(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Low cost centrifugal melt spinning for distributed manufacturing of non-woven media

['Anton Molina', 'Department Of Materials Science', 'Engineering', 'Stanford University', 'Stanford', 'California', 'United States Of America', 'Department Of Bioengineering', 'Pranav Vyas', 'Nikita Khlystov']

Date: 2022-05

Centralized manufacturing and global supply chains have emerged as an efficient strategy for large-scale production of goods throughout the 20th century. However, while this system of production is highly efficient, it is not resilient. The COVID-19 pandemic has seen numerous supply chains fail to adapt to sudden changes in supply and demand, including those for goods critical to the pandemic response such as personal protective equipment. Here, we consider the production of the non-woven polypropylene filtration media used in face filtering respirators (FFRs). The FFR supply chain’s reliance on non-woven media sourced from large, centralized manufacturing facilities led to a supply chain failure. In this study, we present an alternative manufacturing strategy that allows us to move towards a more distributed manufacturing practice that is both scalable and robust. Specifically, we demonstrate that a fiber production technique known as centrifugal melt spinning can be implemented with modified, commercially-available cotton candy machines to produce nano- and microscale non-woven fibers. We evaluate several post processing strategies to transform the produced material into viable filtration media and then characterize these materials by measuring filtration efficiency and breathability, comparing them against equivalent materials used in commercially-available FFRs. Additionally, we demonstrate that waste plastic can be processed with this technique, enabling the development of distributed recycling strategies to address the growing plastic waste crisis. Since this method can be employed at small scales, it allows for the development of an adaptable and rapidly deployable distributed manufacturing network for non-woven materials that is financially accessible to more people than is currently possible.

Here, we demonstrate that CMS can be implemented with simple hardware and used to obtain functional non-woven materials in a way that is commensurate with the requirements of DM. Specifically, we use a modified commercially-available cotton candy machine (CCM) to show that CMS is both a fast and affordable means to produce non-woven materials. In our investigation, we consider the use of different resins on the morphology and performance of the resulting non-woven materials. We explore strategies required for processing the produced material into a functional fabric and for performing quality control in a distributed manufacturing context. The performance of our functional fabrics is compared with commercially-available N95 filters and evaluated according to filtration efficiency (FE) and pressure drop (PD). The purpose of this work is to contribute to the ongoing discussion concerning the limits and opportunities of small- and medium-scale manufacturing for the production of medical equipment and—more generally—technologically advanced materials such as non-wovens.

The polypropylene (PP) micro- and nanoscale fibers used in non-woven filtration media are a challenging technical target for distributed manufacturing. The rheological properties of PP allow it to be transformed into thin fibers with dimensions necessary to achieve direct inertial filtration. The dielectric properties of the material on the other hand allow manufacturers to embed electrical charge that aid with filtration in the diffusive range, thus allowing filtration of much smaller sized particles than is possible through inertial filtration alone [ 1 ]. The existing supply chain for this material relies on large-scale centralized manufacturing, using a process known as melt-blowing. This process operates by passing hot, pressurized air around a heated extrusion die to simultaneously melt and extrude molten polymer into non-woven fibers. The extensive infrastructure needed for die fabrication and high energy costs required to supply compressed, hot air makes this technique efficient but inflexible to surge demand and economically inaccessible. While a typical melt blowing facility can produce filtration media for more than 1 million masks per day, establishing additional manufacturing capacity requires significant investments of both time (∼ months) and capital (∼ million USDs) [ 48 ]. Alternative methods for producing nano- and microscale non-woven fibers that could be operated at small and medium scales include electrospinning and centrifugal melt spinning (CMS) [ 49 , 50 ]. While electrospinning has benefited from an enormous amount of research, it is still limited by low throughput and an inability to work with low dielectric materials like PP [ 49 ]. In contrast, CMS offers an order of magnitude improvement to throughput and is agnostic to the electrical properties of the material, as it eliminates the usage of electrical potential to draw fibers from a polymer pool. In contrast to melt blowing, CMS gains efficiency by decoupling melting and extrusion by using a controllable heating element to melt and centrifugal forces generated by a rotating spinneret to extrude the polymer Fig 1B . Much of the academic work related to the CMS method has used either the FibeRio device [ 50 ] or custom-built devices. Several studies have demonstrated fabrication of nano- and microscale PP fibers using CMS [ 51 – 53 ] and even a capacity to produce fibers from commonly available mixed recycled plastics [ 52 ]. Indeed, CMS has been used as an enabling technology for distributed recycling efforts [ 54 ]; however, the possibility of creating higher value-add, functional materials has not been rigorously discussed in the academic literature.

