IEEE TRANSACTIONS ON MAGNETICS, VOL. 46, NO. 6, JUNE 2010
2187
Novel Solder-Magnetic Particle Composites and Their Reflow Using AC Magnetic Fields Ashfaque Hussain Habib1 , Matthew G. Ondeck1 , Kelsey J. Miller1 , Raja Swaminathan2 , and Michael E. McHenry1 Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, PA 15213 USA Intel Corporation, Chandler, AZ 85226 USA Localized heating of solder interconnects in electronic packaging can help alleviate undesirable thermal stresses during the conventional reflow process, where the package is subjected to high temperatures. Localized heating is possible for conductors by electromagnetic induction of eddy currents which results in Joule heating, when subjected to AC magnetic fields. However, for typical solder pastes with fine solder powders dispersed in a flux medium, the eddy current losses, which have strong size dependence, become too small to cause significant heating at reasonable field amplitudes and frequencies. Magnetic materials exhibit losses from hysteretic and relaxation processes, in AC magnetic fields. We investigated the feasibility of solder paste reflow by localized heating of novel solder magnetic particle composites in AC magnetic fields. A solder paste magnetic nanoparticle (MNP) composite was prepared by mechanical mixing of FeCo MNPs with Type III Sn96.5/Ag3.0/Cu0.5 (SAC305) Pb-free solder paste. The pristine solder paste show an insignificant temperature rise when subjected to AC magnetic field of 500 Oe at 280 kHz, whereas the solder-MNP composite samples with particle concentration of 2 wt% or more were able to reflow due to magnetic heating. Here we report first demonstration of a new approach for solder melting in AC magnetic fields using MNPs. Index Terms—Electromagnetic induction heating, infrared imaging, magnetic nanoparticles, solder reflow.
I. INTRODUCTION ONVENTIONAL methods to make electrical interconnects for electronic packaging depend on multizone hot-air convection or infrared ovens to cause solder paste reflow. In an oven, the printed circuit board (PCB) and components are subjected to high temperatures required for the solder reflow for an extended period of time during the reflow process. Although such a process has the benefits of high throughput in high volume manufacturing, it suffers form some drawbacks due to the exposure of the entire package to the high reflow temperatures, especially warpage induced due to thermal mismatches between different components of the package. Recently, the electronics industry has widely adopted Pb-free solders in view of heath and environmental concerns of Pb toxicity. Pb-free solder alloys, like SAC305 (3wt%Ag; 0.5wt%Cu, remaining Sn), typically have a higher reflow temperature due to their higher melting point and longer reflow window for processing due to poor wetting characteristics compared to Sn-Pb eutectic solders. Consequently, the aforementioned concerns as to conventional reflow techniques are aggravated. Higher reflow temperatures not only increase the PCB warpage, but also increase the risk of delaminating and blistering of the PCB laminates. Localized solder heating at interconnects can help alleviate challenges necessitated by using Pb-free solders. Current localized heating methods include laser reflow [1] and solder jetting [2]. However, these methods have high equipment cost and limited applicability for board level processing of the solder joints that are hidden between the component and the PCB.
