Heat Transfer and Fluid Flow Investigations in PDMS Microchannel Heat Sinks Fabricated by Means of a Low-Cost 3D Printer

Abstract

Polydimethylsiloxane (PDMS), due to its remarkable properties such as optical transparency and ability to easily mold, is one of the most popular polymers used in micro- and nanofluidics. Furthermore, 3D printing technology due to its low cost and simplicity is also gaining a great interest among the microfluidic community. In this work, the potential of 3D printing is shown to produce microfluidic devices, their ability for studying flows and heat transfer of nanofluids, and their applicability as a heat sink device. The low-cost fused deposition modeling 3D printing technique was combined with a PDMS casting technique for the microfluidic device fabrication. The potential of this technique was experimentally demonstrated by fluid flow and heat transfer investigations using different fluids, such as distilled water-, alumina (Al2O3)-, and iron oxide (Fe3O4)-based nanofluids. The simplicity, low-cost, and unique features of the proposed heat sink device may provide a promising way to investigate nanofluids’ flow and heat transfer phenomena that are not possible to be studied by the current traditional systems.

1. Introduction

The continuous investigation on strategies for size reduction while maintaining, or even increasing, power of technological devices demands cooling systems with higher thermal efficiency and smaller sizes. One approach relied on the modification of heat sinks by incorporating microchannels to increase the heat exchange surface. Despite being a well-adopted strategy, its complex configurations proved to be difficult to manufacture [1, 2]. Other strategies focused on the type of the used fluids for the cooling process. Dielectric fluids [1, 3], two-phase fluids [3, 4], and nanofluids (NFs) [5, 6, 7, 8] have been thoroughly investigated. Among these three, the nanofluids, which are fluids comprised of particles, with size ranging from 1 to 100 nm, suspended in a base fluid, have been reported as presenting a better thermal conductivity than the base fluid [8, 9]. Nevertheless, some challenges regarding their usage, such as agglomeration, long-term stability, and high-costs, still need further investigation [8, 10, 11, 12]. The most commonly used nanoparticles (NPs) for NFs are metallic, such as Cu, Ag, Au, and Fe, or non-metallic such as Al₂O₃, CuO, TiO₂ SiC, and carbon nanotubes. Many authors reported that the heat transfer of NFs is influenced by several factors such as shape, dimensions, volume fractions in the suspensions, and the thermal properties of the particle materials [11, 13, 14]. Therefore, enhancement in heat transfer was also reported, but only for small concentrations of NPs [8, 9].

Abareshi et al. [15] have evaluated the thermal conductivity of nanofluids with different volume fractions of Fe₃O₄ NPs, at different temperatures. The thermal conductivity was reported to increase up to 11.5% for a volume fraction of 3 vol% at 40°C. Xia et al. [16] have investigated the heat transfer coefficient of nanofluids using TiO₂ and Al₂O₃ NPs, with different volume fractions. For a volume fraction of 1%, the heat transfer coefficient was significantly increased for both nanofluids, compared with deionized water. Gavili et al. [17] have studied the thermal conductivity of ferrofluids with Fe₃O₄ particles of approximately 10 nm in diameter, suspended in deionized water. With the application of a magnetic field, the thermal conductivity was increased up to 200% for a 5% volume fraction. Kim et al. [18] have investigated the thermal conductivity of alumina- and distilled water-based nanofluids, with concentrations of 0.5, 1, and 2 wt%. The conductivity was found to increase with the increasing of the NPs concentration. Al-Rjoub et al. [19] have tested four cooling liquids: deionized water; distilled water; borax buffer; and Al₂O₃ NPs solution, on a microscale heat exchanger. It was found that the deionized water has presented the lowest heat removal capacity, while the Al₂O₃ solution showed the highest capacity, corresponding to about 69% increase.

