Electrospun PVB/AVE NMs as mask filter layer for win-win effects of filtration and antibacterial activity

Abstract

The COVID-19 pandemic has caused serious social and public health problems. In the field of personal protection, the facial masks can prevent infectious respiratory diseases, safeguard human health, and promote public safety. Herein, we focused on preparing a core filter layer for masks using electrospun polyvinyl butyral/apocynum venetum extract nanofibrous membranes (PVB/AVE NMs), with durable interception efficiency and antibacterial properties. In the spinning solution, AVE acted as a salt to improve electrical conductivity, and achieve long-lasting interception efficiency with adjustable pore size. It also played the role of an antibacterial agent in PVB/AVE NMs to achieve win-win effects. The hydrophobicity of PVB-AVE-6% was 120.9° whereas its filterability reached 98.3% when the pressure drop resistance was 142 Pa. PVB-AVE-6% exhibited intriguing properties with great antibacterial rates of 99.38% and 98.96% against S. aureus and E. coli, respectively. After a prolonged usability test of 8 h, the filtration efficiency of the PVB/AVE masks remained stable at over 97.7%. Furthermore, the antibacterial rates of the PVB/AVE masks on S. aureus and E. coli were 96.87% and 96.20% respectively, after using for 2 d. These results indicate that PVB/AVE NMs improve the protective performance of ordinary disposable masks, which has certain application in air filtration.

Graphical abstract

1. Introduction

Respiratory droplet transmission is the main mode of infecting people during global epidemics of respiratory diseases, which in severe cases can even lead to large-scale infections. Masks are reported to be effective in reducing the spread of pathogens [1]. Generally, ordinary disposable masks comprise three layers: the outer layer prevents the splashing of liquid, the middle layer absorbs particles, and the inner layer resists moisture. There are airborne pollutants such as particulates, and several microorganisms smaller than 0.3 μm which cannot be filtered using conventional air filtration membranes (listed in the supporting information) [[2], [3], [4]]. While the mask is an effective barrier to airborne particulates transmission, commonly used disposable mask materials are not inherently able to deactivate bacteria upon contact [5,6], which may result in pathogenic bacteria attaching to masks and causing cross infection [7,8]. In addition, wearing masks for a long time results in moisture accumulation, enabling bacterial growth [9,10]. The majority of bacteria (90%) remain alive even after 8 h, thus increasing the risk of infection. It is therefore advised to change masks regularly [11,12]. To improve the antibacterial performance of masks, proper methods should be adopted to eliminate germs and reduce the possibility of infection. So far, a series of protection regulations have been issued for the antibacterial performance of masks, such as the group standard T/CIAA 003–2020 “Antimicrobial Masks” and the “To Do Right Job, Dress Right” document [13]. Clearly, the development of innovative functional masks is the need of the future.

The filter material of masks is generally made of non-woven fabrics, commonly prepared via melt-blowing [14,15]. An increasing number of research studies has focused on the development of electrospinning technologies because they are promising in the field of filtration due to the properties of high specific surface area, porous structure, controllable morphology, and multi-functional fibre surface [16,17]. On one hand, electrospinning combined with physical cross-linking methods, electrostatic spraying, and other technologies, can produce micro-nanofibre membranes with special structures [[18], [19], [20]] that can achieve high filtration and low-pressure drop resistance. On the other hand, antibacterial substances such as oxides, metals, and other materials can be added to the spinning solution system to achieve the antibacterial function of the fibre membrane [[21], [22], [23], [24]]. However, the addition of materials increases the difficulty of spinning, leading to the need for simple methods and adjustment of the complex spinning system.

