0
Review Article

# Nanothermal Interface Materials: Technology Review and Recent ResultsOPEN ACCESS

[+] Author and Article Information
Avram Bar-Cohen

Fellow ASME
Agency (DARPA)/Microsystems
Technology Office (MTO),
675 North Randolph Street,
Arlington, VA 22203
e-mail: abc@darpa.mil

Kaiser Matin

Mem. ASME
System Planning Corporation,
3601 Wilson Blvd,
Arlington, VA 22201
e-mail: kaiser.matin.ctr@darpa.mil

Sreekant Narumanchi

Mem. ASME
National Renewable Energy Laboratory,
15013 Denver West Parkway,
Golden, CO 80401
e-mail: sreekant.narumanchi@nrel.gov

Contributed by the Electronic and Photonic Packaging Division of ASME for publication in the JOURNAL OF ELECTRONIC PACKAGING. Manuscript received July 1, 2015; final manuscript received September 11, 2015; published online October 9, 2015. Assoc. Editor: Ashish Gupta.The United States Government retains, and by accepting the article for publication, the publisher acknowledges that the United States Government retains, a non-exclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for United States Government purposes.

J. Electron. Packag 137(4), 040803 (Oct 09, 2015) (17 pages) Paper No: EP-15-1063; doi: 10.1115/1.4031602 History: Received July 01, 2015; Revised September 11, 2015

## Abstract

Thermal interface materials (TIMs) play a critical role in conventionally packaged electronic systems and often represent the highest thermal resistance and/or least reliable element in the heat flow path from the chip to the external ambient. In defense applications, the need to accommodate large differences in the coefficients of thermal expansion (CTE) among the packaging materials, provide for in-field reworkability, and assure physical integrity as well as long-term reliability further exacerbates this situation. Epoxy-based thermoplastic TIMs are compliant and reworkable at low temperature, but their low thermal conductivities pose a significant barrier to the thermal packaging of high-power devices. Alternatively, while solder TIMs offer low thermal interface resistances, their mechanical stiffness and high melting points make them inappropriate for many of these applications. Consequently, Defense Advanced Research Projects Agency (DARPA) initiated a series of studies exploring the potential of nanomaterials and nanostructures to create TIMs with solderlike thermal resistance and thermoplasticlike compliance and reworkability. This paper describes the nano-TIM approaches taken and results obtained by four teams responding to the DARPA challenge of pursuing the development of low thermal resistance of 1 mm2 K/W and high compliance and reliability TIMs. These approaches include the use of metal nanosprings (GE), laminated solder and flexible graphite films (Teledyne), multiwalled carbon nanotubes (CNTs) with layered metallic bonding materials (Raytheon), and open-ended CNTs (Georgia Tech (GT)). Following a detailed description of the specific nano-TIM approaches taken and of the metrology developed and used to measure the very low thermal resistivities, the thermal performance achieved by these nano-TIMs, with constant thermal load, as well as under temperature cycling and in extended life testing (aging), will be presented. It has been found that the nano-TIMs developed by all four teams can provide thermal interface resistivities well below 10 mm2 K/W and that GE's copper nanospring TIMs can consistently achieve thermal interface resistances in the range of 1 mm2 K/W. This paper also introduces efforts undertaken for next generation TIMs to reach thermal interface resistance of just 0.1 mm2 K/W.

## Introduction

The nanothermal interfaces (NTI) thrust of the Thermal Management Technologies (TMT) program kicked off in mid-2009 [1]. The main objective was to develop TIMs based on novel materials and nanostructures that can provide significant reductions in the thermal resistance of the interface layer between the backside of an electronic device and the next layer of the package, typically a heat spreader, heat sink, or coldplate. Increased heat flux of DoD electronics has required significant TIM improvements. Many DoD systems operate at voltages, temperatures, frequencies, or other parameters that are outside of the range specified for reliable long-term operation. As a result, the failure rate for electronic devices under DoD field conditions is often higher than in commercial equipment and, therefore, the NTI program also requires the TIMs to be reworkable. Requiring components to be replaced and “reworked” in the field denotes that the TIMs must have the ability to be applied at modest temperatures and in conventional environments. In addition to providing high thermal conductivity in the through-thickness direction, the NTIs must also accommodate nonplanarity and roughness on the mating surfaces. Finally, the NTIs were required to have long-term reliability (with a minimum of 100 demonstrated temperature cycles) and consistency from chip to chip.

Thermal greases or epoxies, mixed with highly conductive fillers, have relatively low effective thermal conductivity, as the grease or epoxy insulates the filler material. Increasing the concentration of the fillers along the heat flow path may establish preferred paths for heat flow and improve the effective thermal conductivity, but such concentrations strongly increase the viscosity of the grease or paste, as well as the stiffness of the solidified composite and reduce the compliance of the TIM [2,3]. Solder TIMs, on the other hand, offer high thermal conductivity and low thermal interface resistance. But, in typical thicknesses of 25–50 μm, they have low mechanical compliance, which makes them prone to high stress failure in the presence of thermally induced differential expansion caused by a “mismatch” in the CTE across the TIM. In order to meet the compliance requirement, a minimum indium thickness of approximately 200 μm is needed, significantly impeding thermal performance.

Figure 1 displays the TIM thermal resistivity versus thickness meeting the compliance for four NTI technologies, several commercial off-the-shelf (COTS) products, and two theoretical curves for high-conductivity TIM materials. The green lines and points in the figure represent selected thermal greases and epoxies and the red sloped lines on the plot pertain to theoretical resistivities for Indalloy and Indium solder. As shown in the plot, thin bond lines of Indium (80 μm) and Indalloy (35 μm) solder can reach the NTI's thermal resistance goal of 1 mm2 K/W. However, the two commonly bonded packaging materials: silicon, with a CTE of 2.6 × 10−6/K, and copper, with a CTE of 16.7 × 10−6/K, create a large CTE difference and, with thin bond lines, make these TIMs prone to high thermal stress and fatigue failure. The NTI program, therefore, addresses both low thermal resistance and high mechanical compliance, so that the TIMs provide superior thermal performance while maintaining robustness under thermal cycling and aging.

Technology transition is a critical goal of the NTI effort, as well as the overall TMT program, and the final program goals reflect the transition partner needs of each NTI team. Thus, each team has worked with their transition partners to identify a military system that would uniquely benefit from the thermal management technology under development. Toward that end, the transition partners have provided their respective team with the requisite product/system specifications, including form factors, configurations, performance metrics, acquisition standards, and technology qualification procedures. Therefore, modeling, analysis, design, and fabrication of the NTI prototypes were carried out according to these specifications. The prototypes were expected to utilize the unique features of the specific NTI technology and provide performance that materially exceeds the state of the art (SOA) in key parameters. Finally, the projects' experimental and modeling results serve to articulate the TMT “value proposition,” to provide the DoD with the greatest performance value for its investment.

Several materials, such as epoxies and solders, were investigated, but ultimately the performers found that solder was the only acceptable attachment method. Thus, in evaluating the resistivity of the TIM attention must be directed to three distinct “thermal stack” resistances: the “bulk” TIM and an interface material (i.e., solder) on either side. In order to meet the stringent thermal resistance goals of the NTI program, several teams pursued the use of low temperature solder materials, including pure indium, for attaching their nanostructured TIMs. The use of these materials and indium-based solders, in particular, requires additional testing and analysis before transition of the NTI technologies into fieldable DoD platforms.

Table 1 shows the material sets, identifying the interface material, as well as the conducting structure, and TIM approach of the four NTI teams, led by Raytheon, Teledyne, GT, and GE, respectively. The specific approaches taken and the applications targeted are more fully described in a later section of this paper, NTI Technologies and Issues. Each NTI is engineered with unique materials, structures, and configurations to provide maximum benefit to their targeted military system application. The method or material used for attaching the NTIs was a matter of concern throughout the NTI program.