The majority of microchannel heat sink devices that can be found in literature were fabricated in silicon, due to its thermal conductivity. However, the fabrication process of those devices can be laborious and needs extremely expensive facilities. Novel, fast, and low-cost fabrication techniques have been developed by means of different kinds of polymers [20, 21, 22]. PDMS is a silicone elastomer with a set of properties that make it suitable for many applications and is a popular choice for microfluidic devices fabrication [21, 22]. Besides being cheaper than the monocrystalline silicon, it presents a low elasticity change versus temperature, high thermal stability, chemical inertness, dielectric stability, shear stability, high compressibility, and hyperelasticity [23, 24, 25, 26, 27]. Moreover, it is non-toxic and biocompatible [25, 27, 28, 29]. PDMS devices can be manufactured by simple techniques at room temperature, such as replica molding.

The 3D printers are gaining an increased attention by both academic and industrial community to produce microdevices and models at an extremely low cost. Some successful applications can already be found in lab-on-a-chip tools [30], microfluidics [31, 32], and biomedical in vitro devices [27, 32, 33, 34]. There are different kinds of printing methods such as the Fused Deposition Modeling (FDM) and stereolithography [32]. Among those methods, the most popular, simple method with the lowest cost is the FDM technology [32]. For this reason, the FDM process was combined with a PDMS casting technique to produce a PDMS microfluidic device. Hence, the main objective of this work is to show the potential of a FDM 3D printer to produce microfluidic devices and their potential to be used to perform flows and heat transfer studies of nanofluids. To demonstrate the potential of this technique, fluid flow and heat transfer studies were performed by using different fluids such as distilled water-, alumina ( ${\mathit{Al}}_2{O}_3$ )-, and iron oxide ( ${\mathit{Fe}}_3{O}_4$ )-based nanofluids with concentrations of 1 and 2.5%.

2. Materials and methods

This section describes the experimental protocol to develop the nanofluids and the fabrication process to produce the heat sinks, as well as the used experimental setup.

2.1 Nanoparticles preparation

Two different types of NPs were used on the flow and heat experiments: alumina oxide ( ${\mathit{Al}}_2{O}_3$ ) NPs acquired from Sigma Aldrich (ref. 702,129, Sigma Aldrich) and iron oxide ( ${\mathit{Fe}}_3{O}_4$ ) NPs synthesized on our laboratory by co-precipitation. This last method allows to produce magnetic iron oxide NPs in a cost-effective way and is appropriate for mass production. The ${\mathit{Al}}_2{O}_3$ NPs had a size inferior to 50 nm, while the synthesized ${\mathit{Fe}}_3{O}_4$ NPs had an average size of 11 ± 2 nm.

The co-precipitation was initiated with the preparation of the precipitation agent by adding 0.01 g of cetrimonium bromide (CTAB), diluted in 3 mL of distilled water, to 20 mL of ammonium hydroxide ( $N{H}_4\mathit{OH}$ ). Hereinafter, a ferrous solution was prepared by diluting 7.78 g of iron(III) chloride ( $\mathit{Fe}{\mathit{Cl}}_3$ ) and 4.06 g of iron(II) chloride ( $\mathit{Fe}{\mathit{Cl}}_2$ ) in 20 mL of distilled water, in an ultrasonic bath. The solution was subsequently mechanically stirred at, approximately, 1500 rpm. The precipitation solution was then added, dropwise, to the ferrous solution under stirring, on a laminar flow cabinet. The co-precipitation occurred according to the following equation:

$$\mathit{Fe}{\mathit{Cl}}_2.4H_2{O}_{\left(l\right)}+2\mathit{Fe}{\mathit{Cl}}_3.6H_2{O}_{\left(l\right)}+8N{H}_4{\left(\mathit{OH}\right)}_{\left(\mathit{aq}\right)}={\mathit{Fe}}_3{O_4}_{\left(s\right)}+8N{H}_4{\mathit{Cl}}_{\left(\mathit{aq}\right)}+20H_2{O}_{\left(l\right)}$$ E1

To conclude the process, the NPs were washed several times with distilled water with the assistance of a strong magnet.

Figure 1 shows the synthesized NPs and the representative transmission electron microscopy (TEM) image. A detailed description of this process can be found at Cardoso et al. [35].

Figure 1.

Magnetic iron oxide NPs and representative TEM image, adapted from [35, 36].

Despite the TEM images show the aggregates of ${\mathit{Fe}}_3{O}_4$ NPs, in the NF, they were stable and non-aggregated for a period of time after 10 min of ultrasonic bath.

The NPs show a normalized magnetization of ~69 emu.g 1 at ~10 kOe, which corresponds to the saturation magnetization, and also show a superparamagnetic behavior with an extremely low coercivity of 1.6 Oe [35, 36, 37].

2.2 Fabrication of the heat sink microchannel

The heat sink microchannel device was produced based on a scaffold-removal technique [25]. First, the molds were drawn by using the Autodesk Inventor® software and then printed at the FDM 3D printer Ultimaker 2+ (Ultimaker, Netherlands). The first mold was printed with acrylonitrile butadiene styrene (ABS), whereas the second one was printed with polylactic acid (PLA). The fabrication of the molds was performed with a nozzle with a diameter of 0.4 mm, whereas the layer resolution was about 100 μm. The main dimensions of the ABS master mold can be found in Figure 2.

Figure 2.

Schematic representation of the main dimensions of the ABS master mold.

Once the 3D models were printed, PDMS was prepared by adding a PDMS curing agent into the pre-polymer with a mixing ratio of 1:10. The PDMS was poured onto the PLA mold with the ABS master mold inside it. Once the PLA mold was filled with PDMS, the elastomer was cured at room temperature for about 1 day. Finally, the PDMS was removed from the PLA mold and immersed in an acetone bath to remove the ABS for approximately 24 h. Figure 3 shows the schematic diagram of all the main steps to produce the PDMS heat sink device. The overall cost to fabricate the PDMS heat sink device is about 3.8 €. This cost includes the printing of the ABS master mold (∼1 €) and PDMS casting process (∼2.8 €).

Figure 3.

Schematic representation of main steps to fabricate the PDMS heat sink.

Notice that after the PDMS curing process, small holes were made below the inlet and outlet to insert the thermocouples (type K). Figure 4 shows the PDMS heat sink microfluidic device used in the flow and heat experiments.

Figure 4.

PDMS heat sink microfluidic device with the inserted thermocouples.

2.3 Experimental procedures

The PDMS heat sink was placed on top of a hot plate controlled by a 9400-temperature controller (CAL Controls). The temperature of the plate was set to 60°C, whereas the flow rate of the fluids was controlled by a syringe pump (Harvard) connected to the inlet of the heat sink. The temperature at the entrance and exit of the device was acquired through a data acquisition instrument connected to the thermocouples of the device. Wood and polystyrene blocks were used to minimize the heat losses. Figure 5 shows a schematic diagram of the experimental setup. The flow of the ${\mathit{Fe}}_3{O}_4$ nanofluid was analyzed by optical microscopy at a flow rate of 10 μL/min. Note that the temperature measurements presented an uncertainty of ±1°C. The thermographic studies were performed with distilled water at a flow rate of 1, 5, 7.5, 10, and 15 mL/min and a thermographic camera Onca-MWIR-InSb (Xenics Infrared Solutions). The setup and calibration procedures of the camera were performed as in Teodori et al. [38]. Images, with a resolution of 150px × 150px, were taken with a frame rate of 1000 fps.

Figure 5.

Schematic representation of the experimental setup.

To evaluate the influence of the nanofluids properties in the heat sink microfluid device, the tests were performed using distilled water, ${\mathit{Fe}}_3{O}_4$ at a concentration of 1 and 2.5%, and ${\mathit{Al}}_2{O}_3$ at the same concentrations. All the fluids were set to a flow rate of 1 up to 30 mL/min.

2.4 Heat transfer calculations

The properties of the nanofluids were obtained taken into account fundamental equations described on previous studies [39, 40]. The thermal conductivity of the nanofluid was obtained according to the Maxwell model described by the following Equation [39, 40]:

$$K_{\mathit{nf}}=K_{\mathit{bf}}\left(\frac{K_p+2K_{\mathit{bf}}+2\phi \left(K_p-K_{\mathit{bf}}\right)}{K_p+2K_{\mathit{bf}}-2\phi \left(K_p-K_{\mathit{bf}}\right)}\right),$$ E2

where $K_{\mathit{nf}}$ is the nanofluid thermal conductivity, $K_p$ is the NPs thermal conductivity, $K_{\mathit{bf}}$ is the base fluid thermal conductivity, and φ is the NPs concentration.

The nanofluid density and heat capacity were calculated through the weighted average of the individual properties of both the NPs and base fluid. The first is expressed by Eq. (3) [14, 41] and the latter by Eq. (4) [11, 14].

$${\rho }_{\mathit{nf}}={\rho }_p\phi +{\rho }_{\mathit{bf}}\left(1-\phi \right),$$ E3

$$c_{p_{\mathit{nf}}}=\phi {\rho }_p{c}_{p_{\mathit{np}}}+\left(1-\phi \right){\rho }_{\mathit{fb}}{c}_{p_{\mathit{bf}}}$$ E4

In the abovementioned equations, ${\rho }_{\mathit{nf}}$ represents the nanofluid density, ${\rho }_p$ the particle density, ${\rho }_{\mathit{bf}}$ the base fluid density, and $c_{p_{\mathit{np}}}$ and $c_{p_{\mathit{bf}}}$ the specific heat capacity of the NPs and of the base fluid, respectively.

The nanofluid viscosity was determined through the equation proposed by Batchelor [42]:

${\mu }_{\mathit{nf}}=\left(1+2.5\phi +6.2{\phi }^2\right){\mu }_{\mathit{fb}}$ , $0<\phi <10\%$ [5]

where ${\mu }_{\mathit{nf}}$ is the nanofluid viscosity and ${\mu }_{\mathit{fb}}$ the base fluid viscosity. This equation brings in a quadratic dependence with the volume fraction, which provides a better representation of the interaction between the particles on the fluid.

Within the microfluidic device, the heat transfer will occur by convection inside the microchannels and by conduction in the walls between them. Consequently, the mathematical approach that better allowed the evaluation of the heat transfer was described by Ma et al. [10]. The convection heat transfer coefficient was calculated by iterations using the following equations:

$$h=\frac{Q}{N\left(A_b+2\eta A_l\right)\left(T_b-T_{\mathit{\text{avgf}}}\right)},$$ E5

$$m_i=\sqrt{\frac{2h_{i-1}}{k_{\mathit{\text{PDMS}}}.w_p}},$$ E6

$${\eta }_i=\frac{\mathit{th}\left(m_i{z}_c\right)}{m_i{z}_c},$$ E7

where $A_b$ represents the area of microchannel bottom, $A_l$ the area of microchannel sidewall, $T_{\mathit{\text{avgf}}}$ the average temperature of the fluid, $\eta$ the fin efficiency, $w_p$ the average width of fin, and $z_c$ the height of the channel. The main dimensions of the heat sink are illustrated in Figure 2. The heat transfer rate, $Q$  , represents the amount of heat energy taken by the fluid when it flows through the channels and is given by Eq. (9). This parameter was obtained for each mass flow rate, $\dot{m}$ , after the temperature at the inlet, $T_{\mathit{\text{in}}}$ , and at the outlet, $T_{\mathit{\text{out}}}$ , were measured.

$$Q=q{\rho }_f{c}_{\mathit{pf}}\left(T_{\mathit{\text{out}}}-T_{\mathit{\text{in}}}\right),$$ E8

where $q$ , $c_{p_f}$ , and ${\rho }_f$ is the volume flow rate, specific heat capacity, and density of the working fluid, respectively.

3. Results and discussion

In this section, the obtained results are presented and discussed. The temperature measurements and the known properties of the materials were used to calculate the parameters described on the previous section. Using those parameters, the influence of the environment conditions and of the fluid properties in heat transfer was analyzed.

3.1 Influence of the nanofluid properties

As described previously, the prepared nanofluids with iron oxide ( ${\mathit{Fe}}_3{O}_4$ ) and alumina ( ${\mathit{Al}}_2{O}_3$ ) NPs were used to verify the thermal properties influence of the nanofluids in the heat transfer performance of the developed PDMS heat sink device.

Figure 6 shows the temperature difference between the inlet and the outlet as the tested nanofluids flow through the proposed PDMS heat sink. Overall, it is possible to conclude that both tested nanofluids present a bigger temperature difference between the inlet and the outlet in comparison with distilled water. Hence, these measurements indicate that the amount of heat energy absorbed by both nanofluids was bigger than that absorbed by the distilled water. In addition, the amount of heat absorbed by the nanofluids was found to be bigger for smaller flow rates, as shown in Figure 6. These results corroborate the measurements performed by Chein and Chuang [43], where they have investigated the heat performance of nanofluids with CuO NPs in a microchannel heat sink. These results also show that the flow rate affects the amount of heat absorbed by the nanofluids.

Figure 6.

Temperature difference between the inlet and the outlet for the tested nanofluids at the proposed PDMS heat sink microfluidic device.

In Figure 7, an increase of the convective heat transfer coefficient was registered on both nanofluids, in comparison to the base fluid. The increase was more pronounced for the alumina nanofluid due to the greater stability of the nanofluid and bigger thermal conductivity of these kind of NPs [6]. By increasing the convective heat transfer coefficient, an increase of the heat transfer rate is expected. In Figure 8, it is possible to observe an increase in the heat transfer rate for both tested nanofluids in comparison with distilled water. From these results, it is also possible to conclude that for both nanofluids, the convective heat transfer coefficient and heat transfer rate increase with the flow rate, which agrees with the results obtained by Wen and Ding [44].

Figure 7.

Convective heat transfer coefficient of distilled water, alumina, and iron oxide nanofluids as the function of the flow rate.

Figure 8.

Heat transfer rate of distilled water, alumina, and iron oxide nanofluids as the function of the flow rate.

Figure 9 compares the convective heat transfer coefficient for two different concentrations of NPs, i.e., 1 and 2.5% of both ${\mathit{Fe}}_3{O}_4$ and ${\mathit{Al}}_2{O}_3$ . From a macroscopic view, it was noted a better dispersion for the nanofluids containing 1% of NPs. In addition, by increasing the concentration, the heat transfer was not enhanced. In fact, it was noted a decrease of the convective heat transfer coefficient when the concentration of NPs was increased to 2.5%. Although these results are somewhat paradoxical, other researchers, such as Wen and Ding [45] and Putra et al. [46], have reported similar results. The main possible reasons for the seen heat transfer deterioration include both the aggregation and sedimentation of the NPs. However, the possible reasons and mechanism attributed to such phenomena require further research. Currently, our group is carrying out both experimental and numerical work to identify the exact causes for such phenomena.

Figure 9.

Convective heat transfer coefficient of nanofluids as the function of the flow rate, for two different concentrations of alumina and iron oxide NPs.

3.2 Optical and thermal imaging analyses

The biggest advantage of the developed PDMS heat sink device is the ability to visualize the flow phenomena happening inside the microchannels. By using a high-speed video microscopy system, it was possible to visualize several flow phenomena of the nanofluids such as the formation, growing, and breakdown of NPs clusters (see Figure 10). From these observations, it was concluded that one of the main causes for the formation of the clusters was the high roughness of the PDMS surface channels (Figure 11). This verified roughness was caused by the ABS master mold fabricated by the FDM 3D printer. In order to improve the surface roughness, the ABS master molds should undergo an acetone vapor treatment before performing the PDMS casting procedure. More detailed information about this method can be found elsewhere [27].

Figure 10.

Optical images of the formation, growing, and breakdown of a cluster of NPs.

Figure 11.

Optical image of the surface roughness of the heat sink device used in this study.

Another interesting advantage of this PDMS microfluidic device is the ability to visualize both the flow and thermal performance of the system by using a thermographic camera, as shown in Figure 12. Notice that the temperatures acquired were from the surface of the heat sink device and not directly from the working fluid flowing in the microchannels. In fact, PDMS is transparent to visible radiation but partially opaque to the infrared (IR) radiation. Hence, to obtain the temperatures more closely related to the working fluids flowing through the microchannels, the thickness of the upper walls should be reduced in future experiments. A very interesting observation was the ability to detect bubbles that are likely to happen in microfluidic devices. In Figure 13, it is possible to visualize a bubble within the microchannel and the thermal performance of the heat sink device.

Figure 12.

Temperature gradient analyzed through the thermographic camera at three different instants: t = 0 s; t = 60 s, and t = 120 s.

Figure 13.

Formation of a bubble at the entrance of the microchannels, which affects the thermal performance of the microfluidic device.

3.3 Limitations and future directions

In this study, a microchannel device was successfully manufactured and used in microfluidic essays. Nevertheless, some limitations arose throughout the work. Despite the advantages of using PDMS, some properties of the material, such as the low conductivity and partial opacity to the infrared (IR) radiation, were not favorable for the experiments. Also, the walls of the heat sink were rough due to the ABS master mold used in the fabrication technique, and the thickness should be reduced. The stability of the nanofluids has yet to be optimized since the deposition of NPs was detected on the heat sink walls. Future works will aim to improve of those limitations.

4. Conclusion

The main objective of this work was to show the potential of a PDMS heat sink microfluidic device to perform flows and heat transfer studies of nanofluids. The PDMS heat sink device was produced by using the FDM 3D printing process, combined with a PDMS casting technique. This fabrication process allowed to manufacture devices in an easy, low-cost, and reasonable reproductively way. To demonstrate the potential of the produced PDMS heat sink device, fluid flow and heat transfer studies were performed by using two different nanofluids, i.e., alumina ( ${\mathit{Al}}_2{O}_3$ )- and iron oxide ( ${\mathit{Fe}}_3{O}_4$ )-based nanofluids with concentrations of 1 and 2.5%. Overall, it was found that the thermal performance of the working nanofluids is in good agreement with several past studies. For instance, it was noted that the heat energy absorbed by both nanofluids was higher than that absorbed by the distilled water. In addition, it was found that the flow rate affects the amount of heat absorbed by the nanofluids. However, the most interesting and unique results were the optical and thermal imaging results. These results were only possible due to the optical transparency of the PDMS heat sink device. Hence, by using this device, it was possible to visualize several flow phenomena of the nanofluids such as the formation, growing, and breakdown of NPs clusters. From these latter observations, it was possible to conclude that one of the main causes for the formation of the clusters was the high roughness of the PDMS surface channels caused by the surface roughness of the ABS master mold fabricated by the FDM 3D printer. This drawback can be overcome by performing an acetone vapor treatment before performing the PDMS casting. Overall, the simplicity, low-cost, and unique features of the proposed PDMS heat sink microfluidic device may prove a viable alternative tool to investigate nanofluids flow and heat transfer phenomena that are not possible to be performed by the current traditional systems.

Acknowledgments

This work was supported by Fundação para a Ciência e a Tecnologia (FCT) under the strategic grants UID/EMS/04077/2019, UID/EEA/04436/2019, and UID/EMS/00532/2019. The authors are also grateful for the funding of FCT through the projects POCI-01-0145-FEDER-016861, POCI-01-0145-FEDER-028159, NORTE-01-0145-FEDER-029394, and NORTE-01-0145-FEDER-030171, funded by COMPETE2020, NORTE2020, PORTUGAL2020, and FEDER. The authors also acknowledge FCT for partially financing the research under the framework of the project UTAP-EXPL/CTE/0064/2017, financiado no âmbito do Projeto 5665-Parcerias Internacionais de Ciência e Tecnologia, UT Austin Programme. Ana S. Moita also acknowledges FCT for her contract int the context of the recruitment programme FCT Investigator (IF/00810/2015) and exploratory project associated with it.

Conflict of interest

The authors declare no conflict of interest.

A	area
cp	heat capacity
h	convection heat transfer coefficient
K	thermal conductivity
N	number of microchannels
q	volume flow rate
Q	heat rate
R	resistance
Rt	thermic resistance
T	temperature
wp	average width of fin
zc	height of the channel
Greek symbols
η	fin efficiency
μ	viscosity
φ	nanoparticle concentration
ρ	density
Subscript
b	microchannel bottom
bf	base fluid
ext.	exterior
f	fluid
in	inlet
l	microchannel sidewall
nf	nanofluid
out	outlet
p	nanoparticles