A: Distributed manufacturing paradigm allows for flexible, local production of material anywhere in the the world on short notice. B: Schematic illustration showing key steps of RJS process to produce non-woven fiber mats. C: Implementation of RJS using a retrofitted, commercially-available cotton candy machine. D-E: the process deposits fibers in a mat that can be collected and processed into filtration media. F: High speed camera footage showing ejection of material from the spinneret and onset of a Rayleigh-Taylor instability G-H: leading to the formation of nano- and microscale fibers from extrusion holes much larger in size(∼ 500 − 1000μm).

Distributed manufacturing (DM) is a framework that relies on geographically dispersed manufacturing nodes operated at small scales to produce goods locally and equitably, offering an alternative paradigm to centralized manufacturing Fig 1A [ 33 – 35 ]. Small- to medium-scale manufacturing nodes are inherently more flexible and resilient than large-scale, centralized production. For example, redundancy in a manufacturing network minimizes the risk of a single point of failure to supply chains. Additionally, manufacturing at this scale requires less capital and time investment, increasing accessibility in LMIC environments and reducing the financial burden of increasing capacity due to surge demand in mature markets [ 29 ]. The DM approach has been validated in the context of additive manufacturing where the prevalence of 3D printing and digital design tools has enabled rapid and flexible manufacturing capacity to respond to the present crisis at local scales [ 36 – 38 ]. A DM approach to mask manufacturing was proposed during the initial pandemic response [ 39 ]; however, the proposal did not address access to the non-woven filtration material—the main manufacturing bottle-neck. Furthermore, DM provides unique economic opportunities that would be challenging to implement in a centralized model [ 33 , 40 ]. For example, the development of complementary recycling tools has allowed both new approaches for closed-cycle manufacturing [ 41 ] and users to experiment with novel materials from local sources [ 35 , 42 , 43 ]. The development of distributed recycling coupled to DM is a promising route towards increasing efficiency in collection and recycling plastic waste [ 44 – 46 ]. From a sustainability perspective, there is the added benefit of reducing the environmental impact associated with transportation in global supply chains and waste streams [ 47 ]. There is a clear need for an analogous technology for the distributed fabrication of non-woven materials that can be used in a variety of applications, including air filtration.

Non-woven materials represent a class of engineered fabrics that are ubiquitous in modern life with applications including apparel, construction, medicine, and filtration [ 1 – 3 ]. In specific, non-woven filtration media have recently received widespread attention for their use in air filtration devices which provide protection against the inhalation of particulate matter [ 4 ] associated with increasing air pollution due to industrialization and urbanization [ 5 – 7 ], increasing wildfires associated with climate change [ 8 – 11 ], and most recently to deter the spread of COVID-19 [ 12 – 14 ]. In the context of the COVID-19 pandemic, failures in global supply chains for non-woven materials have inspired researchers and left governments desperate to find solutions to meet global demand surges [ 15 , 16 ]. Recent research has identified commonly available materials [ 17 – 19 ] that can be used as an improvised face covering while other work has focused on developing effective reuse and decontamination protocols of existing PPE [ 20 – 25 ]. However, little work has been done to address the production bottle-neck of the non-woven filtration material at the center of these devices [ 13 , 26 ]. Meanwhile, the shortage has been exploited by bad actors who have introduced counterfeit N95 respirators into the marketplace [ 27 ] and has led to countries with domestic manufacturing capacity to enforce export controls at the expense of those without such infrastructure, namely low and middle income countries (LMICs) [ 12 , 28 – 30 ]. Finally, the problem of managing plastic waste from an estimated daily usage of 6.8 billion masks per day must be addressed in an environmentally friendly manner [ 31 , 32 ]. These events highlight the importance of rethinking the production and supply chains associated with functional non-woven materials.

