Laser-Based Additive Manufacturing of Alkali Borosilicate Glass Powder: Influence of Laser-Beam Properties on Component Quality

Abstract

Research and development in the field of glass-based laser additive manufacturing continues to receive significant interest within scientific and industrial contexts. In particular, powder bed fusion by laser radiation (PBF-LB) enables the additive manufacturing of porous and vitrified, complex three-dimensional components. The present study investigates the glass morphology that can be achieved using PBF-LB for components made from alkali borosilicate glass. The investigations focus on the comprehensive analysis of the entire process window, including the characterisation of porous and molten glass morphology. In particular, the influence of different laser-beam diameters, which are achieved through defocusing, and the variation in volume energy density are examined in detail and compared with conventional shaping. It was determined that the process of mechanically stable shaping is constrained to temperatures above the softening temperature and relative component densities within the range of ρrel = 37.8…94.2%. Furthermore, it has been demonstrated that the process-related line-like energy input results in the formation of characteristic vitrification strands. This research contributes to the overall understanding of the producible glass morphology and the process limitations of the PBF-LB process. In addition, the entire range of glass morphologies, ranging from open-pored to closed-melt configurations, could be analysed for the first time.

1. Introduction

Current research focuses on powder bed fusion by laser radiation (PBF-LB), particularly in relation to metallic materials and polymers [1]. In this context, the central focus is on process development and characterisation, as well as on the development and characterisation of suitable powder materials [2,3]. This manufacturing process is also known as ‘selective laser sintering (SLS)’, ‘selective laser melting (SLM^®)’ or ‘laser powder bed fusion (LPBF)’ Specifically, laser-based powder bed fusion is being investigated for the additive manufacturing of porous and dense three-dimensional components [1,2,3,4,5]. In essence, the process entails the layer-by-layer fusion of powder bed areas using laser radiation, in accordance with DIN EN ISO 17296 and ASTM 52900 guidelines [4,5].

The deployment of glass powder materials is currently the focus of research, encompassing diverse glass materials and plant modifications [1]. Various glass materials, including pure fused silica [6,7,8,9,10,11] and multi-component-glasses, e.g., borosilicate glasses [7,12,13,14,15], soda-lime glasses [13,16,17,18,19,20,21] and alkali borosilicate glasses [22,23,24] have been the subject of investigation.

Wang et al. have already described the laser power of the CO₂ laser (P < 50 W) and the particle size (d = 100...300 µm) of the SiO₂ powder as the main influencing factors on the component densities (ρ_max = 44%) [6]. Further research work in the field of SiO₂ glass powder materials (spherical and irregularly shaped, average particle diameter d₅₀ = 20; 59; 175 µm) revealed a substantial influence of the laser-beam diameter (d_LB = 0.327…2 mm) on the component density (ρ_max = 99%) [7,8,9,10,11].

However, material and process-specific investigations were mainly carried out for multi-component glasses, in which predominantly commercial, partially modified system technologies for metallic powder materials and NIR laser sources are used [13,14,16,17,18,19,20,21]. In the case of borosilicate glass powder, the use of near-infrared laser radiation and larger powder particle diameters (d = 125...200 µm instead of d₅₀ = 30 µm) resulted in a 34% improvement in component density [12,13,14,15]. By systematically investigating and optimising the process parameters, the findings and results, which include correlations such as the maximum density in the range of ρ_max = 88–95% by using powders with a particle size of d₅₀ = 44; 64.95; 160 µm, were also achieved for soda-lime glass powders [13,14,15,16,17,18,19,20]. Furthermore, process window limits were identified as a function of powder particle size within which component manufacturing is possible. (Volume energy density E_V = 70…120 J/mm³ by d₅₀ = 44 µm and E_V = 65…110 J/mm³ by d₅₀ = 109 µm) [17,21]. However, it was determined that the reduction in particle diameter resulted in enhanced geometric resolution, due to the possibility of minimising layer heights. For instance, the thickness of the wall could be reduced by a minimum of 67% [17,21].

