Development and Evaluation of an Expedited System for Creation of Single Walled Carbon Nanotube Platforms

Abstract

Single-walled carbon nanotubes (SWNT) have a strong and stable near-infrared (nIR) fluorescence that can be used to selectively detect target analytes, even at the single molecule level, through changes in either their fluorescence intensity or emission peak wavelength. SWNTs have been employed as NIR optical sensors for detecting a variety of analytes. However, high costs, long fabrication times, and poor distributions limit the current methods for immobilizing SWNT sensors on solid substrates. Recently, our group reported a protocol for SWNT immobilization with high fluorescence yield, longevity, fluorescence distribution, and sensor response, unfortunately this process takes 5 days to complete. Herein we report an improved method to immobilize SWNT sensors that only takes 2 days and results in higher fluorescence intensity while maintaining a high level of SWNT distribution. We performed surface morphology and chemical composition tests on the original and new synthesis methods and compared the sensor response rates. The development of this new method of attaching SWNT sensors to a platform allows for creation of a sensing system in just 2 days without sacrificing the advantageous characteristics of the original, 5-day platforms.

Graphical Abstract

1. INTRODUCTION

Single-walled carbon nanotubes (SWNT) are optical sensors for a variety of analytes, such as nitric oxide (NO), reactive oxygen species, insulin, and dopamine [1–13]. When excited, SWNT emit light in the near infrared (nIR) range and respond to their target analyte with a blue or red shift in their wavelength and/or a change in fluorescence intensity [1, 14–17]. Due to their emission wavelength being in the nIR region, where interference from water and blood is minimal, and the fact that SWNT do not photobleach, they are of interest to researchers as sensors for biological applications [18–20].

NO is of great importance in many aspects of cell health, from maintenance of normal tissue homeostasis to repair of tissue injury, which is why a NO sensor is highlighted in this manuscript. Despite the focus in this article on NO, the system should also work for other DNA wrapped SWNT sensors [21–31]. As is true about most analytes used in cell signaling, the concentration and distribution of NO released from cells during healthy and diseased states is largely unknown. Previous researchers have developed models to predict NO release, but a spatial temporal measurement of the analyte was not possible before our recently developed platform, which is based on immobilization of SWNT sensors on a glass substrate through an avidin−biotin interaction [2, 3, 32–35]. The results showed that an even distribution of SWNT sensors could be attached to a glass substrate without altering their fluorescence intensity, response time or longevity. However, this one-time-use platform takes 5 days to build, adding a lot of time onto each experiment. Therefore, we designed an alternative platform building protocol to decrease build time while maintaining the high concentration and even distribution of SWNT sensors on the platform. There are five steps to build the sensor platform, each step was independently optimized based on SWNT fluorescence, distribution, and sensing capabilities. Herein we discuss the optimized steps and the assays used to demonstrate similarity between the previous 5-day method and the newly developed protocol.

2. MATERIALS AND METHODS

2.1. Materials

SWNT ((6,5) chirality, 0.7–0.9 nm tube diameter, carbon < 95%, > 93% carbon as SWNT) was acquired from Sigma Aldrich. The ssDNAs ((AT)₁₅, and Biotinylated-(AT)₁₅) were obtained from Integrated DNA Technologies, Iowa, USA. (3-glycidopropyl)- trimethoxysilane (3-GPTMS), aspartic acid, and hydrogen peroxide (H₂O₂) were obtained from Sigma Aldrich. Horseradish peroxidase solution (HRP), sodium chloride (NaCl), phosphate buffered saline (PBS) and sodium bicarbonate (NaHCO₃) were purchased from Thermo Fisher Scientific. Sulfuric acid was acquired from Millipore Sigma. Acetic acid was purchased from Honeywell and Avidin was obtained from MP Biomedicals. Argon (99.999%) and NO (99%) gases were acquired from Matheson Tri-Gas, Inc. Unless otherwise indicated, all aqueous solutions were prepared using nano pure water.