The NTI metrics, listed in Table 2, are seen to consist of thermal resistivity, temperature cycles, stability, reworkability, application time, shear force, etc., with progressively more difficult metrics as the development teams progress from Phase I to Phase II and then to the application form factor of Phase III. It is to be seen that Phase III goals include surface roughness tolerance, maximum processing temperature and area applied.

Phase III resistivity goal, of less than 1 mm2 K/W, posed a significant metrological challenge to the performer teams. The GT team developed a method using infrared microscopy for measuring the total thermal resistance across multiple interfaces. The method is capable of measuring samples of wide ranging resistances with thicknesses varying from 50 μm to 250 μm. The National Renewable Energy Laboratory (NREL), within the DOE, was selected to perform TIM characterization for the NTI program and have developed significant expertise and infrastructure for the thermal characterization of low resistivity TIMs. The NREL performed the following tasks to investigate the performance of each NTI technology under conditions from Table 2: (1) one-dimensional steady-state resistance measurement of NTI samples held between copper blocks, (2) thermal cycling of the NTI samples, and (3) thermal aging of the NTI samples held between copper and silicon, with both the one-dimensional steady-state testing and acoustic microscopy, before and after the extended “aging.”

As part of the characterization—and “qualification”—of the NTI samples, all the teams were asked to evaluate the reworkability of their NTI's in the following two ways: (1) re-attach the mating surfaces five or more times with “fresh” NTI's and document the resulting sequence of thermal resistances and (2) attempt to reuse a single NTI sample up to five times—dismantling and re-attaching the mating surfaces each time—and report the results. It was realized that some of the NTI technologies might not be suited for reuse and that each attempt to reuse the NTI might result in either (a) the destruction of the sample upon removal or (b) the failure to reattach. However, the teams were asked to attempt to reuse their samples and document the changes in the thermal resistance, so that the features and limitations of their approaches were understood.

## NTI Technologies and Issues

###### Raytheon Corporation.

In collaboration with the Georgia Institute of Technology and Purdue University, the Raytheon NTI team developed nanothermal interface materials (nTIMs) based upon metallically bonded, vertically aligned carbon nanotubes (VACNTs), of the type shown in Fig. 2. The short, vertically aligned CNTs are grown on both sides of a thin interposer foil and interfaced with substrate materials via metallic bonding and attached using Sn63, Indalloy 121, or Indalloy 256 solder on both sides [49]. These nTIMs provide the thermal performance of a solder joint while maintaining the compliance typical of a low-conductivity, filled-epoxy, or grease. The Raytheon team has achieved a factor of three improvement in interfacial resistance relative to the SOA commercial TIMs.

A high-precision, one-dimensional, steady-state conduction test facility was utilized to measure the performance of the nTIM samples and, more importantly, to correlate performance to the controllable parameters. Hundreds of samples have been tested, utilizing myriad permutations of these parameters, and contributing to a deeper understanding and optimization of the CNT growth characteristics and application processing conditions.

With the nTIM, junction temperature reductions of 30 °C or more are possible. This increases the device efficiency by approximately 10%, reducing prime power consumption. It also increases the power output by approximately 10%, improving range and sensitivity. Finally, the expected 30 K temperature reduction can increase reliability by approximately 10× , reducing life-cycle/ownership cost.

Figure 3 shows that the Raytheon nTIM offers benefits at all levels electronic device stack-up. An nTIM1 will offer thermal performance equivalent to the SOA in die-attach but with improved mechanical compliance/die stresses. An nTIM2 will significantly improve heat flow from the chip carrier to the heat spreader and enable an increased device footprint for a given stackup CTE mismatch. Use of an nTIM3 to attach the heat spreader to the heat sink will offer improved thermal performance, enhanced surface conformability, and will allow process repeatability and rework.

Figure 4 shows that the nTIM's thermal performance, at less than 0.02 cm2 °C/W at 100 μm thickness, is an order of magnitude better than a gap pad, ten times better than thermal grease, and approximately five times better than silver-filled epoxy at an increased bondline thickness than the latter two TIMs. This CNT-based nTIM has also demonstrated stability after thermal cycling and high-temperature baking with multiple samples. During Phase III, the Raytheon team improved bonding methods and increased the nTIM footprint. They demonstrated the nTIM in Raytheon product representative settings and explored potential collaboration with commercial ventures for larger-scale production.

###### GT.

The GT NTI team developed processing techniques that use separated steps of CNT synthesis and low-temperature transfer to a mating surface. Due to their high thermal conductivity (up to 3000 W/mK), GT's CNTs are a natural choice for the TIM for electronic components with high heat dissipation [1020]. Early NTI attempts used CNTs as fillers, to form high thermal conductivity fluids or TIM composites. However, this approach has proven not to be effective, due to the random dispersion and the intermittent contact among the CNTs. A more advanced approach involves growing CNTs vertically on a silicon wafer. These vertically aligned CNTs can then be attached to a copper heat spreader to form a thermal interface structure.

GT's nano-TIM is based on well-aligned, open-ended VACNTs that have superior thermal conductivity in the heat transfer direction. The approach is different than Raytheon's as it has no interposer. GT applied extensive measurement capability to their effort, includes the use of infrared microscopy for thermal resistance measurements [14]. Thermally conductive adhesives are used as the attachment material for the NTI examples shown in Fig. 5, but use has been made of Ag ink, indium, and other materials. The GT's VACNTs were grown by chemical vapor deposition on both sides of a compliant interposer foil and interfaced with various substrates via a metallic bond. Two metallic bonding techniques were explored: (1) palladium nanoparticle and (2) gold diffusion.

###### General Electric (GE) Global Research.

GE's compliant nTIM allows for thin solder bondlines by using a compliant structure of copper nanosprings, fabricated via glancing angle deposition (GLAD) on W or Si with a Cu capping layer, to accommodate the thermal expansion difference of their mated materials [2123]. Process parameters that control the spring structure are: substrate rotation speed controlling spring radius, growth time controlling spring height, and the rotation speed and growth time controlling the number of turns and, thereby, stiffness. The capping layers are electroless plated Ni to minimize corrosion and oxidation. They are attached to the substrate using solder on both sides: Indium and Sn62 have been explored. Nanosprings are 100× more compliant than solders, so the thermal stresses are carried by the nanosprings rather than the solders, enabling thinner solder layers. TIMs generally fail due to thermal strain during thermal cycling but nanosprings can stretch under thermal strain.

The GE nanospring nTIM is more than 100× higher in compliance than the solder bond, allowing for thin bondlines with thermal resistance less than 0.01 cm2 °C/W. The best performance among commercially available TIMs is that of an indium solder layer that is reflowed and wetted to the heat sink and silicon surfaces. In typical applications the resistance is limited by the bondline thickness, which, due to thermal fatigue, must be kept to around 200 μm. A thermally conductive, thermal stress relieving layer within the bondline, i.e., the approximately 100 μm high nanospring structure, enables much thinner solder bondlines. The GE NTI's bondlines use thin mating layers with the CTE close to the respective substrates to minimize thermal strain. A schematic figure of the structure is shown in Fig. 6 where the TIM is used to bond a silicon die to a copper heat sink. The nanospring length is the factor that needs to be changed for variation in the amount of thermal strain that has to be managed for a reliable interface material. Larger bonding area, CTE mismatch, and temperature cycling range will require longer nanospring extension but this has not been characterized for reliability. The nanospring length is varied by the spring pitch and turns. The pitch and radius are fixed by the deposition angle so the pitch and radius increase linearly. The stiffness decreases with increasing height and with material modulus.

