Abstract
The use of porous polymer materials for vascular prostheses demonstrates promising results. Fleece-like non-woven structures can be generated by atomization of dissolved polycarbonate urethanes. This article focuses on the manufacturing process for the fleece structures. The solution is atomized through a high-volume low-pressure nozzle. The solvent evaporates during time of flight to target, so that small fibres are formed from each drop of solution. By using different rotating molds and positioning systems, tubular shapes or open surfaces can be generated. The manufacturing process is described in detail. The usability of different grades of polycarbonate urethane is tested and the time necessary for evaporation of the solvent from the mesh is assessed. The influence of several basic parameters on the mechanical properties of the generated non-woven material is analysed and discussed.
Introduction
The use of non-woven polymer textiles for blood-contacting implants like grafts has shown promising results in the past [1,2]. Nonetheless, the manufacturing of these fleece-like structures remains a challenge, as established techniques like Electrospinning do not allow producing small fibres with a high yield [3–5]. Lui et al. [6], Reneker et al. [7], Reneker [8,9] and Benavides et al. [10] use processes that augment or replace the electrical forces with gas-dynamic forces to overcome the limitation of Electrospinning. Their research shows promising results, yet is limited to specially designed gas nozzles which are not commercially available. All these processes use polymers that are dissolved in solvents.
Recent research at the Institute of Applied Medical Engineering (AME), using a standard High Volume Low Pressure (HVLP) airbrush gun, showed promising results in yield and fibre morphology in first trials [11]. This process of spray atomization of dissolved polymers uses a thermoplastic polyurethane that is dissolved in a volatile solvent. The solution is then atomized by the spray gun, without using any electrical forces. The solvent evaporates from the spray and thin fibres are formed, which are then deposited on a rotating mold. This allows forming medical devices like grafts. Polyurethane is used because of its high biocompatibility and biostability [12–14]. It was demonstrated that the described process allows manufacturing of microporous structures which are permeable for heparinized blood until it starts clotting the pores [11].
Research on this process was continued and process analysis was done to identify parameters that influence the morphology and mechanical properties of the produced mesh. The intend of this paper is to present the results of this process analysis.
Materials and methods
Spraying process
The basic process of spraying polycarbonate urethane (PCU) non-wovens is very similar to the well-known airbrush process. The polymer solution is atomized by a pneumatic atomizer and sprayed on a surface. An HVLP spray gun is used, where air and liquid leave the nozzle coaxially and are mixed outside, directly in front of the gun. Because of the quickly evaporating solvent, the formed droplets rapidly dry out in the turbulent airstream and are therefore not dissipated into smaller drops. Instead, they are elongated into thin fibres. Depending on factors like distance of the nozzle to the target, solvent concentration, solvent type and air temperature, the fibres are nearly dry when they hit the mold that they are sprayed on. A fleece-like surface with random fibre orientation is therefore formed. Figure 1 shows a microscopic image of a sprayed sample. Fibre diameter ranges from 1–10 µm, with some larger agglomerations of not completely atomized solution droplets in between.
Microscopic image of a sprayed sample (PC3575A, 1.6 bar, 0.5 ml/min, 12 cm).
Several requirements must be met to achieve a sprayable solution that allows the formation of fibres by spraying: the solvent must have a low boiling point and the created solution must have a sufficiently low viscosity so that it can be used in the spray gun. All polymers must be dissolved in the solvent and the solvent must evaporate completely, not leaving any remains in the created fibres. Chloroform is therefore used as solvent for all spraying experiments because of its low boiling point, its ability to dissolve the used polyurethane and its inflammability. The created solutions are not filtered prior to use.
Spraying system
Stock spraying equipment is used, which is mounted to a custom two-axis portal system, and a rotating mold. By using the portal system, the nozzle can be moved sideways and therefore coat the complete length of the rotating mold. Distance to the mold can also be varied, allowing to either change the characteristics of the formed mesh or to have the gun follow the contour of the mold.
The spray gun is a pneumatically controlled SATA Minijet 3000 (SATA GmbH & Co. KG, Kornwestheim, Germany) with a 0.3 mm nozzle. It is supplied with air from two pressure regulators, combined with pressure gauges. One regulator is used for the atomizing air and the other for the control air. Magnet valves are used to switch the air lines. Pressurized air is supplied by the in-house air system. Liquid is supplied to the nozzle by a syringe pump that operates a standard 50 ml syringe. 6 mm od nylon tubing is used to connect supply and nozzle. This setup ensures a controlled and reproducible flow of solution, which is necessary for an analysis of the influence of flow to the spraying results.