Results

Centrifugal melt spinning Commercial CCMs consist of an electrically heated spinneret with a material reservoir that is spun using a vertically mounted electric motor Fig 1C [55]. The spinneret is rotated while being simultaneously heated to generate the centrifugal forces necessary for the extrusion of the molten material contained within. The molten polymer extrudes through tiny orifices in the spinneret in a radially outward direction in the spinneret’s frame of reference. The molten polymer mass undergoes an extensional flow and thins out into smaller diameters [56]. Simultaneous cooling due to ambient temperature gradients and surrounding air flows causes solidification of the stretched melt flow, which ultimately defines the diameter of the resulting fiber. These fibers then accumulate as a sheet on the walls of the cylindrical enclosure around the spinneret. With all the necessary principles for CMS present in the CCM, we decided to modify such a device by replacing the wire screen on the spinneret with a solid aluminum ring having several orifices (8–24) with uniform diameters ranging from 0.016”—0.038” (0.4064–0.9652 mm) (S1 Fig). These orifices allow for well-controlled extrusion of molten polymer and allow us to obtain a fiber diameter range required for filtration media applications (0.1–10 μm). To prevent self spooling of extruded fibers by the spinneret shaft, we introduced a simple cardboard cover shield, which allowed formation of continuous flat fiber sheets on a conical surface around the spinneret Fig 1D. Fibers accumulate as an annular sheet with the inner circumference adhered to the spinneret and the outer circumference adhered to the collection cylinder. The material used for further testing is obtained by cutting open the sheet radially and flattening it out on a surface to remove regions of the sheet which display defects associated with static collection during batch assembly. We observe two types of processing defects. First, material near the outer circumference has a lower density due to a constant mass flux being deposited over an area which scales with radius2. Second, material near the inner circumference becomes fused due to its close proximity to the heated spinneret. The material collected between these two defect regions is shown in Fig 1E and was extruded in ∼2 min. and weighs ∼ 24–26 g, containing enough material for ∼ 12 − 13 masks. Through high speed imaging of the machine in operation, we also observed the breakup of pre-solidified jets into smaller droplets due to the presence of a Rayleigh-Plateau instability, often resulting in the formation of microspheres. We consider the presence of microspheres to be a source of contamination for filtration media since they can be dislodged by air currents. This can be controlled by using higher viscosity or higher surface tension materials, or utilising electric fields for drawing fibers (Fig 1F–1H), as has been reported previously [51, 57, 58]. Another source of microsphere formation is the breaking up of a polymer flow stream at the orifice tip itself. Having a replaceable aluminium ring design allowed us to easily debug and tune the geometry of the fiber being produced by simply swapping the ring with a different orifice size. In general, we observe that larger orifices lead to larger diameters; however, we finally settled on an orifice size of 0.024 inches for the data presented in this study. One of the limitations of using a commercially-available CCM, is the lack of precise temperature control and access to only a single RPM value. As a result our study considers fibers produced at T = 160 − 200°C and 3,500 RPM. This range in measured temperatures is a result of a changing amount of material in the reservoir. As material is depleted, the constant-voltage heat source continues to provide the same energy input leading to higher observed temperatures. Application of an electric field during the fiber extrusion process has been used to minimize the production of microspheres [57], reduce fiber diameter, and to impart an electric charge to the material—which is a common strategy for increasing the FE of the non-woven material [59–61]. The presence of an electric field during fiber extension has also been shown to produce charges embedded within the fiber volume that are more stable against environmental conditions such as humidity and temperature compared to surface charges [59, 62]. To implement similar strategies, we connected a -5kV potential source with the negative terminal placed on the collection drum. Since the spinneret was electrically grounded, a field was established with the spinneret at ground potential and the collection drum at -5kV negative potential, allowing for polarization of the extruded molten polymer. We characterized the charge of the material with a handheld, electrostatic surface DC voltmeter. The readings of the voltmeter only provide a crude description of the charge distribution on the surface of and inside the material [63]. Measurements performed right after the material is produced lead to the most consistent readings ranging from -1 to -10 kV. The surface potential is relatively uniform (varying about 30%) along the surface and is similar on both sides of the collected fiber sheet. We also observed that surface contact with other dielectric materials used for handling and storing the fibers lead to high variance in the charge measurement of the sheet. Such large variance is similar to previous work in the literature using similar characterization methods [64]. This suggests that most of the collected charge was either able to conduct or get exchanged through triboelectric charge transfer. The low charge retention could be attributed to the conduction properties of the material, predominantly surface accumulation of charges instead of the bulk or the energy level of the localized bulk charge trapping sites [65, 66]. We tested this hypothesis by utilising a custom made corona discharge device using a van de Graaff generator and friction induced triboelectric charge transfer from commonly available polystyrene packaging material. Such treatments resulted in temporary enhancement of surface charges which have been recently utilized for rejuvenation of filtration properties post decontamination of N95 masks [64, 67–69], but the material maintained aggressive charge exchange properties with the surfaces in contact.

