Hierarchical carbon-nanotube/quartz-fiber films with gradient nanostructures for high efficiency and long service life air filters

Abstract

Hierarchical and gradient nanostructures are important to exploit the full potential of nanofibers in filtration applications. The introduction of a gradient into carbon nanotube (CNT)/fiber hierarchical structures could result in a change of the particle capturing properties. Here, we show the fabrication of hierarchical carbon nanotube (CNT)/quartz-fiber (QF) filters with gradient nanostructures where the content of CNTs decreases exponentially along the thickness direction of the filters. The loading of catalysts for the growth of CNTs in the QF filter has been achieved using an aerosol technique, which can be carried out on a large-scale. With only 1.17 wt% CNT, the penetration of the CNT/QF filter at the most penetrating particle size (MPPS) has been reduced by one order of magnitude, while the pressure drop only increases about 6% with respect to that of the pristine QF filter, leading to an obvious higher quality factor (Q_f) for the CNT/QF filter. More importantly, the service life of the CNT/QF filter with the CNT-rich side downstream has increased by 64% when compared with the pristine QF filter. In contrast, when the CNT-rich side is placed upstream, the service life of CNT/QF filter is only 41.7% of that observed when placing the CNT-rich side downstream. Scanning electron microscopy (SEM) images reveal that the gradient nanostructure of the CNT/QF filter, together with the CNT/QF hierarchical structure play very important roles in the simultaneous enhancement of the filtration efficiency and the service life of the air filters.

---

Introduction

Air filtration is of primary importance in many fields such as cabin environment, nuclear industry, indoor environment, clean room and engine emission reduction. Fibrous filtration is the most commonly used method to remove airborne particles with high efficiency, especially for particulate matters smaller than 1 μm. Practical fibrous filters do not act as a sieve because the sieve structure tends to be clogged quickly, resulting in an excessively high pressure drop. In fact, fibrous filters usually have high porosity and the inter-fiber distances are larger than the size of particles. Aerosol particles are removed from the gas flow mainly based on following mechanisms: direct interception, Brown diffusion and the forces of inertia, gravity, or electrical attraction. The first two predominate the filtration process of sub-micron aerosol particles.^1–3

Filtration efficiency and pressure drop are the two most important parameters of fibrous filters. They remain constant at the beginning phase of filtration. During the filtration process, aerosol particles will gradually accumulate on the fiber surface, causing clogging of the filter.^4–7 This results in a sharp increase in the pressure drop and changes in filtration efficiency. Due to the “clogging effect”, filters have to be disposed of when their pressure drop reaches a certain value, resulting in a limited service life of an air filter. Service life is also an important parameter for an air filter, particularly in many special fields such as aircraft, nuclear power station, and semiconductor industry, where the replacement of air filters is very expensive. Numerous studies have been devoted to the filtration efficiency and pressure drop of air filters. In contrast, the clogging effect remains poorly understood and reports on filter structure design towards extending the service life of air filters are very limited.^8,9

Recently, nanofibers have gained great interest in air filtration.^10–16 Compared with microfibers, the most significant advantage of nanofibers is the high specific surface area,¹⁷ which will be helpful for promoting contact between the aerosols and fibers. Besides, according to the classic filtration theory, when the diameter of the fiber is comparable with the mean free path of air molecules (about 66 nm under ambient conditions), the flow around a single fiber will be “slip flow”, “transition flow” or even “free molecule flow”, which means the disturbance of fibers on the flow pattern will be very small or even negligible.^18,19 Therefore, air filters made of nanofibers will theoretically have a high efficiency as well as low pressure drop. Currently, the most popular nanofibers used in air filtration are polymer nanofibers.^10,11,20,21 Compared with polymer nanofibers, CNTs have much smaller diameters, a much higher specific surface area and much better mechanical properties.^39,40 These advantages make CNTs a perfect candidate for air filtration. Recently there have been a few reports on the application of CNTs in air filtration.^13,22–29 The commonly used strategies for fabricating CNT-based air filters include coating CNT films on microfiber filters,¹³ which provide mechanical support for the CNT layer or using free-standing thin CNT films directly.^27–29 When nanofibers, especially CNTs, are used as filter media, a high filtration efficiency of the nanofiberfibre-based air filter is always accompanied with a high pressure drop and a short service life. This is because nanofibers are more inclined to have high packing density and the pores between nanofibers are usually too small for aerosol particles. During the air filtration process, a nanofibre-based air filter will be easily clogged, resulting in a rapid increase of pressure drop. Besides, it is also difficult to fabricate nanofiber-based air filters on a large-scale and maintain their high structural stability under the air force during air filtration.