Given the stability of naturally extracted substances, their application in functional materials has received widespread attention. Many researchers have explored the versatility of natural extracts, hoping to enhance the functionality and durability of materials through different techniques and achieve unique functions [[25], [26], [27]]. Our research group has been studying apocynum venetum for a while; it is a plant that has high adaptability to saline and acid environments [[28], [29], [30]] and its roots, stems, and leaves contain several antibacterial substances exhibiting bactericidal, anti-inflammatory, and medicinal properties [31]. The long-lasting antimicrobial properties of apocynum venetum extract (AVE) are linked to the unique structure of its compounds, whereas some of the features of plants need to be further explored in practice. With the right extraction techniques, AVE can be blended with other materials and achieve more versatile composites. Cellulose nanofibres (CNFs), PLA/CNF films, and CNFs reinforced chitosan-based composite hydrogels are a few of the materials reported in the literature [32,33]. In this study, Apocynum venetum leaves were combined with polyvinyl butyral (PVB) to prepare a filter membrane material for mask filtration. PVB is a polymer prepared from polyvinyl alcohol and butyraldehyde, which is a white powder at room temperature and is non-toxic with no pungent odour [34,35]. Compared to commonly used nanofibrous materials as mask filters, such as polyamide (PA), polyvinylidene fluoride (PVDF), and polyacrylonitrile (PAN), PVB has excellent solubility and spinnability, and is usually soluble in organic solvents such as alcohols, ethers, and ketones, demonstrating its environmentally friendly properties [36,37]. In this study, based on the properties of PVB and AVE, we attempted to fabricate structure-controlled functional membranes and explore their mechanism of action. Our results provide an opportunity to construct a simple method for achieving the industrial use of PVB/AVE NMs.

2. Experimental

2.1. Materials

All reagents were of analytical grade, and used without further purification. The list of reagents is included in the supporting information.

2.2. Extraction of AVE

Apocynum venetum leaves were washed thoroughly with distilled water, dried, and ground. The ground powder was refluxed in 72.4% ethanol at 60 °C for 3.3 h and then filtered. The filtrate was evaporated using a rotary evaporator (RE-52A, Shanghai Arong Instrument Co. Ltd., China.) at 45 °C. The sample was freeze-dried overnight in a freeze-dryer and then ground to a powder. The detailed procedure is described in the supplementary material.

2.3. Preparation of polyvinyl butyral/Apocynum venetum extract nanofibrous membranes (PVB/AVE NMs)

Electrospinning solutions of different AVE concentrations (0, 4, 6, and 8 wt%) were prepared by dissolving AVE in anhydrous ethanol at 37 °C under stirring for 1 h, while keeping the concentration of PVB at 8 wt% in all solutions. The spinning parameters were set as follows: the flow rate was 1.5 mL/h, the tip-to-collector distance was 16 cm, the voltage was set at 25 kV, high pressure was applied at 1/3 of the tip, and the receiver drum parameter was set to 400 r min⁻¹. The nanofibre membranes were designated as PVB-AVE-0%, PVB-AVE-4%, PVB-AVE-6%, and PVB-AVE-8%, indicating the different AVE concentrations.

2.4. Evaluation of filtration performance

A particulate filtration efficiency tester (FYY268, Wenzhou Fangyuan Instrument Co. Ltd., China) was used to measure the air filtration efficiency and pressure drop resistance of the fibre membranes. The filter test used neutralized NaCl aerosol particles of 300 nm mass median diameter at the air velocity of 85 L min⁻¹. Prepared samples were in the 100 cm² test area. The quality factor (QF) was calculated using equation (1). The detailed experimental procedure is included in the supporting information.

$$Qf=-\frac{in\left(1-\epsilon \right)}{\Delta \rho }$$ (1)

Where QF is the quality factor of the test mask filter material, ε is the filtration efficiency of the test mask filter material, ρ is the pressure drop resistance of the mask filter material.

2.5. Antibacterial activity determination

Gram-negative Escherichia coli (E. coli) (ATCC 25922) and gram-positive Staphylococcus aureus (S. aureus) (ATCC 25923) were selected as representative bacteria for the PVB/AVE NMs antimicrobial tests in accordance with ASTM E2149-13a. The reduction rate was calculated using equation (2). The detailed experimental procedure is described in the supporting information.

$$\mathrm{R}\mathrm{e}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e},\%\left(\mathrm{C}\mathrm{F}\mathrm{U}{\mathrm{m}\mathrm{L}}^{-1}\right)=\frac{\left(A-B\right)}{A}\times 100$$ (2)

where A = CFU mL⁻¹ for the control sample, and B = CFU mL⁻¹ for the treated sample.