The two-axis portal system is built using parts from A-Drive Technology GmbH, Taunusstein, Germany. It consists of a scratch-built lower sled that is powered by an A-Drive stepper motor with fixed spindle, and a second perpendicular axis, which is made from an A-Drive linear rail system. The mold is installed in the system by using a drill chuck that is connected to a Faulhaber electric motor and a rotating cone as support. It can be made from any material that is resistant against the used solvent. The whole spraying system is installed in a glove box (SylaTech Analysetechnik GmbH, Walzbachtal, Germany), which is vented by ambient air at 100 m3/h. This prevents the toxic solvent vapours from contaminating the lab.
We use rotating molds to facilitate the handling system, but the spraying process itself is able to coat any more or less flat surface if a proper handling system like a robot arm is used for the gun. As long as there is a sufficient supply of solution to the spray gun, the process can be kept running for an infinite time. It is also possible to temporarily stop the process to change the syringe in the pump.
Variable parameters.
Spraying system and nozzle.
Atomized polymers
Test setup
A rotating aluminum cylinder with 22 mm diameter was used as mold for all tests.
All available grades of Carbothane were examined for their usability for the spraying process. For this, it was tested at which concentrations of PCU in chloroform a sprayable solution is obtained. The grade that results in a broader usable concentration range was then used for the parameter study.
Drying time until dissipation of the remaining chloroform was investigated using a precise laboratory scale as well as using thermogravimetric analysis (TGA). Samples were sprayed from a distance of 180 mm using 7.5 wt% PC3575A in chloroform and 1 ml/min of flow at 1 bar of atomizing pressure. The standard mold was used for the TGA analysis. The samples for the scale were sprayed on a 8 mm hollow aluminum tube. This was done to allow placing the coated mold on the scale right after spraying. Its weight was recorded in regular intervals. To reduce external effects, the scale was operated in a climate chamber set to 30℃. Orange gel created a low humidity.
TGA measurements were done at the DWI – Leibnitz-Institute for interactive materials e.V (Aachen, Germany) with kind support by Prof. Popescu. The TGA used a nitrogen atmosphere with a heating rate of 10℃/min from 30 to 200℃. Samples were tested right after spraying and 24 and 48 h later.
Factors and their variation for the design of experiments.
Atomization pressure was chosen because it is supposed that it has as significant influence on droplet size and therefore on the structure of the mesh. The same is believed to be true for the spraying distance. Material flow is of interest because it directly relates to the economic efficiency of the process. All sprayed solutions consisted of 7.5 wt% PCU dissolved in chloroform. Rotational speed of the mold and horizontal speed of the gun are also kept constant.
Target factors were chosen as follows:
Tensile strength (N/mm2) Modulus of elasticity (N/mm2) Elongation at fracture (%)
These were measured using a Z2.5 tensile tester, manufactured by Zwick GmbH & Co. KG, Ulm, Germany. Geometry of the samples for the tensile test is shown in Figure 3. The samples were die-cut from the mesh. Samples were taken in axial and in circumferential direction of the mesh. Parameter for the tensile tests was 50 mm/min speed for all samples. Modulus of elasticity was calculated between 1 and 5% of elongation.
Sample geometry for tensile tests.
For all spraying tests, the standard 22 mm diameter rotating mold was used. Fifteen samples were produced per factor combination and direction of sample orientation (circumferential to mold or axial). The results of the tensile tests are given as average and standard deviation for each factor combination. Results of the analysis of variance are given as effect (change in measured values) when changing the factor from − to +. Effects are displayed in relation to statistical significance for each factor with a 95% confidence level. All calculations were done according to Kleppmann [20].
Results
All results were obtained using the spraying apparatus as described above.
Sprayable polymer solutions
Each PCU grade was tested dissolved at different concentrations in chloroform. A too high concentration results either in the PCU not being completely dissolved or in a solution that is too viscous to be sprayed reproducibly. Figure 4 shows the concentration range (in %wt of PCU in chloroform) at which a sprayable solution is obtained.
Concentration range of sprayable PCU solution.
