Photothermal Desorption of Single-Walled Carbon Nanotubes and Coconut Shell-Activated Carbons Using a Continuous Light Source for Application in Air Sampling

Abstract

INTRODUCTION

Every year millions of tons of volatile organic compounds (VOCs) are released into the environment by anthropogenic sources globally and domestically (Middleton, 1995; Goldstein and Galbally, 2007; U.S. EPA, 2008), with a portion of these emissions resulting in occupational exposure of workers. Validated methods have been developed to measure these compounds to ensure compliance with the permissible exposure limits set forth by the U.S. Occupational Safety and Health Administration (OSHA) (U.S. EPA, 1999; NIOSH, 2003; OSHA, 2008, 2010).

For VOC monitoring, the most common and preferred sampling technique is sorbent-based preconcentration in which the analyte is collected on a solid sorbent. Sorbent samples have good sample stability, are compact and convenient; in many cases, the preferred sorbent is high surface area granularized activated carbon (AC). Sorbent samples must be desorbed prior to analysis and desorption can be performed by solvent extraction or heat treatment. Solvent desorption is simple, provides high desorption efficiencies, and allows multiple analyses per sample. However, it requires significant lab preparation, uses toxic solvents, and has inherent sensitivity limitations due to the small fraction of sample injected into the gas chromatograph (GC), 1 μl out of 1000 μl such as in OSHA method 111 or NIOSH method 1500 (NIOSH, 2003; OSHA, 2008). Thermal desorption requires no lab preparation, is solvent free, and simple in principle: heat is applied to the sorbent under purge gas flow that releases the trapped analyte. The sensitivity of thermal desorption is excellent since the entire sample is delivered to the GC, but for many substances, more utility would be gained by delivering only a portion of the sample to the GC and reserving the rest of the sample for multiple analyses. Some drawbacks of this technique are the high cost associated with thermal desorption units and samplers, the added system complexity, and the added analysis time. Flow splitting thermal desorption units have been developed but are expensive and still require complete desorption of the original sample with the unanalyzed portion recollected on another sorbent tube (CDS Analytical, 2007).

In this study, we explore the use of high-intensity light irradiation to thermally desorb an analyte from a preloaded sorbent, single-walled carbon nanotubes (SWNT), and AC. Light energy is absorbed directly by the sorbent and converted to heat. SWNT can have high surface areas (Peigney et al., 2001; Cinke et al., 2002; Li et al., 2004; Byl et al., 2006) and are excellent light absorbers (Kataura et al., 1999; Ajayan et al., 2002; Bockrath et al., 2002; Tseng et al., 2007; Mizuno et al., 2009; Tabakman et al., 2010). SWNT are explored in parallel with AC due to their emerging potential as a sorbent for environmental samplers (Zheng et al., 2006; Hussain et al., 2008a; Takada et al., 2010a,b; Hussain and Mitra, 2011) and the ability to functionalize SWNT to alter chemical affinities (Georgakilas et al., 2002; Hilding et al., 2003; Tagmatarchis and Prato, 2004; Hussain et al., 2009). While traditional thermal desorption heats the sorbent via conduction from a heating sleeve through the sampler body, using light reverses the heat flow. Heat is generated by the sorbent and conducted from the sorbent to the sampler body and the external environment. Reversing heat flow and isolating the sample could allow more precise control of heating than is possible with a simple resistive heater.

This work explores the use of high-intensity broad spectrum light to photothermally desorb a representative VOC from two sorbents, AC and SWNT. Differences in desorption between AC and SWNT substrates and the utilization of field grade PID to measure desorption are also examined.

