HelixJet: An innovative plasma source for next ‐ generation additive manufacturing (3D printing)

A novel plasma source (HelixJet) for use in additive manufacturing (AM)/3D printing is proposed. The HelixJet is a capacitively coupled radio frequency plasma with a double ‐ helix electrode configuration that generates a surprisingly stable and homogeneous glow plasma at low flow rates of argon and its mixtures at atmospheric pressure. The HelixJet was tested on three polyamide powders usually used to produce parts by laser sintering, a powder ‐ based AM process, to form local deposits. The chemical composition of such plasma ‐ printed samples is compared with thermally produced and laser ‐ sintered samples with respect to differences in morphology that result from the different thermal cycles on several length scales.

photopolymerisation (VP). One of the most common PBF-based polymer-AM methods is laser sintering (LS). InLS,alayerofpowderisspreadoverabuildplatform, following which selected regions of this layer are selectively melted by a laser over cross sections determined in the CAD file. [2] By far the most exploited class of materials for 3D printing are polymers. This is not surprising given that polymers can be 3D printed by a number of different processes, with many AM techniques utilising thermoplastic or thermoset polymers as build materials. Plasma treatments are already integrated into many of these processes, for example, to improve adhesion between layers produced by ME or for better surface quality of PBF-fabricated parts. In addition, plasma treatment of powders before PBF leads to a reduction in porosity due to better wetting of solid surfaces by the melt. [3] Plasma spraying of polymers might also be considered as DED, as these systems have the ability to process large build volumes, at relatively high deposition rates and uses an articulated energy source with a strong energy coupling to the feedstock. The feedstock can be deposited directly onto existing components and has the potential to change chemical composition within a build to produce functionally graded materials (see ASTM F3187-16). However, in ASTM F3187-16, simple coating deposition systems are not considered to be DED for 3D printing because the latter must also have the ability to build layer-bylayer according to the CAD file.
It was shown decades ago that polyamide 11 (PA 11) can be plasma-sprayed into coatings using a plasma arc source, where the PA powder is fed into the flame of the plasma torch. [4] However, high arc powers (30 kW) were needed to melt the particles leading to a substantial weight loss of the feedstock due to thermal degradation. Both melting and degradation showed strong size effects resulting in major defects (e.g., unmelted particles, cracks, and pores). The best balance was found for PA 11 particles with sizes between 40 and 60 µm; but when compared with bulk material, the plasma-sprayed polyamide coatings exhibited a lower tensile strength and a brittle stressstrain characteristic-this brittle behaviour can also be observed in LS parts. Even for a given particle size, there are large temperature gradients within this arc plasma jet meaning that satisfactory coatings can only be produced for certain angles of feedstock injection. It follows that to obtain high quality coatings, a plasma source with a homogenous temperature distribution along its central axis is needed to ensure a high degree of melting.
This together with the requirement for layer-uponlayer deposition, places two further demands on the plasma source: (a) the ability to provide a well-defined lateral deposition spot (pixel) and (b) a temperature profile along the jet axis that allows the deposition of subsequent layers without causing thermal degradation. These conditions cannot be fulfilled by an arc plasma or similar sources, meaning innovative plasma sources are needed. The self-organisation phenomena observed in a nonthermal atmospheric pressure plasma jet have generated fresh ideas for homogenising the contracted discharge within the helical dynamics of filaments. [5] In a self-organised plasma jet, filaments are helically shaped and rotate with a velocity that depends on the inclination angle or helicity of filaments. Hence, the idea of doublehelix electrodes is proposed for the new plasma jet, HelixJet. To provide a proof of principle, the treatment of polyamide 12 (PA 12) by HelixJet was chosen because PA 12 is widely used in the LS process.
