep compliance J c,recovery compliance J r and recovery ratio for the four kinds of pastesSample J c(Pa-1)J r(Pa-1)(J c-J r)(Pa-1)(J c-J r)/J c (%)SM17.98910-57.30910-5 6.80910-68.5 SM2 1.15910-2 1.11910-2 4.00910-4 3.5 SM3 6.36910-4 5.97910-4 3.90910-5 6.1 SM4 3.83910-6 2.60910-6 1.23910-632.1recovery were preferred to minimize print defects,e.g. smear and insufficiency.Duraira et al.[24]found that the pastes with lower compliances would have fewer tenden-cies to bleed out underneath the stencil during printing process and therefore reduce the tendency for the paste to be smeared.So the creep and recovery test shows that to make excellent grid by screen printing,the low-tempera-ture silver paste should have low compliances(J c and J r). 4ConclusionBy comparatively investigating the rheological properties and the corresponding screen-printing performance of four different low-temperature silver pastes with viscosity test, thixotropy test,oscillatory stress sweep test and creep-recovery test,it is found that a good quality paste for printing the grid withfine grids and high aspect ratio should have an optimum shear thinning property to allow it flow through the printing mesh,and a suitable thixotropic nature for leveling but without slumping after printing, Further,an obvious viscoelastic behavior can prevent the paste bleeding between the screen and the substrate.A high recovery ratio and a low compliance can also avoid the paste bleeding and smearing during the printing process. Such correlationship between the screen-printing perfor-mance and the rheological properties of the low-tempera-ture silver paste can be referred to improve the electrode grid fabrication for silicon heterojunction solar cells.The comparative study presents clearly that the rheo-logical properties of the low-temperature silver pastes are related to their components.Poor thixotropic and visco-elastic properties of paste SM2show that metal particles with large size can inhibit the recovery of the paste after printing.Nano-scale metal particles are helpful to improve the paste rheological properties.Further,although paste SM4has the highest metal concentration,it presents the best printability,which indicates that the organic solvent plays another important role on the paste rheological properties.In other words,appropriate organic solvents can give the paste good rheological properties for printing,and thus allow the paste to adopt high metal concentration, which will be very beneficial to enhance the conductance of the printed grids.Acknowledgments This work was supported in part by the863high technology research program of China under Grant2011AA050502 and the National Natural Science Foundation of China under Grant 61274061.References1.M.Tanaka,M.Taguchi,T.Matsuyama,T.Sawada,S.Tsuda,S.Nakano,H.Hanafusa,Y.Kuwano,Jpn.J.Appl.Phys.31,3518 (1992)2.R.M.Swanson,in Proceedings of the31th Photovoltaic Spe-cialists Conference(PVSC)(Lake Buena Vista,USA,2005), p.8893.M.Taguchi, A.Yano,S.Tohoda,K.Matsuyama,IEEE J.Photovolt.4,96(2014)4.H.Jiang,W.Shi,A.Shaikh,E.Graddy,in Proceedings of the37th IEEE Photovoltaic Specialists Conference(PVSC)(Wash-ington State Convention Center,2011),p.14565.Y.Veschetti,J.C.Muller,J.Damon-Lacoste,P.R.i Cabarrocas,A.S.Gudovskikh,J.P.Kleider,P.J.Ribeyron, E.Rolland,inProceedings of the31th Photovoltaic Specialists Conference (PVSC)(Lake Buena Vista,USA,2005),p.11316.D.P.Chen,L.Zhao,H.W.Diao,W.B.Zhang,G.Wang,W.J.Wang,J.Mater.Sci.Mater.Electron.25,2657(2014)7.J.Hoornstra,A.Weeber,H.de Moor,W.Sinke,in Proceedingsof the14th European Photovoltaic Solar Energy Conference(EU PVSEC)(Barcelona,Spain,1997),p.8238.M.Neidert,W.Zhang,D.Zhang,A.Kipka,in Proceedings of the33th IEEE Photovoltaic Specialists Conference(PVSC)(San Diego,California,2008),p.19.G.J.Jackson,R.Durairaj,N.N.Ekere,in Proceedings of27thIEEE/SEMI International Electronic Manufacturing Technology Symposium(the Fairmont Hotel,San Jose,2002),p.22310.H.A.Barnes,J.Non-Newtonian Fluid Mech.70,1(1995)11.R.Faddoul,N.Reverdy-Bruas,J.Bourel,J.Mater.Sci.Mater.Electron.23,1415(2012)12.S.Burnside,S.Winkel,K.Brooks,V.Shklover,et al.J.Mater.Sci.Mater.Electron.11,355(2000)13.R.Faddoul,N.Reverdy-Bruas,A.Blayo,Mater.Sci.Eng.B177,1053(2012)14.M.Pospischil,J.Specht,M.Konig,M.Horteis,C.Mohr,F.Clement,D.Biro,IEEE J.Photovolt.4,498(2014)15.S.Mallik,J.Thieme,R.Bauer,N.N.Ekere,A.Seman,R.Bhatti,R.Durairaj,in Proceedings of the11th Electronics Packaging Technology Conference(Tang Cheng Hotel,Xi’an,China,2009), p.86916.T.A.Nguty,N.N.Ekere,A.Adebayo,in Proceedings of the24thIEEE/CPMT International Electronics Manufacturing Technol-ogy Symposium(Austin,Texas,USA,1999),p.30417.S.Mallik,Doctoral dissertation,University of Greenwich,200918.R.Durairaj,S.Ramesh,S.Mallik,A.Seman,N.Ekere,Mater.Des.10,3812(2009)19.C.N.Chen,C.T.Huang,W.J.Tseng,M.H.Wei,Appl.Surf.Sci.257,650(2010) Fig.10Recovery behaviours of the four kinds of Ag pastes。