In relation to the application of alkali borosilicate glass for PBF-LB, CO₂ laser radiation is utilised [22,23,24]. It was determined that the use of commercial NIR-PBF-LB systems is not possible due to the low absorption levels exhibited [24]. Alkali borosilicate glasses are distinguished by their capacity for segregation and extraction of alkali and borate components [24]. The present investigations are focusing on the generation of defined pore sizes in the µm and nm range by combining PBF-LB with segregation and extraction of sodium borosilicate glass (SBG) [22,23]. A homogeneous powder bed could be generated and quantitatively described for different particle size diameters and distributions (d₅₀ = 28; 68 µm) [22,23,24]. In addition, a differential thermal analysis (DTA) connected to the PBF-LB was used to retroactively determine process temperatures between T = 660…1000 °C [22] and confirmed by thermographic measurements [23]. Further ana-lyses have demonstrated that the PBF-LB-related minimisation of the borate-rich domains results in the formation of smaller nm pores [22]. In principle, bimodal pore systems with pore diameters ranging from d_pore = 10…1000 µm could be manufactured by PBF-LB, and bimodal pore systems with diameters ranging d_pore = 3…50 nm could be produced by extraction, with a total porosity of 73% [22].

The present study determines the glass morphologies for alkali borosilicate glass that can be achieved using PBF-LB. The investigations focus on the comprehensive analysis of the entire process window, including the characterisation of the porous and molten glass morphology, as well as the assessment of their dependence on shape and dimensional accuracy.

2. Development and Characterisation of the Experimental Setup

A prototype system technology has been utilised [6,7,9,24]. In this configuration, the CO₂-laser radiation is guided via deflecting mirrors to a scanning system that incorporates a focusing lens and is focused onto the building platform, which is surrounded by a heating chamber (see Figure 1).

In principle, the system technology has implemented a beam path that enables direction-independent laser material processing through the realisation of circular polarisation of the laser beam. Moreover, the heating chamber, in which the building platform is integrated, enables homogeneous heating of the powder bed up to T = 400 °C on the surface [23]. The initial

Table 1. Technical specifications of the laser source, the beam path and the heating chamber.

Laser Source
Operating mode	Continuous wave (frequency = 25 kHz)
Wavelength	10.6 µm
Laser Power	>100 W
Beam Diameter	2 mm
Divergence angle	0.7 mrad
Ray Path
Telescope magnification	5×
Scanning System
max. scanning speed	20 m/s
repeat accuracy	<2 µrad
Focusing lens
focal length	298 mm
calc. focus diameter	322 µm
Heating Chamber
max. heating temperature	1000 °C
Building platform
material	Quarzal®
diameter	120 mm
vertical travel distance	200 mm

lists the technical data of the different components.

This system technology enables the laser-beam diameter to be varied on the building platform in a theoretical range of d_LB = 0.322…4 mm. This is due to the height-adjustable scanning system and depends on the focussing optics used. The investigations are based on three theoretically defined beam diameters. (Focus diameter, d_LB = 1.0 mm and d_LB = 2.0 mm) The application of these requires the characterisation of beam caustics. The beam caustic is defined as a function of the beam diameter in relation to the propagation axis, as measured from the focusing optics to the focusing plane. The position and size of the focal diameter and resulting defocusing heights (z_dh) are determined by measuring and analysing beam caustics using the 1/e² method, as illustrated in Figure 2a.

The minimum measurable focus diameter is thus d_F = 317 µm and a defocusing height of z_dh = 20 mm (d_1.0 = 1.07 mm) and z_dh = 40 mm (d_2.0 = 2.06 mm) is determined for the selected beam diameters. In order to achieve the desired beam diameter, defocussing is required. This is achieved by increasing the distance between the focussing lens and the building platform. In accordance with the increased beam diameter, a corresponding decrease in intensity can be observed, as demonstrated in Figure 2b.

Furthermore, it is necessary to determine the usable laser power. In principle, the power is reduced by the beam guidance optics by ΣΔP < 10% due to reflection and absorption losses. Within the focal plane area, an actual power difference of ΔP < 15% could be measured by power meter (StarLite, OPHIR^®, Jerusalem, Israel). This is considered in the calculation of the volume energy density. The calculation of volume energy density (E_V) is dependent on the process parameters laser power (P), scanning speed (v_S), layer thickness (h_L) and hatching space (h_S), using Equation (1).

$${\mathrm{E}}_{\mathrm{V}}={\displaystyle \frac{\mathrm{P}}{{\mathrm{V}}_{\mathrm{S}}·{\mathrm{h}}_{\mathrm{L}}·{\mathrm{h}}_{\mathrm{S}}}}$$

(1)

Additionally, a thermographic camera (sc655, FLIR Systems (Thousand Oaks, CA, USA)) has been integrated into the system technology to analyse the heat affect zone. The process temperature is recorded in the fourth layer, and the average process temperature is determined from a measuring line (length = 10 mm, width = 0.39 mm = 1 pixel) aligned orthogonally to the scan direction in the centre of the component. Furthermore, the material-dependent emissivity is determined using a calibration source (9150, WIKA Alexander Wiegand (Klingenberg, Germany)) with a value of ε = 0.95.