2.2. Dispersion of SWNT with DNA Oligos

The wrapping of SWNT with (AT)₁₅ was prepared as previously described [15, 36, 37]. Briefly, ssDNA (2 mg mL⁻¹) ((AT)₁₅: Biotinylated-(AT)₁₅, 1:1) and SWNT at a 2:1 DNA: SWNT mass ratio were added into 0.1 M NaCl solution. The solutions were bath-sonicated (Bransonic, M2800H) for 10 min followed by 40 min ultrasonication treatment (QSonica Q125 Sonicator equipped with a 3 mm probe tip). The solution was centrifugated (Eppendorf Centrifuge 5415D) for 180 minutes at 16,100 RCF, the supernatant was collected and diluted 10-fold. The concentration was determined by UV/vis spectrometry (Beckman Coulter DU730) as previously described [1].

2.3. Preparation of Sensor Arrays

The procedure, adapted from our recently reported paper, was modified to shorten the preparation time of avidin derivatized glass slides [32]. Detailed experimental procedure is illustrated in Figure 1. There are five steps involved and each step was modified to shorten the preparation time. Glass microscope slides (Corning) were cut into a rectangular shape, approximately 2.3 × 2.5 cm. All glass slides were cleaned with nano pure water and ethanol, respectively prior to use.

Figure 1.

Illustration of timeline for original protocol and alternative approaches to shorten the preparation time of avidin derivatized glass slides.

Surface oxidation step:

Originally, glass slides were immersed in piranha solution (H₂SO₄:H₂O₂, 7:3) for 14 h to create hydroxyl groups (-OH) on the surface [32, 38]. As an alternative to this step, glass slides were either 1) immersed in piranha solution of 1 h or 2) rinsed with 2 M NaOH and 0.5 M HCl followed by immersing them into piranha solution for 45 minutes at 75 ˚C [39]. The slides were then washed three times with nano pure water and ethanol followed by centrifuged for 1 min at 1600 g for a fast dry.

3-GPTMS functionalization on the surface step:

One group of oxidized glass slides was immersed in a solution of 1% 3-GPTMS, 16 mM acetic acid and 95% ethanol for 24 h and followed by three ethanol rinses. Alternatively, oxidized glass slides were immersed into 1% of 3-GPTMS in toluene for 30 min, or 10% of 3-GPTMS in cyclopentanone for 20 minutes followed by rinsing with toluene or cyclopentanone, respectively [40, 41].

Curing step:

In order to generate a 3-GPTMS network on the glass slides, samples were cured at various temperatures and times. The slides were cured in an oven for 1 h at 150 °C, 3 h at 150 °C, or 1 h at 100 °C [42]. After curing, glass slides were allowed to cool to room temperature before the next step.

Avidin immobilization on the surface step:

The surface of the slide was coated with 60 μL Avidin (1 mg mL⁻¹) solution, containing 10 mM NaHCO₃, which was then covered with a plastic coverslip and incubated for 1, 3, or 24 h in a 37 °C incubator [43]. To quench any remaining epoxide groups, 60 μL of 2 mM aspartic acid in 0.5 M NaHCO₃ buffer was applied to the surface of the slide and covered with a plastic coverslip for 30 min. After removing the coverslips, the slides were washed with nano pure water, and dried with argon gas.

Biotinylated-SWNT sensor immobilization step:

Next, 60 μL of (AT)₁₅-SWNT was dropped onto glass slides and incubated for 1, 3 and 24 h. Following incubation, unbounded (AT)₁₅-SWNT was removed by immersing the slide in nano pure water for 1 min. Finally, (AT)₁₅-SWNT functionalized glass slides were placed in PBS (pH 7.4) for 2 h and then stored in custom-made individual holders.

2.4. Preparation of Nitric Oxide Solution and Determination of its Concentration

A nitric oxide (NO) solution was prepared as previously described [1]. Briefly, 10 mL of PBS (pH 7.4) was added to a 10 mL round bottom flask, which was then sealed with a sleeve stopper. Two needles were placed in the sleeve stopper for input and outlet gas flow. Next, argon gas was bubbled for 20 min to de-oxygenate the PBS, followed by NO bubbling for 20 min. The needles were then removed, and the NO solution remained in the sealed flask and was used as a stock solution. Mixing the NO stock solution with Argon-purged PBS buffer (20 min) yielded NO dilutions. A new NO stock solution was made daily.