Fiber morphology and processing As compared to the fiber sheets obtained through the commercial melt-blowing process, those obtained through our method have lower density of fibers and hence require additional post processing before they can be evaluated as candidates for air filtration media, an approach that is distinct from one of the recent studies in this space [26]. Here, we consider two approaches for densification: 1) calendaring and 2) compaction. We compare the fiber morphology of the resulting materials against commercially-available N95 FFR filter media (Fig 2A and 2B) through scanning electron microscopy (SEM). Our calendaring process was carried out using cold lamination rollers. The material is supported between two layers of spun-bound PP to prevent adhesion of the material to the rollers. The support layer is then removed for subsequent testing. The resulting material shows an increase in density but still lower than that of the reference material. Meanwhile compaction was carried out with and without the application of heat (130°C). We find that application of heat is important to increase the density of the material and produces a densification similar to that of the reference material. Compaction without heat produces the least dense of the samples considered here. Both of these techniques allow for the construction of multi-ply filters which makes them more mechanically robust. Additionally, multi-layer constructions of filtration media are important since a failure in a single layer will not lead to a failure of the entire filtration device, since a defect at a position in one layer can be compensated by continuous material deposition in the other layers. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Fiber processing and characterization. A: Produced fibers must be processed into dense mats before they can be used as a filtration media. We evaluate two methods: i) calendaring and ii) compaction. Photographs show as-produced material (left) and material after compaction (right). B: SEM characterization of large-scale features of non-woven filtration media produced using Pinnacle 1112 PP homopolymer (MFI = 12 g/10 min). Insets show a macroscopic section of material obtained after each densification process compared with material obtained from a commercial N95 mask; scale bar represents to 1 cm. C: SEM characterization (top) enables comparison of fiber morphology between commercial N95 and fibers produced from using a modified CCM (image obtained from calendared sample shown in part B). Histograms (bottom) of fiber diameters show that both samples share a similar long-tailed distribution of fiber diameters. The black curve is a continuous probability distribution derived from the experimental data. Insets show the same distribution plotted on a logarithmic axis. Fiber diameters were measured from the sample at several different locations using 150 fiber counts. https://doi.org/10.1371/journal.pone.0264933.g002 Comparison of the material microstructure (Fig 2B) provides further insights. We find that the sample subjected to calendaring is most similar at the microscale to the N95 reference material. Heat compaction produces a fusion of the fiber network which results in extremely large pressure drops (data not presented). This can be understood by noting a significant reduction in pore size compared with the other samples in our study. Reduction in the applied shear forces during calendaring also contributes to preserving randomness in the fiber network, which has been realized by industrial processes for ensuring high FEs through complex carding processes [70]. We finally examine the morphology of individual fibers for calendared samples (Fig 2C). We see that the N95 reference material has an average fiber diameter of 3.7 ± 2.6μm. However, the distribution of fiber diameters shows that while most of the fibers are on the single micron scale, the distribution extends to include much larger fibers. This wide distribution allows for larger fibers to act as a support for the smaller fibers which are typically more fragile. The fibers produced by our method using PP resin with MFI = 12g/10min have slightly larger average fiber diameters (3.4 ± 3.0μm) and a similar long-tailed distribution that is skewed more heavily towards large diameter fibers. These morphological similarities with the reference material at all length scales make these materials promising candidates for use as air filtration media. Additionally, we processed PP resins with increasing MFIs which are typically used in traditional melt-blowing manufacturing (S3 Fig). We find that resins with MFI = 50, 500, 1550g/10min give fibers with diameters 5.2 ± 3.8μm, 4.1 ± 3.9μm, and 3.8 ± 5.2μm, respectively. In general, increasing MFI results in fiber diameter distributions that have a higher fraction of small diameter fibers (∼ 1μm). However, we note the presence of artifacts associated with batch processing. The continuous sheet of fibers collected after a run shows variations in the fiber diameters with the size of the fibers decreasing with newer fibers deposited on upper layers. One possible explanation for this is the variable amount of heat absorbed by the material depending on the time it spends in the spinneret. The material to extrude last suffers from higher polymer decomposition and has altered mechanical properties. Moreover, smaller amounts of material in the spinneret also prevents the orifices from being completely filled during the extrusion, which reduces the effective orifice diameter participating in the extrusion, thus altering the nominal diameter and mechanical properties of the fibers expected during a run.