3. Determination of Process-Relevant Material Properties

In the context of laser-based powder bed fusion, the morphological and rheological powder properties are of significance in generating a homogeneous powder bed. Additionally, the thermal material properties, particularly the specific temperature-dependent flow behaviour, are crucial for characterising the resulting glass morphology.

Sodium borosilicate glass, which belongs to the alkali borosilicate glass category, is utilised as the raw material. The glass powder material was subjected to two cycles of melt quenching in order to produce a homogeneous glass composition. The composition was then determined by ICP-OES, yielding the following results: SiO₂ = 61.8 mol-%, B₂O₃ = 30.5 mol-%, Na₂O = 6.8 mol-%, Al₂O₃ = 0.9 mol-% and impurities < 0.1 mol-%. [22]

In preparation for the PBF-LB process, the powder material is subject to grinding and sifting, resulting in a powder with irregularly shaped particles (see Figure 3a).

Due to the irregular shape of the particles, the measurement of particle size is repeated four times using laser diffractometry in order to take into account measurement uncertainties due to the deviation from the ideal spherical shape (in accordance with [25]), see Figure 3b. This results in a particle size distribution (PSD) with d_P < 150 µm and with the specific particle diameters d₁₀ = 5.0 µm ± 0.77 µm, d₅₀ = 34.7 µm ± 3.50 µm, d₉₀ = 79.4 µm ± 1.02 µm, at which 10%, 50% and 90% of the powder particles of the total quantity are present, respectively. As posited by Azema et al., the calculation of the size range S_PSD = 0.88 (S_PSD = (d₉₀ − d₁₀)/(d₉₀ + d₁₀)) results in a broad poly-dispersive s-shaped particle size distribution, which in turn results in a high packing density [26]. The high particle density is intended to produce a homogeneous powder bed, which is a basic prerequisite for defect-free PBF-LB components. However, due to the irregular particle shape and the resulting lower flowability, there is the potential for limitations in terms of transport suitability.

The rheological powder properties and subjective assessments of the generated powder bed homogeneity have been employed as indicators of the process suitability of the respective powder [7,13,17,21,22,23,24]. However, contrast analysis [23], a quantitative analysis technique, enables powder bed homogeneity (PH) to be described as the degree of powder coverage of the powder bed, depending on the single-layer height and the powder transport system within the plant technology used. For this purpose, the powder material to be examined is applied to a high-contrast prepared building platform (see Figure 4a) at the layer height to be analysed using the transport system. Mechanical post-processing is employed to attain the same surface roughness as on the original building platform. This is achieved in order to prevent any influence on the flow properties of the powder material. The powder-covered area of the building platform (see Figure 4b) is then analysed graphically (see Figure 4c). As illustrated in Figure 4c, the presence of defects within the powder layer results in diminished powder homogeneity, consequently leading to a decline in building quality.

In order to eliminate interferences, for example, caused by shadows cast by the system technology or changing ambient conditions, it is essential to ensure homogeneous illumination of the area to be analysed. To this purpose, a ring light is mounted above the building platform, thus reducing the deviation of the illuminance from the centre of the building platform to less than 5%. Contrast analysis constitutes a viable methodology for the evaluation of powder materials with regard to their process suitability for PBF-LB [22,23]. In the case of the powder material under consideration, the results of the contrast analysis indicate a powder homogeneity level of PH ≥ 99% for a single layer with a height of h_L ≥ 150 µm, thus confirming its process suitability (see Figure 4d).

Furthermore, the results demonstrate that the homogeneity of the powder bed increases with the height of the single layers. This is to be expected, as particle size-related voids do not occur with increasing single-layer height. As indicated by the preceding particle size analysis, the powder material contains particles (d₉₀ = 79.4 µm) which are larger than the minimum layer height examined (h_L = 50 µm). This fact, as well as agglomeration or contamination caused by the squeegee process, leads to voids, see Figure 5a. Contrast analysis can thus be applied to determine the minimum layer height at which no additional defects occur on the powder surface (h_L = 150 µm), as illustrated in Figure 5a.

Due to the process and material, an uneven powder surface always occurs, which is limited by the packing density of the powder material used. This unevenness cannot be determined by means of contrast analysis. Hence, contrast analysis is considered a suitable method for determining the minimum layer height under real operating conditions for an existing powder material.

Furthermore, the relevance of the contrast analysis with regard to the minimum possible layer height is evaluated using white light interferometry (WLI), s. Figure 6.