C
Eddy current induced Joule heating that occurs in conductors subjected to AC magnetic field is an effective method to achieve localized heating in solders. Li et al. [3] have recently investigated such an approach for reflow of solder balls. They reported the feasibility of achieving reflow temperatures within diameter in short times when exposed solder balls of 760 to AC magnetic fields using eddy current heating. Since induced eddy currents strength strongly depends on sample dimensions, eddy current power losses diminish drastically for smaller solder balls. For solder pastes which consist of solder or less dispersed in a powders with particle diameter of 50 flux medium, these losses are negligible and do not result in significant heating. The inability to effectively reflow solder paste limits the applicability of this technique. Magnetic materials subjected to an AC magnetic field can also dissipate magnetic energy via hysteretic and relaxation processes [4], [5]. Unlike eddy current losses these can result in significant power loss in submicron and nanosized magnetic materials [6]. We investigate incorporation of magnetic nanoparticles (MNPs) in solder pastes to form a composite that can be melted locally by AC magnetic fields to cause solder paste reflow. Magnetic heating will enable solder reflow in AC magnetic fields of moderate amplitude and frequency which would help to minimize any unnecessary eddy current based heating in the package. In order to ensure good electrical and mechanical properties of solder joints, it is desirable to limit the magnetic particle concentration. In view of this consideration, choice of highly magnetic permeability FeCo MNPs is ideal. Here, we present the feasibility of reflow of novel solder-MNP composite paste by localized heating in AC magnetic fields. II. EXPERIMENTAL
Manuscript received October 31, 2009; accepted February 21, 2010. Current version published May 19, 2010. Corresponding author: A. H. Habib (e-mail:
[email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TMAG.2010.2044640
Solder-MNP composite paste were made by mechanical mixing of FeCo MNPs into Type III Sn96.5/Ag3.0/Cu0.5 (SAC305) Pb-free solder paste (EFD, Inc.) comprised of 25 solder powders dispersed in RMA (rosin mildly to 45 activated) solder flux. The MNPs were thoroughly mixed with
0018-9464/$26.00 © 2010 IEEE
2188
IEEE TRANSACTIONS ON MAGNETICS, VOL. 46, NO. 6, JUNE 2010
Fig. 2. RF heating setup along with the infrared thermal imaging camera to record temperature profile during heating experiments.
used to obtain thermal maps to determine spatial and temporal temperature profile of the samples during the experiments. Fig. 1. TEM micrograph of plasma torch synthesized FeCo MNPs used for fabrication of solder-MNP composites, inset shows the particle size distribution of the MNPs.
solder paste by mechanical stirring them together with the help of a stirrer for a sufficiently long time to break up any agglomerates in the MNPs and ensure homogeneous dispersion of particles in the paste. Solder-MNP composite materials with varying nanoparticle concentration (2, 3, 4, 5, and 10 wt%) of FeCo MNPs dispersed in SAC305 paste were prepared for the study.
C. SEM Characterization Solder-MNP composite paste with 4 wt% of particle loading was placed between two Ni-Au surface finish test coupons and melted by induction heating. The reflowed sample was crosssectioned and polished to prepare SEM specimen. Solder/Ni-Au finish interface and bulk solder matrix was examined under a FEI XL-30 scanning electron microscope (SEM) equipped with energy dispersive X-ray (EDX) detector to ascertain the effect of MNPs on the wettability, interfacial and bulk properties of the solder material. III. RESULTS AND DISCUSSION
A. FeCo MNPs Synthesis FeCo nanoparticles were synthesized from pre-alloyed FeCo via thermal plasma synthesis [7]. precursor powders A RF (radio frequency) plasma torch (Tekna Plasma Systems Inc.) was operated with a 50 kW/4 MHz power supply (Lepel High Frequency Laboratories, Inc.) to generate Ar plasma. Precursor powders were fed at a feed rate of 2 g/min into the plasma to obtain MNPs with particle size ranging from 10 to 100 nm and mean size of 32 nm. Fig. 1 shows transmission electron microscope (TEM) micrograph of the FeCo MNPs, inset shows their particle size distribution. B. RF Heating and Thermal Imaging Setup Fig. 2 shows the experimental setup used to measure the heating response of the solder-MNP composite material in an AC magnetic field. The setup includes a 2.4 kW RF power generator (Ameritherm, Inc.) used to deliver energy to a series-resonant circuit with a water-cooled capacitor bank and a three-turn induction coil. The unit operated at 280 kHz with a 300 A current resulting in a magnetic field strength of 500 Oe in the induction coil. Optical thermometry employing an infrared (IR) thermal imaging camera (Process Sensor Corporation) was