Integrated with GE's thermal ground plane (TGP) [20], the GE NTI is designed to be integrated into GE's Intelligent Platforms GRA111 VPX line-replaceable module. Integration of the NTI and TGP will yield a 40% reduction in temperature rise, enable 66.7% higher power and processing speed, and provide low thermal resistance and reliable bonding to a low-cost copper lid. TGP and NTI together could improve the thermal performance of the GRA111 and increase its mean time between failures.

The GE team was the first to reach Phase III resistivity goal of 1 mm2 K/W. However, this remarkably low value necessitated careful calibration and repeated measurements, with laser flash equipment, before this result could be confirmed. More information about performance testing is in the NTI Performance section of this paper.

###### Teledyne.

Teledyne's laminated, carbon-based NTIs using vertically aligned graphite–metal composites is shown in Fig. 7. They are based on highly oriented two-dimensional graphite nanoplatelets (GNPs) embedded in Indalloy 121 solder. The GNPs are compressed into flexible graphite sheets and are then introduced into the solder, with the high-conductivity planes aligned along the primary heat transfer direction. With this structure, it is easy to form straight, highly conductive paths along the desired heat flow direction [24]. Moreover, the metallic host matrix has a much higher thermal conductivity than that of a polymer host matrix and, due to its excellent flow characteristics in the molten state, can drastically improve the quality of physical contact between the fillers and the substrate surfaces. The bonding between the graphite and the metallic host is also stronger than the bonding between GNPs and nonmetallic materials. Furthermore, compared to organic hosts, solders have much lower phonon spectra mismatch with the GNPs and CNTs, and thus, offer a significantly higher interface conductance which greatly improves the overall thermal performance [25]. CTE mismatch can be controlled by ratio of GNP: solder, CNT spacing, or GNP volume fraction. Also solder and graphite sheet thickness can be adjusted to control compliance [26]. Teledyne reported that their NTI achieved an overall thermal resistivity as low as 1 mm2 K/W, at a bonding temperature approximately 230 °C, and compression pressure of 30 PSI using improved fabrication processes for the graphite-based TIMs [27]. Due to its unique construction, the Teledyne NTI can function as both a TIM and an effective heat spreader, eliminating the need for costly high conductivity spreaders, made of diamond or Boron Nitride, and simplifying the packaging process.

## NTI Testing Methods

To experimentally characterize the thermal properties and thermal performance of TIMs, a variety of techniques have been developed and utilized, including both steady-state and transient techniques [NREL paper]. In this section, we mainly focus on the TIMs which were characterized by metrology measurements provided at NREL using several different techniques.

###### Xenon Flash Technique.

The xenon flash technique was used to evaluate the performance of NTI bonded samples, using a Netzsch LFA 447 Nanoflash instrument. A xenon flash pulse irradiates one surface of the test sample and the generated thermal energy then penetrates and flows toward the underside of the sample as shown in Fig. 8. An infrared detector with a 7.8-mm aperture records the sample's temperature rise as a function of time. The technique is demonstrated in Fig. 8. Under an adiabatic condition, the thermal diffusivity is directly calculated using the temperature rise profile via the following equation:

Display Formula

(1)$α=0.1388 l2t50$
where a is the thermal diffusivity, l is the thickness of the test sample, and t50 is the time at which 50% of the temperature rise has occurred on the back side of the sample.

To determine the thermal conductivity, the density and heat capacity of the test samples must also be known. Using measured values or from previous knowledge of a sample's bulk density and specific heat, its thermal conductivity can be calculated, as shown in the following equation: Display Formula

(2)$λ(T)=α(T)⋅ρ(T)⋅Cp(T)$
where T is the temperature, λ is the thermal conductivity, ρ is the bulk density, and Cp is the specific heat.

When the thermal conductivity is known and w with knowledge of the sample's bondline thickness, the thermal resistance of the interface layer can be calculated Display Formula

(3)$R=xλ$
where R is the thermal resistance, x is the bondline thickness, and λ is the thermal conductivity.

NREL has developed a steady-state test stand for measuring the thermal resistance of TIMs (TIMs and NTIs). The operation of this steady-state test stand follows the method outlined in ASTM D5470-12.

The basic configuration of the test apparatus is shown in Fig. 9, along with the picture of an actual test apparatus that was used to test interface materials. The heater cartridges are embedded into an aluminum hot plate, while silicone oil is circulated through an aluminum cold plate.

Four thermistors (or resistance temperature detectors) are embedded in the metering blocks which have the TIM/NTI between them. The metering block is made of oxygen-free copper with a thin nickel coating to prevent oxidation and also to prevent the block from erosion or corrosion. The average roughness of the surface of the blocks (facing the TIM/NTI) is about 0.5 μm. Through measurement of heat flux and temperature difference across the sample, the thermal resistance is computed.

###### Phase-Sensitive Transient Thermoreflectance (PSTTR).

The PSTTR technique employs two localized fast lasers to help determine the critical thermal properties of materials and interfaces, such as thermal conductivity and thermal resistance. A modulated pump laser is directed on the sample surface to induce a temperature rise and the consequent thermal wave travels through the sample. A probe laser is used to detect the temporal temperature variation on the back surface, which responds to the periodicity of the heating on the front surface of the sample. By extracting the inherent phase information, the thermal properties are derived. Compared with the xenon flash technique and steady-state ASTM test stand, the PSTTR technique is sensitive to small temperature changes (within a few degrees) and has a high resolution which enables measurement of thermal resistances lower than 1 mm2 · K/W [28,29]. The experimental configuration of the PSTTR is depicted in Fig. 10.

###### Reliability Testing and Characterization.

The NTI samples provided by the four DARPA teams were subjected to temperature cycling as well as thermal aging at an elevated temperature. In thermal aging tests, the samples were exposed to a temperature of 130 °C for 300 hrs. In thermal cycling tests, the samples were subjected to temperatures from −40 °C to 80 °C at low (3 °C/min) and high (25 °C/min) ramp rates. The thermal cycle testing array, shown in Fig. 11, was used to study thermally induced stresses in the novel bonded interfaces. The chambers contain workspace volumes of 50 × 28 × 30 cm and are capable of operating from −70 °C to 180 °C.

Acoustic microscopy images were taken of the NTI bondlines for qualitative evaluations of the interfaces, before and during cycling and aging tests. The C-mode scanning acoustic microscope (C-SAM), shown in Fig. 12, is an instrument that uses ultrasound in the frequency range of 5–230 MHz to nondestructively inspect samples for defects.

The technique relies on the acoustic impedance mismatch of materials, and can generate images within a sample based on scattering, refraction, and/or absorption of the acoustic signal within various material layers. Ultrasound cannot travel through air or a vacuum; therefore, it is capable of finding defects within a sample such as voids, cracks, and/or delaminations.

## NTI Performance

The full suite of NREL measurement techniques was used to evaluate the thermal and reliability characteristics of the sample NTI's produced by the DARPA performers. They include aligned CNTs (Raytheon and GT), a laminated graphite and solder composite (Teledyne), and copper nanosprings (GE). GE has established an assembly technique that forms metal nanosprings by the GLAD process. The number of springs, diameter of spring wire, radius of winding, number of windings, and overall spring length can be controlled by the GLAD process. This makes it possible to engineer the desired shear and compressive compliance within the nano-interface material while also optimizing for minimal thermal resistance. Teledyne developed a bonding process that vertically aligns graphite platelets within the contact area between two surfaces. The platelets are first aligned and compressed into thin layers before a solder binds the graphite layers to each other and to the surfaces. The Georgia Institute of Technology led an effort to develop a low-temperature process that grows and aligns CNTs as a thin interface material.

Two rounds of testing, in which standardized test samples were bonded by the performers for accelerated thermal cycling, were conducted by NREL. The first round of testing was performed on the samples from GT, GE, and Teledyne. NREL did not receive samples from Raytheon for the first round of testing. The measured thermal resistance results warranted improvements in material performance, and hence, the performers were given an opportunity to modify their materials before a second round of testing was conducted. For the second round of testing, NREL received samples from Raytheon, GE, and Teledyne while GT opted not to participate in this phase of the testing. In each round of testing, initial thermal performance was characterized by the xenon flash transient measurement technique. Xenon flash measurements were taken at periodic intervals during and after the completion of accelerated testing. First round testing of GT will be presented here only.