The figure demonstrates that, as expected, rougher surface structures are achieved with low layer heights. Increases in layer height result in a smoother powder bed and a reduction in defects. The findings of this study indicate that contrast analysis is a suitable method for evaluating the usability of the powder material in relation to the layer height. The contrast analysis is regarded as offering distinct advantages due to its practical and cost-effective in situ implementation.

In addition, it is evident that the support layer necessary for each build job [6] results in a powder bed height of h_PB = 0.5 mm. This, in turn, has the additional effect of producing a homogeneous powder layer for the LBF-LB process.

The shaping and structuring of glass materials can be achieved by both sintering and melting processes, and is characterised by temperature-dependent viscosity. The glass is melted at a viscosity of ƞ= 10² dPa∙s. However, the sintering process occurs at considerably lower temperatures, typically above the softening point (ƞ = 10^7.6 dPa∙s), and is characterised by the incomplete melting of the glass powder material. Due to the higher viscosity and the resultant lower process temperatures, the sintering process exhibits a mould-retaining morphology, concomitant with a porous glass morphology [28,29]. As illustrated in Figure 7, the viscosity–temperature curve includes the characteristic viscosity fixed points of the SBG material.

The calculation was carried out in accordance with the Vogel–Fulcher–Tammann equation [28], based on the measured values from [24]. In order to assess the temperature-dependent shaping of the powder material, this is determined in the range of the viscosity fixed points by means of heating microscopy. The deformation temperature (T_D, i.e., the temperature at which the initial changes in shape become visible, e.g., through edge rounding) and the hemispheric temperature (T_H, i.e., the temperature that characterises the onset of the melting range) are analysed in accordance with DIN 51730 [30] and Scholze et al. [31]. The diameter of the used tablet-shaped powder compacts is measured at d_HTM = 10.2 mm, with a corresponding height of h_HTM = 2.62 mm.

Additionally conventional melting tests are carried out for these process-relevant temperatures in order to analyse the resulting glass morphology using SEM. This enables a comparative analysis of the resulting change in shape and the resulting glass morphology as a function of temperature, which is shown in

Table 2. Determined shaping of the glass material (HTM) and glass morphology (SEM, magnification: 200× T_G; 50× T_S, T_D, T_H, T_P) for the respective temperature points.

	TG	TS	TD	TH	TP
T [°C]	413	659	716	866	891
lg ƞ [dPa∙s]	13.3	7.6	6.7 (calc.)	4.9 (calc.)	4.0
HTM
SEM

.

The investigation revealed no detectable morphological change concerning the glass transition temperature (T_G) using HTM, a phenomenon that can be confirmed by the SEM image. This is due to the fact that mechanical stability cannot be achieved at this temperature, and consequently, no alterations to the powder particles can be detected. However, shrinkage of the sample body was detected at the softening point (T_S) using HTM imaging. In addition, the SEM analysis revealed particle adhesion and unstable mechanical stability. Furthermore, from the deformation temperature (T_D) onwards, it was possible to produce dimensionally stable samples. It was observed that these samples exhibited increased porosity, which decreased as the temperature increased.

The investigation into temperature-dependent changes in shape indicates that the initial material shrinkage occurs within the range of the glass transition temperature (T_G= 413 °C, ƞ= 10^13.3 dPa∙s) and the softening temperature (T_S = 659 °C, ƞ = 10^7.6 dPa∙s). Beyond the softening temperature, an open-porous glass morphology becomes visible, resulting in unstable mechanical properties due to marginal particle adhesion. As the temperature rises, a concomitant decrease in viscosity (T_D = 716 °C, ƞ = 10^6.73 dPa∙s) leads to the discernible onset of deformation, characterised by edge rounding and a macroscopically porous melt glass morphology. By further temperature increase (T_H = 866 °C, ƞ = 10^4.91 dPa∙s), the theoretical yield point (ƞ = 10^5.0 dPa∙s) is almost reached and the melt glass morphology with discrete closed pores is visible on the macroscopic scale. These pores decrease in size as the temperature increases, as illustrated by the working temperature (T_P). It can be observed that, due to the viscous flow of the material, changes in shape are induced from the deformation temperature onwards, which are no longer dimensionally stable after T_H.