PC3575A results in the broadest range of sprayable solutions. It can still be sprayed at a concentration of 12 wt%. PC3585A shows similar properties. Because of the higher amount of hard segments in this polymer, the resulting solution has a higher viscosity compared to PC3575A. The higher viscosity leads to a higher chance of plugging the spray nozzle. With PC3595A, the concentration window moves farther into the direction of low values. It is still sprayable over a broad range of concentration, but the created mesh shows increased amounts of not atomized droplets and the risk of plugging is again higher. PC3555D has a narrower process window. At higher concentrations, it can only be atomized using high gas pressure and also the risk of plugging the nozzle is greatly increased because of the solutions high viscosity. At lower concentrations, spraying is possible, but the resulting product is more of a foil than a mesh. This is caused by the high amount of chloroform in the solution that leads to a longer drying time and thus still wet droplets when the spray hits the mold. It is therefore, even though it has a reasonably wide process window for sprayability, very difficult to repeatedly create meshes with this material. PC3572D has the smallest process window of all Carbothane grades. It is not possible to create a homogeneous solution with more than 4 wt% of PCU in chloroform, as gel-like PCU blocks remain undissolved at higher concentration. Below 3 wt%, the viscosity of the solution drops rapidly, so that no fibres are formed anymore during spraying. Also, the drying time is increased because of the high amount of chloroform.
All further experiments were only conducted with Type 3575A. The reason for this is that it possesses the widest process window and we decided to concentrate on one type of polymer.
Drying time
Chloroform is a harmful solvent, therefore it is important to assess the time it takes for the solvent to evaporate from the sprayed fleece. Drying time was assessed using a scale as well as TGA. Figure 5 shows the results of the measurements with the scale. The shown weight is the mold with fleece.
Weight of sample during drying.
A clear loss in weight is visible during the first hour of drying. The loss is reduced during the next two hours. No change in weight is measurable after four hours.
Figure 6 shows the results of the TGA measurements. Samples were removed from the mesh right after spraying and analysed after different storage times. The zero-day sample was examined 30 min after spraying, as it needed to be transported to the DWI before analysis. Sample weight is only the fleece for these tests.
Results of TGA measurements.
The TGA results for zero days show a clear weight loss at 70℃. This correlates with the boiling temperature of chloroform (61℃). Afterwards, the weight increases again. This can be explained by oxidation effects from the oxygen that is trapped in the mesh’s pores. The weight of the sample starts to drop again at 160℃. At this temperature, first degradation processes of the PCU start. The one-day sample does not show a loss of weight anymore. This indicates that all chloroform has already evaporated. Weight gain until 160℃ is visible again. After two days, the sample shows no weight loss again, but the gain in weight is reduced. It is possible that oxidation reactions with the ambient oxygen have ended. Degradation still starts at 160℃.
Both measurements, scale and TGA, indicate that all chloroform remains evaporate within the first 24 h after the fleece is sprayed.
Influence of atomizing pressure, spraying distance and material flow on the sprayed mesh
The influence of three select parameters was tested using only PC3575A as polymer. This decision was taken because results from the assessment of the sprayable solutions showed that the process is most stable when using this polymer at 7.5 wt% of concentration in chloroform.
Figure 7 shows the results of the analysis of variance for the samples that were taken in axial direction. Figure 8 shows the results for the circumferential samples. All effects are for changing the factor from − to + with corresponding values as displayed in Table 3. All graphs for each effect are scaled so that the statistical significance with 95% confidence level is at the same level.
Effects in axial direction. Effects in circumferential direction.
The graphs show that the change of the distance has the most significant effect on all three target factors. The change of atomizing pressure does not have a significant effect in axial direction and only a slight effect in circumferential direction. The change of flow does have a significant effect on all target factors in both directions, but the effect on the modulus of elasticity is very low.
The individual results for each factor combination in axial direction are displayed in Figure 9. The results in circumferential direction are displayed in Figure 10. Again, values for + and − can be found in Table 3.
Mechanical properties in axial direction. Mechanical properties in circumferential direction.
These images do also show the effect of the change of distance. Samples that are sprayed with a high distance generally show lower measurement values than samples that are sprayed with a low distance.
Discussion
Results show that it is possible to manufacture non-woven polymer textiles using commercially available equipment. Compared to the established process of Electrospinning, a higher throughput of material is possible. With Electrospinning, a flow rate of 2 ml/h per spinneret is possible [4]. Our spraying process is capable of using a flow rate of 120 ml/h. Drawbacks of our process are the larger diameter of the produced fibres and the non-uniform fibre diameter (1–10 µm), when compared to Electrospinning (≤1 µm) [5], as well as the agglomerations of not completely atomized solution droplets.