PA 12 is the most common polymer in LS due to its mechanical properties and its favourable thermal characteristics for sintering. These characteristics (e.g., a relatively low melting temperature and a large window between the melting and crystallisation temperatures) render PA 12 as one of the most reproducible materials for use in LS. However, there are issues to overcome in terms of part properties and the overall efficiency of the process. [6] Though some parts produced by LS tend to perform similarly to more traditional techniques, such as injection moulding, in some cases (e.g., tensile strength), they perform poorly in others (e.g., elongation at break). In particular, the levels of porosity encountered within the parts tend to lead to a more brittle failure under tension, meaning, it is common to observe ductility that is an order of magnitude lower than for injection moulding. [7] It has been suggested that porosity is responsible for the large variability in LS parts, [8] and a number of attempts have been made to reduce this, including the plasma treatment of powders before sintering. [3,[9][10][11] Although such plasma powder treatment processes c o u l db es c a l e dt ot h ek i l o g r a m sq u a n t i t i e sn e e d e dp e rL S run, this concept has not been taken up widely; [12] presumably for two reasons: (a) The degree of surface functionalisation resulting from the treatment is markedly reduced within hours [12] and (b) the time and temperature need to be tightly controlled to avoid the degradation of the powders. [10,11] Even without plasma pretreatment, the powder degradation variations resulting from different time-temperature cycles are thought to be responsible for property variations in LS parts. [6] The LS process requires preheating of the base powder to a temperature slightly below the melting temperature to prevent "curl" of parts. [13] As a consequence, changes to the molecular weight and crystallinity of unsintered powder also occur. [14,15] At a certain point, the powder can no longer be processed, meaning this preheat stage adds time, cost, and complexity to the process. [14] The potential to process and deposit the powder directly where needed, rather than through selective sintering of a layer, would therefore provide a clear benefit.

| Powders and LS
Several grades of commercially available polyamide 12 (PA 12) were selected for this study: PA2201, PA2200, and PA2221, all from EOS GmbH. Samples produced from these powders are labelled in this study as 1, 2, and 3, respectively. The producer specifies mechanical and thermal properties, for example, strain at break: 15 % (PA1), 4-18 % (PA2), 4-16 % (PA3), vicat softening temperature: 163°C (PA1 and PA2), melting temperature: 176°C (PA1 and PA2) and 187°C (PA3). The principal differences between these powders are in terms of processability and part properties. The common properties of PA 12 have been applied for simulation of particle heating: density of bulk material 1.01 g/cm 3 ,t h e r m a l conductivity 0.51 W/(m K) and specific heat capacity 1.75 J/(gK). [16] Note that mass density varies depending on its form (powder or bulk). Thus, the specified density of powder is 0.93 g/cm 3 (PA1 and PA2) and 0.97 g/cm 3 (PA3).
Parts were produced on an EOS Formiga P100 LS system, using the default parameters for PA 12. They exhibit a range of morphologies with different degree of order at macroscopic and microscopic scales. Figure 1b shows the typical powder shapes and sizes on the surface of a LS object (sphere in Figure 1a). Figure 1c demonstrates the polycrystalline structures combining amorphous zones with lamellas of crystalline PA 12.
To compare the structure and chemistry of the plasmamanufactured samples, several samples were produced by LS (labelled with L), and melting of powders on a glass substrate for one hour at 190°C (labelled with H). For the LS process, the following parameters were applied: Power 21.0 W, spacing 0.25 mm, scan speed 2500 mm/s, and beam offset 0.150 mm, leading to a manufacturing speed of 20 cm 3 /hr.

| Plasma source and processing of powders
The HelixJet is an atmospheric pressure capacitively coupled plasma source operated with radio frequency (RF) power applied to two double-helix electrodes. The electrodes are placed outside a quartz tube fed by the working gas, in this case argon. The specification of the HelixJet and operation conditions used in this study are listed in Table 1. The description of the design, corresponding electric field and thermal properties of the plasma source are presented in the results section.