The investigations demonstrate that the PBF-LB process is feasible from the softening temperature, but only achieves mechanical stability at the deformation temperature. From this temperature range, open-pored glass melt glass morphology is created, which change into a closed glass morphology as the temperature rises, although the shape retention is reduced at the same time. The resultant open porosity manifests in a partially unstable component, forming at low temperatures and a low-porosity molten glass morphology with reduced shape retention at high temperatures. Theoretically, PBF-LB has the capacity to produce a molten glass morphology in the range of T_S ≥ T_PBF-LB < T_M, which exhibit both open and discrete closed porosity. It is necessary to analyse the process limits with regard to mechanical strength, as well as dimensional and shape deviations.

4. Results on Process Development and Resulting Component Quality of the PBF-LB Process for SBG Materials

As part of the PBF-LB experiments, a comprehensive analysis of the entire process window will be carried out. This includes the characterisation of porous and molten glass morphology and the evaluation of their shape and dimensional accuracy. As part of the process characterisation studies, the volume energy density (E_V) is varied for three different beam diameters. The hatch spacing and the preheating temperature (T_PB) are not varied, whereby this procedure is based on initial preliminary investigations [23] and the minimum possible layer height is kept constant for all tests on the basis of a powder analysis with regard to the powder bed homogeneity (see Section 3). In addition to the beam diameter and laser power, the scanning speed is also varied. In previous publications on the PBF-LB of multi-component glasses, the focus was on very low scanning speeds (v_S ≤ 0.50 m/s [7,13,21,24]), as the objective was to investigate the production of transparent, vitreous glass components. An analysis of the scanning velocities in the range of v_S = 0.1…0.40 m/s was carried out by Koppka et al. with the aim of generating a defined porous glass morphology [22].

In the present investigations, the fundamental glass morphology that can be manufactured using the PBF-LB process will be subjected to analysis with respect to achievable shape accuracy. As a result, it is essential to evaluate the entire process window area with regard to the generation of open-pored glass morphology through to a glazed, closed glass morphology. For this purpose, the varying and constant process parameters are shown in

Table 3. Varied and constant process parameters.

Varied Setting Parameters
dLB [mm]	0.317/1.0/2.0
P [W]	10/30
vS [m/s]	10…2000
EV [J/mm3]	1.1…222.2
Constant setting parameters
hs[µm]	90
hL [µm]	150
TPB [°C]	354 ± 11

.

The disc-shaped PBF-LB components are manufactured using a unidirectional beam guidance for a target diameter of d = 14 mm and a number of layers of n = 8, see Figure 1. In addition, the entire construction platform is covered with a porous, lattice-like SiO₂ support layer by PBF-LB. This increases the stability of the powder bed (see Section 3) and can be removed mechanically after the component has been finished. In principle, the support layer is an essential element for achieving a stable process, thus preventing misalignment of the individual layers that is related to powder application. As demonstrated in Figure 8, the limits of the achievable process window of the manufactured components are derived from the varied process parameters, which are represented by the volume energy density as a function of the laser-beam diameter.

Independent of the various influencing factors, the process window for the analysed parameter range has been determined to be 1.1 J/mm³ > E < 222.2 J/mm³. In this range, components can be detached from the building platform without defects, i.e., they are mechanically stable and can therefore be evaluated.

For the largest beam diameter (d_2.0) within the low energy density range, the present process window demonstrates a distinct limitation, a phenomenon that is amplified at higher power settings (P = 30 W), s. Figure 7. Thus, for higher laser power, defect-free components are attainable from 37 J/mm³, which is already achieved from 2.5 J/mm³ for the other laser-beam diameters, see

Table 4. Achieved process window limits for the respective beam diameter and laser power level.

	P	E
dF	10 W	1.1…74.1 J/mm3
	30 W	1.5…74.1 J/mm3
d1.0	10 W	1.6…24.7 J/mm3
	30 W	2.4…222.2 J/mm3
d2.0	10 W	7.4…74.1 J/mm3
	30 W	37.0…222.2 J/mm3

. It is postulated that the density range within which a different glass morphology can be produced is more limited for the highest beam diameter compared to the other two beam diameters.

The analysis of the process window for low laser powers indicates that a shift in this range towards lower energy densities can also be observed for the beam diameter d_1.0, s. Figure 8 and Table 4. In this particular context, the investigation of the dependence of the achievable density range on laser power is of interest. It is possible to produce defect-free components in a wide energy density range using the focussed laser radiation, irrespective of laser power. However, the maximum energy density that can be utilised is considerably lower than that of the other beam diameters.

In the following analysis, the achievable glass morphology is examined as a function of various variables, including the laser-beam diameter and the energy input. Furthermore, the significance of component density, dimensional and shape accuracy is considered.