###### First Round Results.

The thermal performance and reliability of the performers' interface materials were characterized by utilizing 10-mm × 10-mm cross-sectional footprint samples of silicon bonded to copper via the NTIs, as shown in Fig. 13. The surface roughness was not measured. The silicon diodes used for creating the bonded samples were 350 μm thick and were provided with a backside metallization of aluminum/titanium/nickel/silver. The copper coupons were 1 mm thick and were not provided with any metallization. Surface preparation and additional metallization processing were allowed for the teams to optimize the bond strength with their interface materials. After bonding, bondline thicknesses varied amongst the performers' samples from 70 to 325 μm. These samples were made to simulate actual thermos-mechanical behavior of electronic packages.

Bonded samples were evaluated for thermal performance, using a Netzsch LFA 447 Nanoflash instrument. The Nanoflash operates following the ASTM E-1461-13 test standard [18]. A xenon flash pulse directs energy toward the underside of a test sample. An infrared detector with a 7.8 mm aperture records the sample's top-side rise in temperature as a function of time.

Prior to testing, all samples were sprayed with DGF-123 Dry Graphite Film Spray. This uniform graphite coating allows for consistent absorptivity of the xenon flash pulse and emissivity to the infrared detector between test samples. Initial thermal resistance measurements are summarized in Table 3 and Fig. 14.

The GT NTI design was targeted to produce a thermal resistivity of 3.6 mm2 K/W. Approximately half of the samples were measured to have thermal resistances lower than 5 mm2 K/W; however, a significant number of samples were measured with much higher resistivity values. This result indicates that while GT's approach has the potential to produce a low thermal resistance material, significant synthesis variations are present in the production process.

In addition to transient thermal measurements with the nanoflash apparatus, acoustic microscopy images were taken of the bondlines for qualitative evaluations of the interfaces. In general, darker areas indicate a strong bond between the silicon and copper coupons while lighter areas denote the likely presence of voiding or delamination. The GT NTI samples typically had large areas of discontinuity in bond quality (Fig. 15). The presence of lighter, poorer bond areas in these samples correlated with higher thermal resistance measurements. For reference, a sample bonded with lead-solder is shown with a high percentage of voiding.

###### Reliability Testing and Characterization.

The bonded samples were subjected to accelerated tests in the form of temperature cycling as well as thermal aging at an elevated temperature. In total, 15 samples were tested as shown in Table 3. In thermal aging tests, the samples were exposed to a temperature of 130 °C for 300 hrs. In thermal cycling tests, the samples were subjected to temperatures from −40 °C to 80 °C at low (3 °C/min) and high (25 °C/min) ramp rates. Transient thermal measurements with the Nanoflash apparatus were performed to characterize the thermal performance of all samples prior to, during, and after accelerated testing. Acoustic microscopy was used to monitor the condition of the interfaces during the same analysis intervals. Samples thermally aged at 130 °C were inspected every 100 hrs.

###### Aging Tests.

After 300 hrs, the GT NTI samples all showed an increase in thermal resistance, as shown in Table 4 and Fig. 16, though 2 of the 4 samples (#2 and #4) experienced just a 50% increase in resistivity over that time period. Thus, the lowest resistivity sample that initially was measured at 4.6 mm2 K/W only reached 7 mm2 K/W when the aging test concluded. The thermal resistance of the bondline within samples approached 100 mm2 K/W for just one case, suggesting that a failure of the interface would occur shortly, if thermal aging continued.

###### Thermal Cycling—Low Ramp Rate Tests.

Under low ramp rate thermal cycling conditions, samples were subjected to temperature variations from −40 °C to 80 °C and transitioned between the extremes at a low, 3 °C/min, ramp rate. Samples were inspected every ten cycles with transient thermal measurements and acoustic imaging. The GT NTI samples showed an approximately 50% increase in thermal resistance for samples that were initially measured at 4 mm2 K/W. Samples that were initially measured at 10 mm2 K/W or higher showed a significantly larger increase in thermal resistivity, approaching three times the initial value. The GT typical sample results are summarized in Table 5 and Fig. 17.

###### Thermal Cycling—High Ramp Rate Results.

The NTI samples were subjected to a second accelerated test with temperature variations again cycling between −40 °C and 80 °C, but with ramp rates greater than 25 °C/min to impart a more severe thermal shock condition onto the samples. All the samples from GT showed a significant increase in thermal resistivity after thermal cycling under high ramp conditions, as shown in Table 6 and Fig. 18. However, it is to be noted that the GT samples with a higher initial resistivity appeared to deteriorate less, with sample #11, for example experiencing only a 30% increase, while the samples with the lower initial resistivity (#9 and #12) deteriorated by a factor of ten and 3.5, respectively.

###### Second Round Results.

Despite the encouraging results achieved in the thermal testing of the NTI samples supplied to NREL, the accelerated testing results for these samples did not meet the DARPA NTI performance criteria. Consequently, the participating teams were given an opportunity to improve their NTI performance through changes in the synthesis process. Samples were obtained from Raytheon, GE, and Teledyne for this second round of testing. The GT team declined to participate in this second round of testing. Raytheon sent two sets of samples—one in which the NTI was bonded between the Si and Cu coupons, and another in which the NTIs were bonded between SiC and CuMo coupons. GE made adjustments to their samples by using a different metallization on the Cu and the NTI, which would prevent formation of a brittle intermetallic layer. The authors are not aware of the changes made by Teledyne in the synthesis process in order to enhance the thermal performance and reliability of the Teledyne NTIs.

###### Initial Thermal Resistivity Measurements.

The initial thermal resistance results of all samples for the second round of testing are summarized in Table 7, breakdown of Raytheon samples from different sources are also listed.

It is clear from the above figure (Fig. 19) that the initial thermal resistivity of the GE and Teledyne samples significantly improved, as compared to the first round. Almost all of the GE samples had less than 1 mm2 K/W initial thermal resistivity with just two out of 13 being slightly higher than the target value of 1 mm2 K/W. The Teledyne NTI samples exhibited resistance values between 1.8 and 4.1 mm2 K/W. However, a significant inconsistency was noted with the Raytheon samples, with the best sample displaying a thermal resistivity of 2.9 mm2 K/W, but more than half of the samples exhibiting thermal resistivities greater than 10 mm2 K/W. Acoustic images representative of the general condition of samples from each performer are shown in Fig. 20. GE (center) shows full coverage bonding while Teledyne and Raytheon show voiding or delamination. Teledyne shows more bonding dark areas.

###### Reliability Testing and Characterization.

The samples were subjected to the exact same accelerated testing profiles as in the first round of testing—thermal aging, thermal cycling with a low ramp rate, and thermal cycling with a high ramp rate. Acoustic microscopic images were taken at periodic intervals to monitor the structural integrity of the interface layers under accelerated testing.

###### Thermal Aging Tests.

The Raytheon NTI samples demonstrated a mixed response to thermal aging, as shown in Table 8 and Fig. 21. In both the silicon–carbide-to-copper–molybdenum (R-SiC) samples and the silicon-to-copper (R-Si) samples, different rates of increase of thermal resistivity were observed. A couple of R-SiC samples can be considered to have failed, with their thermal resistivities measuring higher than 100 mm2 K/W after 300 hrs of aging. It is to be noted that a single Raytheon SiC-CuMo NTI sample (#29) did display a low and nearly invariant resistivity, starting with an initial value of 2.9 mm2 K/W and ending at 3.3 mm2 K/W.