METHODS

Commercially available coconut shell AC and SWNT were loaded with a known mass of toluene vapor and irradiated with broad spectrum light at different light powers for several minutes. Desorbed toluene was quantified with a field grade PID and compared across lamp settings and materials. Three test chambers for photothermal desorption were evaluated for each sorbent: a static chamber (500ml, Fig. 1a), a ‘large’ dynamic chamber (500ml, Fig. 1b), and a ‘small’ dynamic chamber (75ml, Fig. 1c). Toluene adsorption isotherms were obtained to compare sorbents and determine a maximum sample loading that does not saturate either sorbent. Samples were loaded with toluene vapor by injecting liquid toluene onto the interior sidewall of the chamber instead of directly on the sorbent and desorbed by irradiation with light for at least 4min. Toluene was selected as a representative VOC because of its similarity to benzene and its use in similar studies (Zheng et al., 2006; Hussain et al., 2008b, 2010; Balanay et al., 2011).

Open in a separate window

1

Schematic of test chambers used for adsorption and desorption experiments. (a) 500-ml static desorption chamber. (b) 500-ml flow desorption chamber. (c) 75-ml flow desorption chamber. (d) 120-ml adsorption chamber.

RESULTS

Light power and material characterization

Light power measurements for the 500-ml desorption chambers with 8, 10, 12, and 15V applied to the lamp were 94, 149, 224, and 361 mW cm⁻², respectively. Light power for the 75-ml desorption chamber with 15V applied to the lamp was 752 mW cm⁻².

Specific surface area and pore size distribution are similar between AC and SWNT. AC has slightly larger surface area (S _BET, m² g⁻¹) and microporosity (V _µP, cm³ g⁻¹, pore volume <20 Å) but notably lower total porosity (V _TP, cm³ g⁻¹) (Table 1). From the plot of pore size distribution (Fig. 2), we can see that AC has a sharper peak with smaller pore diameter in the micropore range (<20 Å) than SWNT, while SWNT has greater total porosity than AC. Grain size of the AC was stated by the manufacturer as 20/40 mesh, and this translates to 400–800 µm; grain size of the SWNT was estimated to be 0.5–50 µm by SEM imaging (Fig. 3).

Table 1.

BET surface area, micropore volume, and total pore volume for SWNT and AC samples

	S BET (m2 g−1)	V µP (cm3 g−1)	V TP (cm3 g−1)
SWNT	980	0.39	0.59
AC	1030	0.41	0.45

Open in a separate window

BET, Brunauer, Emmett and Teller.

Open in a separate window

2

Pore size distribution of AC (hollow diamonds) and SWNT (solid squares).

Open in a separate window

3

SEM image of SWNT powder shows a majority of particles between 0.5 and 4 μm with larger particles reaching 20–50 µm.

Toluene adsorption isotherms

The toluene adsorption isotherms at 23°C for both materials are shown in Fig. 4. The adsorption capacity of SWNT was lower than AC. Both adsorption isotherms (AC and SWNT) are quite similar up to 139mg g⁻¹ (red horizontal line in Fig. 4); therefore, the maximum loading of both sorbents was set below this level. High mass loading was used for initial characterization experiments and later reduced to realistic levels as the system performance became known.

Open in a separate window

4

Adsorption isotherm (23°C, 1 atm) for AC (hollow diamonds) and SWNT (solid squares) samples. The horizontal line represents the upper range of linear adsorption for SWNT. This was the maximum loading level used for both materials.

Sample loading

Samples of AC and SWNT (50mg) were loaded by indirect injection of pure toluene (6.525mg), as described above. Loading samples at this level approach the upper range of linear adsorption for SWNT as shown in Fig. 3. The chamber containing the samples of AC or SWNT was monitored with the PID during loading to determine the time necessary to achieve complete adsorption. In both materials, 99% adsorption was observed by 30min, and >99.4% adsorption by 60min, as determined by dividing the mass of toluene in the chamber headspace (from p.p.m. measurements with PID) by the known injected amount.