Many concepts for continuous powder feeding that would be compatible with the HelixJet exist, for example, gravitybased systems for LS. [17] Continuous feeding in the context of plasma powder treatments was extensively reviewed recently. [12] Here, a single shot of powder (10 mg) was fed into the plasma by a dosing spoon in the upper part of the HelixJet. Thus, as the powder dose falls, the gas flow drives the particles through the plasma column. Because of the particle dimensions, the velocity of the gas determines the residence time of the particles in the plasma. In this case, the residence time is in the range of 50 ms (plasma column of 6 cm, 50 W, and 1 slm). The discharge was switched off immediately after deposition of the single dose to eliminate prolonged exposure by the plasma. The treatment speed of 200 mg/s was achieved by this setting. The samples prepared with the plasma process are labelled with P.

| Diagnostics and characterisation methods
Morphological investigationso fd e p o s i t so nt h em ms c a l e were carried out by light photography. The microscale and nanoscale morphology was investigated by scanning electron microscopy in secondary electron (SE) and backscattered electron (BE) imaging modes. High-resolution top views were provided with a JSM 7500F (Jeol, Japan), which employs a field emission gun and two detectors used in this study: SE in-lens detector (maximum specified resolution of 1.0 nm) and low-angle BE detector with maximum resolution of at least 5 nm. The cross sections were observed in a Helios Nanolab G3 UC in a low-voltage ultrahigh resolution mode with no conductive coating.
Chemical analysis included a combination of Fouriertransform infrared (FTIR) spectroscopy and X-ray photoelectron spectroscopy (XPS). An FTIR spectrometer (Perkin Elmer) with a diamond-based attenuated total reflection (ATR) module allowed the analysis of samples in the mid-infrared range (700-4000 cm −1 ). Elemental composition and bonding was analysed by XPS, via an Axis Ultra DLD (Kratos Analytical, UK) with Al K α radiation (15 kV, 10 mA for survey spectra and elemental scans, and 15 kV, 15 mA for highly resolved measured C 1s and O 1s peaks). Three spots at different positions on each sample were analysed and averaged. Data acquisition and processing were carried out using CasaXPS (Casa Software Ltd.). After subtraction of the Shirley background the O 1s peaks were fitted using a Gaussian-Lorentzian GL (30) peak shape.
Two high-speed (HS) cameras were used for the characterisation of the HelixJet. The homogeneity of the discharge was investigated with VIS photography in the range of 200-900 nm at exposure times from 3 to 100 ns (660 × 660 pixels), by means of a PI-MAX4 (Princeton Instruments). The HS camera was oriented in the discharge tube to visualise the radial profile of plasma emission. An infrared HS camera SC6801 (FLIR) was employed for the investigation of surface temperature in the calibrated range from −10°C to 350°C. For thermography of the HelixJet, a 50-mm IR objective was utilised and in particular, the surface temperature of the quartz tube was investigated (565 fps, 640 × 512 pixels). For the temporally resolved measurements of particle processing in the plasma effluent, a 1:1 IR microscope objective (2,094 fps, 160 × 128 pixels) was applied achieving a spatial resolution of about 20 µm. This allowed the capture of the smallest droplets of polyamide particles (original average size of~60 µm) with a time step of 0.5 ms.