4.1. Component Density

The relative density (ρ_rel) is determined by measuring the bulk density of the manufactured PBF-LB components (liquid displacement method A2, DIN EN ISO 18754 [32]). The relative density is calculated by determining the ratio of the bulk density to the pure density. The pure density of the glass material (ρ = 2.21 g/cm³) was determined by Krenkel et al. [24]. The bulk density is calculated by determining the mass of the dry sample (m₁), the immersed sample (m₂) and the sample impregnated with n-heptane under vacuum (m₃) taking into account the density of the immersion liquid at test temperature (n-heptane, ρ₁), see Equation (2).

$${\mathsf{\rho }}_{bulk}={\displaystyle \frac{m_1}{m_3-m_2}}{\rho }_1$$

(2)

The measurement is repeated twice for all defect-free components [32].

The parameter settings analysed enabled the attainment of a density range of ρ_{rel =} 37.8…94.2%, as demonstrated in Figure 9.

In essence, it can be demonstrated that smaller beam diameters (d_F) can attain higher component densities within the same volume energy densities, irrespective of the laser power. For instance, at a volume energy density of E_V = 11 J/mm³, a component density of ρ_dF = 91% to ρ_d2.0 = 67% is achieved at P = 10 W in each case. This is to be expected, due to the fact that the intensity of laser radiation decreases with increasing beam diameter, see Figure 3.

Furthermore, it was determined that a linear increase with the energy input could be observed for all used beam diameters up to a density of ρ_rel ≈ 90%. However, beyond this point, a phenomenon of saturation was observed. For the combination of the largest beam diameter and the highest laser power, it is not possible to produce components with ρ_rel < 90%. Moreover, it can be derived that, by using the focal diameter, almost identical component densities can be achieved with the same energy densities, regardless of the laser power. For instance, when an energy density of E_V = 7.4 J/mm³ is considered, and consequently a scanning speed of v_S = 0.1 m/s for P = 10 W and v_S =0.9 m/s for P = 30 W, it is evident that merely a component density difference of Δρ = +0.6% for P = 30 W is ascertained. For explanation of the different component densities, an analysis of the glass morphology achieved for different volume energy inputs is carried out depending on the laser power, see

Table 5. Comparison of the resulting glass morphology of different energy densities in comparison to the laser power level for the focused-beam diameter. (Measured by SEM, scan direction = vertical). In order to enable a comparison of the particle size of the starting material, an additional symbolic scale for d₉₀ is integrated.

	EV = 2.9 J/mm3	EV = 7.4 J/mm3	EV = 14.8 J/mm3	EV = 74.1 J/mm3
dF (by P = 10 W)
	ρrel = 56.2% vS = 0.25 m/s	ρrel = 82.2% vS = 0.1 m/s	ρrel = 92.9% vS = 0.05 m/s	ρrel = 91.1% vS = 0.01 m/s
dF (by P = 30 W)
	ρrel = 60.3% vS = 0.75 m/s	ρrel = 82.8% vS = 0.3 m/s	ρrel = 91.1% vS = 0.15 m/s	ρrel = 90.3% vS = 0.03 m/s

.

The microscopic illustration that the laser-induced line-like energy input leads to the formation of a vitrified glass morphology, which can be differentiated in size, shape and arrangement by varying the energy density, regardless of the laser power. Components with a low density, for example, have finely structured periodic pulsed glazed strands arranged in a net-like pattern and therefore have large pore sizes. The samples exhibit minimal vitrification, with a sinter-like glass morphology. This leads to a vitrification glass morphology that partially correspond to the maximum powder particle size and exhibit incompletely fused powder adhesions. For instance, the thickness of the glazing strands is d_gs =117…366 µm for E_V = 2.9 J/mm³ and thus in the range of the focus diameter (d_F = 317 µm). For larger beam diameters, periodically pulsating vitrification strands are also observed in this size range. Consequently, it can be assumed that the characteristics of the vitrification strands depend primarily on the particle size and less on the laser-beam diameter.

In addition, SEM and X-ray diffractometry are used to analyse the resulting vitrification strands of the PBF-LB components with regard to their resulting glass structure, see Figure 10.

The analysis demonstrated the presence of an X-ray amorphous glass structure, a consequence of the utilisation of grounded glass powder. In the context of this microstructure analysis, the presence of phase boundaries could not be identified through SEM observation. These results are consistent with those of Datsiou et al. [21], who also detected an amorphous glass structure in the PBF-LB samples. The discrete peak at 26.14 °C could be indicative of contamination due to processing and/or manufacturing. This hypothesis requires detailed verification in follow-up investigations.