Highly promising results were obtained for the GE samples under thermal aging (Table 9 and Fig. 22). Only after 100 hrs of aging did the thermal resistances of all samples start to increase, albeit rather slightly, and the resistivity of one sample (#27) remained invariant through 200 hrs at 1.1 mm2 K/W. The thermal resistivity of all the GE NTI's remained under 3.4 mm2 K/W after 300 hrs of aging, in this second round of testing.

Apart from one sample, in this second round of testing, all the Teledyne NTI samples failed during or before thermal aging testing due to poor structural coherence within the bondline thickness, as shown in Table 10 and Fig. 23. The initial thermal resistances of these samples did not serve as a predictive indicator of failure under thermal aging test conditions. The best sample, T-09, showed nearly invariant behavior for the first 200 hrs varying in resistivity from 3.7 mm2 K/W to 4.1 mm2 K/W, but then rising to 4.9 mm2 K/W at 300 hrs.

###### Thermal Cycling—Low Ramp Rate Tests.

Under thermal cycling (low ramp rate) testing, most of the Raytheon samples showed a significant amount of degradation as the number of cycles increased, as shown in Table 11 and Fig. 24. One sample each from the R-Si and R-SiC batches performed relatively better than other samples; however, the thermal resistances of both were measured above 5 mm2 K/W after 50 thermal cycles. It is to be noted that Raytheon sample #16, an NTI bonding silicon and copper, and sample #25, an NTI bonding SiC and CuMo, do display acceptable 50-cycle results, 20.9–20.7 mm2 K/W and 5.4–7.2 mm2 K/W, respectively, but failed to meet the DARPA resistivity goal of 1 mm2 K/W.

Similar to what was observed with the thermal aging results, the GE samples again demonstrated an excellent thermal performance and reliability under thermal cycling (low ramp rate). Of the four samples that underwent thermal cycling, the thermal resistances of two samples remained under 1 mm2 K/W while that of the other two were only slightly higher after 50 thermal cycles. Also, it was observed that thermal cycling (low ramp rate) had very little impact on these samples as the thermal resistance values remained more or less the same as the initial ones (Table 12 and Fig. 25).

Although the Teledyne samples did not perform as well as the GE samples under low ramp rate thermal cycling, the measured thermal resistivity values were much better than achieved in the first round of samples. As shown in Table 13 and Fig. 26, after 50 thermal cycles, three out of four samples had thermal resistivities lower than 10 mm2 K/W, and sample T-05 varied from 1.8 to just 2.2 mm2 K/W over the test period.

###### Thermal Cycling—High Ramp Rate Tests.

Under high ramp rate thermal cycling, the Raytheon samples followed a similar trend to that observed under other accelerated test conditions, with the thermal resistivities of most of the samples indicating major signs of failure at the end of accelerated testing. The thermal resistivity values of the second-round Raytheon samples are shown in Table 14 and Fig. 27. Once again, however, two of the Raytheon samples—#21 and #22, bonding SiC to CuMo—displayed relatively robust resistivity behavior, increasing by less than 50% over the 50 high-ramp-rate cycles, but—in both cases—starting with resistivities of 5.2 and 11.1 mm2 K/W, respectively, that were already higher than the DARPA goal.

The thermal resistivity values of the second-round GE samples, shown in Table 15 and Fig. 28, confirmed their superior thermal performance and reliability under various accelerated test conditions. All four samples met the thermal resistance target of 1 mm2 K/W after 50 hrs of thermal cycling. Similar to the results obtained for thermal cycling under the low ramp rate conditions, the GE samples performed well under thermal cycling with the high ramp rate, exhibiting minimal variation in thermal resistances.

The second round high-ramp-rate test of Teledyne NTIs revealed similar results to those achieved in the first round, with two samples approaching failure and displaying resistivities of 42 and 82 mm2 K/W, respectively, under thermal cycling loads, but two other samples remaining under 10 mm2 K/W after 50 thermal cycles (Table 16 and Fig. 29). On average, the second round of Teledyne samples had higher thermal performance than the first round of samples. But the small sample size and lack of consistency in these results make it difficult to substantiate that conclusion.

###### Next Generation nTIMs.

Current DARPA efforts are investigating TIMs with extremely low, solderlike thermal resistance along with high, epoxylike mechanical compliance with the goal of providing 1 mm2 K/W. Next generation effort seeks to advance SOA TIMs by reducing the thermal resistance by another factor of 10, without compromising the mechanical compliance of the TIM. DARPA awarded two Young Faculty Awards (YFAs) for developing next generation nano-enabled TIMs. The goal for this was more aggressive with a target of 0.1 mm2 K/W. TAMU (Texas A&M) and CMU (Carnegie Mellon) approaches and results will be discussed in this section as they look very promising.

###### TAMU Effort.

The Texas A&M team used functionalized BN/Copper nanoribbons/sheets in solder alloys for their next-generation TIM, as shown in Fig. 30. The functionalized nanosheets are attached to the metallic solder and other metallic alloys by soft ligands. Exfoliated BN nanosheets were functionalized by soft organic ligands and, using an electro-codeposition approach, functionalized BN nanosheets were dispersed in the copper matrix and metal–inorganic–organic TIMs nanocomposites were reliably produced [3033]. In systematic TTR testing at NREL, with several different ligands, TIM wafers with a 45–50 μm bondline thickness the overall thermal resistivity was found to be 0.33–0.45 mm2 K/W depending on the ligand, corresponding to an effective thermal conductivity of 240 W/m K to 280 W/m K. The Young's modulus was found to equal 15 GPa and the Hardness equal to 117 MPa.

Figure 31 depicts the relationship of thermal conductivity and elastic modulus for a range of packaging materials. It may be seen that the Cu/f-BNNS nanocomposite TIMs, produced by the TAMU team, occupy a previously unexplored part of this chart, offering very large improvements in thermal conductivity at low values of elastic modulii, compared to the commercially available polymer(epoxy)-based, solder TIMs, and thermal greases.

###### CMU Effort.

Carnegie Mellon's next generation TIM is based on the use of conductive and compliant ordered nanostructures, such as copper or silver nanowire arrays with embedded metal nanoparticles [27,3437]. Figure 32 reflects the CMU approach, which displays large-scale anodic aluminum oxide templates with tunable sizes were used to grow copper and silver nanowire arrays via electrochemical deposition. Scanning electron microscopy studies showed that the nanowires have a large aspect ratio (e.g., 100–1000) and a high packing density. The high thermal conductivity of the metal nanowires and use of an open array with filler particles offers the promise of efficient heat transfer and high mechanical compliance and makes CMU a prime candidate for achieving an overall nTIM thermal resistance of less than 0.1 mm2 · K/W.

Preliminary thermal measurements have demonstrated a best thermal resistivity of approximately 0.3 mm2 K/W. The AFM-measured Young's moduli for the copper and silver nanowire arrays were found to range from 6 MPa to 25 MPa, and from 5 MPa to 19 MPa, respectively, more than three orders of magnitude smaller those of bulk copper and silver.

## Conclusions

The goal of the NTI program was to improve the thermal bottleneck in DoD systems by lowering TIM resistance to 1 mm2 K/W ensuring thermoplastic like compliance and reworkability. This paper summarized the work of four performers: GT, Raytheon, GE, and Teledyne from their samples tested at NREL. It is found all the performers improved substantially from the SOA. GT achieved a thermal resistance best value of 3.9 mm2 K/W, Raytheon best values are close to 2.9 mm2 K/W, Teledyne achieved a value close to 1.8 mm2 K/W, and GE a best value of 0.7 mm2 K/W. To that end in goal four performers provided samples that improved substantially on the SOA. GE NTIs were the only ones that consistently met the DARPA goals offering 1 mm2 K/W. YFA efforts are paving the way for further reductions in resistivity toward 0.1 mm2 K/W.