| Homogeneity of the HelixJet and electrical field simulation
The double-helix electrodes (see Figure 2a) were constructed perpendicular to the inclined filaments (ortholines) observed in the nonthermal atmospheric pressure plasma jet to accelerate the gliding of filaments in the new plasma source, HelixJet. [5] Although the helical structure of the electrodes induces a complicated 3D structure of the electric field, initial experiments demonstrated that a noncontracted, mostly homogeneous discharge is generated in the HelixJet,asshowninFigure2c,e.Toexplainthisapparent contradiction, a first study was devoted to the determination of the electric field E(r)=−∇ϕ and its homogeneity. For this purpose, the Poisson equation with the vacuum permittivity ε 0 was solved in the solution domain shown in Figure 2a. Within this first approach, the space charge density ρ(r) was neglected and appropriate values ε r = 1 and 4.3 were used for the relative permittivity ε r in the plasma zone and in the capillary, respectively. The solution domain is restricted in the axial direction to one turn of the electrodes, which are wound onto the capillary with a helix angle of 33°. Note that the ortholines to the electrode helix have the helix angle 57°, which corresponds to the helicity of filaments described in the previous study. Period boundary conditions were applied to link the bottom and top sides of the solution domain, and the voltages V = 0 and V = 1 kV were set to the grounded (blue mesh in Figure 2a) and powered (red mesh in Figure 2a) electrodes, respectively. The condition E · n = 0 was used at the surfaces between the electrodes, where n denotes the normal vector from the surfaces. TheresultsobtainedarepresentedinFigure2aandshow the 3D structure of the electric field using the isosurfaces A, B, C 1 ,a n dC 2 ,w h i c hr e p r e s e n tc o n s t a n tv a l u e so ft h e magnitude of the electric field |E|aslistedinthecaption.The colour of the surfaces represents the electric potential ϕ. Representative axial (y =0) and radial (z =5.4mm) cuts of the electric field |E|a r es h o w ni nF i g u r e2 ba n dF i g u r e2 d , respectively. The dielectric and plasma regions are separated  Figure 2b by vertical lines at |x| = 3.8 mm. These results demonstrate that the modulations of the electric field within the inner zone are limited to <50 %. This weak variation of the electric field magnitude may be an indication of the establishment of a homogeneous plasma in the inner zone as observed experimentally.
The HS imaging experiments of the discharge confirm that no filaments were formed in the HelixJet, but a homogeneous glow discharge column was developed in the tube. Only the modulation of emission due to the frequency of the RF power remains. The homogeneity of the plasma column was proven by HS imaging with several sequences. Figure 2c shows an image that was obtained by applying the smallest exposure time of 3 ns.

| Thermal characterisation of the HelixJet
Temperature control is the key to plasma deposition of polymeric materials with reproducible properties. [18]  Thus, a neutral gas temperature, T g , of approximately 427°C can be expected in the active zone of the discharge, which is substantially higher than the melting temperature of 187°C of PA 12. However, as particles are introduced and melted, PA 12 molecules evaporate and then quench the discharge. Thus, the T g drops and the melting of particles is consequently stopped and the discharge can develop again. Therefore, we assume a dynamical equilibrium of the kinetic temperature near to the melting point of the polymer. This hypothesis was investigated by HS microscopic thermography of particles and corroborated by modelling.
The particle temperatures were measured as they leave the HelixJet by focusing a HS IR camera at the area labelled as "zoom" in Figure 3b. The results are shown in Figure 4. The temperature of the tube edge is approximately 200°C and the particles leave the edge with surface temperatures between 160°C and 190°C (Figure 4c). The movement of one particle is demonstrated in Figure 4d by HS thermography at a frame rate of 2 kfps. The velocity of the particle is 0.63 m/s and its core temperature decreases with −12°C/ms over a distance of 2 mm. Because of such a high cooling rate, it is likely that crystallisation occurs by homogeneous nucleation at the lower temperatures. [19] The particle velocity and cooling rate are linked to the gas velocity. The latter depends on the applied gas flow rate Q at the inlet of the plasma jet and T g can be controlled by the RF power coupled into the plasma. The measured particle velocity is in very good agreement with the results of a fluid dynamics model of the laminar gas flow in the tube of the HelixJet. The model shows that the gas velocity is largest at the centre of the plasma jet and decays parabolically towards the inner wall of the capillary. Maximum gas velocities in the range from 0.1 m/s (Q = 0.1 slm, T g = 300 K) to 2.7 m/s (Q = 1 slm, T g = 1000 K) are obtained by numerical solution of the stationary incompressible momentum Navier-Stokes equation using COMSOL Multiphysics®. Due to the low dimension of the particles, gravity effects are small compared with friction and the particles in the jet are driven purely by the gas flow. With this assumption, residence times in the range of 0.05 to 1.5 s can be estimated from the calculated gas velocities. To achieve a high throughput, this residence time should be minimised to such an extent that the required particle-melting temperature can still be achieved.
Therefore, to study the heating of particles in the argon gas stream, the time-dependent heat equation was solved numerically using COMSOL Multiphysics ® . The simulation of the heating of particles with a diameter of 60 µm in the plasma with a predefined T g predicts a thermalisation of the particles within 0.1 s. Hence, a fly path of particles in the active plasma of approximately 1 cm would be sufficient for future experiments.
Finally, the results of the IR imaging (see Figure 4c,d) suggest that the temperature of the particles does not rise substantially higher than the melting point confirming our initial hypothesis. The feedback mechanism between particles and plasma controls the evaporation of material and protects the particles against excessive thermal degradation, which is likely to play an important role in achieving deposits with reproducible properties.
The results from the heat-transfer model shown in Figure 4a,b demonstrate the change of the temperature distribution caused by ingesting a particle at the initial laboratory temperature into the hot gas (T g = 350°C). As can be seen in Figure 4a, the introduction of the particle leads to a quick drop of the gas temperature close to the particle. After approximately 0.7 ms, the particle core temperature reaches 35°C while the particle surface has a temperature of about 46°C at this instant (cf. Figure 4a). The melting point of 187°C at the particle surface is reached after approximately 10 ms, at which point the temperature of the particle core is 180°C (cf. Figure 4b).
The results above demonstrate how the process conditions can be adjusted and predicted from a combination of modelling, easily accessible parameters such as the flow rate and power coupled into the plasma The HS IR camera was focused at the area labelled "zoom" for IR measurement of the particle temperature (compare with Figure 4). HS IR, high-speed infrared 6o f1 3 | enabling a model-based optimisation of the process for different particle sizes and powder types.

| Morphological characterisation of plasma-manufactured samples
The typical sample, P1, was deposited by HelixJet and is shown in Figure 5a (compare to L1, Figure 1a). The plasma deposit is confined to an area that closely corresponds to the diameter of the quartz tube (7 mm) and has a typical Gaussian profile with the width 2σ = 7.97 ± 0.05 mm and the maximum at the axis of the jet corresponding to the gas flow profile (Figure 5c). The process excels with a maximum deposition rate of 2-3 µm/ms (200 mg/s). Some micron-scale roughness related to the powder size is visible here, which hints that particles approaching the surface are either partially melted or the melted droplets are cooled quickly before landing, which is consistent with the results of section 3.2. The cross-section of the sample (Figure 5b) visualised by secondary electron imaging (SEI) evidence two important features. Although the original form of particles or droplets is partially conserved, they are melted together. Moreover, the morphology indicates a sufficient cohesion of the produced sample suggested by lamellas, which are spread over several microns in the vertical fracture surface (Figure 5b) in dark orange.
The comparison of microscale and nanoscale morphology between L, P, and H samples is demonstrated by SEM analysis (Figure 6). The purely thermally treated powder surface (H samples) exhibit less surface roughness as the original form of the particles disappeared completely over the field of view with a width of 1.2 mm (Figure 6a,c,e), which represent the surface topography. Temperature dynamics of particles. Particle and gas temperature obtained from heat-transfer simulations after t 1 = 6.6 × 10 −4 s (a) and t 2 = 9.3 × 10 −3 s (b), both for a background temperature of 350°C; high-speed IR (infrared) thermography of particles in the effluent at 2 kfps: particle temperature from zoomed area indicated in Figure 3b (c), tracked single-particle movement (d). Note that in (c) and (d) different sizes of particles are caused by IR imaging optics being focused to the axis of the jet. Particles out of focus appear with different size and with incorrect temperatures These images were collected with a secondary electron (SE) detector at a low primary beam accelerating voltage of 1 kV. Figure 6a reveals that in the LS sample all powder particles are still visible, whereas the plasmaprinted area (cf. Figure 6c) contains only a few clearly identifiable powder particles that are embedded in large smooth areas. Thus, on this length scale, the plasmaprinted material combines features that are characteristic of LS and thermal annealing, respectively.
This shows promising results for the establishment of 3D plasma printing. First, the partially melted particles can serve as coalescence centres and thus ensure confinement of the deposited materials to the plasma jet exposed area. Second, a smoother compact composite is likely to achieve higher mechanical propertie sa si ti sk n o w ni nL St h a tt h e mechanical properties of parts improve when the proportion of non-melted particles is reduced. [20] Indeed, the connection between particles is made via partially melted material. [21] However, global shrinkage possesses a challenge in LS. [22] The coherent connection in plasma printing has been demonstrated in Figures 5b and 6c. Moreover, local connections (see inset in Figure 6c) by folded structures between particles indicate an efficient mechanical relaxation, which might prevent the substantial shrinkage of final parts by plasma printing.
The mechanical properties are determined by the molecular arrangement of PA 12 on microscopic scales, which is affected by the degree of melting. [19] Therefore, BEI was used to reveal any small density differences in the polymers as a result of local crystallinity variations. [23] Due to their higher density, crystalline areas appear brighter than amorphous areas. [23] Figure 6b,d,f show BE images covering4×3µm 2 from each sample. Due to the design of BE detectors, BEI requires a higher acceleration voltage of the incident electron beam (15 kV). As a result, Figure  6b,d,f contain some information that may have originated from several microns below the surface. In these BEI, false colouring was applied so that the bright (poly)-crystalline areas appear in orange tones while dark amorphous areas are in cyan tones. The coloured micrographs reveal that the spatial extent of ordered and disordered areas varies markedly depending on the deposition method. The H samples exhibit the highest degree of crystallinity (numerous appearances of spherulites), whereas L samples exhibit the smallest degree. Moreover, in all BEI, the presence of nanoparticles, which are assumed as nucleation centres of polymerisation, have been observed (dark dots). Their content, as well as the structure features in PA 12, are closely linked to chemical properties, in particular to the degree of crosslinking and polymerisation. Hence, the chemical analysis of samples is fully discussed in the next section.

| Chemical characterisation of samples
The chemical surface composition was determined by XPS. The atom percentages of the elements carbon, oxygen, nitrogen, and silicon are summarised in Table 2 for the powders (PA1, PA2, and PA3), LS sample (L1), plasma deposits (P1, P2, and P3), and deposits obtained by heating (H1, H2, and H3). Note that hydrogen cannot be determined by XPS. XPS revealed the overall similarity of all samples in terms of stoichiometry to PA 12. In FIGURE 5 Top view photo (a) taken from particle deposition (sample PA1) on a black smooth plastic substrate; secondary electron imaging (SEI) of the local deposit cross-section (b); and thickness profile of the deposit (c) along the red line marked in (a). The saddleshaped cross-section (b) results from the original shape of two adjacent melted particles. The false-colour scale of SEI was applied to enhance fine gray scale variations: blue, low SEI emission (dark); orange, high SEI emission (bright) particular, nitrogen concentration approaches the ideal content of 6-7 atomic %, and small quantities of additives were observed: silicon (0-3 at.%) and phosphorus (below 1 at.%), as would be expected in powders designed for LS. In terms of oxygen content only, the heat-processed materials coincide with the content theoretically expected from the molecular structure of PA 12 while powders before processing exhibit an excess of oxygen. The plasma deposits contain less oxygen than the powders but more than the heat-processed samples. These observations could be explained by postpolymerisation as polyamide FIGURE 6 Scanning electron micrographs of a laser-sintered sample L (a,b), plasma-printed sample P (c,d), and sample prepared by the heat treatment H (e,f). All samples are imaged at low magnification with a secondary electron detector, SEI (a,c,e), and at high magnification with a backscattered electron detector, BEI (b,d,f). False-colour BE imaging was applied to better visualise the polycrystal (BEI bright orange) and amorphous (BEI dark cyan) zones molecules are terminated with hydroxyl groups (R-OH), which polycondense and produce water. [14,24] To test this hypothesis, the binding components of oxygen in the XPS O 1s spectrum were further investigated (see Figure 7 and Table 3). The processing of powders consistently reduced the hydroxyl groups in all samples to 8-30 % (related to carbonyl groups) compared with the raw powders with 44-50 % (C-OH/C═O), with shifts of Amide I and intensity of Amide A in FTIR spectra (Figure 8). This implies postpolymerisation and crosslinking of polymer chains in deposits, consistent with the predominance of polycondensation in PA 12. [15] Evidently, these reactions take place during all applied deposition methods. In particular, LS samples (L1) and samples prepared by plasma processing (P1-3) exhibit a similar degree of postpolymerisation resulting in the oxygen concentration of 8-9 at.%, whereas the thermally processes material contains the expected 7 at.%. Note that the effect occurs on different time scales of 50 ms and 1 hr during plasma treatment and thermal processing, respectively. Hence, plasma processing accelerates the molecular kinetics dramatically. LS would be expected to produce a sample with a comparable mass to P1 (10 mg) in approximately 1 s. In parallel to the polymerisation kinetics, the polycrystalline structure of the samples is also affected during processing.
A main structural feature in polyamides is the organisation of the chains into well-defined two-dimensional (2D) hydrogen-bonded sheets, which are held together in 3D by van der Waals' interactions with similarities to β-sheets arrangements in silks. [26] The sheet alignment (antiparallel or parallel) is linked to that form in which the PA 12 crystallites. Crystallisation in the α-form (antiparallel) and γ-forms (parallel) is possible, [27] with the γ-forms being preferred. [28] The structural arrangement in the powders and as result of the different deposition methods can be probed by IR spectroscopy. In particular, the Amide I, II, and A as shown in Figure 8 are significant for qualitative interpretation of these structure changes, which correlate with XPS results.
The Amide I (Figure 8a) reflects the backbone conformation and the hydrogen bonding pattern. The mode consists primarily of stretching vibration of C═O (76%) with small contribution of CN and CCN deformation. The position varies from 1635 cm −1 for the α-form and 1640 cm −1 for γ-form. [28] In Table 3, the position of Amide I is compared with the binding composition obtained with XPS. From both Table 3 and Figure 8, it can be seen that all thermally prepared samples have the peak at the smallest wavenumber of 1633 cm −1 , which is consistent with the αform and that they have the smallest concentration of C-OH groups, whereas plasma prepared samples show peak broadening towards the higher wavenumbers (peaks at 1634 and 1636 cm −1 ). Thus, the latter are likely to comprise of a combination of α and γ forms. and (c) samples prepared by heat processing. For all materials, the two components of the peak were deconvoluted and the corresponding composition with respect to C-O-H (532.9 ± 0.2 eV) and O═C (531.3 ± 0.2 eV) bonding was quantified (see Table 3). PA, polyamide; XPS, X-ray photoelectron spectroscopy    (a,d). The intervals of wavenumber in orange, cyan, and grey (a) mark regions of the ordered, disordered, and free hydrogen bonding conformation, respectively. [25] FTIR, Fourier-transform infrared; LS, laser sintering; PA, polyamide modifications of material polycrystallinity, crosslinking, and in the densification of the polymeric network. On the basis of the feasibility experiments and modelling, an upscaling of the plasma manufacturing process using a HelixJet to speeds in the order of 1 cm 3 /min seems possible, although further work is needed to transform the HelixJet into a full-AM tool, for instance, introducing a continuous powder feed system and computerised control. However, the potential for constructing larger parts than possible in LS, in air and without the enormous quantities of waste powders, which drive up the cost, should provide the impetus to carry out such work. Moreover, the combination of HelixJet with laser beam manipulation might enhance the precision, speed, and controlled material structuring from microscopic to macroscopic level with a broad impact for many applications.