Measurements are shown indicating that the drying rate of nanochannels can be enhanced by up to 3 orders of magnitude relative to drying by vapor diffusion, and that the drying rate is independent of the relative humidity of the environment up to a relative humidity of more than 90%. Micromachined Pyrex glass nanochannels of 72 nm height and with sharp corners (corner angles 7 degrees) were used. Available theory shows that the sharp corners function as a low-resistance pathway for liquid water, siphoning (wicking) the water to a location close to the channel exit before it evaporates. The described phenomena are of importance for the understanding of drying processes in industry and agriculture. The introduction of sharp corners or grooves can furthermore be beneficial for the functioning of microheat pipes and capillary-pumped loops. DOI: 10.1103/PhysRevLett.95.256107 PACS numbers: 68.03.Fg Understanding the drying mechanism of porous materials is of importance in many industries such as the food, paper, pharmaceutical, and textile industry [1][2][3]. It has previously been observed that microporous media dried approximately 1 order of magnitude faster than can be expected from vapor diffusion alone [4,5]. Flow in liquid water films held on surfaces and flow of water held in corners or grooves were thought to cause this acceleration [6]. The contribution of film flow to drying has been experimentally investigated in cylindrical nanocapillaries, where it caused a tenfold increase of drying rate, [7] but the contribution of corner flow has never been investigated. Here we report on experiments using noncylindrical micromachined nanochannels to quantify corner flow.Drying results from three water transport mechanisms: vapor diffusion, film flow, and corner flow. To specifically investigate corner flow we designed an array of high aspect ratio (widthheight) noncylindrical channels of equal height but different width (Fig. 1). The three water transport mechanisms schematically are shown in Fig. 2. When corner flow dominates the drying process in a channel, the drying rate will depend on the inverse of the channel width, because the number of corners is independent of width but the total water volume inside the channel proportional with width. Drying due to film flow (for widthheight) and vapor diffusion in contrast does not depend on the channel width. Arrays of Pyrex channels, open on two sides, were manufactured in a clean room. Channels were wet etched (hydrofluoric acid) into one Pyrex wafer using a photolithographic mask. This wafer was bonded by thermal fusion to a second wafer which had access holes for filling. The channels were 4 mm long, 72:4 0:8 nm high (determined by AFM) and the width in the array differed from 2 to 30 m. The channel shape was an isosceles trapezoid of very high aspect ratio (width=height > 40) (Fig. 1 bottom). The angle of the sharp corner was determined to be 6:6 0:7 degrees by SEM measurements of bonded chips and by AFM measurements prior to bonding of the Pyrex plates. The shar...

Thermal Interface Materials (TIMs) are particulate composite materials widely used in the microelectronics industry to reduce the thermal resistance between the device and heat sink. Predictive modeling using fundamental physical principles is critical to developing new TIMs since it can be used to quantify the effect of particle volume fraction and arrangements on the effective thermal conductivity. The existing analytical descriptions of thermal transport in particulate systems do not accurately account for the effect of inter-particle interactions, especially in the intermediate volume fractions of 30%–80%. An efficient Random Network Model (RNM) that captures the near-percolation transport in these particle-filled systems, taking into account the inter-particle interactions and random size distributions, was previously developed by the authors. The RNM is computationally efficient compared to full field simulations and was demonstrated to match to within 5% of the full field simulations and to within 15% of the experimentally measured values. The RNM approach uses a cylindrical region to approximate the thermal transport within the filler particles and to capture the inter-particle interactions. This approximation is less accurate when the polydispersivity of the particulate system increases. In the present paper, a novel semi-spherical approximation to the conductance of the fillers is presented as an alternative to the cylindrical region approximation used earlier. The new semi-spherical model is compared to the cylindrical model in two and three dimensions. In two dimensions, the semi-spherical model and the cylindrical model were compared with Finite Element Model (FEM) results. The comparison showed that the temperature distribution of the semi-spherical model matched more closely to the FEM model than the temperature distribution of the cylinder model when the radius ratio of the two particles increases. In three dimension microstructures, the semi-spherical model and the cylindrical model were compared under various volume fractions. The comparison showed that thermal conductivities of the semi-spherical model were always higher than thermal conductivities of the cylindrical model and were in better agreement with existing experimental data for particulate TIMs at 58% volume loading.