A good filter should have a high filtration efficiency, low pressure drop and long service life. In previous work, we have fabricated a depth-type hierarchical CNT/quartz fibre (QF) filter through the in situ growth of CNTs on QF fibres, which show a greatly enhanced filtration efficiency with a small increase in pressure drop.²³ In this work, we further develop the hierarchical CNT/QF filter into a hierarchical CNT/QF filter with gradient nanostructures along the thickness direction. We have also developed a method to utilize the large-scale fabrication of the CNT/QF filter with gradient nanostructures using an aerosol technique. We found that the introduction of CNTs into the QF filter could significantly increase the filtration efficiency of the air filter. More importantly, the service life of the CNT/QF filter was extended by placing the CNT-rich side downstream. Conversely, the service life of the CNT/QF filter was greatly shorted when placing the CNT-rich side upstream.

Experimental

Materials

The QF filters with a diameter of 150 mm were purchased from Membrane Solutions Inc., and used directly as the substrate for CNT growth. Fe(NO₃)₃·9H₂O (purity > 99.99%) and Al(NO₃)₃·9H₂O (purity > 99.99%) were purchased from Tianjin Damao chemical reagent factory, China. Bis(2-ethylhexyl) sebacate (DEHS, AR, purity > 97%) was purchased from Aladdin reagent Co., Ltd., China. NaCl (purity > 99.5%) was purchased from Modern Oriental Fine Chemistry Co., Ltd., China. High purity nitrogen (N₂, purity > 99.999%), argon (Ar, purity > 99.999%), hydrogen (H₂, purity > 99.999%) and ethylene (C₂H₄, purity > 99.5%) were supplied from Beiwen Gas Co., Ltd., China and used as-received.

Material synthesis

Catalyst loading on QF filters using an aerosol technique. An aerosol technique was used to load the catalyst on the QF filter. Firstly, catalyst aerosols were generated by atomizing a mixed aqueous solution of 0.01 g ml⁻¹ Fe(NO₃)₃·9H₂O and Al(NO₃)₃·9H₂O (the molar ratio of Fe³⁺ : Al³⁺ = 0.8 : 1) using a constant output atomizer (model 3076, TSI). The concentration and particle size distribution of the catalyst aerosols are shown in Fig. S1.† The catalyst aerosol particles were dried using a diffusion drier and then passed through the QF filter with an area of 100 cm² for 30 min, N₂ was used as the carrier gas and the gas flow rate was 0.22 cm s⁻¹. The catalyst loading process was illustrated in Fig. S2.†

Growth of CNTs on QF filters with gradient nanostructures. The growth of CNTs with gradient nanostructures upon the QF filters was accomplished using a chemical vapour deposition (CVD) method. Before the CVD process, the catalyst-loaded QF filter was annealed at 800 °C for 15 min in Ar at a flow rate of 300 ml min⁻¹. After that, a gas flow of Ar (300 ml min⁻¹), H₂ (200 ml min⁻¹) and ethylene (200 ml min⁻¹) was introduced into a quartz tube (inner diameter 73 mm) from the beginning of the CVD process. The furnace was heated to 800 °C over 40 min and maintained at 800 °C for 60 min. Then, the furnace was cooled down to room temperature in an Ar (300 ml min⁻¹) and H₂ (10 ml min⁻¹) atmosphere. The CNT/QF filters with area of 100 cm² were fabricated.

Characterization of the materials

The morphology of the pristine QF filter and CNT/QF filter were characterized using scanning electron microscopy (SEM) (JSM 7401F) and transmission electron microscopy (TEM) (JEM 2010). Mercury intrusion porosimetry (for determining the pore distributions in the filters) was performed using a Micromeritics AutoPore IV 9500 system (Micromeritics Corp., Norcross, GA). Brunauer–Emmett–Teller (BET) surface areas were determined using a Quadra Sorb Station (Quantachrome Instruments Corp., Florida, USA). Thermogravimetric analysis (TGA) measurements were carried out on a thermogravimetric analyzer instrument (TGA Q500) at a scan rate of 10 °C min⁻¹ from 30 to 1000 °C with a mixed flow of O₂ (50 ml min⁻¹) and N₂ (20 ml min⁻¹).