2.6. Characterisation

The AVE components were detected using high performance liquid chromatography (HPLC, 1200 series, Agilent Technology Co., America). The morphologies of PVB/AVE NMs and bacterial cell membrane were observed by scanning electron microscope (SEM, TGY3100, Hefei Guoyi Quantum Technology Co. Ltd., China) after coating with gold sputtering. The diameter of the nanofibres was measured using an image analysis software. The pore size structure of PVB/AVE NMs was measured using a bubble method (CFP1100A, Porous Materials Inc., America). The chemical composition of PVB/AVE NMs was analysed using an fourier transform infrared spectrometer (FT-IR spectrometer, Bruker Vertex 70, Bruker Corporation, Germany) in the range of 500–4000 cm⁻¹. The wettabilities were characterized using an Optical Contact Angle Measuring Device (OCA15EC, DataPhysics Instruments Co., Germany). The tensile stress, tensile strain, and other mechanical properties of PVB/AVE NMs were tested using a tensile testing machine (XQ-1C, Shanghai New Fiber Instrument Co. Ltd., China). A pretension of 0.1 CN was applied to the samples and the clamping distance parameter was set at 10 mm.

3. Results and discussion

3.1. Preparation and optimisation of PVB/AVE NMs

The preparation and optimisation procedures of PVB/AVE NMs are presented in Fig. 1 (a). AVE was used in the preparation of PVB/AVE NMs. AVE is mainly composed of flavonoids (Fig. S1), which contains gallic acid, catechin, hyperoside, and rutin, with the rutin content being higher than the other compounds. AVE influenced the surface charge density of the electric-jet, allowing the fibres to be stretched and refined with the support of strong electric force. Furthermore, it acted as a highly effective natural antibacterial agent [38]. AVE was efficiently extracted from apocynum and the membranes were prepared through electrospinning by adding AVE to the solution. The optimisation of PVB/AVE NMs was therefore conducted based on the following effects: (1) High air filtration efficiency and low pressure drop resistance, (2) high level of bactericidal action efficiency.

Fig. 1.

(a) Design and fabrication of PVB/AVE NMs. SEM images of (b) PVB-AVE-0%, (c) PVB-AVE-4%, (d) PVB-AVE-6%, and (e) PVB-AVE-8%. Diameter images of (f) PVB-AVE-0%, (g) PVB-AVE-4%, (h) PVB-AVE-6%, (i) PVB-AVE-8%. (j) Pore size distribution of PVB/AVE NMs with different AVE concentrations. (k) Average pore size and pore size ratio of PVB/AVE NMs. (l) FTIR spectra of PVB/AVE NMs.

The conductivity of the PVB/AVE spinning solutions increased from 2.8 to 18.2 mS/cm at different AVE concentrations (Table 1 ), which is due to the addition of AVE. AVE as a salt changed the conductivity of the solution and affected the viscosity, while the surface tension was affected to a lesser extent because the solvent type did not change. From Fig. 1(b)–(i), it can be seen that PVB/AVE NMs exhibited different morphologies and diameter distributions when the AVE concentration changed from 0 to 8 wt%. All the membranes exhibited smooth and bead-free surfaces, with a randomly arranged three-dimensional structure. When the content of AVE increased from 0 to 6 wt%, the average PVB/AVE NMs nanofibre diameter decreased from 0.61 ± 0.14 to 0.32 ± 0.07 μm. For an AVE content of 8 wt%, the diameter increased to 0.54 ± 0.30 μm and was distributed between 3 orders of magnitude, presenting a multipolar size. The main reason for this phenomenon may be related to the properties of the spinning solution; the increased charge density led to greater electrical forces promoting the stretching of the electrospinning jet, which improved the homogeneity of the fibre diameters and refined it. However, the excessive viscosity hindered the fluidity of the polymer macromolecular chain, resulting in an inconsistent drafting process, and further influencing the diameter of the fibres [39,40]. Therefore, the morphology of the nanofibre membrane was mainly influenced by the electrical conductivity, surface tension, and viscosity of the spinning solution. When a certain equilibrium is reached, uniform fibres can be formed.

Table 1.

Properties of PVB spinning solution with different AVE concentrations.