## Acknowledgements

The authors would like to thank DARPA for the financial support for this work. The authors would also like to thank the Principal Investigators: Dave Altman (Raytheon), Jack Moon (GT), Yuan Zhao (Teledyne), and Dave Shaddock (GE), as well as their research teams, for their contributions to the NTI program and their collaboration with the authors in preparation of this paper. We would also like to thank Dr. Sheng Shen and Dr. Mustafa Akbulut for their contributions as DARPA YFAs. The views, opinions, and/or findings contained in this article/presentation are those of the author/presenter and should not be interpreted as representing the official views or policies, either expressed or implied, of the Defense Advanced Research Projects Agency or the Department of Defense. Distribution Statement A, Approved for Public Release, Distribution Unlimited.

## Nomenclature

• cp =

specific heat (J/g K)

• R =

thermal resistance (mm2 K/W)

• t =

time (s)

• T =

temperature (°C)

Greek Symbols
• α =

thermal diffusivity (cm2/s)

• λ =

thermal conductivity (W/m K)

• ρ =

bulk density (g/cm3)

## References

Bloshchok, K. P. , and Bar-Cohen, A. , 2012, “ Advanced Thermal Management Technologies for Defense Electronics,” Defense Transformation and Net-Centric Systems 2012, R. Suresh , ed., Proc. SPIE, 8405, p. 84050I.
Xu, J. , and Fisher, T. S. , 2006, “ Enhancement of Thermal Interface Materials With Carbon Nanotube Arrays,” Int. J. Heat Mass Transfer, 49(9), pp. 1658–1666.
Prasher, R. , 2008, “ Thermal Boundary Resistance and Thermal Conductivity of Multiwalled Carbon Nanotubes,” Phys. Rev. B, 77(7), p. 075424.
Zhen, H. , Fisher, T. , and Murthy, J. , 2010, “ Simulation of Phonon Transmission Through Graphene and Graphene Nanoribbons With a Green's Function Method,” J. Appl. Phys., 108(9), p. 094319.
Zhen, H. , Fisher, T. , and Murthy, J. , 2011, “ An Atomistic Study of Thermal Conductance Across a Metal–Graphene Nanoribbon Interface,” J. Appl. Phys., 109(7), p. 074305.
Zhen, H. , Fisher, T. , and Murthy, J. , 2010, “ Simulation of Thermal Conductance Across Dimensionally Mismatched Graphene Interfaces,” J. Appl. Phys., 108(11), p. 114310.
Wasniewski, J. R. , Altman, D. H. , Hodson, S. L. , Fisher, T. S. , Bulusu, A. , Graham, S. , and Cola, B. A. , 2012, “ Characterization of Metallically Bonded Carbon Nanotube-Based Thermal Interface Materials Using a High Accuracy 1D Steady-State Technique,” ASME J. Electron. Packag., 134(2), p. 020901.
Singh, D. , Murthy, J. Y. , and Fisher, T. S. , 2011, “ Mechanism of Thermal Conductivity Reduction in Few-Layer Graphene,” J. Appl. Phys., 110(4), p. 044317.
Singh, D. , Murthy, J. Y. , and Fisher, T. S. , 2011, “ Spectral Phonon Conduction and Dominant Scattering Pathways in Graphene,” J. Appl. Phys., 110(9), p. 094312.
Guo, L. , Hodson, S. L. , Fisher, T. S. , and Xu, X. , 2012, “ Heat Transfer Across Metal–Dielectric Interfaces During Ultrafast-Laser Heating,” ASME J. Heat Transfer, 134(4), p. 042402.
Nguyen, J. J. , Bougher, T. L. , Pour Shahid Saeed Abadi, P. , Sharma, A. , Graham, S. , and Cola, B. A. , 2013, “ Postgrowth Microwave Treatment to Align Carbon Nanotubes,” ASME J. Micro Nano-Manuf., 1(1), p. 014501.
Yao, Y. , Moon, K.-S. , McNamara, A. , and Wong, C.-P. , 2013, “ Water Vapor Treatment for Decreasing the Adhesion Between Vertically Aligned Carbon Nanotubes and the Growth Substrate,” Chem. Vap. Deposition, 19(7–9), pp. 224–227.
Ginga, N. J. , and Sitaraman, S. K. , 2013, “ The Experimental Measurement of Effective Compressive Modulus of Carbon Nanotube Forests and the Nature of Deformation,” Carbon, 53, pp. 237–244.
Cao, A. , and Qu, J. , 2012, “ Kapitza Conductance of Symmetric Tilt Grain Boundaries in Graphene,” J. Appl. Phys., 111(5), p. 053529.
Pour Shahid Saeed Abadi, P. , Hutchens, S. B. , Greer, J. R. , Cola, B. A. , and Graham, S. , 2013, “ Buckling-Driven Delamination of Carbon Nanotube Forests,” Appl. Phys. Lett., 102(22), p. 223103.
Thess, A. , Lee, R. , Nikolaev, P. , Dai, H. , Petit, P. , Robert, J. , Xu, C. , Lee, Y. H. , Kim, S. G. , and Rinzler, A. G. , 1996, “ Crystalline Ropes of Metallic Carbon Nanotubes,” Science, 273(5274), pp. 483–487. [PubMed]
Cao, A. , and Qu, J. , 2012, “ Size Dependent Thermal Conductivity of Single-Walled Carbon Nanotubes,” J. Appl. Phys., 112(1), p. 013503.
Gao, F. , Qu, J. , and Yao, M. , 2011, “ Interfacial Thermal Resistance Between Metallic Carbon Nanotube and Cu Substrate,” J. Appl. Phys., 110(12), p. 124314.
Ginga, N. J. , Chen, W. , and Sitaraman, S. K. , 2014, “ Waviness Reduces Effective Modulus of Carbon Nanotube Forests by Several Orders of Magnitude,” Carbon, 66, pp. 57–66.
Yao, Y. , Li, Z. , and Wong, C.-P. , 2013, “ Quality Control of Vertically Aligned Carbon Nanotubes Grown by Chemical Vapor Deposition,” IEEE Trans. Compon. Packag. Manuf. Technol., 3(11), pp. 1804–1810.
Shaddock, D. , Weaver, S. , Chasiotis, I. , Shah, B. , and Zhong, D. , 2011, “ Development of a Compliant Nanothermal Interface Material,” ASME Paper No. IPACK2011-52015.
Avram, B.-C. , Kaiser, M. , Nicholas, J. , and Sharar, D. , “ Two-Phase Thermal Ground Planes: Technology Development and Parametric Results,” ASME J. Electron. Packag., 137(1), p. 010801.
General Electric, 2013, “ Article Including Thermal Interface Element and Method of Preparation,” U.S. Patent No. 8405996 B2.
Zhao, Y. , Strauss, D. , Liao, T. , Chen, Y. C. , and Chen, C. L. , 2011, “ Development of a High Performance Thermal Interface Material With Vertically Aligned Graphite Platelets,” ASME Paper No. AJTEC2011-44169.
Collins, K. C. , Chen, S. , and Chen, G. , 2010, “ Effects of Surface Chemistry on Thermal Conductance at Aluminum–Diamond Interfaces,” Appl. Phys. Lett., 97(8), p. 083102.
Chen, H.-H. , Zhao, Y. , and Chen, C.-L. , 2013, “ Experimental Study of Coefficient of Thermal Expansion of Alleged Graphite Thermal Interface Materials,” Front. Heat Mass Transfer, 4, p. 013004.
Li, P. , Shi, J. , Ng, L. , and Shen, S. , “ All-Metal Nanostructured Thermal Interface Materials With High Thermal Conductivity and High Mechanical Compliance,” (in preparation).
DeVoto, D. , Paret, P. , Mihalic, M. , Narumanchi, S. , Bar-Cohen, A. , and Matin, K. , 2014, “ Thermal Performance and Reliability Characterization of Bonded Interface Materials,” IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), Orlando, FL, May 27–30, pp. 409–417.
Tong, T. , Zhao, Y. , Delzeit, L. , Kashani, A. , Meyyappan, M. , and Majumdar, A. , 2007, “ Dense Vertically Aligned Multiwalled Carbon Nanotube Arrays as Thermal Interface Materials,” IEEE Trans. Compon. Packag. Technol., 30(1), pp. 92–100.
Akbulut, M. , Yegin, C. , and Nagabandi, N. , “ Ultra High-Performance TIMs Nanofabricated Via Electrocodeposition of Functionalized Boron Nitride Nanosheets and Metal Matrix,” U.S. Patent (in preparation).
Yegin, C. , Nagabandi, N. , and Akbulut, M. , “ Functionalized Nanosheet–Metal Nanocomposite TIMs With Superior Thermal and Mechanical Properties,” Adv. Mater. (in press).
Nagabandi, N. , Yegin, C. , and Akbulut, M. , “ Effect of Surface Chemistry on Thermal and Mechanical Properties of Nanosheet/Metal Nanocomposite TIMS,” Chem. Mater. (in press).
Chen, I. C. , and Akbulut, M. , “ Adsorption Kinetics and Thermodynamics of Graphene Oxide on Cationic Self-Assembled Monolayers,” Langmuir (in press).
Shen, S. , 2014, “ Metal Nanowire Thermal Interface Materials,” Invention Disclosure, Patent No. 15-136.
Shen, S. , Henry, A. , Tong, J. , Zheng, R. T. , and Chen, G. , 2010, “ Polyethylene Nanofibers With Very High Thermal Conductivities,” Nat. Nanotechnol., 5, pp. 251−255. [PubMed]
Shen, S. , Narayanaswamy, A. , and Chen, G. , 2009, “ Surface Phonon Polariton Mediated Energy Transfer Between Nanoscale Gaps,” Nano Lett., 9(8), pp. 2909–2913. [PubMed]
Shen, S. , Narayanaswamy, A. , Goh, S. , and Chen, G. , 2008, “ Thermal Conductance of Bimaterial Microcantilevers,” Appl. Phys. Lett., 92(6), p. 063509.
View article in PDF format.