Testing filtration performance

To measure the filtration performance, the filter was placed into a filter holder with an inner area of 100 cm² at a face velocity of 5.31 cm s⁻¹. N₂ was used as the carrier gas. Aerosols were generated by atomizing a DEHS solution using a constant output atomizer (model 3076, TSI) operating in the recirculation mode. Aerosol particles were then dried using a diffusion drier. The aerosol concentration and size distribution upstream and downstream of the filter were measured using a scanning mobility particle sizer (SMPS, model 3936, TSI), which consists of an electrostatic classifier (model 3080, TSI) equipped with a long differential mobility (model 3081, TSI) and ultra condensational particle counter (UCPC 3776 low, TSI). The pressure drop for the filters was measured using a pressure transmitter (P3000T, Bestace). A typical DEHS aerosol concentration and size distribution upstream of the filter is shown in Fig. S3.† To measure the filtration performance changes in filtration under continuous aerosol loading, the concentration of aerosol was measured every 3 min, and the pressure drop of the filter was recorded every 30 seconds. At least three experiments were performed to ensure reproducibility of the results.

To measure the morphology of aerosols loading on the fiber surface of the air filter, a 0.1 g ml⁻¹ NaCl aqueous solution was atomized using a constant output atomizer (model 3076, TSI) and then dried using a diffusion drier. NaCl aerosols were carried by N₂ and passed through filters at a velocity of 9.18 cm s⁻¹ for a period of 60 seconds and 6 min, respectively. A typical NaCl aerosol concentration and size distribution upstream of the filter is shown in Fig. S4.†

Results and discussion

Depositing catalysts on QFs on a large-scale is the first step towards the large-scale fabrication of CNT/QF filters. Although various methods for depositing catalysts on structured surfaces have been reported,³⁰ herein we have developed a facile method for depositing catalysts on porous structures, as shown in Scheme 1. According to the principles of air filtration, aerosol particles will be captured on the surface of the fibers and their contents will decrease gradually in the filter along the direction of air flow (Scheme 1a). Therefore, if catalyst aerosols are filtrated by a fibrous filter, they will deposit on the fiber surface and form a thin layer around the fiber (Scheme 1b-2). After annealing, catalyst nanoparticles will form on the fiber surface (Scheme 1b-3). Then, CNTs can grow on the fiber surface after the CVD process (Scheme 1b-4). The CNT content will decrease gradually in the thickness direction of the filter, which is consistent with the concentration distribution of catalyst aerosols in the filter (Scheme 1a). For air filtration, the CNT-rich side of the CNT/QF filter will be placed downstream to remove air pollutants.

Scheme 1 The synthetic process used to prepare the CNT/Quartz fiber filter. (a) Schematic illustrating the catalyst loading on a Quartz fiber filter using an aerosol technique and the growth of CNTs with its constant decrease along the thickness of the filter to form a CNT-based air filter with gradient structure. The CNT-rich side will be placed facing downstream for air filtration. (b) Schematic further illustrating the CNT growth process on the QF filter. 1 stands for the pristine QF filter, 2 stands for the catalyst layer formed on the QFs after catalyst nanoaerosol filtration by the QF filter, 3 stands for the catalyst nanoparticles formed on the QFs after annealing, 4 stands for CNT growth on the QF after the CVD process.