Sample	Viscosity(mPa.s)	Conductivity(mS/cm)	Surface Tension(mN/m)
PVB	295.1	2.8	26.1
PVB-AVE-4%	308.7	6.9	26.2
PVB-AVE-6%	337.5	10.6	26.0
PVB-AVE-8%	370.2	18.2	26.2

The pore size was affected by the fibre diameter and arrangement, influencing the filtration effect. When AVE increased from 0 to 6 wt% (Fig. 1(j) and (k)), the pore size distribution of PVB/AVE NMs changed and the highest corresponding peak was reduced from 3.5 to 1.9 μm. Furthermore, the average pore size decreased from 3.6 to 2.4 μm, whereas the porosity increased from 78.8% to 85.8%. However, when the concentration of AVE increased to 8 wt%, the pore size of PVB-AVE-8% exhibited three different peaks (1.8, 2.2, and 2.9 μm), resulting in an average pore size of 2.4 μm and a porosity of 85.0%.

Fig. 1(l) shows the FT-IR spectra of PVB-AVE-6%, which was chosen as a representative sample and was compared to the raw materials to identify the components of the membranes. PVB-AVE-6% and PVB NMs exhibit the same characteristic peaks near 3415 and 2868 cm⁻¹, which were ascribed to the stretching vibrations of –OH and the asymmetric stretching vibrations of –CH₃, respectively. The bands at 1734 and 113 cm⁻¹ correspond to -C-O stretching vibrations and symmetric stretching vibrations [41]. In addition, the characteristic peaks near 1608, 1513, and 1267 cm⁻¹ were basically the same, while the characteristic peaks at 1608 and 1513 cm⁻¹ were attributed to the stretching vibration superposition of C O and the benzene ring (C C), respectively. At 1267 cm⁻¹, the absorption peak generated by the stretching vibration of -C-O, confirmed that AVE has been successfully added to PVB NMs [42].

3.2. Wetting and mechanical properties of PVB/AVE NMs

Water-repellence properties are important in nanofibre membranes for capturing fine particles attached to the droplet [43]. Hydrophilic filter materials swell in a humid environment by absorbing water vapours or droplets transmitted from the air, blocking the aperture channels of the nanofibre membranes, resulting in low mechanical force and high resistance to air pressure drop. On the contrary, hydrophobic filter materials are durable and can prevent the spread of droplets from outside intrusion [44]. The methyl orange experiment is described in the supporting information (Fig. S2). It also clarified that PVB/AVE NMs has good hydrophobicity. As seen in Fig. 2 (a), the water contact angle (WCA) of PVB/AVE NMs decreases from 135.2° to 113.0° as the AVE concentration increased from 0 to 8 wt%. This is because AVE contains polar substances; when the membrane comes in contact with water molecules, the surface energy increases, resulting in a decrease in the WCA of the nanofibre membrane. The hydrophilicity of PVB/AVE NMs was enhanced by AVE, whereas the WCA still remained above 90°, meaning that PVB/AVE NMs exhibit a certain hydrophobicity and are reliable filter materials for long-term use. The hydrophobicity of the membrane repelled the deposition of droplets on the mask, and thus improved its filtration efficiency [45].

Fig. 2.

(a) Water contact angle of PVB/AVE NMs, (b) Mechanical properties of PVB/AVE NMs.

According to the aforementioned findings, nanofibre membranes should exhibit certain mechanical properties to ensure their durability and applicability as the core layer of masks. In Fig. 2(b), upon addition of 0–8 wt% AVE, the tensile stress of PVB/AVE NMs increased from 5.0 to 6.4 MPa, and then decreased to 5.7 MPa. Furthermore, the tensile strain decreased from 42.4% to 40.1%, followed by a further increase to 46.6%. This was mainly attributed to the refinement of the fibre diameter and pore diameter channel, which is beneficial for increasing the total amount of fibre accumulation per unit volume. Moreover, the structure between the fibres was tight and the tensile strength of the fibre membrane was improved after stretching by an external force. When 8 wt% AVE was added, the uneven fibre diameter and pore size distribution led to inconsistent fibre forces that reduced the unit cross-sectional bonding point of the fibres. This increased the weak-link of PVB/AVE NMs, making it prone to slippage when subjected to external forces [46,47]. To conclude, all membranes exhibited certain mechanical properties, with PVB-AVE-6% being superior to the others.