## References

Bloshchok, K. P. , and Bar-Cohen, A. , 2012, “ Advanced Thermal Management Technologies for Defense Electronics,” Defense Transformation and Net-Centric Systems 2012, R. Suresh , ed., Proc. SPIE, 8405, p. 84050I.
Xu, J. , and Fisher, T. S. , 2006, “ Enhancement of Thermal Interface Materials With Carbon Nanotube Arrays,” Int. J. Heat Mass Transfer, 49(9), pp. 1658–1666.
Prasher, R. , 2008, “ Thermal Boundary Resistance and Thermal Conductivity of Multiwalled Carbon Nanotubes,” Phys. Rev. B, 77(7), p. 075424.
Zhen, H. , Fisher, T. , and Murthy, J. , 2010, “ Simulation of Phonon Transmission Through Graphene and Graphene Nanoribbons With a Green's Function Method,” J. Appl. Phys., 108(9), p. 094319.
Zhen, H. , Fisher, T. , and Murthy, J. , 2011, “ An Atomistic Study of Thermal Conductance Across a Metal–Graphene Nanoribbon Interface,” J. Appl. Phys., 109(7), p. 074305.
Zhen, H. , Fisher, T. , and Murthy, J. , 2010, “ Simulation of Thermal Conductance Across Dimensionally Mismatched Graphene Interfaces,” J. Appl. Phys., 108(11), p. 114310.
Wasniewski, J. R. , Altman, D. H. , Hodson, S. L. , Fisher, T. S. , Bulusu, A. , Graham, S. , and Cola, B. A. , 2012, “ Characterization of Metallically Bonded Carbon Nanotube-Based Thermal Interface Materials Using a High Accuracy 1D Steady-State Technique,” ASME J. Electron. Packag., 134(2), p. 020901.
Singh, D. , Murthy, J. Y. , and Fisher, T. S. , 2011, “ Mechanism of Thermal Conductivity Reduction in Few-Layer Graphene,” J. Appl. Phys., 110(4), p. 044317.
Singh, D. , Murthy, J. Y. , and Fisher, T. S. , 2011, “ Spectral Phonon Conduction and Dominant Scattering Pathways in Graphene,” J. Appl. Phys., 110(9), p. 094312.
Guo, L. , Hodson, S. L. , Fisher, T. S. , and Xu, X. , 2012, “ Heat Transfer Across Metal–Dielectric Interfaces During Ultrafast-Laser Heating,” ASME J. Heat Transfer, 134(4), p. 042402.
Nguyen, J. J. , Bougher, T. L. , Pour Shahid Saeed Abadi, P. , Sharma, A. , Graham, S. , and Cola, B. A. , 2013, “ Postgrowth Microwave Treatment to Align Carbon Nanotubes,” ASME J. Micro Nano-Manuf., 1(1), p. 014501.
Yao, Y. , Moon, K.-S. , McNamara, A. , and Wong, C.-P. , 2013, “ Water Vapor Treatment for Decreasing the Adhesion Between Vertically Aligned Carbon Nanotubes and the Growth Substrate,” Chem. Vap. Deposition, 19(7–9), pp. 224–227.
Ginga, N. J. , and Sitaraman, S. K. , 2013, “ The Experimental Measurement of Effective Compressive Modulus of Carbon Nanotube Forests and the Nature of Deformation,” Carbon, 53, pp. 237–244.
Cao, A. , and Qu, J. , 2012, “ Kapitza Conductance of Symmetric Tilt Grain Boundaries in Graphene,” J. Appl. Phys., 111(5), p. 053529.
Pour Shahid Saeed Abadi, P. , Hutchens, S. B. , Greer, J. R. , Cola, B. A. , and Graham, S. , 2013, “ Buckling-Driven Delamination of Carbon Nanotube Forests,” Appl. Phys. Lett., 102(22), p. 223103.
Thess, A. , Lee, R. , Nikolaev, P. , Dai, H. , Petit, P. , Robert, J. , Xu, C. , Lee, Y. H. , Kim, S. G. , and Rinzler, A. G. , 1996, “ Crystalline Ropes of Metallic Carbon Nanotubes,” Science, 273(5274), pp. 483–487. [PubMed]
Cao, A. , and Qu, J. , 2012, “ Size Dependent Thermal Conductivity of Single-Walled Carbon Nanotubes,” J. Appl. Phys., 112(1), p. 013503.
Gao, F. , Qu, J. , and Yao, M. , 2011, “ Interfacial Thermal Resistance Between Metallic Carbon Nanotube and Cu Substrate,” J. Appl. Phys., 110(12), p. 124314.
Ginga, N. J. , Chen, W. , and Sitaraman, S. K. , 2014, “ Waviness Reduces Effective Modulus of Carbon Nanotube Forests by Several Orders of Magnitude,” Carbon, 66, pp. 57–66.
Yao, Y. , Li, Z. , and Wong, C.-P. , 2013, “ Quality Control of Vertically Aligned Carbon Nanotubes Grown by Chemical Vapor Deposition,” IEEE Trans. Compon. Packag. Manuf. Technol., 3(11), pp. 1804–1810.
Shaddock, D. , Weaver, S. , Chasiotis, I. , Shah, B. , and Zhong, D. , 2011, “ Development of a Compliant Nanothermal Interface Material,” ASME Paper No. IPACK2011-52015.
Avram, B.-C. , Kaiser, M. , Nicholas, J. , and Sharar, D. , “ Two-Phase Thermal Ground Planes: Technology Development and Parametric Results,” ASME J. Electron. Packag., 137(1), p. 010801.
General Electric, 2013, “ Article Including Thermal Interface Element and Method of Preparation,” U.S. Patent No. 8405996 B2.
Zhao, Y. , Strauss, D. , Liao, T. , Chen, Y. C. , and Chen, C. L. , 2011, “ Development of a High Performance Thermal Interface Material With Vertically Aligned Graphite Platelets,” ASME Paper No. AJTEC2011-44169.
Collins, K. C. , Chen, S. , and Chen, G. , 2010, “ Effects of Surface Chemistry on Thermal Conductance at Aluminum–Diamond Interfaces,” Appl. Phys. Lett., 97(8), p. 083102.
Chen, H.-H. , Zhao, Y. , and Chen, C.-L. , 2013, “ Experimental Study of Coefficient of Thermal Expansion of Alleged Graphite Thermal Interface Materials,” Front. Heat Mass Transfer, 4, p. 013004.
Li, P. , Shi, J. , Ng, L. , and Shen, S. , “ All-Metal Nanostructured Thermal Interface Materials With High Thermal Conductivity and High Mechanical Compliance,” (in preparation).
DeVoto, D. , Paret, P. , Mihalic, M. , Narumanchi, S. , Bar-Cohen, A. , and Matin, K. , 2014, “ Thermal Performance and Reliability Characterization of Bonded Interface Materials,” IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), Orlando, FL, May 27–30, pp. 409–417.
Tong, T. , Zhao, Y. , Delzeit, L. , Kashani, A. , Meyyappan, M. , and Majumdar, A. , 2007, “ Dense Vertically Aligned Multiwalled Carbon Nanotube Arrays as Thermal Interface Materials,” IEEE Trans. Compon. Packag. Technol., 30(1), pp. 92–100.
Akbulut, M. , Yegin, C. , and Nagabandi, N. , “ Ultra High-Performance TIMs Nanofabricated Via Electrocodeposition of Functionalized Boron Nitride Nanosheets and Metal Matrix,” U.S. Patent (in preparation).
Yegin, C. , Nagabandi, N. , and Akbulut, M. , “ Functionalized Nanosheet–Metal Nanocomposite TIMs With Superior Thermal and Mechanical Properties,” Adv. Mater. (in press).
Nagabandi, N. , Yegin, C. , and Akbulut, M. , “ Effect of Surface Chemistry on Thermal and Mechanical Properties of Nanosheet/Metal Nanocomposite TIMS,” Chem. Mater. (in press).
Chen, I. C. , and Akbulut, M. , “ Adsorption Kinetics and Thermodynamics of Graphene Oxide on Cationic Self-Assembled Monolayers,” Langmuir (in press).
Shen, S. , 2014, “ Metal Nanowire Thermal Interface Materials,” Invention Disclosure, Patent No. 15-136.
Shen, S. , Henry, A. , Tong, J. , Zheng, R. T. , and Chen, G. , 2010, “ Polyethylene Nanofibers With Very High Thermal Conductivities,” Nat. Nanotechnol., 5, pp. 251−255. [PubMed]
Shen, S. , Narayanaswamy, A. , and Chen, G. , 2009, “ Surface Phonon Polariton Mediated Energy Transfer Between Nanoscale Gaps,” Nano Lett., 9(8), pp. 2909–2913. [PubMed]
Shen, S. , Narayanaswamy, A. , Goh, S. , and Chen, G. , 2008, “ Thermal Conductance of Bimaterial Microcantilevers,” Appl. Phys. Lett., 92(6), p. 063509.