Thermal interface materials (TIMs) are particulate composite materials widely used in the microelectronics industry to reduce the thermal resistance between the device and heat sink. Predictive modeling using fundamental physical principles is critical to developing new TIMs since it can be used to quantify the effect of particle volume fraction and arrangements on the effective thermal conductivity. The existing analytical descriptions of thermal transport in particulate systems do not accurately account for the effect of interparticle interactions, especially in the intermediate volume fractions of 30–80%. An efficient random network model (RNM) that captures the near-percolation transport in these particle-filled systems, taking into account the interparticle interactions and random size distributions, was previously developed by Kanuparthi et al. The RNM approach uses a cylindrical region to approximate the thermal transport within the filler particles and to capture the interparticle interactions. However, this approximation is less accurate when the polydispersivity of the particulate system increases. In addition, the accuracy of the RNM is dependent on the parameters inherent in an analytical description of thermal transport between two spherical particles and their numerical approximation into the network model. In the current paper, a novel semispherical approximation to the conductance of the fillers is presented as an alternative to the cylindrical region approximation used earlier. Compared with the cylindrical model, the thermal conductivities of the semispherical model are more closely to the finite element (FE) results. Based on the FE analysis, the network model is improved by developing an approximation of the critical cylindrical region between two spherical particles over which energy is transported. Comparing the RNM results with FE results and experimental data, a linear relationship of the critical parameter with the thermal conductivity ratio and the volume fraction was found that provides a more accurate prediction of the effective thermal conductivity of the particulate TIMs.

Thermal interface materials (TIMs) are particulate composite materials widely used in the microelectronics industry to reduce the thermal resistance between the device and the heat sink. Predictive modeling using fundamental physical principles is critical to developing new TIMs, since it can be used to quantify the effect of polydispersivity, volume fraction and arrangements on the effective thermal conductivity. A random network model that can efficiently capture the near-percolation transport in these particle-filled systems was developed by the authors, which can take into account the interparticle interactions and random size distributions. In this paper, a Java-based code is used to generate the microstructures at different volume fraction and different particle-size distribution (PSD). COMSOL was used to investigate the impact of polydispersivity on the effective thermal conductivity of particulate TIMs. The log-normal distribution was used to capture the filler PSD. From the simulation results, there exists an optimum value of the polydispersivity which has the largest thermal conductivity for a given volume fraction.

Predictive modeling using fundamental physical principles is critical to developing new Thermal Interface Materials (TIMs) since it can be used to quantify the effect of particle volume fraction and arrangements on the effective thermal conductivity.In our prior work, we described a Random Network Model (RNM) [2] that can efficiently capture the near-percolation transport in these particle-filled systems, taking into account the inter-particle interactions and random size distributions. The accuracy of the RNM is dependent on the parameters inherent in analytical description of thermal transport between two spherical particles, and their numerical approximation into a network model. In the present study, based on a finite element analysis, the network model is improved by developing an approximation of the critical cylindrical region between two spherical particles over which energy is transported. Taking the finite element (FE) results (generated by COMSOL TM ) as a reference, this critical parameter demonstrated a much larger effect on the thermal conductance of a single cell than other parameters (thermal conductivity ratio of fillers and matrix, ratio of radii of two neighboring particles and distance between two neighboring particles). Thirty different microstructures were generated using COMSOL TM at volume fractions of 40%, 50%, 58% and 70% respectively. Comparing RNM results with FE results and experimental data, a linear relationship of the critical parameter and the volume fraction was found that provides a more accurate prediction of the effective thermal conductivity of the particulate thermal interface material for a given volume fraction.