3.3. Comprehensive filtration evaluation of PVB/AVE NMs

The filtration mechanism and the SEM images of PVB/AVE NMs are presented in Fig. 3 (a). The morphological changes were analysed before and after filtration. Typical filtration mechanisms of fibre materials involve diffusion effects, electrostatic deposition, gravitational deposition, inertial impact, and interception effects. The capture efficiency towards particles is usually influenced by the gas velocity and fibre diameters. When air containing NaCl particles passes through PVB/AVE NMs, the particles get intercepted on the electrospun membrane. The membrane exhibits strong adhesion and interception abilities toward the aerosol particles, which are tightly bound to the nanofibres via Van Der Waals Forces [48,49]. In our study, the surface morphology of the fibrous membranes did not change, and no fracture occurred during filtration, indicating that the fibrous membranes have good mechanical properties. Combined with the structural analysis of the fibrous membranes, we concluded that a network structure can attain high porosity, which can further improve the ability to capture particles as well as the permeability of the masks.

Fig. 3.

(a) PVB/AVE NMs filtration mechanism diagram with SEM images of PVB/AVE NMs before and after filtration, (b) Filtration efficiency and pressure drop resistance, (c) Quality factor, (d) Filtration performance of PVB/AVE NMs of different gram weight.

To verify the filtration performance of PVB/AVE NMs, the membranes were treated with NaCl particles at an air flow velocity of 85 L min⁻¹. In Fig. 3(b), the filtration efficiency of PVB/AVE NMs increased, followed by a decrease upon the addition of AVE. PVB-AVE-6% exhibited high filtration efficiency (98.3%), whereas the pressure drop resistance of the membrane was 142 Pa. The QF is used to access the comprehensive filtration performance related to membrane filter efficiency and pressure drop [50], with high QF values indicating excellent integrated filtration performance. As demonstrated in Fig. 3(c), upon addition of AVE, the QF value tends to increase and then gradually decrease. PVB-AVE-6% exhibited the highest QF among the four samples as well as the best comprehensive filtration efficiency, which was determined by the small aperture channel and uniform pore size distribution. Furthermore, the filtration efficiency and pressure drop curves of PVB-AVE-6% were analysed with different basis weights as shown in Fig. 3(d). When the weight increased from 0.574 to 2.984 g m⁻², the filtration efficiency increased from 72.1% to 99.8%, and the pressure drop resistance increased from 89 to 252 Pa. This finding was attributed to the excessive density of the fibres blocking the air flow passage and increasing the air resistance. As the weight increased, the growth rate of the filtration efficiency plateaued. The pressure drop continued to be high due to tiny particles adhering to each other and congesting the pore channels, further increasing the risk of being struck [51]. Therefore, PVB-AVE-6% exhibited the best filtration performance at the weight of 2.103 g m⁻².

3.4. Analysis of the antibacterial properties of PVB/AVE NMs

The antibacterial mechanism of PVB/AVE NMs is shown in Fig. 4 (a). The analysis revealed that: (1) AVE contains gallic acid, hyperoside, catechin, and rutin, all of which contain phenolic hydroxyl groups in their structure which can separate the phospholipid bilayer of bacterial cell membranes. This increases the permeability of cell membranes, leading to fluidization and expansion of the membrane lipids, resulting in leakage of membrane ions, nutrients, and contents. (2) Weakly acidic flavonoid compounds contain a large number of hydrophobic benzene rings that cause proteins to coagulate or deform, leading to bacterial death [52]. In addition, the hydroxyl group on the benzene ring of flavonoids can bind to carboxyl and amino groups contained in bacterial protein macromolecules, exhibiting a bacteriostatic effect [53].

Fig. 4.

Results of colony count experiment (a) Schematic illustration of antibacterial mechanism of PVB/AVE NMs, (b) PVB-AVE-0%, (c)PVB-AVE-6%. SEM images of S. aureus and E. coli on (d) PVB-AVE-0% and (e) PVB-AVE-6%.

The antibacterial performance of PVB-AVE-6% was evaluated by the colony counting method based on previous results. As shown in Fig. 4(b), a large number of bacterial colonies are present in the control sample plate of pure PVB NMs, confirming that PVB-AVE-0% exhibits no antibacterial properties. In contrast, a few bacterial colonies are present in the PVB-AVE-6% sample as shown in Fig. 4(c). The antibacterial rate of PVB-AVE-6% against S. aureus was 99.4%, which was better than that for E. coli (99.0%). Meanwhile, both E. coli and S. aureus were affected, demonstrating the broad spectrum of bactericidal properties of PVB-AVE-6%. Therefore, PVB-AVE-6% exhibited efficient antibacterial performance, enough to inhibit the bacterial substances attached to it.