As established in Section 3, the conventional melting experiments, it is only possible to produce a mechanical stable component above the softening temperature. For mechanical separation of the PBF-LB components from the building platform, the stability is necessary in order to be able to analyse the manufactured components. It can therefore be assumed that the mechanically less stable open-pored glass morphology, which corresponds to a sintered morphology, can be produced by PBF-LB in the temperature range < T_S, but cannot be detached from the building platform and analysed.

In contrast, high volume energy densities result in the formation of linear, highly pronounced vitrification strands, whose width and degree of fusion increase steadily. The orientation of these vitrification strands is consistent with the direction of the scan. Additionally, it can be observed that almost an identical glass morphology can be achieved at constant volume energy densities. These densities are achieved by varying the scanning speed and adjusting the laser power. This indicates that by applying higher laser power (P = 30 W), a higher building rate can be attained, and the duration of single-layer production can be diminished by a third, with a volume energy density of E_V = 14.8 J/mm³.

The saturation of component density at a volume energy density of E_V = 11 J/mm³, and the nevertheless present change in the characteristics of the vitrification strands, indicate the strong influence of the process-related linear energy input. The resulting heat-affected zone (HAZ) in the powder material is dependent on the intensity, as illustrated in Figure 3, and thus in particular on the beam diameter. The characteristics of this heat-affected zone are presented in

Table 6. Comparison of generated glass morphology (SEM) and induced heat-affected zone (thermographic camera) for each beam diameter at maximum energy density.

	dF (by P = 30 W)	d1.0 (by P = 30 W)	d2.0 (by P = 30 W)
	E = 74.4 J/mm3, ρrel = 90.3%, TPBF-LB = 1291 °C	E = 222.2 J/mm3, ρrel = 90.8%, TPBF-LB = 1243 °C	E = 222.2 J/mm3, ρrel = 92.3% TPBF-LB = 1282 °C
Surface
HAZ

, which provides a comparative analysis of the three beam diameters and their respective glass morphology.

As can be observed in the table, the width and length of the heat-affected zone increase with increasing laser-beam diameter, as expected, whereas the energy input is constant. Although the intensity of the laser radiation will naturally decrease as the laser-beam diameter increases, similar process temperatures are achieved This is due to the increasing degree of overlap (O) of the beam diameters (O_dF = 64.7%, O_d1.0 = 88.6%, and O_d2.0 = 95.5%, calculated according to Hecht [33]). This results in a more homogeneous softening of the powder material due to the more extensive heating, which can be seen in a direct comparison of the SEM images (Table 5). In addition, the viscosity decreases with increasing temperature, which also supports the levelling of the glazing strands.

In order to achieve a comprehensive analysis of the formation processes of the vitrification strands, the interaction mechanisms between powder material and laser radiation are investigated. This is carried out using a high-speed camera, which is used simultaneously to the thermographic process. Figure 11 presents an exemplary image for illustrating the characteristic molten pool behaviour at E_V = 222.2 J/mm³ for the average laser-beam diameter (d_1.0).

The illustration demonstrates the interaction area between the laser beam and the glass powder material and the resulting molten pool. Furthermore, the phenomenon of evaporation is observable. The simultaneous determination of the process temperature of T_PBF-LB = 1243 °C and the powder bed temperature T_PB = 354 °C allows the derivation of a theoretical viscosity range of ƞ= 10^15.47…10^2.05 dPa s, compare Section 3 and Figure 7. The combination of parameters under investigation is intended to induce process temperatures within the melting range of the glass material. From a phenomenological perspective, this classification is also consistent with conventional HTM investigations of the starting material and the resulting evaporation phenomena. In subsequent studies, it is imperative that this hypothesis be validated and additional investigations be conducted into energy densities and powder materials. The aim is to further explore the characteristics of the melt pool by means of appropriate metrological characterisation and simulations of the heat conduction effects in the glass powder bed, analogous to the results known from other materials (e.g., [34,35,36,37,38]). For example, the reduction in pores and the achievement of a more homogeneous surface quality as well as an increase in component density has also been proven for metallic materials, by varying the laser-beam diameter [33,34].

Due to the distinct heat-affected zone and the theoretical shape deviations caused by the temperature-specific flow behaviour of the material (Section 3), the actual dimensional accuracy is also investigated.

4.2. Dimensional Accuracy

The achieved dimensional accuracy is considered as a function of energy input and beam diameter and, as expected, decreases with increasing energy input and laser-beam diameter, see Figure 12.