Material performance We evaluated the FEs and PDs of the processed materials using a setup similar to that used in previous studies on filtration materials (S2 Fig) [71]. FE measurements were made using incense smoke as a source of particles with a range from 0.01 to 5.0 μm with a flow rate of 2.8 L/min. This rate is set by the particle counter (S2A Fig). We note that this flow rate is substantially lower than what is specified by conventional testing standards [17, 20]. PD represents the air resistance across the filtration media with lower values indicating higher breathability. All measurements were made with a flow rate of 5 L/min. Samples are excised from calendared material with diameter 17.25 mm (S2B Fig). Under these test conditions, we show that several of our produced materials have FEs comparable to the N95 reference material of the same dimensions subject to the same experimental conditions (Fig 3A). In specific, the N95 sample had a measured FE of 97.34 ± 0.57% compared with 94.48 ± 1.23% of our best material (MFI = 12g/10 min, 3ply). While several of our samples performed quite well, there are important trade-offs in material performance that are dependent on material processing. For example, our best material (MFI = 12g/10min, 3ply) has a pressure drop of 16.28 ± 3.43 cm H 2 O which just underperforms that of the N95 reference 3.69 ± 0.39 cm H 2 O. However, by reducing the number of plys in the construction, we can reduce the pressure drop to 6.76 ± 0.74 mm H 2 O/cm2 with only a slight sacrifice in filtration efficiency 87.26±1.77%. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Performance of non-woven filtration media. A: Filtration efficiency and B: pressure drop for several different materials produced via CMS plotted against the grammage of the sample. The numbers in the legend indicate the melt flow indices of the polymers. C: Phase plot of filtration efficiency vs pressure drop with marker size representing grammage of the sample. The markers represent mean reading from N ≥ 3 samples with a triplicate experiment for each sample. The error bars represent standard error of the mean on each side for both vertical and horizontal axes. The dashed lines represent the corresponding measurements for the filter material extracted from N95 FFRs. All the samples were prepared using 30g of polymer material except for those with explicitly mentioned values of 12g. https://doi.org/10.1371/journal.pone.0264933.g003 To better understand the design space associated with material processing, we consider the effects of density on both FE and PD. We find that density does not have a significant impact on FE (Fig 3B). However, we do observe a significant reduction in FE for single ply samples with decreasing density. We can understand this observation by recognising that any defects in the non-woven material will lead to a failure of the filter. This failure mode can easily be remedied by adding a second ply which allows defects in one layer to be compensated by functional regions of the other layer. We also considered the effect of the relative orientation of sheets for two-ply filters. We find that there are no significant effects, confirming the presence of sufficiently random fiber orientations observed in the microstructure of single layers. We also observe a significant dependence of PD on density. Increasing density results in reduced breathability. Our data suggests that two-ply materials find a good balance between FE and breathability. More generally, we have shown that a modified cotton candy machine can be used to produce functional fabrics which might be useful in air filtration applications. While the performance metrics are less than those of the N95 reference material, they are comparable with those of community-based mask manufacturing efforts [72].

[END]

[1] Url: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0264933

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.