The preliminary tests showed that Carbothane PC3575A has the widest range of concentration in chloroform, at which it is still possible to create fibres by atomization. The higher the shore hardness of the used PCU, the smaller the window of usable concentration becomes. Clotting of the spray gun also becomes an issue with higher shore hardness or higher polymer concentration. This get as far as that PC3572D is only sprayable in a very thin concentration range. To achieve a stable process, a wide process window is necessary. It is therefore currently not possible to use PC3595A, PC3555D and PC3572D for manufacturing processes, as we were not able to repeatedly create a stable spraying process with these materials due to clotting of the spray gun. If the application does allow for using PC3575A, it is recommended using this polymer, as it demonstrated a stable process over a sufficiently wide range of concentration. If necessary, PC3585A can be used for spraying. But it must be noted, that this process is less stable than using PC3575A because of clotting of the spray gun. We therefore did not conduct a parameter study with any material except PC3575A.
A limitation of our process is the use of potentially harmful chloroform as solvent. This limitation also affects other processes of gas-dynamic fibre creation or Electrospinning, as these processes rely on dissolved polymers as well [5–10]. Results of the measurement of evaporation of the chloroform from the created mesh show that it is possible to remove the remaining solvent from the material without technical difficulties. No weight loss because of chloroform evaporation was visible in the TGA after 24 h of drying time. Results from the scale even indicate that the chloroform has evaporated after 4 h of storage in a ventilated location. As no chemical analysis for traces of chloroform in the dried mesh was done, a complete removal of all chloroform cannot be ensured. Yet both results indicate that storing the material in a ventilated place for 24 h is sufficient to remove all traces of the chloroform. Such a ventilated place might well be a drying oven or similar. Drying time can be reduced by using an elevated temperature for storage, given the low boiling point of chloroform.
Even though we assumed the atomizing pressure to have the highest influence on the results, it could be shown that this parameter has only a slight influence in circumferential direction of the mold. The larger effect of atomizing pressure on the mechanical properties in circumferential direction, compared to axial direction, can be explained by the airflow around the cylindrical mold. Fibres are aligned more in circumferential direction by the airflow around the mold if the atomizing pressure and thus the velocity of the air are higher. A higher atomization pressure results in lower values for the mechanical properties of the mesh. We do not have a convincing explanation for the generally low influence of atomizing pressure at the moment. Flow has a significant effect on the results, but to a lower amount as distance. For higher flow, the mechanical properties of the sprayed meshes are also higher. One exception is the modulus of elasticity, where higher flow results in lower values for the axial measurements. But it has to be remarked that this effect is small and only slightly above the significance line. Reason for the effect of higher mechanical properties with higher flow can be more fibres that are deposited on the mold at the same time, leading to more wet fibres at the same time, thus creating more interconnections and therefore stiffer meshes. The change of distance of the nozzle to the mold has the highest impact on the results. A close mold results in mechanical properties that are twice as high as with a far mold. The effects of a change of distance are not only the highest ones absolutely, but they are also the highest if compared to the significance threshold. It is obvious that the mechanical properties of the mesh can be manipulated the easiest way by modifying the spraying distance. One can therefore set spraying pressure and flow to values that maximize efficiency and process stability and then fine-tune the process to achieve a desired mesh by changing the distance from gun to mold. The mechanical properties increase with decreasing the distance. The cause for this effect can be an increase in interconnections between fibres in the mesh because of less drying time during flight from the gun to the mold.
Conclusion
We could show that it is possible to use different shore hardnesses of Carbothane for the process of atomization of polymers. In general, grades with lower hardness are better suitable for the process because of their better solubility in chloroform and because of resulting solutions that have a usable viscosity over a wider range of concentrations. The risk of clotting of the spray gun is also reduced with lower hardness grades. The time that is necessary to dry a sprayed mesh until the remaining chloroform has evaporated is within a range that is usable for technical applications of the process. Twenty-four hours after manufacturing of the mesh, no remaining chloroform could be detected. The process of spray atomization of dissolved polymers can be used to manufacture non-woven meshes with a range of different mechanical properties, depending on set process parameters. It is possible to change the tensile strength, modulus of elasticity and the elongation of the meshes without using different grades of polymer. The process is therefore well usable for the application of manufacturing medical microfibre non-wovens because with using only one type of approved polymer it is possible to manufacture parts with different mechanical properties.
Results also indicate that the process is not limited to Carbothane. Any polymer that can be dissolved in a solvent with a low boiling point should in theory be usable, as long as the resulting solution is within a usable window of viscosity. Other polymers should therefore be included in future research.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.