Qiang et al. previously detailed how to measure NO concentration [44]. Briefly, NO solution was mixed with a horseradish peroxidase solution (HRP) (final concentration 1.36 μM). The absorbance values of the NO/HRP mixture were taken between 300 and 650 nm, and the corresponding second-order derivative spectra were generated using the Savitzky–Golay algorithm. The peak-to-peak approach was used to measure the amplitudes between the peak and its adjacent valley in second order derivative spectra.

2.5. Optical Characterization

A custom-built nIR microscope was used to collect SWNT fluorescence intensity and spectrum (Photon, etc., Montreal, QC, Canada). The microscope uses a 2 W laser (561 nm) to excite the samples, a volume Bragg grating to collect the emission signal, and an InGaAs camera to capture the data (Xenics, Beverly, MA, USA). To scan across the slide surfaces and capture the intensity values for each pixel at 990 nm, custom software (PhySpec, PhotonEtc.) was employed. The resulting images of 990 nm fluorescence intensity were processed with MATLAB and then converted into colorized heatmaps to display and compare the mean spatial distribution of SWNT-derived fluorescence emission. Heatmaps were computed by subtracting the average summed matrix of control group A from each individual matrix. In order to quantify both the fluorescent yield and spatial distribution, the fluorescence intensity and mean deviation, Ra, were calculated. Mean deviation is given by:

$$Ra=\frac{1}{\#ofpixels}\sum _{i=1}^n|y-y_i|$$

Where $\mathit{y}$ is the average intensity of the entire image and ${\mathit{y}}_i$ is the intensity of each pixel.

2.6. Morphology and Chemical Characterization

The surface morphology was characterized with a previously reported procedure [32] and an optimized procedure using a focused ion beam scanning electron microscope (FIB/SEM) dual-beam workstation (Helios 660 NanoLab, FEI) at 5 V with a working distance of 4 mm. SEM- Energy-dispersive X-ray spectroscopy (EDX) was used to characterize surface chemical composition.

RESULTS and DISCUSSION

The immobilization of the SWNT sensors is a key factor when evaluating the change in SWNT fluorescence caused by analyte exposure. The immobilization can be performed on solid substrates (e.g., glass slides, nanofibers, papers and textile) and historically (3-aminopropyl) triethoxysilane (APTES)-based deposition methods have been widely used [1, 3, 33]. However, the APTES-based method provides a low concentration of SWNT adhered to the surface, resulting in low fluorescence yield, which requires the collection of data via high laser power and/or long exposure times, or the collection of data for each sensor independently, rather than the collection of data for a complete field of view. Recently, we demonstrated an avidin-biotin interaction that results in an appreciable increase in SWNT retention compared to a conventional APTES technique while increasing the solid platform’s spatial distribution of SWNT [32]. However, as shown in Figure 1, the processing time takes five days, adding additional time to any tests that require its use. A systematic evaluation has been conducted to shorten the processing time while preserving the spatial distribution of the SWNT on the platform.

Since the original protocol consists of a five step processes, we independently modified each step while keeping the other steps the same as those in the original protocol. Five slides were used for each condition and five measurements were acquired per slide to increase data acquisition and ensure even distribution across the slides. Each modification was compared with the original protocol, with a focus on fluorescence intensity and surface distribution. The first step, generation of hydroxyl groups on the glass slide surfaces required 14 hour treatment with piranha solution at room temperature in the original protocol [32]. In previous studies, Kamra et al. [39] and Nam et al. [38] demonstrated the generation of hydroxyl groups by rinsing glass slides with NaOH and HCl solutions then treating with piranha solution for 45 min at 75 °C or incubation with piranha solution for 1 hour at room temperature, respectively. We treated a set of the slides with these two new methods and the method previously used for platform fabrication. Steps two to five from the original protocol were applied for each set of slides to compare how the -OH generation step impacted the spatial fluorescence distribution on glass. Figure 2a, 2b and 2c show the representative fluorescence images of three sets of slides taken at 990 nm (the characteristic wavelength for (6,5) (AT)₁₅ SWNT sensors). The samples prepared with piranha solution treatment of 14 h at 25 °C and 45 min. at 75 °C produced average fluorescent intensity values of 918.5±69.0 and 906.4±113.3 a.u., respectively (Figure 2d). An unpaired t-test shows that the variations in peak values are not statistically significant (p = 0.7548). The average 990 nm intensity value for one hour of piranha treatment was found to be 904.5±78.8 and when it was compared to 14 h treatment, it showed an insignificant difference in peak values (p = 0.2819). SWNT distribution was calculated by measuring the arithmetic average roughness, R_a, and is shown in Figure 2e. For the development of an evenly distributed platform, a lower R_a, which represents a smoother surface, is preferred. While the samples prepared with piranha treatment of 14 h at 25 °C and 45 min at 75 °C showed statistically significant differences in Ra values (p= 0.0012) compared to the original protocol, the difference in R_a values between samples prepared with 14 h and 1 h treatment is not statistically significant (p= 0.2048). These results suggest that instead of 14 h process, we can employ 1 h piranha solution treatment to shorten the fabrication time for SWNT-based optical sensor while preserving florescence intensity and the SWNT distribution.

Figure 2.

Representative fluorescence images of (AT)₁₅ SWNT sensor immobilized glass slides prepared via (a) 14 h at room temperature (control) (b) 45 min at 75 °C and (c) 1 h at room temperature taken at 990 nm. (d) Mean fluorescence intensities of as-prepared sensors which showed no significant differences (labeled as ns) (e) Mean fluorescence distribution of as-prepared sensors showed a statistically significant difference between samples prepared with piranha solution treatment of 14 h at 25 °C and 45 min at 75 °C while there is a statistically no difference between samples prepared with piranha treatment of 14 and 1 h at room temperature. (** p<0.01)

The second step in the protocol is to functionalize the surface with 3-GPTMS to generate an epoxy silane network. In the literature, 3-GPTMS was used with different solvents with non-polar capabilities such as ethanol, toluene, and cyclopentanone [40, 41], therefore we explored the usage of various solvents and conditions to investigate the possibility of decreasing incubation time. Detailed experimental conditions are described in the Materials and Methods section. Since the results from step 1 showed that 1 h piranha treatment at room temperature does not alter the florescence intensity and roughness in comparison to the original protocol, this condition was used for optimizing the next steps. In step 2, piranha treated surfaces were functionalized using 3-GPTMS in ethanol (control) for 24 h, toluene for 30 min., and cyclopentanone for 20 min. Steps 3 – 5 from the original protocol were then used to compare the impact of these changes to step 2. Percent fluorescence intensities are shown in Figure 3a. The results showed that there is no statistically significant difference between samples prepared in ethanol and cyclopentanone (p= 0.9503). But, when toluene was employed, the array showed a much higher, and statistically significant, fluorescence intensity in comparison to the ethanol (previous method) and toluene (p<0.0001) conditions. Interestingly, despite the differences in fluorescence intensity, there were no significant differences in fluorescence distribution between the samples (p= 0.9194 for ethanol and toluene and p= 0.8748 for ethanol and cyclopentanone) (Figure 3b). These results suggest that the toluene-based system can be employed to shorten preparation time to fabricate SWNT-based sensor platforms without altering the SWNT distribution and with the added bonus of increasing platform fluorescence.

Figure 3.

Percent fluorescence intensity and distribution of SWNT for the 3-GPTMS functionalization step (a and b), curing step (c and d), avidin (e and f) and SWNT (g and h) immobilization steps.

The third step involves curing the 3-GPTMS to form a network. The slides were cured at either 3 h at 150 °C (control), 1 h at 150 °C, or 1 h at 100 °C. There is no significant difference in fluorescence intensity for curing times of 3 h or 1 h at 150 °C (p= 0.1289) (Figure 3c), but there is a significant difference between the 3 h at 150 °C and 1 h at 100 °C conditions (p<0.0001), with the lower temperature resulting in a significantly lower fluorescence intensity. As shown in Figure 3d, glass slides prepared for 1 h, both at 150 °C and 100 °C, show a statistically significantly lower Ra value in comparison to the 3 h condition (control), indicating increased SWNT distribution on the platform for the conditions cured for shorter time periods. The significant increase in Ra value as well as the lack in change of fluorescence intensity between the 3 h and 1 h 150 °C conditions show that the curing step can be performed at 1 h without negatively impacting the platform, in fact it will improve the platform by increasing SWNT distribution.

In step three, the 3-GPTMS network created an epoxide functionalized surface that can react with a variety of nucleophiles, such as amine, hydroxyl, and thiol groups, under mild experimental conditions to create a very stable covalent bond [45]. In step four, avidin, a tetrameric biotin-binding protein, is covalently immobilized on the glass substrates by incubation at 37 °C for various lengths of time (24 hours (control), 3 hours, or 1 hour). As shown in Figure 3e, there is a dramatic change in percent fluorescence intensity when 1 and 3 h of incubation was employed compared to 24 h incubation time (p= 0.0006 for 3 h and p= 0.0037 for 1 h), with an increase of more than 50% in fluorescence intensity when 3 hours of incubation was used compared to the control condition of 24 hours. This result suggests that incubation time for avidin attachment plays a critical role in SWNT loading for platform fabrication. Ra values of samples prepared with 1, 3, and 24 h of avidin incubation showed no significant difference (1 h p= 0.4244 and 3 h p= 0.5573) (Figure 3f). Prolonged incubation of avidin on glass slides can have detrimental effects on both the stability and functionality of avidin, as well as increase the likelihood of nonspecific binding events that may be washed away during the subsequent cleaning step. Avidin, known for its ability to simultaneously bind four biotin molecules, plays a crucial role in fluorescence distribution within samples. Any loss of avidin from the surface can significantly impact the fluorescence distribution and compromise the accuracy of the results. Therefore, optimizing the incubation time is essential to strike a balance between achieving adequate binding and maintaining the stability of avidin, thus ensuring accurate and reliable experimental outcomes.

In step 5, biotinylated SWNT sensors were immobilized onto an avidin functionalized surface through the highly potent avidin/biotin interaction, which is widely recognized as the strongest non-covalent interaction. Previous studies have reported varying reaction times, spanning from a few minutes to as long as 24 h, and operating within pH levels ranging from 7 to 9 [46, 47]. In our experimental setup, we employed different incubation times: a control group incubated for 24 hours, while other groups were subjected to 3-hour or 1-hour incubations. Figure 3g illustrates that utilizing a longer incubation time led to a significantly higher fluorescence intensity (p<0.0001 for 1 hour and p=0.0223 for 3 hours). Conversely, when comparing the Ra values, no notable distinction was observed between 3 hours and 24 hours (p=0.8804). However, when employing a 1-hour incubation time, a lower Ra value was observed (p=0.0044). Since having a significantly higher initial fluorescence intensity is important for the SWNT sensor platform, the 24-hour option is chosen for moving forward. After going through all five steps of the original protocol, we were able to drastically decrease the fabrication time of the SWNT sensor array, bringing it from 5 days to 2 days.

After optimization of the fabrication method through fluorescence intensity and SWNT distribution, additional tests were run to ensure that there are no changes in the function of the sensor due to the changes in the platform building steps. The surface morphology of the platforms fabricated with the original and newly developed method was studied with SEM (Figure 4). In both cases, the SWNT sensor is randomly distributed on the glass slide, resulting in an interconnected network. SEM also supported the Ra analysis by showing that the SWNT sensors are evenly distributed, not localized at certain locations on the platform. To further investigate the SWNT distribution on the substrate, EDX elemental mapping was performed. Figure 5a and 5b show overlapping images of Carbon (c and d), Nitrogen (e and f), and Oxygen (g and h) mappings of the original and new fabrication methods, respectively. The EDX elemental mapping shows that the change in the protocol did not alter the Carbon, Nitrogen, or Oxygen percentages on the glass slides, with percentages at 45%, 17%, and 38%, respectively.

Figure 4.

SEM images of SWNT platforms fabricated using the original protocol (a and b), and newly developed method (c and d) at different magnifications.

Figure 5.

EDX elemental mapping images for (a, c, e and g) platforms fabricated with the original protocol and (b, d, f and h) the newly developed method. a, b: Overlap of the three conditions, c, d: Carbon, e, f: Nitrogen and g, h: Oxygen.

FT-IR and Raman spectroscopy techniques were employed to assess the presence of Single-Walled Carbon Nanotubes (SWNT) on the substrate. Initially, FT-IR spectroscopy was utilized to examine the chemical composition of the sensory array both before and after the optimization process. The complete FT-IR spectrum, as depicted in Figure S1, exhibits a series of sharp peaks within the range of 1740–1859 cm⁻¹, which can be attributed to the presence of carbonyl groups. These carbonyl groups are a consequence of the presence of avidin and biotinylated SWNT solutions. Notably, aside from these carbonyl-related peaks, there were no discernible features in the spectrum. Figure S2 provides a closer look at the FT-IR spectra within the range of 2400 to 3300 cm^-1. Within this region, specific peaks at 2939, 2864, and 2797 cm⁻¹ were observed in both spectra. These peaks can be attributed to distinct C–H bond stretching vibrations associated with the avidin that had been immobilized on the substrate surface and ssDNA which was wrapped on SWNT [48]. Figure S3 presents the Raman spectra for both sets of samples. Within these spectra, distinctive peaks at 1316, 1589, and 2632 cm⁻¹ are identified, corresponding to the D, G, and 2D peaks of SWNT, respectively [49–51]. To assess the overall number of defects in the SWNT, the intensity ratio of the G band to the D band, denoted as the G/D ratio, is utilized. In the original method and the new protocol, the G/D ratios are 3.99 and 5.92, respectively. This disparity implies that the original method exhibits a higher G/D ratio, indicating a greater presence of non-specific binding, which in turn results in more defects on the SWNTs compared to the new method.

The full width half maximum (FWHM) values of the spectra were also investigated. FWHM is important since it tells us about the shape of the spectrum and how similar spectra are to one another. The fluorescence spectra with a reference point of 990 nm in order to conduct an accurate comparison were normalized. Then, it merged these normalized spectra to provide graphical representations. An unpaired t-test was run to carefully evaluate any differences. The findings showed that there was no statistically significant difference between the two spectra in terms of FWHM (p = 0.3973). This result indicates that expediting up the protocol had no impact on the FWHM of the SWNT emission peak.

The SWNT platforms’ sensing capabilities were tested for their response to NO to ensure that the new fabrication method does not alter the sensing function of the SWNT. As shown in Figure 6c, fluorescence intensities were measured for each sample before and after the addition of 30 μM NO. The NO quenching profiles shows that NO quenching is the same for both platforms, demonstrating that the new fabrication method does not alter the real time nature of the SWNT sensors.

Figure 6.

FWHM values for sensory arrays prepared using the (a) original and (b) new fabrication methods. (c) Fluorescence quenching behavior of SWNT sensors upon introduction of 30 μm NO, comparing results from the original method (in black) and the newly developed fabrication method (in red). (d) Examination of stability and reproducibility for both the original and new methods. Examples of sample heat maps from (e) the original method and (f) the new method.

Our sensing platform’s primary objective is to detect changes in extracellular analyte concentrations accurately. To monitor these changes within the extracellular environment, it is crucial for the sensors to remain firmly attached to the substrate, even when subjected to the rinsing and media replacement cycles in cell culture experiments. Consequently, we conducted an extensive investigation into the adhesion of SWNT over an extended duration, encompassing multiple solution exchanges for both sample sets.

These treated slides were placed within individual slide holders, each containing 3 mL of PBS to ensure complete surface coverage. On a daily basis, we carefully replaced the supernatant with fresh PBS, maintaining a consistent environment. Over a period of 14 days, we monitored the changes of percent florescence intensity, as illustrated in Figure 6d. Fluorescence measurements were conducted, and subsequently, the average fluorescence values were calculated. Remarkably, both sets of samples, prepared using the original and new methods, exhibited exceptional stability, retaining approximately 90.45% ± 1.55% and 91.19% ± 1.97%, respectively, of their initial fluorescence levels. An example of heat maps of samples from original and new methods are shown in Figure 6e and 6f, respectively.

CONCLUSIONS

We reported a systematic study of SWNT platform fabrication in an attempt to decrease fabrication time without significantly altering the SWNT fluorescence or distribution. The data shows that we can actually increase SWNT fluorescence and distribution while decreasing the fabrication time from 5 days to 2 days. SEM and EDX results show that the two platforms have similar surface morphology and surface chemical composition. Furthermore, both platform fabrication methods result in similar NO quenching profiles. This new fabrication protocol will decrease preparation time for future studies of NO as well as other SWNT based sensor systems.