In addition, SEM images revealed morphological changes of S. aureus and E. coli attached to PVB-AVE-6%, and the antibacterial mechanism was further studied. As shown in Fig. 4(d) and (e), the germs in contact with the control samples have a smooth and complete surface, indicating that PVB did not damage the structure of the bacterial cell. However, after being treated with PVB-AVE-6%, the bacterial content spilled out, resulting in depression, wrinkles, and even cracked bacterial cell membranes, indicating bacterial death. The zone-of-inhibition test of fibre membranes is presented in the supporting information (Fig. S3).

3.5. Analysis of the practical application and mechanism of masks

Fig. 5(a) shows the front and side view of a subject wearing a PVB/AVE mask. In this study, the samples were ordinary disposable masks with a core filter layer of 25 g cm⁻² melt-blown polypropylene. The filter material of the control mask was PVB-AVE-6%. Fig. 5(b) shows the filtration efficiencies of the selected masks worn by the subject for 8 h; the filtering efficiency of the PVB/AVE mask decreased from 98.3% to 97.7%, while that of disposable masks decreased from 90.9% to 77.5%. As shown in Fig. 5(c), the filtration efficiency of ordinary disposable masks is 90.9% at a gas flow rate of 85 L min⁻¹, lower than that of the PVB/AVE mask (98.3%), whereas the pressure drop resistance is 102 Pa, lower than that of the PVB/AVE mask. The QF of ordinary disposable masks and PVB/AVE masks were 0.0235 and 0.0288, respectively. These results show that the comprehensive filtration performance of PVB/AVE masks is superior to that of disposable masks.

Fig. 5.

(a) Front and side view of subjects wearing PVB/AVE masks, (b) Actual use performance results (change of filtration efficiency within 8hrs), (c) Comprehensive filtering performance result diagram, (d) Antibacterial map of blank control group, (e) Antimicrobial map of disposable mask, (f) Antibacterial map of PVB/AVE mask.

The antimicrobial performance of PVB/AVE NMs compared to other membranes is shown in Table S1, with PVB/AVE NMs being in the middle to upper range. A comprehensive test was conducted to test the durability of the masks. Subjects wore a disposable mask and a PVB/AVE mask for 2 days (8 h each day), after which the middle layer of the worn mask was removed for analysis. As shown in Fig. 5(d)–(f), a large number of colonies appeared on the surface of the control plate, suggesting that it had no inhibitory effect against E. coli and S. aureus. On the contrary, a small number of colonies appeared on the plate of the PVB/AVE mask, with the antibacterial rates of E. coli and S. aureus being 96.20% and 96.87%, respectively, indicating that PVB/AVE still exhibited antibacterial properties even after two days of use.

4. Conclusions

Multifunctional, mechanically stable, and highly-controlled electrospun PVB/AVE NMs were successfully fabricated by electrospinning methods, and used in protective masks to further enhance their efficiency. These nanofibrous membranes posed well-defined fibrous structures with high filtration efficiency, low air resistance, good water repellency, and excellent mechanical stability. Moreover, the PVB/AVE NMs exhibited long-lasting durability, high filtration efficiency (98.32%), and efficient bacteriostasis performance (99.38% for S. aureus and 98.96% for E. coli) after 8 h. The win-win situation was achieved due to the unique characteristics of AVE that offered excellent points of innovation. Using the mask filter material layer prepared by PVB/AVE materials furnished a straightforward and efficient route that deserved certain value in the development of personal protective filter masks.

Author contributions

Z. Y. S. Lou and L. Wang contributed equally to the experiments’ design, manufacture, analysis, and writing of the publication. K. F. Yu performed some collations. Q. F. Wei and H. Tanveer contributed to part of material characterization and the analyses of data and the discussions. X. Xia, H. M. Zhou, has made substantial contributions to the project administration and correction of the drafting the work.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

No data was used for the research described in the article.