Table 2 shows fraction of adsorption at 15-min intervals during the first hour of loading. Figure 5 is a representative sample loading (50mg SWNT, 6.525mg toluene). Adsorption using this technique was also confirmed by chemical extraction. Recoveries for AC loaded with 6.525 and 0.435mg toluene were 98.8% (±1.8) and 99.0 (±1.3), respectively (n = 5); recoveries for SWNT loaded with 6.525 and 0.435mg toluene were 100.5% (±2.1) and 99.0 (±1.9), respectively (n = 5).

Table 2.

Degree of adsorption for 50mg AC and SWNT samples at 15, 30, 45, and 60min post-injection of 6.525mg toluene, n = 3

Sorbent		Adsorbed fraction at
		15 min	30 min	45 min	60 min
SWNT, n = 3	Average (SD)	0.964 (0.009)	0.991 (0.002)	0.993 (0.001)	0.994 (0.001)
AC, n = 3	Average (SD)	0.957 (0.001)	0.992 (<0.001)	0.995 (0.001)	0.996 (0.001)

Open in a separate window

Open in a separate window

5

Representative sample loading, 50mg SWNT loaded with 6.525mg toluene.

Desorption

Desorption of AC and SWNT samples by irradiation at varied lamp voltages (8, 10, 12, and 15V) using the 500-ml static chamber are shown in Fig. 6. The desorbed percentage of the loaded toluene mass calculated from the headspace concentration within the chamber is shown in the plot. As expected, desorption increases with input energy, with the highest lamp voltage (15V) yielding the greatest desorption. Desorption was greater in SWNT than in AC at each lamp voltage but seems to differ even more so at 15V. This experiment was used as a screening process to characterize desorption at varied lamp settings; the 15-V lamp setting is explored in more detail later. Since this is a closed system, the desorbed mass accumulated in the headspace of the chamber and established a right-shifted dynamic equilibrium but could not achieve complete desorption with the applied light power. During experimental trials, when irradiated at 15V, desorption concentration peaked after 5–6min and then declined. It was determined that the declining concentration was due to the PID getting hot during sample irradiation since it was inside the chamber with the sample. To correct for this effect, the PID was shielded with reflective foil and a correction curve was developed for sensitivity loss during irradiation, which was applied to later trials.

Open in a separate window

6

Screening of 50mg AC (a) and SWNT (b) samples loaded with 6.525mg toluene and desorbed by irradiation at varied lamp voltages (8, 10, 12, and 15V) inside the 500-ml static chamber.

The same screening process as above was performed using the 500-ml ‘large’ flow chamber (Fig. 7). Air flow rate was maintained at 0.14 l min⁻¹ (0.3 air changes per minute) and lamp voltages of 8, 10, 12, and 15V were applied. A peak in concentration emerges around 5min as lamp power increases, suggesting that a greater degree of desorption occurs faster with more light power. Figure 8 summarizes both screening experiments (static and dynamic flow) in units of total mass desorbed and percent desorbed.

Open in a separate window

7

Screening of 50mg samples loaded with 6.525mg toluene and desorbed by irradiation at varied lamp voltages (8, 10, 12, and 15V) inside the 500-ml flow chamber, Q = 0.140 l min⁻¹. (a) AC and (b) SWNT.

Open in a separate window

8

Screening of 50mg samples (SWNT and AC) using the 500-ml flow chamber and irradiation at varied lamp voltages. Flow through chamber = 0.14 l min⁻¹, each sample loaded with 6.525mg toluene and irradiated at 8, 10, 12, or 15V.

Figure 9 shows percent desorption of 50mg samples of AC and SWNT loaded with 6.525, 4.350, and 2.175mg toluene and irradiated at 15V in the static chamber for 4min; output is corrected for PID sensitivity loss (n = 6–7). After 4min of irradiation, the concentration inside the static chamber plateaus and minimal desorption is achieved beyond this point (Fig. 6). The effect of loading level on percent desorption after 4min of irradiation was assessed by one-way analysis of variance (ANOVA) and found to be significant for both AC and SWNT (<0.0001). The ANOVA was followed by Tukey’s post hoc test to determine which loading levels were significantly different. In AC, the lowest loading level (2.175mg) was significantly different (lower) than the higher levels (4.350 and 6.525mg), but there was not a significant difference between the higher loading levels. In the SWNT, each loading level is significantly different from the others (α = 0.05) with an increasing trend as loading level increases. These statistical observations are evident in Fig. 9.

Open in a separate window

9

Percent desorption of 50mg samples loaded with 6.525, 4.350, or 2.175mg toluene and irradiated for 4min at 15-V lamp setting in 500-ml static chamber (n = 6–7); results are corrected for PID sensitivity loss. (a) AC, percent desorption is significantly different at only the lowest loading level. (b) SWNT, percent desorption is significantly different at all loading levels.

The mass of toluene collected by a 3M diffusive sampler at 10 p.p.m. for 8h would be approximately 0.6mg. A similar mass (0.435mg) was loaded for trials with the 75-ml dynamic desorption chamber (n = 3). In the 75-ml flow chamber, the air change rate was much higher due to the smaller volume and light power was much greater than in the 500-ml chamber because the sample was closer to the light source (Fig. 1c). On average, 0.199mg (±0.014) and 0.316mg (±0.009) of the loaded toluene was desorbed from AC and SWNT samples, respectively, when irradiated 4min at the 15-V lamp setting. Figure 10 summarizes results from all trials with repeated measures at the 15-V lamp setting. Percent desorption in all configurations was significantly greater than controls and SWNT desorption was significantly greater than AC in all configurations. In the large chamber, when airflow was increased (1.6 l min⁻¹, 3.2 air changes per minute), desorption was greater than at 0.14 l min⁻¹ (AC, 0.06; SWNT, <0.001) and static test conditions (AC, 0.005; SWNT, <0.001).

Open in a separate window

10

Desorption after 4min of irradiation at 15V, 50mg samples, loaded with 6.525mg toluene for 500-ml chamber and 435mg toluene for 75-ml chamber. t-Test performed on each pair of AC/SWNT test conditions and SWNT was significantly greater in all pairs (S, P < 0.0001; LF, P < 0.001; HF, P = 0.0219; 75ml, P = 0.0006). Additionally, all values were statistically different than controls. C = control, S = static chamber, LF = ‘large’ 500-ml chamber at low ventilation rate (0.140 l min⁻¹), HF = ‘large’ 500-ml chamber at high ventilation rate (1.6 l min⁻¹), and 75ml = ‘small’ 75-ml chamber with high ventilation rate (0.9 l min⁻¹).

DISCUSSION

Of the two sorbents evaluated, the materials are quite comparable, with AC having greater adsorption capacity and SWNT being more sensitive to photothermal desorption at all levels. The greater abundance of micropores in the AC accounts for the greater toluene adsorption capacity; however, the greater total porosity of SWNT suggests potential for greater adsorption capacity of larger molecules such as semivolatile organic compounds. Some observed differences could be due to differences in material grain size. Though SWNT were heavily agglomerated, particle diameters were still quite small (5–40 µm as determined by SEM imaging), whereas particle diameters for the AC were quite large (400–800 µm). The finer particle size of the SWNT was easier to spread evenly across the sample tray possibly resulting in greater light absorption; however, heat transfer through multiple layers of small particles is less efficient than through a monolithic large granule, so the effects may be offsetting.

Attempts were made to measure the temperature of the sample during irradiation using an infrared thermocouple but were non-quantitative; however, in the 75-ml chamber, the sample tray was held in place with a neodymium magnet on the sample pedestal (Fig. 1c). This magnet lost its magnetism with both sorbents after only a few trials. The Curie temperature (temperature at which magnetism is lost) of neodymium is 320°C; therefore, both AC and SWNT samples must have reached this temperature, suggesting differences due to grain size were minimal.