A. Power Loss Theory in MNPs Eddy current losses per unit volume for conducting particles placed in AC magnetic fields are given by (1) where is the resistivity, is the frequency, is the amplitude of magnetic field, and is particle diameter. Due to small size of MNPs, eddy current losses can be ignored when accounting for their heating in AC magnetic fields. Further, MNPs become superparamagnetic (SPM) below a certain critical size, which do not exhibit any DC hysteresis response and associated hysteretic losses when subjected to cyclic magnetic fields. In such MNPs, the dominant mechanism of power loss when subjected to AC magnetic fields is via relaxation processes [4]. Thermal relaxation of magnetic moment for immobilized particles (i.e., lack of Brownian motion) is governed by Néel rotation of magnetization vector against their magnetic anisotropy energy. Néel relaxation time is given by (2)
HABIB et al.: NOVEL SOLDER-MAGNETIC PARTICLE COMPOSITES AND THEIR REFLOW USING AC MAGNETIC FIELDS
2189
where is time constant, is the Boltzmann constant, is the absolute temperature, is the is the magnetic particle magnetic anisotropy constant, and volume. Magnetic power loss per unit volume via relaxation process in AC magnetic field of amplitude and frequency is (3) is the permeability of free space where is the imaginary part of the complex magnetic susceptiand bility, , that can be expressed as
(4) is the static susceptibility, which for small field amwhere plitudes can be approximated by a Langevin relationship (5) where the Langevin parameter and are given by
the initial susceptibility
(6)
Fig. 3. Thermal images showing the temperature profile in (a) solder balls and (b) solder paste at various times.
where is the saturation magnetization of MNPs. Peak power loss for a given frequency of AC magnetic field is dictated by an optimum relaxation time which is dependent on particle size. Therefore, it is desirable to have a narrow particle size distribution with appropriate relaxation time for best results. Power loss in MNPs scales with saturation magnetization of the particles. As Fe-Co alloy system has the highest saturation magnetization [8], [9], FeCo MNPs were selected for solder-MNP composite paste fabrication. Reflow of FeCo MNPs-based solder-MNP composite paste in AC magnetic fields can be achieved in short times with minimum particle concentration. Alternatively, it is possible to achieve significant power loss to cause solder reflow even with polydisperse particles at reasonably small concentrations. This is of interest since synthesis of monodisperse nanoparticles on a large scale is not easy and is quite expensive. B. RF Heating Experiments Bulk solder material reflowed quickly in a few seconds when subjected to AC magnetic field of 200 to 500 Oe at 280 kHz. Under similar operating conditions SAC305 solder balls of 380 diameter did not show sufficient temperature rise to cause solder reflow. Particularly for Type III SAC305 solder paste samples placed in AC magnetic fields an insignificant temperature rise was observed. Fig. 3(a) and (b) show the thermal images of the solder balls and pristine solder paste samples taken at various time intervals in an AC magnetic field of 500 Oe at 280 kHz. The sharp drop in the eddy current power loss with shrinking particle size is evident from these images which show temperature profile of samples placed in center of the induction coil. Compared to pristine solder paste, solder-MNP composite paste showed increased temperature rise, when subjected to
Fig. 4. Temperature profile for different wt% solder-MNP composite paste in AC magnetic field of 500 Oe at 280 kHz, inset shows the time to reflow for the different wt% samples.
field, due to MNP power loss. Sample with 2 wt% of FeCo MNPs showed sufficient power loss to cause temperature of composite to rise above SAC305 solder melting temperature leading to solder paste reflow. With higher wt% composites, the reflow temperatures were attained in shorter times. Fig. 4 shows comparative temperature profiles for the different wt% samples subjected to AC magnetic field till reflow temperatures were attained. Due to the localized heating, rapid sample cool down can be seen, once the field is removed. The 10 wt% sample showed migration of the MNPs to the reflowed solder surface due to the magnetic field gradient; such segregation of the MNPs to the interface can deteriorate solder
2190
IEEE TRANSACTIONS ON MAGNETICS, VOL. 46, NO. 6, JUNE 2010
C. SEM Characterization Fig. 6 shows an SEM image of a polished cross section from the 4 wt% solder-MNP sample reflowed between Ni-Au surface finish test coupons. The image shows that the presence of MNPs in the solder paste does not affect wettablity of solder on the Ni-Au surface finish. The reflowed solder shows excellent interfacial contact with the Ni-Au surface. EDX analysis of the sample showed the presence of Fe and Co near the interface. FeCo MNPs were not detected in the bulk of the solder matrix. However, the reflowed samples did show a magnetic response indicating the presence of MNPs within the matrix. Due to the localized heating of the solder-MNP composite and the high post-melting cooling rates the formation of intermetallic compounds (IMCs) at the interface is minimized. Fig. 5. Thermal images of (a) 4 wt% solder-MNP composite paste and (b) 4 wt% solder-MNP composite paste between two Ni-Au finish test coupon, during induction reflow.
IV. CONCLUSION We have demonstrated the feasibility of localized heating of solder-MNP composite paste for their reflow using an applied AC magnetic field. By varying the concentration and/or the composition of the magnetic particles in the solder paste one can control the time required to attain reflow temperatures within such composites. It should be further possible to improve the heating rates within the composite material by optimizing the particle size of the MNPs and operating field frequency. Use of such solder-MNP composite paste for their localized reflow using AC magnetic fields can be deployed with ease at industrial level for forming solder interconnects at a large scale in a cost effective manner. ACKNOWLEDGMENT This work was supported by Intel Corporation and the National Science Foundation.
Fig. 6. SEM micrograph of cross section from solder-MNP composite reflowed between two Ni-Au surface finish test coupon at the interface.
properties. Overall 4 wt% samples showed the optimum performance with respect to reflow time and good reflowed solder surface. Fig. 5(a) shows the thermal images of 4 wt% solder-MNP composite paste under AC magnetic field at different time interval. A Ni-Au surface finish test coupon comprising of a Cu clad FR4 board having a Ni-Au surface layer was used to ascertain influence of MNPs on wettability of solder over Ni-Au surface. Fig. 5(b) shows the induction reflow of solder-MNP composite paste placed in-between two Ni-Au surface coupons. The localized heating of the solder-MNP composite is clearly evident from the thermal images wherein the composite exhibit a temperature rise to solder reflow temperatures while the substrate surface remains at room temperature.
REFERENCES [1] J.-H. Lee, Y.-H. Lee, and Y.-S. Kim, “Fluxless laser reflow bumping of Sn-Pb eutectic solder,” Scripta Mater., vol. 42, no. 8, pp. 789–783, Mar. 2000. [2] D. J. Hayes, D. B. Wallace, M. T. Boldman, and R. E. Marusak, “Picoliter solder droplet dispensing,” Int. J. Microcircuits Electron. Packag., vol. 16, no. 3, pp. 173–180, 1993. [3] M. Li, H. Xu, S.-W. R. Lee, J. Kim, and D. Kim, “Eddy current induced heating for the solder reflow of area array packages,” IEEE Trans. Advanced Packag., vol. 31, no. 2, pp. 399–403, May 2008. [4] A. H. Habib, C. L. Ondeck, P. Chaudhary, M. R. Bockstaller, and M. E. McHenry, J. Appl. Phys., vol. 103, p. 07A307, 2008. [5] C. A. Sawyer, A. H. Habib, K. J. Miller, K. N. Collier, C. L. Ondeck, and M. E. McHenry, J. Appl Phys., vol. 105, p. 07B320, 2009. [6] K. J. Miller, K. N. Collier, H. B. Soll-Morris, R. Swaminathan, and M. E. McHenry, J. Appl. Phys., vol. 105, p. 07E714, 2009. [7] Z. Turgut, N. T. Nuhfer, H. R. Piehler, and M. E. McHenry, J. Appl. Phys., vol. 85, p. 4406, 1999. [8] M. E. McHenry, M. A. Willard, and D. E. Laughlin, Progress in Mat. Sci., vol. 44, p. 291, 1999. [9] M. E. McHenry, M. A. Willard, H. Iwanabe, R. A. Sutton, Z. Turgut, A. Hsiao, and D. E. Laughlin, Bull. Mat. Sci., vol. 22, p. 495, 1999.