Sintering Kinetics of Inkjet-Printed Conductive SilverLines on Insulating Plastic Substrate

WENCHAO ZHOU, FREDERICK A. LIST, CHAD E. DUTY,and SUDARSANAM S. BABU

This paper focuses on sintering kinetics of inkjet-printed lines containing silver nanoparticlesdeposited on a plastic substrate. Upon heat treatment, the change of resistance in the printedlines was measured as a function of time and sintering temperatures from 423 K to 473 K(150 �C to 200 �C). A new phenomenon was observed that a critical temperature existed for thesintering process, beyond which there was no further reduction in resistance. Experimentalevidence and analysis show the critical temperature is associated with the boiling point of thesolvent. New sintering mechanisms have been proposed to explain the observed phenomenon,including accelerated diffusion facilitated by the existence of liquid solution based on the theoryof liquid phase sintering, and particle collision and coalescence caused by the induced liquidflows in the solution. The proposed theory suggest new means can be devised to improve thesintering results for inkjet-printed lines and other applications.

DOI: 10.1007/s11663-014-0288-4� The Minerals, Metals & Materials Society and ASM International 2015

I. INTRODUCTION

EVER since its invention in 1936,[1] lithographytechnology has been widely adopted in the printingcircuit board (PCB) industry for the fabrication ofconductive lines. To meet the requirements of flexibleand embedded electronics such as smaller feature size,and reduced weight, volume, and environmental impact,many manufacturing processes have been developed,including curved layer fused deposition modeling(CLFDM),[2] laser-based direct write (LDW),[3] lasermicro-cladding electronic pastes (LMCEP),[4] screenprinting,[5] offset lithography,[6] aerosol jet printing,[7]

and inkjet deposition.[8–13] Inkjet deposition stands outfor its low cost, high resolution, wide choice ofconductive materials and substrates, no requirement ofphysical mask, and digital control of ejection anddeposition. Various conductive inks have been studiedand successfully printed such as molten metal,[14–16]

conductive polymers,[17,18] organometallic com-pounds,[19,20] metal precursors,[21] and metallic nano-particle suspensions.[10–12] The focus of our ongoingresearch is to evaluate the feasibility of integrating inkjetprinting with other additive manufacturing processes,such as fused deposition modeling (FDM). The moti-

vation is to develop 3D hybrid structures[22] that arecapable of supporting electrical circuits for functionalapplications envisioned by automotive and aerospaceindustries. In this regard, the use of a metallic nano-particle suspension has a greater potential[23,24] becauseit can significantly reduce the high processing temper-atures required for standard lithographic fabrication.Nanoparticle inks can typically be processed at temper-atures below 573 K (300 �C), which is compatible withmost of the thermoplastic materials envisioned byFDM. However, we have identified the followingchallenges: (i) printed line instabilities and (ii) rate ofdrying and sintering processes (see Figure 1).Figure 1 illustrates the types of printed line instabil-

ities generally encountered when depositing low-viscos-ity ink on a rough and porous surface typical of FDM.Thermoplastic materials used for FDM have low surfaceenergy and interconnected porosity that allows thedeposited ink to penetrate several layers deep into theFDM structure (Figure 1(b)), potentially causing ashort with neighboring circuits. Even if the surfaceporosity is eliminated, the oscillating surface of adjacentdeposited beads draws ink to the low points, generatinglocal discontinuities (Figure 1(c)). Two surface treatingtechniques have been developed to address this chal-lenge.[25] In the first technique, surface ironing was usedto seal and smoothen the surfaces. The second methodused a heated needle to ‘‘plow’’ a predefined pattern ofchannels in the substrate that were designed to containthe deposited inks. Each of these techniques providedconsistent geometry of printed lines.The second challenge is obtaining conductivity from

the printed lines due to both the particle segregationduring the droplet drying process and the lack ofparticle sintering. There is extensive literature withreference to the drying process that leads to segregation

WENCHAO ZHOU, Assistant Professor, is with the Departmentof Mechanical Engineering, University of Arkansas, Fayetteville.Contact e-mail: zhouw@uark.edu FREDERICK A. LIST, SeniorStaff, is with the Manfacturing Demonstration Facility, Oak RidgeNational Laboratory, Oak Ridge, TN 37831. CHAD E. DUTY,Group Leader, and SUDARSANAM S. BABU, Professor andGovernoros Chair on Advanced Manufacturing, are with theDepartment of Mechanical, Aerospace and Biomedical Engineering,The University of Tennessee, and also with the ManfacturingDemonstration Facility, Oak Ridge National Laboratory.

Manuscript submitted December 10, 2014.Article published online January 24, 2015.

1542—VOLUME 46B, JUNE 2015 METALLURGICAL AND MATERIALS TRANSACTIONS B

of nanoparticles in the fluid, often referred to as the‘‘coffee ring’’ effect.[26] Kim et al.[27] showed anapproach for eliminating the ‘‘coffee ring’’ effect byadding a liquid component with higher boiling point butlower surface tension to induce a reverse Marangoniflow. Layani et al. reported a method that utilizes the‘‘coffee ring’’ effect to obtain transparent conductivepatterns using inkjet printing. The sintering process isequally important for achieving conductivity. Metal inkstypically consist of nanoparticles capped with organicstabilizer to prevent the agglomeration of the nanopar-ticles, which can be removed upon heating. Manystudies have also been performed on the sinteringprocess. Greer et al.[28] demonstrated highly conductivefilms can be made at low curing temperatures by thermalannealing of silver nanoparticles. Wakuda et al.[29]

demonstrated a new method for sintering silver nano-particles at room temperature in an air atmosphere byremoving the dispersant from the nanoparticles. Magd-assi and coworkers developed a process to trigger thesintering process at room temperature by modificationof the surface properties of the nanoparticles throughcharge neutralization and desorption of stabilizer.[30,31]

Olkkonen et al.[32] demonstrated an approach forsintering inkjet-printed conductive lines using boiling

salt water. Long et al. developed a rapid sinteringmethod for silver nanoparticles by immersing them inelectrolyte solutions.[33] However, there has been verylimited research on the sintering kinetics of inkjet-printed particles on polymer substrates as a function oftemperature and time. Such a scientific understandingfor synchronizing the curing process during FDM withthat of inkjet sintering is required to successfullycombine a hybrid additively manufactured structurewith embedded electrical circuits. Therefore, the goal ofthis research focused on understanding on the role of thesintering kinetics of inkjet-printed silver nanoparticleson polymer substrates.

II. EXPERIMENTS AND DISCUSSION

A Fuji Dimatix Materials Printer was used to printconductive paths on an electrically insulating substrateusing the Cabot Conductive Ink CCI-300. In order toeliminate the influence of the irregularity of the FDMsubstrate surface, commercial smooth Ultem PEI plasticfilm with a thickness of 127 lm was used as substrate.The composition and corresponding properties of theCabot ink is listed in Table I.[34] Conductive lines wereprinted at the same conditions with a length of 2 cm,

Fig. 1—Challenges faced during inkjet printing of silver nanoparticles on polymeric substrates made by FDM additive manufacturing process.(a) Schematic illustration of surface textures; (b) permeation of nanoink through layers leading to discontinuities; (c) instability of fluid leadingto breakage of lines; (d) appearance of printed line after sintering showing lack of sintering in certain regions.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 46B, JUNE 2015—1543

droplet spacing of 10 lm, and a width of 100 lm. Theprinted lines were then put on a hot plate setting atdifferent temperatures for sintering and the resistancesof them were measured at different time during thesintering process.

The measured resistance of the printed lines ispresented in Figure 2(a) as a function of time at differenttemperatures. The data shows the stochastic nature ofresistance variation at a given time for a given temper-ature due to the difficulty of maintaining consistency ofthe printing process. A best-fit line through the datareveals some surprising features of measured resistance.At 423 K (150 �C), the resistance decreases from42 Ohm as a function of time slowly, and reaches aplateau at ~24 Ohms after 3600 seconds. The resistancedrops further with an increase in temperature from423 K to 453 K (150 �C to 180 �C). A plateau of~13 Ohms was reached at 453 K (180 �C). Surprisingly,the curves of resistance change over time overlap with

each other at the sintering temperatures of 453 K and473 K (180 �C and 200 �C). This is the first time in theliterature that the existence of such a critical tempera-ture is noted for the sintering of silver nanoparticles ininkjet-printed lines. To explain this phenomenon, thesame data was put in the form of Arrhenius plot shownin Figure 2(b) from the lowest to the highest time. Theplots show that the resistance reduction from 423 K to453 K (150 �C to 180 �C), conforms to a processcontrolled by thermal activation. The activation energydecreases over time, which can be explained by theremoval of the ink stabilizer upon heating and thus theincrease of the contact area between nanoparticles.However, the trend stops at 473 K (200 �C). Wehypothesized that the trend change of resistance couldbe attributed either to changes in cross sectional area ofthe printed lines or to a change in resistivity due to thesintering mechanism.In order to test the sensitivity of printed line cross

sectional area with resistance, lines were printed withdifferent cross sections by modifying the line width (100and 500 lm) and droplet spacing (0 to 40 lm). Theresistivity of these lines was measured after sintering at473 K (200 �C) for 10 minutes. The average area of thethree cross sections (the widest, narrowest, and medium)for each printed line was calculated. As expected, thecross section area decreased with an increase in dropletspacing (Figure 3(a)). By measuring the resistance of theprinted lines, the resistivity of the printed lines can becalculated with their cross section area and length. Theresistivity data (Figure 3(b)) shows poor correlationwith the cross-section area. For example, the resistivityof printed lines with cross-sectional areas around 10�4

to 10�3 mm2 made by two different widths show nosignificant variations.The above data indicate that the observed phenom-

enon of reduction in resistance in Figure 2 should berelated to the sintering mechanism of nanoparticles. Inthe literature, sintering of nanoparticles in an inkjet-printed line is considered to be driven by solid-statediffusion along the surfaces. However, this mechanismcannot explain the shutdown of the process at 473 K(200 �C). The coincidence of the critical temperature[~473 K (200 �C)] and the boiling point of the solvent[ethylene glycol: 470.45 K (197.3 �C) shown in Table I]inspired us to hypothesize that the liquid solvent mightplay an important role in the sintering process of thesilver nanoparticles. This hypothesis is also supportedby the previously developed sintering methods usingboiling salt water[32] and electrolyte solutions.[33] There-fore, we proposed a transition for the sintering mech-anism of nanoparticles as illustrated in Figure 4 toexplain the observed phenomenon of the critical tem-

Table I. Chemical Composition and Corresponding Properties of the Cabot Ink CCI-300

Chemical Name Ethanol Ethylene Glycol Silver

Weight percent 10 to 70 10 to 40 19 to 21Boiling point [351.52 K (78.37 �C)] [470.45 (197.3 �C)] —Viscosity [298 K (25 �C)] 1.26 mPa s 14 mPa s —Surface tension [298 K (25 �C)] 21.97 mN/m 47.3 mN/m —

Fig. 2—(a) Measured resistance of printed lines as a function of sin-tering temperature and time. (b) Analyses of the data in the form ofArrhenius plot. Each column of points represents the change ofresistance overtime at the specified temperature.

1544—VOLUME 46B, JUNE 2015 METALLURGICAL AND MATERIALS TRANSACTIONS B

perature. Since the conductivity of electrons requiremetallurgical linkages between nanoparticles, it isimportant to consider the details of the sinteringreaction before and after evaporation of the liquid inthe ink. In this regard, the sintering process is verysimilar to oxide inclusion coarsening observed in liquidsteel making or steel weld pools.[35]

In our experiments, the ink is made up of silvernanoparticles dispersed in a mixture of ethanoland ethylene glycol. During the initial condition(Figure 4(a)), the nanoparticles may contact with eachother stochastically depending on the printing condi-tions. Based on the theory of liquid phase sintering,[36] ifthe liquid is motionless, the sintering of silver particles ismainly driven by accelerated diffusion in the liquid, suchas by solution precipitation or by Ostwald ripeningprocess. However, this should not be the only role theliquid component plays in the sintering process. Previ-ous research has shown that strong flows may be presentduring the drying and sintering process of the printedlines,[26,27] such as Marangoni flow. Inspired by thesimilarity between the sintering process and the oxideinclusion coarsening observed in liquid steel making orsteel weld pools,[35] we realize the motion of thenanoparticles caused by the liquid flow can significantly

increase their chance of contact with each other andfacilitate the sintering process by collision and coales-cence.[35] During the droplet drying process, severalflows can be induced, including the evaporation flow,the replenishing flow caused by the difference of theevaporation rates between the edge and the center,[26]

the buoyant flow, and the Marangoni flow caused by thesurface tension gradient.[27] All these flows are influ-enced by temperature. The rate of change in sizedistribution [dn(r)/dt] of the particles in the liquid flowcan be described by the following equation.

dnðrÞdt

¼

0:5

Zr

0

n r1ð Þn r2ð Þ 43

rþ rð Þ3 r

r2

� �2

gradV dr1

!0@

1A�

Zr

0

n r1ð Þn r2ð Þ 43

rþ rð Þ3 r

r2

� �2

gradV dr2

!0@

1A

266666666664

377777777775;

½1�

where gradV is the velocity gradient, and n(r1) and n(r2)are the number of particles per unit volume size r1 andr2. The above equation demonstrates that a rapidincrease in particle size requires large velocity gradients.Since all the induced flows are influenced by tempera-ture, temperature plays a significant role in determiningthe gradV term.When the temperature is low, the evaporate rate is

low and the velocities of the evaporation flow, thebuoyant flow, and the induced replenishing flow are alsolow. Marangoni flow is also reduced because of thesmall surface tension gradient at low temperature. Inthis case, the sintering process is primarily facilitated bydiffusion of silver atoms in liquid as illustrated inFigure 4(a). As the temperature increases, the evapora-tion rate increases, which causes all the flow velocities toincrease and enables collision and coalescence of silvernanoparticles. At temperatures between 423 K and453 K (150 �C and 180 �C), large velocity gradientsfacilitate the sintering process primarily by collision andcoalescence. As the liquid evaporates, collisions occurless often and the resistance drop becomes less sensitiveto temperature, which explains the decrease of theactivation energy (i.e., Q/R decreases from 2588.6 to1550.8) during the later stage of the sintering process asshown in Figure 2(b). When the temperature increasesabove 470 K (197 �C), the liquid is completely evapo-rated and the rate of increase in silver particle size bycollision and coalescence is reduced to zero. Therefore,any further increase in sintering has to be supported bythe diffusion of silver along the nanoparticle grainboundary surface, similar to the classical sinteringphenomenon[37–39] as shown in Figure 4(c). Althoughmore elaborated experiments with more sophisticatedequipment should be conducted to fully validate theproposed sintering mechanisms, the experimental evi-dence and analysis presented in this paper nevertheless

Fig. 3—(a) Measured cross-sectional area of printed silver lines onUltem substrate as a function of droplet spacing used during inkjetprinting process; (b) measured resistivity as a function of cross-sec-tional area shows poor correlation.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 46B, JUNE 2015—1545

strongly support a connection between the observedcritical temperature and the boiling point of the solvent.As already demonstrated to be useful for the sinteringprocess using boiling salt water[32] and electrolytesolutions,[33] the role of the liquid in the sinteringprocess should be further explored in order to developnew sintering methods for envisioned applications.

III. CONCLUSIONS

In conclusion, the current research suggests twomechanisms for accelerating the sintering kinetics ofsilver nanoparticles and improving the conductivity forlow temperature sintering in inkjet printing, including (1)promoting diffusion of silver atoms in organic solvents asin conventional liquid phase sintering, and (2) imposingor inducing a liquid flow with large velocity gradients.The results presented in this paper suggest new meanscan be devised to improve the sintering results for inkjet-printed lines and other similar applications.

ACKNOWLEDGMENTS

This work is sponsored by the U.S. Department ofEnergy, Office of Energy Efficiency and Renewable

Energy, Advanced Manufacturing Office, under con-tract DE-AC05-00OR22725 with UT-Battelle, LLC.

REFERENCES1. K. Petherbridge, P. Evans, and D. Harrison: Circuit World, 2005,

vol. 31 (1), pp. 41–45.2. O. Diegel, S. Singamneni, B. Huang, and I. Gibson: Key Eng.

Mater., 2011, vols. 467–469, pp. 662–67.3. A. Pique, S.A. Mathews, B. Pratap, R.C.Y. Auyeung, B.J. Karns,

and S. Lakeou: Microelectron. Eng., 2006, vol. 83 (11–12),pp. 2527–33.

4. Z. Xiaoyan, L. Xiangyou, L. Jingwei, and Q. Xiaojing: IEEETrans. Adv. Packag., 2006, vol. 29 (2), pp. 291–94.

5. M.J. Er, ed: New Trends in Technologies: Devices, Computer,Communication and Industrial Systems. Sciyo. 2010, p. 444.

6. P.M. Harrey, P.S.A. Evans, B.J. Ramsey, and D. Harrison: inEnvironmentally Conscious Design and Inverse Manufacturing,1999. Proceedings. First International Symposium on EcoDesign‘99. 1999.

7. C. Werner, D. Godlinski, V. Zollmer, and M. Busse: J. Mater.Sci.: Mater. Electron., 2013, vol. 24 (11), pp. 4367–77.

8. J.B. Szczech, C.M. Megaridis, D.R. Gamota, and Z. Jie: IEEETrans. Electron. Packag. Manuf., 2002, vol. 25 (1), pp. 26–33.

9. K. Cheng, M.-H. Yang, W.W.W. Chiu, C.-Y. Huang, J. Chang,T.-F. Ying, and Y. Yang: Macromol. Rapid Commun., 2005,vol. 26 (4), pp. 247–64.

10. H.-H. Lee, K.-S. Chou, and K.-C. Huang: Nanotechnology, 2005,vol. 16, pp. 2436–41.

Fig. 4—Schematic Illustration of possible mechanisms for the sintering phenomenon during inkjet printing. (a) Initial distribution of the silvernanoparticles dispersed in the stagnant liquid. Coarsening of silver particles may not happen at low temperature due to the requirement foratomic/ionic diffusion of silver around the particles or in liquid, respectively. (b) With an increase in temperature above 353 K (80 �C), the evap-oration of alcohol may induce fluid flow and drive the particles together leading to coarsening by collision and coalescence. Silver atoms/ionsmay diffuse along the surface. (c) With the full evaporation of liquid media (here ethylene glycol), the sintering can only occur on the particlesurface that have made contact already. Lack of liquid media will reduce or shut down the sintering reaction due to limited solid-state diffusion.

1546—VOLUME 46B, JUNE 2015 METALLURGICAL AND MATERIALS TRANSACTIONS B

11. B.K. Park, D. Kim, S. Jeong, J. Moon, and J.S. Kim: Thin SolidFilms, 2007, vol. 515 (19), pp. 7706–11.

12. T.H.J. van Osch, J. Perelaer, A.W.M. de Laat, and U.S. Schubert:Adv. Mater., 2008, vol. 20 (2), pp. 343–45.

13. K. Woo, D. Kim, J.S. Kim, S. Lim, and J. Moon: Langmuir, 2008,vol. 25 (1), pp. 429–33.

14. F Gao and AA Sonin: Proc. Math. Phys. Sci., 1922, vol. 1994(444), pp. 533–54.

15. M. Orme and R.F. Smith: J. Manuf. Sci. Eng., 2000, vol. 122 (3),pp. 484–93.

16. Q. Liu and M. Orme: J. Mater. Process. Technol., 2001, vol. 115(3), pp. 271–83.

17. H. Sirringhaus, T. Kawase, R.H. Friend, T. Shimoda, M.Inbasekaran, W. Wu, and E.P. Woo: Science, 2000, vol. 290(5499), pp. 2123–26.

18. B. Chen, T. Cui, Y. Liu, and K. Varahramyan: Solid-State Elec-tron., 2003, vol. 47 (5), pp. 841–47.

19. T. Cuk, S.M. Troian, C.M. Hong, and S. Wagner: Appl. Phys.Lett., 2000, vol. 77 (13), pp. 2063–65.

20. G.G. Rozenberg, E. Bresler, S.P. Speakman, C. Jeynes, and J.H.G.Steinke: Appl. Phys. Lett., 2002, vol. 81 (27), p. 5249.

21. Z. Liu, Y. Su, and K. Varahramyan: Thin Solid Films, 2005,vol. 478 (1–2), pp. 275–79.

22. M. Ashby and Y. Brechet: Acta Mater., 2003, vol. 51 (19), pp. 5801–21.

23. S.B. Fuller, E.J. Wilhelm, and J.M. Jacobson: J. Microelectro-mech. Syst., 2002, vol. 11 (1), pp. 54–60.

24. J. Perelaer, B.J. de Gans, and U.S. Schubert: Adv. Mater., 2006,vol. 18 (16), pp. 2101–04.

25. W. Zhou, F.A. List, C.E. Duty, and S.S. Babu: Rapid Prototyp. J.2014. In press.

26. R.D. Deegan, O. Bakajin, T.F. Dupont, G. Huber, S.R. Nagel,and T.A. Witten: Nature, 1997, vol. 389 (6653), pp. 27–829.

27. D. Kim, S. Jeong, B.K. Park, and J. Moon: Appl. Phys. Lett.,2006, vol. 89 (26), p. 264101.

28. J.R. Greer and R.A. Street: Acta Mater., 2007, vol. 55 (18),pp. 6345–49.

29. D. Wakuda, M. Hatamura, and K. Suganuma: Chem. Phys. Lett.,2007, vol. 441 (4), pp. 305–08.

30. M. Grouchko, A. Kamyshny, C.F. Mihailescu, D.F. Anghel, andS. Magdassi: ACS Nano, 2011, vol. 5 (4), pp. 3354–59.

31. S. Magdassi, M. Grouchko, O. Berezin, and A. Kamyshny: ACSNano, 2010, vol. 4 (4), pp. 1943–48.

32. J. Olkkonen, J. Leppaniemi, T. Mattila, and K. Eiroma: J. Mater.Chem. C, 2014, vol. 2 (18), pp. 3577–82.

33. Y. Long, J. Wu, H. Wang, X. Zhang, N. Zhao, and J. Xu: J.Mater. Chem., 2011, vol. 21 (13), pp. 4875–81.

34. L. Basirico: Ph.D. Thesis, University of Cagliari, 2012.35. S. Babu, S. David, and T. DebRoy: Sci. Technol. Weld. Join.,

1996, vol. 1 (1), pp. 17–27.36. R.M. German, P. Suri, and S.J. Park: J. Mater. Sci., 2009, vol. 44

(1), pp. 1–39.37. D.L. Johnson and T. Glarke: Acta Metall., 1964, vol. 12 (10),

pp. 1173–79.38. R. Eadie, G. Weatherly, and K. Aust: Acta Metall., 1978, vol. 26

(5), pp. 759–67.39. W. Lee, R. Eadie, G. Weatherly, and K. Aust: Acta Metall., 1978,

vol. 26 (12), pp. 1837–43.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 46B, JUNE 2015—1547