Fig. 1 shows the process of CNTs growth on a QF filter and the morphologies of the CNT/QF filter. The pristine QF filter is formed by randomly stacked QFs with diameters in the range of 0.3–5 μm. A typical macroporous structure can be seen in Fig. 1a. After catalyst aerosols are filtrated by the QF filter, they will deposit on the surface of the QFs and form thin layers (Fig. 1b). These catalyst layers are non-uniformly coated on the QFs as indicated by black arrows (Fig. 1b), which can be seen more clearly in Fig. 1c, the catalyst layers will transform into catalyst nanoparticles on the QFs after annealing. The non-uniform coverage of catalyst nanoparticles on the QFs can be attributed to the disturbance of the airflow by the nearby fibers. Therefore, the catalyst nanoparticles can only come into partial contact with the QFs. As a result, CNTs are grown in bunches on the QFs since catalysts cover the QFs non-uniformly (Fig. 1d). The pristine QF filter has a high filtration efficiency and very few catalysts could penetrate through the filter. Therefore, there are very few catalysts deposited on the downstream side of the QF filter. Consequently, on downstream side there are almost no CNTs grown (Fig. 1e). The CNTs synthesized in this work are multi-walled CNTs with diameters in the range of 10–30 nm (Fig. 1f).

Fig. 1 SEM (a–e) and TEM (f) images of the QF and CNT/QF filters. (a) the pristine QF filter; (b) QF filter with a deposited catalyst layer (as indicated by black arrows); (c) QF filter with a deposited catalyst layer after annealing (inset shows the high-magnification SEM image of the catalyst nanoparticles); (d) the as-obtained CNT/QF filter after the CVD process (black arrow shows the part of fiber where no CNTs grow); (e) the side with almost no CNTs; (f) TEM image of the as-grown CNTs (inset shows the high-magnification TEM image of the end of a CNT).

The fabricated CNT/QF filter has gradient nanostructures, as shown in Fig. 2. The as-prepared CNT/QF filter can be divided into three separate layers along the thickness direction, which include the top layer (the upstream layer upon catalyst loading), the middle layer and the bottom layer (Fig. 2a). Fig. 2b, d and f show the high-magnification SEM images of the upper layer in Fig. 2a from top to bottom along the direction of air flow. It can be found that the amount of CNTs in the QF filter shows a sharp decrease in this direction (indicated by the red dash circle).

Fig. 2 SEM images of the cross-sectional view of the CNT/QF filter. (a) Low-magnification SEM image of the cross-sectional view of CNT/QF filter (the blue, purple and grey dashed box shows the top layer, the middle layer and the bottom layer of the CNT/QF filter respectively); (b, d and f) high-magnification SEM images of the upper layer in (a) from top to bottom (the red dashed circle represents the CNTs on the QF filter); (c) TGA traces for the pristine QF filter, CNT/QF filter, and the top, middle, and bottom layer of the CNT/QF filter; (e) exponential fit trace of the CNT content for the top, middle, and bottom layer of the CNT/QF filter.

Thermogravimetric analysis (TGA) measurements were used to study the content of CNTs in the CNT/QF filter and its distribution in each layer (Fig. 2c). Both CNT/QF filter and its different layers have a weight loss between 400–670 °C, which can be attributed to the decomposition of CNTs. The total content of CNTs in the CNT/QF filter was calculated to be 1.17%, while the CNT content in the top, middle and bottom layer was calculated to be 2.97%, 0.68% and 0.56%, respectively (Table 1). The TGA results have confirmed that the CNT/QF filters have gradient nanostructures with CNT content decreasing along the thickness direction. The growth of CNTs will mainly be concentrated on the top layer of the QF filter, which is in the upstream side during catalyst loading. Brown et al.³¹ derived an expression for the mass deposition profile of a homogeneous fibrous filter under continuous loading of monodisperse aerosols. They found the concentration of deposited aerosols exponentially decreased along the thickness direction from the upstream layer to the downstream layer for a non-clogging filter. In this work, the pristine QF filter can be considered a homogeneous fibrous filter. During the catalyst deposition process, the concentration of catalyst aerosols and the velocity of the carrier gas are kept at a low level so that the clogging of QF filter is avoided. This can also be confirmed in Fig. 1b, where catalyst aerosols only form thin layers around the QF rather than porous thick layer on the filter surface. Therefore, Brown's expression can be applied to estimate the catalyst deposition profile for the QF filter in spite of the polydisperse catalyst aerosols used in this work.^8,9 The concentration of catalyst decreases exponentially along the thickness direction, which results in an exponential decrease in CNT content in the same direction, as shown in Fig. 2e.

Table 1 Comparison of the structural properties and filtration performance of the QF filter and the CNT/QF filter

	QF filter	CNT/QF filter
CNT content (wt%)	0	1.17 (total)	2.97 (upper layer)
			0.68 (middle layer)
			0.56 (bottom layer)
Thickness (mm)	0.37	0.43
Density (g cm^−3)	0.22	0.20
Porosity (%)	90	90.1
Filter specific area (m^2 m^−3)	5.3 × 10^5	8.6 × 10^5
Pressure drop (Pa) at 5.31 cm s^−1	407	435
Penetration at MPPS (MPPS)	2.45 × 10^−4 (88.2 nm)	4.13 × 10^−5 (63.8 nm)
Qf for MPPS (kPa^−1)	20.43	23.21

---

The structural properties and filtration performance of the QF filter and the CNT/QF filter are summarized in Table 1. The penetration at the most penetrating particle size (MPPS) of the CNT/QF filter is reduced by nearly one order of magnitude when compared with the QF filter, which indicates an obviously higher filtration efficiency of the CNT/QF filter than the QF filter. Besides, the MPPS of the filter has decreased from 88.2 nm to 63.8 nm after the CNTs were grown on the QF filter. The shift of MPPS towards smaller particles can be attributed to an enhancement of the diffusion mechanism due to the introduction of CNTs.¹² At the same time, the pressure drop of the CNT/QF filter only increased by 6.9% when compared with the QF filter. The quality factor (Q_f) is usually taken as the criterion for comparing the filtration performance of different filters, which is expressed by:

where P is the penetration and Δp is the pressure drop for the air filters. According to classical filtration theory, a higher Q_f indicates the better performance of the filter.² The CNT/QF filter has a much higher Q_f than the QF filter in all the particle sizes tested (Fig. S5†), which indicates the deposition of CNTs on the QF filter has enhanced the filtration performance of the raw filter. This is consistent with the result that we have reported before.²³ The role of the CNTs in improving filtration efficiency and the dust holding capacity of the raw filters have been investigated using SEM and the results shown in Fig. S6.†

The dynamic filtration process of the CNT/QF filter and the QF filter are shown in Fig. 3. Since DEHS aerosols are liquid aerosols, the filtration of liquid aerosols by oleophilic fibrous filters under continuous aerosol loading can be divided into four separate steps.^32,33 During the first step, the filter will be wetted by the deposited liquid particles and a thin film will form around the fiber. This film has very little interference on the airflow. However, it will reduce the fiber collection area. As a result, the pressure drop will increase slowly whereas the penetration will increase rapidly.⁶ As liquid aerosols continue to accumulate on the fiber surface, the filtration process enters into the second step. The thin aerosol film will be broken up due to its Plateau-Rayleigh instability^34,35 and droplets will form on the fiber surface. These droplets can further coalesce to form bigger drops and bridges between fibers,³⁶ leading to an exponential increase in penetration, which can also be attributed to a decrease in fiber collection area. The pressure drop will increase rapidly during this step. In the third stage, the interstices between fibers will be occupied by large drops in a short time, which results in a massive increase in pressure drop. At the same time, the gas interstitial velocity will increase and contribute to a diminution of the penetration by enhanced impaction collection.⁴ At the end of clogging, the filter will reach an equilibrium state. In this stage, a pseudo-stationary state is established between collection and drainage of the liquid droplets. The pressure drop and filter penetration will be constant.

Fig. 3 (a) The rate of pressure drop increase versus filtration test time for the CNT/QF filter with different placement position and the QF filter under continuous aerosol loading; (b) the decrease in rate of efficiency at MPPS versus filtration test time for the CNT/QF filter with different placement position and the QF filter under continuous aerosol loading.

Both the QF filter and the CNT/QF filter can be well wetted by DEHS droplets (Fig. S7†), so their dynamic filtration processes are in compliance with the above rules, as shown in Fig. 3a and b. The pressure drop of the QF filter and the CNT/QF filter is significantly increased under continuous aerosol loading during the filtration test (Fig. 3a). At the same time, the filtration efficiency decreases with test time (Fig. 3b). These characteristics suggest that the filtration test is undergoing the first and second step of the dynamic filtration process. It is found that the CNT/QF filter with the CNT-rich side positioned downstream have an obviously slower pressure drop rise than the QF filter (Fig. 3a). Although the evaluation of the service life of air filter is a challenging problem due to the changing external environment and the influence of environmental factors on filter loading such as aerosol concentration, particle size distribution, and temperature,¹⁰ it is generally recommended that HEPA filters (high efficiency particulate air filter with efficiency higher that 99.97% at MPPS) should be replaced when the pressure drop reaches twice of their initial value.³⁷ Therefore, in this work we use the time required when the filter pressure drop reached two times that of the initial value to evaluate the service life of each filter; detailed information regarding this is shown in Table S1.† For the QF filter, the service life is only 11 min due to the high concentration of testing aerosols used in this work, while for the CNT/QF filter with the CNT-rich side positioned downstream, the service life is 18 min, which is 1.64 times that of the QF filter. However, if the CNT-rich side is placed upstream, the pressure drop of the CNT/QF filter increases very quickly and the service life is only 7.5 min. The different placement positions of the CNT/QF filter will result in a 2.4 times difference in its service life.

At the same time, it is found that the filtration efficiency of the QF filter shows a sharp decline under continuous aerosol loading. It rapidly no longer meets the standard of a HEPA filter (Fig. 3b). In contrast, the filtration efficiency of the CNT/QF filter with the CNT-rich side positioned downstream remains at a very high level that is above the standard of a HEPA filter, as shown in Fig. 3b and Table S1.† If the CNT-rich side is placed upstream, the decrease in filtration efficiency of the CNT/QF filter becomes faster, but is still lower than that of the QF filter. Therefore, we can conclude that the CNT/QF filter with the CNT-rich side positioned downstream is a much better air filter than the CNT/QF filter with the CNT-rich side positioned upstream, even if they have the same filtration performance and Q_f at the initial stage of air filtration. While comparing the filtration performance of the pure QF filter and the CNT/QF filter, we can further conclude that the filter with a higher Q_f is not necessarily a better filter since it may have a much shorter service life. The evaluation of filter performance should be based on both the Q_f of the filter during the initial filtration stage and its service life during the dynamic filtration process.

NaCl aerosols are used to study the loading behaviour of the QF filter and CNT/QF filter, since DEHS aerosols are not suitable for SEM observation. The results are shown in Fig. 4. From Fig. 4a and b, it can be found that there are much more NaCl aerosols deposited on the filter surface when the CNT-rich side is placed upstream. This proves that when the high filtration efficiency layer is placed upstream, the filters are easily clogged. It is noteworthy that there is a trend to fabricate air filters through coating a nanofiber layer on the surface of macrofiber layers.^10,13,38 However, this is not a rational structure since the nanofiber layer will accelerate the “clogging effect” and reduce the service life of the filter, even if it will improve the filtration efficiency and Q_f during the initial stage. When aerosols are continuous deposited on the QF (Fig. 4c), they will accumulate into large aggregates and clog the pores between fibers (indicated by a black arrow). However, when CNTs are grown on the QFs (Fig. 4d), aerosol aggregates will be much smaller (indicated by a black arrow), since CNTs have a high specific surface area for aerosol loading. As a result, the resistance to air flow will also be smaller. This explains why the CNT/QF filter with the CNT-rich side positioned downstream has a much higher service life than the QF filter.

Fig. 4 SEM images of the filters with NaCl aerosol particles deposition for 6 min. (a) NaCl aerosol particles deposited on the surface of the QF filter; (b) NaCl aerosol particles deposited on the surface of the CNT/QF filter; (c) NaCl aerosol particles accumulated on a single QF (the black arrow shows the formation of large aerosol particle aggregation on a single QF); (d) NaCl aerosol particles accumulated on an isolated CNT/QF fiber (the black arrow shows the formation of small aerosol particle aggregation on an isolated CNT/QF fiber).

Based on the above results, it can be seen that the growth of CNTs with gradient distribution is very important for the performance of the CNT/QF filter, which combines high filtration efficiency, high Q_f, and high service life. The influence of the CNTs and their structural arrangement on the filtration performance can be illustrated in Fig. 5. For the CNT/QF filter, CNTs are grown with a gradient structure on the QF filter. If the CNT-rich side is placed upstream (Fig. 5a), the surface layer will have a high filtration efficiency and the majority of DEHS aerosols will be trapped on the surface of the filter. As a result, the “clogging effect” will be accelerated. In this work, DEHS aerosols will form large drops on the fiber surface very quickly, resulting in a rapid increase in pressure drop and decrease in filtration efficiency. On the contrary, if the CNT-rich side is placed downstream (Fig. 5b), the upstream layer has a low filtration efficiency and there will be much less DEHS aerosols trapped on the surface of filter. Thus, the “clogging process” will be slowed down. The rest of the aerosols, which have penetrated through the surface layer will be removed by the CNT-rich layer with high efficiency. At the same time, the rest of the aerosols, which have penetrated through the upstream layer have a very limited concentration. Therefore, their clogging effects to the CNT-rich layer will be drastically weakened. As a result, the CNT/QF filter with the CNT-rich side positioned downstream will have both a high efficiency and a high service life. When compared with the QF filter (Fig. 5c), the advantages of the CNT/QF filter with the CNT-rich side placed downstream lies in the high specific surface area of the CNTs. Therefore, the rapid formation of large aerosol aggregates or drops, which occupy the voids between fibers will be avoided. As a result, the increase in pressure drop and the decrease in filtration efficiency of the CNT/QF filter will be slower than that observed with the QF filter.

Fig. 5 An illustration of aerosols accumulating in the CNT/QF filter with different placement position and the QF filter.

Conclusions

In summary, hierarchical CNT/QF filters with gradient nanostructures have been fabricated with the assistance of an aerosol technique. When compared with the pristine QF filter, the CNT/QF filter has a significantly improved filtration efficiency and an obviously higher Q_f. Furthermore, the CNT/QF filter with the CNT-rich side positioned downstream shows a dramatically longer service life than the QF filter. We also found that when the CNT-rich side was placed upstream, the service life will be greatly shortened. Although the CNT/QF filter showed the same filtration performance at the clean state when being placed in an opposite direction, their performance during the dynamic filtration process are obviously different. This study provides a new strategy towards the fabrication of high performance air filters through taking advantages of each structural unit in a hierarchical structure, which includes microfibers with high structural stability and CNTs with a high specific surface area, combing the high dust holding capacity of the upstream QF layer and the high filtration efficiency of the downstream CNT-rich layer.

Acknowledgements

The work is supported by the Boeing Company and the National Science Foundation of China (21203107), National Natural Science Funds of China (50976058), and National Key Basic Research and Development Program (2013CB228506).

Notes and references

1. R. C. Brown, Air filtration: an integrated approach to the theory and applications of fibrous filters, Pergamon press Oxford, England, 1993.
2. C. Y. Chen, Chem. Rev., 1955, 55, 595–623.
3. C. N. Davies, Air filtration, Academic press, London, 1973.
4. P. Contal, J. Simao, D. Thomas, T. Frising, S. Callé, J. C. Appert-Collin and D. Bémer, J. Aerosol Sci., 2004, 35, 263–278.
5. C. B. Song, H. S. Park and K. W. Lee, Powder Technol., 2006, 163, 152–159.
6. D. C. Walsh, J. I. T. Stenhouse, K. L. Scurrah and A. Graef, J. Aerosol Sci., 1996, 27(suppl. 1), S617–S618.
7. S. Payet, D. Boulaud, G. Madelaine and A. Renoux, J. Aerosol Sci., 1992, 23, 723–735.
8. W. W.-F. Leung and C.-H. Hung, Sep. Purif. Technol., 2012, 92, 174–180.
9. W. W.-F. Leung, C.-H. Hung and P.-T. Yuen, Aerosol Sci. Technol., 2009, 43, 1174–1183.
10. K. Graham, M. Ouyang, T. Raether, T. Grafe, B. McDonald and P. Knauf, Polymeric nanofibers in air filtration applications, 2002.
11. H. Misslitz, K. Kreger and H.-W. Schmidt, Small, 2013, 9, 2025.
12. A. Podgórski, A. Bałazy and L. Gradoń, Chem. Eng. Sci., 2006, 61, 6804–6815.
13. G. Viswanathan, D. B. Kane and P. J. Lipowicz, Adv. Mater., 2004, 16, 2045–2049.
14. O. Yildiz and P. D. Bradford, Carbon, 2013, 64, 295–304.
15. R. Balamurugan, S. Sundarrajan and S. Ramakrishna, Membranes, 2011, 1, 232–248.
16. R. S. Barhate and S. Ramakrishna, J. Membr. Sci., 2007, 296, 1–8.
17. P. Gibson, H. Schreuder-Gibson and D. Rivin, Colloids Surf., A, 2001, 187–188, 469–481.
18. B. Maze, H. Vahedi Tafreshi, Q. Wang and B. Pourdeyhimi, J. Aerosol Sci., 2007, 38, 550–571.
19. Z. Zhang, Ph.D. Thesis, University of Minnesota, 1989.
20. Y. Ahn, S. Park, G. Kim, Y. Hwang, C. Lee, H. Shin and J. Lee, Curr. Appl. Phys., 2006, 6, 1030–1035.
21. N. Wang, X. Wang, B. Ding, J. Yu and G. Sun, J. Mater. Chem., 2012, 22, 1445–1452.
22. N. Halonen, A. Rautio, A.-R. Leino, T. Kyllonen, G. Toth, J. Lappalainen, K. Kordás, M. Huuhtanen, R. L. Keiski and A. Sápi, ACS Nano, 2010, 4, 2003–2008.
23. P. Li, Y. Zong, Y. Zhang, M. Yang, R. Zhang, S. Li and F. Wei, Nanoscale, 2013, 5, 3367–3372.
24. A. G. Nasibulin, A. Kaskela, K. Mustonen, A. S. Anisimov, V. Ruiz, S. Kivisto, S. Rackauskas, M. Y. Timmermans, M. Pudas and B. Aitchison, ACS Nano, 2011, 5, 3214–3221.
25. H. Parham, S. Bates, Y. Xia and Y. Zhu, Carbon, 2013, 54, 215–223.
26. J. H. Park, K. Y. Yoon, H. Na, Y. S. Kim, J. Hwang, J. Kim and Y. H. Yoon, Sci. Total Environ., 2011, 409, 4132–4138.
27. S. J. Park and D. G. Lee, Carbon, 2006, 44, 1930–1935.
28. S. J. Park and D. G. Lee, Curr. Appl. Phys., 2006, 6, e182–e186.
29. O. Yildiz and P. D. Bradford, Carbon, 2013, 64, 295–304.
30. V. Meille, Appl. Catal., A, 2006, 315, 1–17.
31. R. C. Brown and D. Wake, J. Aerosol Sci., 1999, 30, 227–234.
32. T. Frising, D. Thomas, D. Bémer and P. Contal, Chem. Eng. Sci., 2005, 60, 2751–2762.
33. A. Bredin and B. J. Mullins, Sep. Purif. Technol., 2012, 90, 53–63.
34. B. J. Mullins and G. Kasper, Chem. Eng. Sci., 2006, 61, 6223–6227.
35. R.-J. Roe, J. Colloid Interface Sci., 1975, 50, 70–79.
36. P. C. Raynor and D. Leith, J. Aerosol Sci., 2000, 31, 19–34.
37. Z. Xu, Fundamentals of air cleaning technology and its application in cleanrooms, Springer, 2014.
38. D. Cho, A. Naydich, M. W. Frey and Y. L. Joo, Polymer, 2013, 54, 2364–2372.
39. X. Li, J. Yang, Y. Hu, J. Wang, Y. Li, M. Cai, R. Li and X. Sun, J. Mater. Chem., 2012, 22, 18847–18853.
40. Q. Zhang, J. Q. Huang, W. Z. Qian, Y. Y. Zhang and F. Wei, Small, 2013, 9, 1237–1265.

---

Footnote

† Electronic supplementary information (ESI) available: Typical size distribution of atomized catalyst aerosols used for catalyst loading on the QF filters; a schematic of the catalyst loading process of the QF filter and a digital photo of the as-prepared CNT/QF filter; typical size distribution of the atomized polydisperse DEHS aerosols used for the air filtration test; typical size distribution of the atomized polydisperse NaCl aerosols used for loading on the CNT/QF filter and the QF filter; the filtration performance of the QF filter and the CNT/QF filter; SEM images of the filters with NaCl aerosol particles deposition for 60 seconds; the contact angles test of DEHS droplets on the QF filter and the CNT/QF filter. See DOI: 10.1039/c4ra08746a

---