## Figures

Fig. 1

NTI thermal resistance versus thickness compared to COTS technologies

Fig. 2

Raytheon nTIM device overview. Multiwalled CNTs grown on both sides of a graphene or metallic foil; ends of CNTs are metalized to enhance adhesion.

Fig. 3

Raytheon nTIM application

Fig. 4

Comparison of Raytheon NTI performance with alternate TIMs

Fig. 5

GE Compliant NTI TIM [19]

Fig. 6

GT NTI overview

Fig. 7

Teledyne NTI overview

Fig. 8

Xenon flash technique

Fig. 9

Fig. 10

PSTTR technique experimental setup

Fig. 11

Thermal cycling and aging testing array

Fig. 12

C-SAM

Fig. 13

Silicon and copper coupons

Fig. 14

Initial sample thermal resistance (mm2 K/W)

Fig. 15

Acoustic images of samples from GT (left) and lead solder as a reference (right)

Fig. 16

GT thermal aging results (mm2 K/W)

Fig. 17

GT thermal cycling (low ramp rate) results (mm2 K/W)

Fig. 18

GT thermal cycling (high ramp rate) results (mm2 K/W)

Fig. 19

Initial sample thermal resistance (mm2 K/W)

Fig. 20

Acoustic images of samples from Raytheon (left), GE (center), and Teledyne (right)

Fig. 21

Raytheon thermal aging results (mm2 K/W)

Fig. 22

GE thermal aging results (mm2 K/W)

Fig. 23

Teledyne thermal aging results (mm2 K/W)

Fig. 24

Raytheon thermal cycling (low ramp rate) results (mm2K/W)

Fig. 25

GE thermal cycling (low ramp rate) results (mm2 K/W)

Fig. 26

Teledyne thermal cycling (low ramp rate) results (mm2K/W)

Fig. 27

Raytheon thermal cycling (high ramp rate) results (mm2 K/W)

Fig. 28

GE thermal cycling (high ramp rate) results (mm2 K/W)

Fig. 29

Teledyne thermal cycling (high ramp rate) results (mm2K/W)

Fig. 30

High thermal conductivity fillers in metal matrix

Fig. 31

Comparison of Cu/fBNNS with SOA TIMS

Fig. 32

Metal nanowire array TIM [25]

## Tables

Table 1 NTI team approach
Table 2 NTI metrics
Note: Additional comments: device/substrate is assumed to be Si/Cu, although other combinations are allowed for specific applications. Surface preparation should be no more complex than simple O2 plasma. Storage should not be more complex than temperature of −40 °C in dry package. After processing, TIM shall not require additional long-term static pressure beyond requirements for assembly and fixtures.
aUnits defined as thermal resistance (C/W) normalized by area. Results must include Rth of TIM and both interfaces.
bTen samples tested, all below thermal resistance specification for phase at all times. Ramps at < 3 °C/min (slow) and > 25 °C/min (fast).
cDisassemble and reassemble substrate/TIM/substrate N times and meet all other program requirements for Phase.
dSteady operation at 130 °C and meet Rth requirement for Phase. Degradation is based on Rth measurements before and after test for specified duration at 130 °C.
eApplication time includes all surface prep, attachment, and curing.
f1 cm2 die must sustain shear force requirement while maintaining 100% attach.
Table 3 Initial sample thermal resistance (mm2 K/W)
Table 4 GT thermal aging typical results (mm2 K/W) 300 hrs at 130 °C
Table 5 GT thermal cycling (low ramp rate) results (mm2 K/W)
Table 6 GT thermal cycling (high ramp rate) results (mm2 K/W)
Table 7 Initial thermal resistance results
Table 8 Raytheon thermal aging results (mm2 K/W) 300 hrs at 130 °C
Table 9 GE thermal aging results (mm2 K/W) 300 hrs at 130 °C
Table 10 Teledyne thermal aging results (mm2 K/W) 300 hrs at 130 °C
Table 11 Raytheon thermal cycling (low ramp rate) results (mm2 K/W)
Table 12 GE thermal cycling (low ramp rate) results (mm2 K/W)
Table 13 Teledyne thermal cycling (low ramp rate) results (mm2 K/W)
Table 14 Raytheon thermal cycling (high ramp rate) results (mm2 K/W)
Table 15 GE thermal cycling (high ramp rate) results (mm2 K/W)
Table 16 Teledyne thermal cycling (high ramp rate) results (mm2 K/W)

## Discussions

Some tools below are only available to our subscribers or users with an online account.

### Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections