Bulk and localized plasmonic heating in nanogold doped polymers

Metal nanoparticles (NPs) exhibit a surface plasmon resonance frequency^1–4 that is affected by NP size, shape, composition, state of aggregation, and environment.^5–14 Nonpropagating surface plasmon modes within the electromagnetic spectrum enable the conversion of optical energy to thermal energy with a coincident temperature change within the surrounding media. Previous hypotheses state that gold NPs (GNPs) with diameters above 2 nm have a large extinction cross section, enabling up to 100% light-to-heat conversion efficiency.¹⁵ Incorporation of metallic NPs in a polymer matrix has been used for efficient heat generation¹⁶ and as localized heat sources;¹⁷ and metallic NPs linked with polymer have been used to monitor the temperature change.¹⁸ Biomedical applications for metallic NPs in polymer is a growing field with applications including iron oxide NPs used as drug carriers to target cancer tissues,¹⁹ gold NPs used for heat ablation of tumors upon light exposure,^16,20,21 enhanced light capture of antibacterial nanogold doped polymers,²² and antibacterial polymers containing stabilized silver NPs embedded in polymers for antibacterial applications.²³ The photothermal process experienced by metallic NPs has also been used as a nanoscale experimental tool,^16,24–26 for controlled chemical release,²⁷ vapor deposition,^28–30 catalysis,³¹ and phase transition monitoring.³² 

In our work, we have incorporated ∼3-nm-diameter GNPs and 40 nm by 15 nm gold nanorods (GNRs) into polymer blend thin films and electrospun (ES) fibers to utilize the nanomaterial plasmonic response for localized heating. The plasmonic response of gold NPs involves electrons that oscillate at resonance with incident light on the particle surface.¹⁵ As the plasmon frequency is typically comparable with the visible light spectrum, it can be predicted that the use of plasmonic NPs or nanorods would enhance the photothermal conversion efficiency of materials, especially when illuminated at these wavelengths. In previous work, we embedded inexpensive GNPs in ES fiber mats to create a light-activated, on-demand drug delivery system and measured a temperature change up to 9 °C at very low concentrations.³³ The GNRs used in this work exhibit a tunable longitudinal surface plasmon resonance that depends on the GNR aspect ratio.^34–38 Metal nanomaterials in polymer can be activated remotely through light illumination and tuned specifically based on the wavelength of the excitation source.¹⁷ The plasmonic photothermal effect of metallic nanomaterials in a polymer is dependent on thermal properties of the immediate environment surrounding the nanomaterial,^16,39–42 as well as the overall system.¹⁷ In polymer thin films, metallic nanomaterials activated with light distribute heat throughout the thin film. However, ES polymer fibers provide an enhanced surface area for light exposure and confine the surrounding polymer closer to the particle surface.¹⁷ 

Here, we present polyethylene glycol/polyethylene oxide (PEG/PEO) blend polymer thin films and ES fiber mats doped with GNPs or GNRs. Although heating in the bulk polymer films was low, it was assumed the local heating/melting around each GNP could be much higher. In this work, mathematical modeling was used to describe the GNP distribution and heat/melt profile surrounding each GNP in the polymer blends. Using resistance temperature detectors (RTDs), the temperature change of the GNP-doped thin films was compared to that of GNP-doped ES fibers under laser interrogation. In addition, heating in bulk polymer blend films containing GNPs or GNRs was mapped using an infrared camera system with light-emitting diodes (LEDs) at 530 and 810 nm. Efficiency enhancement of the polymer materials due to the nanogold dopants was calculated, considering convective heat transfer that occurred during sample illumination.

Transmission electron microscopy (TEM) samples were prepared by the Langmuir-Schaefer technique. The GNRs were dispersed into chloroform, and a small amount was carefully deposited on top of water in a vial. As the chloroform evaporated, the GNRs were able to self-assemble on the surface of the water; a TEM grid was then dipped onto the GNRs on water, parallel to the interface. TEM images were collected with a JEOL 2100 (JEOL, Peabody, MA) at 200 keV.

PEG/PEO (PEG: 8000 MW, PEO: 200 000 MW) polymer blends were prepared by simple mixing in a beaker with a stir bar on a hot plate. The hot plate was set to 100 °C with stirring at 300 rpm. In our previous work, we embedded ES polymer mats with GNPs for localized heating in the fibers.³³ However, while PEG-only materials doped with GNPs increased to the highest temperature measured (8.9 °C), ES PEG produced a pastelike substance that was not easily peeled from the deposition surface for distribution. Various PEG/PEO blends were prepared and tested, and the 3:2 PEG to PEO blend used for this work was determined to provide distributable material properties while heating measured was still significant (7.4 °C). When doped with gold nanomaterials, dry polymers were directly added to the gold nanomaterial colloid in de-ionized water.

GNPs in reagent grade de-ionized water were purchased commercially at www.purestcolloids.com/mesogold for $0.10/ml. GNPs purchased did not contain gold ions, were sterilized prior to purchase, were stabilized with citrate, and contained an average NP diameter of 3.2 nm at a concentration of 10 ppm when blended with polymers.

GNR synthesis has been described in detail previously.^34,43 GNRs are synthesized by a secondary (seeded) growth procedure that allows control over the aspect ratio. Control of the aspect ratio allows for direct tuning of the longitudinal surface plasmon resonance exhibited by the rods.

Electrospinning was performed on a vertical Spraybase® tabletop unit. Materials were electrospun directly onto resistance temperature detectors (RTDs) for monitoring the subsequent temperature change. RTD wires were protected during spinning by parafilm. Electrospinning parameters included a separation distance of 9 cm, an applied voltage of 6.8 kV, polymer delivery at 0.4 ml/h for 5 min, and a 20-gauge stainless steel needle.

Thin films were prepared by spin coating at 10 000 rpm for 1 min and heating for 10 s on a hot plate set to 100 °C. The substrate used during thin film deposition was a polyester transparency film. Thin film deposition onto RTDs was carried out in the same manner; however, RTD wires were protected by parafilm during deposition.

Scanning electron microscopy was performed on a Tescan Mira 3. Samples were gold coated in a sputter coater and imaged at a z-height of 10 mm.

A forward-looking infrared (FLIR) camera system (FLIR A35sc) was used to map heating throughout polymer thin films doped with GNPs or GNRs. The irradiance power of the 530 and 810 nm LEDs used during interrogation was 2000 W/m² with a spot size diameter of approximately 8–9 mm. Power levels for each LED source were set to 2000 W/m² with an absolute optical power meter from Thorlabs. FLIR ResearchIR software was used to measure the temperature values of the sample under test. The maximum measured temperature in the heated region was compared to the sample temperature preillumination to determine the resultant heating magnitude. Temperature measurements using a per-pixel analysis provided a temperature resolution of 0.001 °C.

Resistance measurements were taken under laser illumination on a 2450 Keithley source meter. Reference values were taken from PEG/PEO only films or fibers under the laser. The laser used for interrogation was green with a wavelength of 532 nm and a full-scale power of 20 mW with a spot size diameter of 1 mm. The laser beam was contained completely on the RTD, almost completely covering the surface. The RTD used was 2.0 mm width by 2.0 mm length with a nominal resistance of 1000 Ω. Resistance measurements were converted to temperature using the Callendar–Van Dusen equation

where $Rref$ is the reference resistance measured, A and B are Callendar–Van Dusen constants (3908 × 10⁻³ and 5775 × 10⁻⁷, respectively), and T is the temperature.

From conservation of energy theory, the localized temperature change in close proximity to the plasmonic NPs must be much higher than the bulk temperature change. Each NP can be viewed as a heat source with photonic energy being converted to thermal energy through loss mechanisms in the plasmonic resonance of the NPs. The following theoretical analysis is an effort to estimate the mechanism of localized heating in close spatial proximity to the GNPs.

In the case where spherical colloidal GNPs were used, the volume of the GNP $VNP$ is given by

where $rNP$ is the radius of the GNP. For $rNP=1.5nm$⁠, $VNP=14.1nm3$⁠. Next, the mass of the GNP $mNP$ becomes

where $ρNP=19320kg/m3$⁠. The mass of a single GNP becomes $mNP=2.73×10−22kg$⁠. We also know that the concentration of NPs for the colloidal gold solution was 10 ppm (10 mg/l). The mass of PEO and PEG used in each 5 ml of composite solution was 0.2344 and 0.3908 g, respectively. Because of the similarities of relevant material properties between PEO and PEG, the properties of PEG were used in place for all PEO/PEG blends with minimal error, and the subscript PEG refers to the blended polymer. The mass fraction of GNPs in the composite material $η$ is given by

where the mass of Au in 5 ml of solution and the mass of PEG were given by $mAu_5ml$ and $mPEG$⁠, respectively. For the GNP + PEG/PEO composite, $η=8.00×10−5$⁠.

For the colloidal gold composite material, the number of NPs per cubic micrometer of material $#NPs$ was calculated to be ∼318. It is helpful to calculate the mean separation between NPs in the composite. For simplicity, it was assumed that the GNPs were arranged in a cubic structure with one NP at each cube corner. Therefore, the particles have a center-to-center separation of ∼147 nm. Based on the plasmonic theory and the evanescent field strength of plasmonic energy,⁴⁴ the coupling between adjacent GNPs could occur when the separation between particles is <λ/2. For a peak absorption of λ = 530 nm, λ/2 = 265; therefore, the particles center-to-center separation of ∼147 nm suggests that it is reasonable to assume that coupling between adjacent GNPs is present in the analyzed samples.

Conservation of energy analysis in the initial photothermal heating before bulk heating is observed provides a useful tool to estimate localized temperature changes around the GNPs. The energy associated with the heat capacity of a solid $qhc$ is given by

where m, $cp$⁠, and $ΔT$ are the mass, specific heat, and change in temperature, respectively, of the solid to be analyzed. For gold, $cp=129J/(kg°C)$⁠, and for PEG at 19 °C $cp=1472J/(kg°C)$⁠. For a measured bulk composite temperature change of 0.2 °C with illumination from 100 mW/cm², if all the energy associated with the bulk temperature change was concentrated in the GNP, the GNP would experience a $ΔT=28000°C$⁠. This temperature change is obviously not physically reasonable given that PEG degrades at a temperature close to 60 °C, and no PEG degradation is observed. It is, therefore, necessary to analyze a spherical heat zone centered at the NP whereby the polymer can absorb energy through heat capacity associated with solid temperature change and energy associated with the heat of fusion in a phase change of the polymer.

The localized heating around GNPs was studied by Govorov and Richardson,¹⁶ where the change in temperature for photothermal heating from NPs $ΔT$ is given by

where Q is the heat generation from the NP, $kPEG$ is the thermal conductivity of the polymer, and r is the radial distance from the center of the GNP. This equation can serve as a useful tool to estimate the temperature distribution surrounding a GNP.

For an analysis of localized heating in a spherical area around the GNP, an effective volume $Veff$ will be defined as

where $reff$ represents the effective radius of a polymer heated zone. It is assumed that because of the small volume associated with the GNPs the energy associated with heating the GNP and the surrounding polymer can be approximated by assuming a spherical polymer volume with no NP at the center. The effective temperate change of this heated polymer sphere $ΔTeff$ is given by

where $ΔTmeasured$ is the measured bulk temperature change of the composite film. When the melting point of PEG is taken as 55 °C, the maximum temperature change achievable by the polymer before the onset of melting $ΔTsolid=32°C$⁠. Equating the energy associated with the bulk film measured temperature change to that of a uniformly heated sphere gives the effective radius of this heated polymer sphere $reff_solid$⁠,

The value of $reff_solid=16.7nm$⁠. This effective radius is reasonable and represents the radius of a heated polymer zone if no melting occurs. If melting occurs like reported by Maity et al.,³⁸ the heat of fusion $qf$ must be considered,

where m is the mass of the melted polymer and $hf$ (1.717 × 10⁵J/kg for PEG) is the heat of fusion of the polymer. When the heated spherical zone accounts for melting with a uniform temperature profile, the effective radius of melted polymer $reff$becomes

The solution for the effective radius considering polymer melting gives $reff=10nm$⁠, which is a reasonable melted volume and provides some level of confidence that no bulk melting of the composite film will be observed. An average, idealized GNP separation of 147 nm in the composite film indicates that the photothermal heating at each GNP will not be large enough for the melted zone of polymer centered at each GNP to extend to adjacent GNPs. In fact, on a volumetric basis, only 0.13% of the total composite sample is melted based on the present analysis.

As an analysis of the energy split between the solid heat capacity and the heat of fusion, the fraction of energy that is attributed to heating the polymer to the onset of melting $κ$ is given by

where $κ=0.22$ for the current analysis, meaning that 78% of the energy absorbed by the polymer is associated with the phase change of melted polymer.

The most compelling result of the present analysis is that with a measured bulk temperature change of 0.2 °C, a spherical volume of polymer with a 10-nm radius could be expected to melt. In cases where drugs were to be delivered through a core-shell ES fiber, the bulk temperature change induced can be rather small to induce an adequate melted heating zone on the nanoscale. Figure 1 shows a plot of Eq. (12) for the measured bulk temperature changes extending to 30 °C, just below the melting point of PEG when heated from a room temperature of 23 °C.

GNP-doped fibers [Fig. 2(a)] and films [Fig. 2(b)] were deposited directly onto RTDs, and measured resistance under laser interrogation was used to determine the change in temperature of the sample. PEO/PEG-only thin films or ES fibers were used as the reference material for these studies. GNPs encased in polymer fibers are surrounded by less material than in a thin film, and it has been predicted that heat flow away from the metallic nanomaterial is slow due to the small cross-sectional area.¹⁷ Based on these predictions, it was assumed the ES fiber mat would heat to a higher temperature than the bulk GNP-doped thin film. Results of these studies supported this prediction, showing a 72% increase in the temperature change in the GNP-doped ES fiber sample (ΔT = 1.72 °C) as compared to the GNP-doped thin film (ΔT = 1 °C).

A FLIR infrared camera system was used to map heating of polymer films real-time (Fig. 3). In all cases, the maximum temperature was reached in less than 10 s, and no additional heating was observed in the films at least up to 5 min. GNPs used had a maximum plasmon resonance at 522 nm, while maximum plasmon resonance of the GNRs varied depending on the aspect ratio tuned during synthesis (Fig. 4).^34,43 Tables I and II show the results of FLIR heating measurements.

Heat mapping results collected on the FLIR system were used to calculate the efficiency of the nanogold doped polymer films. Using a heat transfer analysis, convective heat flux for each of the films was determined, Eq. (13), and the efficiency was calculated, Eq. (14). The FLIR setup placed the sample in ambient conditions in the air. Under LED interrogation, the sample underwent convection on two sides. In Fig. 5, the sample setup and variables assumed in these calculations are shown.

Convective heat transfer for the films was determined by

where h is the convective heat transfer coefficient (5.0 W/m² K), A is the surface area of the illumination spot size (63.6 mm²), and $ΔT1$ is the change in temperature between the thin film sample and ambient environment. The free or natural convective heat transfer coefficient is a difficult value to estimate; and therefore, a reasonable value was assumed for these calculations.^45–47 Efficiency of the thin films $ε$ is reported in Table III and calculated using the following formula:

where $q˙photon$ was the irradiance power of the LED light source (2000 W/m² × 0.0635 m² = 127 mW). The irradiance power of the LED source was adjusted for the amount of light from the diode that was collected by the sample. In these equations, radiative heat transfer was assumed to be negligible due to the temperatures involved. It is also important to note that a majority of the illuminating light is not absorbed by the sample.

It was assumed that because the GNRs used exhibited a maximum surface plasmon resonance between 740 and 820 nm, the GNR-doped polymer materials would experience more efficient photothermal heating when illuminated at 810 nm as compared to illumination at 530 nm. Results from efficiency calculations show that while the (740, −) and (785, +) GNRs did not experience a change in efficiency when interrogated at either 530 or 810 nm, the GNR samples with maximum plasmonic resonance between 785 and 820 nm did experience increased efficiency when interrogated with the 810 nm LED. As expected in the case of the GNPs, maximum heating occurred under the 530 nm LED due to the GNP maximum plasmon resonance being at 522 nm and therefore closer to the interrogation LED wavelength. The change in efficiency compared to the polymer blend control is also included in Table II and shows significant efficiency enhancements in the nanogold doped films as compared to that of the polymer blend-only films. GNR-doped film efficiencies were observed to be up to 6.6 times that of the polymer blend-only film when interrogated at 810 nm, while the GNP-doped film efficiency increased by 1.8 times under the 530 nm LED.

Metallic nanomaterials in polymers exhibit a plasmonic response that can be used for applications ranging from photocatalysis to biomedical treatments. Here, the distribution and heating of GNPs in a polymer were modeled to better understand the relationship between bulk material heating and the localized heating and melting near the surface of the GNP. Results showed that a bulk temperature change of only 0.2 °C corresponds to a 10-nm radius melted spherical volume of polymer centered at each GNP. These results indicate that in cases where drugs are to be delivered through an ES drug delivery system, the bulk temperature change can remain small while still inducing an effective melt radius for drug delivery. In this work, a bulk polymer GNP film system was also compared to an ES GNP fiber system. It was demonstrated that by minimizing the material dimensions closer to that of the heated sphere by electrospinning high surface area nanoscale fibers, the temperature change of the bulk material increased by 72%. The GNPs used in this study were sourced commercially at low concentration to provide an economically attractive treatment delivery system. In addition to these materials, GNRs synthesized with a particular aspect ratio that tuned the GNR maximum surface plasmon resonance were also studied. GNR- and GNP-doped polymer thin films were interrogated with 530 and 810 nm LEDs and heat mapped using an FLIR camera system. Based on the bulk heating results from the FLIR system measurements, photothermal efficiency calculations of the nanogold doped polymer films were performed. Significantly, GNR-doped film efficiencies recorded were up to 6.6 times (558.6% increase) that of the polymer blend-only film when interrogated at 810 nm, while the GNP-doped film efficiency increased by 1.8 times (75.7% increase) under the 530 nm LED.

Research was sponsored by the Combat Capabilities Development Command Army Research Laboratory and was accomplished under Cooperative Agreement No. W911NF-15-2-0020. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Combat Capabilities Development Command Army Research Laboratory or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein. This material is based upon work supported by the National Science Foundation (NSF) under Grant No. OIA-1757351. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.