While both materials are amenable to photothermal desorption, it is clear that SWNT is more sensitive to this technique having significantly greater desorption in all test conditions and as much as 80% greater when highly loaded. With SWNT, each loading level played a significant role in percent yield, whereas with AC, significant differences in loading were observed between 2.175 and 4.350mg loadings but not above this level (Fig. 9). These differences may diminish when sampling at low loadings typical of environmental and occupational sampling.

Using the 75-ml chamber, 70.5% of the initial mass was desorbed with 4min of irradiation at the 15-V lamp setting; after extinguishing the lamp, an additional 1% was desorbed over the following 10min. This is due to the residual toluene in the dead volume of the chamber and latent heat of the system accumulated during irradiation. Desorbing to a partial end point introduces variability and some sample loss; likely, this is a reason why traditional thermal desorption has not sought partial desorption end points previously. The real advantage of using a photothermal process is that heat is generated by the sorbent and not conducted to the sorbent. Reversing the direction of heat flow and heating without contact has the potential to allow precise control of desorption.

One intent of this study was to demonstrate the potential of using light as a thermal desorption technique that achieves partial desorption end points. We also investigated differences between sorbents (AC and SWNT) and the suitability of field grade monitoring equipment (PID) to measure desorption. We have clearly demonstrated the reliability of achieving partial desorption end points in both substrates since the relative standard errors of trials ranged between 3.3 and 12.4% for AC and 1.6 and 6.1% for SWNT. We also observe that SNWT desorbed a greater fraction with consistently lower variability than AC. Achieving partial desorption could be useful in occupational and environmental sampling by allowing field hygienists to prescreen samples using common equipment such as hand-held PIDs prior to submitting to an analytical lab. This would greatly accelerate the time to knowledge of integrated samples while still allowing full lab analysis of the remaining sample. The need for duplicate and triplicate field samples could be eliminated by only analyzing a portion of the original sample. Analytical labs could use this technique to improve sensitivity over analyses currently using chemical desorption. With current thermal desorption techniques, the sample must be completely desorbed; some systems allow the flow to be split with one portion recaptured and the other analyzed. These systems are quite expensive and can be subject to sample loss from leaks and sorbent collection efficiencies. Retaining the unanalyzed portion of the sample on the original sorbent seems a more reliable approach than completely desorbing the sample, passing it through various plumbing and valves and recollecting it on another sorbent sampler.

In this study, only one analyte was explored to eliminate the need for a chromatographic detection system. It is likely that the temperature ramp of the sorbent will not result in uniform desorption of all analytes in time. While specific desorption efficiencies could be determined for all potential analytes and mixtures thereof, a better approach would be to heat the sorbent above the desorption temperature of all analytes very rapidly. This could be accomplished by irradiating with very high energy density light over a short period of time, such as with a photo flash. This is currently being explored. A 20-J photo flash has the potential to pulse the temperature of 50mg of SWNT by 340 °C within milliseconds (the duration of the photo flash) based on specific heat capacity measurements by Hepplestone et al. (2006) and light absorption properties of SWNT films and forests (Savage et al., 2003; Zhang et al., 2005; Mizuno et al., 2009). Photothermal desorption with a photo flash would be easily field portable and could be used to release small quanta of adsorbed contaminants nearly instantaneously.

CONCLUSIONS

Photothermal desorption is a viable technique to be included in the analysis of occupational and environmental air samples. In this pilot study, we demonstrated the potential viability of photothermal desorption since it produced repeatable desorption that was significantly greater than controls. Desorption of toluene from SWNT was significantly greater than AC and a field grade PID has the sensitivity to detect contaminants from an integrated sampler loaded at occupationally relevant levels (0.435mg toluene). Coupling this desorption technique to a sorbent sampler specifically designed for photothermal desorption (similar to a diffusive sampler) could facilitate infield estimations of total VOC using equipment routinely employed by hygienists while retaining enough sample for full laboratory confirmation using the same sample.

FUNDING

National Institute for Occupational Safety and Health (2T42OH008436-03).

REFERENCES