In consideration of the correlation, maximum deviations from the component diameter of Δd_max < 1 mm can only be achieved with the focused-beam diameter. For the beam diameters d_F and d_1.0, no differences in laser power are observed. As described before, for the beam diameter d_2.0 with P = 30 W, only components with a density ρ_rel > 90% can be achieved with a relatively high energy input. This relationship, as expected, leads to the highest deviation in dimensional accuracy.

The ensuing

Table 7. Macroscopic representation (digital microscope; VHX 7000, Keyence (Osaka, Japan)) of the produced PBF-LB components depending on the minimum and maximum energy density for the different laser-beam diameter. Red border marks the target diameter of d_t = 14 mm.

	E = Minimal	E = Maximum
dF (by P = 30 W)
	E = 1.5 J/mm3 | ρrel = 45.2%| Δdmax = −0.42 mm	E = 74.07 J/mm3 | ρrel = 89.3% | Δdmax = 1.05 mm
d1.0 (by P = 30 W)
	E = 2.3 J/mm3 | ρ = 44.9%| Δdmax = 1.56 mm	E = 222.2 J/mm3 | ρ = 90.8% | Δdmax = 5.76 mm
d2.0 (by P = 10 W; 30 W)
	E = 7.4 J/mm3 | ρrel = 57.6%| Δdmax = 3.23 mm	E = 222.2 J/mm3 | ρrel = 92.3%| Δdmax = 6.63 mm

illustrates the attained component quality, using the different beam diameters at P = 30 W, for the minimum and maximum energy input. The translucent whitish lattice structure is the result of the support geometry used, which creates a ground and thus an unpolished rough surface through mechanical cutting.

As previously outlined in Section 3, an open-pore glass morphology exhibits a small melt glass strands and thus minimal shape deviations. This fact can also be demonstrated for PBF-LB, since the lower energy input results in less deviation from the target diameter. Due to the small heat affected zone this deviation is, as expected, smallest for the focused-beam diameter. Consequently, the most significant deviation from dimensional accuracy is observed for the highest energy density in conjunction with the largest beam diameter. Simultaneously, the most homogeneous surface morphology is produced. In addition to the dimensional accuracy, the increase in glazing strands and the resultant increase in component density with the energy input is also evident in this instance. For high energy inputs, a glazed and partially transparent glass surface is visible but the lowest dimensional accuracy is achieved as a consequence. Achieving precise dimensional accuracy specifications is dependent on the utilisation of parameters that result in low component density, or the application of adequate post-processing methods. In subsequent investigations, it is imperative to enhance the quality of components, with particular emphasis on dimensional accuracy and mechanical stability, through the implementation of statistical process optimisation.

4.3. Bending Strength

Glass is a brittle material whose flexural strength (σ) is significantly lower in comparison to metallic materials. An exemplar of this phenomenon is BOROFLOAT 33^® (σ = 25 MPa), which exhibits a chemical composition similar to NBS glass.

The strength is influenced by a variety of factors, including in particular the molecular binding forces resulting from the chemical composition, as well as inhomogeneities, defects and contamination in the material under investigation. Furthermore, the test conditions (geometric shape, workpiece and ambient temperatures, type and speed of loading, etc.) have a significant influence on the results. Consequently, it is necessary to maintain these conditions at a constant level in accordance with the relevant standards [1].

In the following, initial bending tests are carried out for the focal diameter (d_F) with regard to the highest and lowest volume energy density. The biaxial bending test, in accordance with ISO 6872 [39], is applied and evaluated for rotationally symmetrical and inorganic (ceramic) components. The results are visualised in Figure 13.

The evaluation of the data indicates that a low volume energy density (ρ_rel = 56.2%) results in an average bending strength of σ₀ = 4.51 MPa, while a high-volume energy density results in a value of σ₀ = 19.86 MPa. The findings indicate that volume energy density is a substantial factor in determining the average bending strength. The high dispersion at high volume energy density is particularly pronounced and can be explained by the uneven surface topography typical of the process and the resulting porosity, as discussed in Section 4.1 and Section 4.2.

In principle, the flexural strength values for multi-component glass (σ < 7 MPa [21]) previously achieved using the PBF-LB process were significantly exceeded. Nevertheless, the bending strength of these additively manufactured components is significantly lower than that of commercially manufactured components (e.g., BOROFLOAT^®). In subsequent investigations, it is imperative to analyse the resulting bending strengths as a function of the beam diameter. Furthermore, it is essential to enhance the bending strength through process parameter optimisation and post-processing.

5. Conclusions

The objective of identifying the glass morphology achievable using PBF-LB, as well as the influence of laser-beam diameter, energy input, and thermal material properties, was achieved. A significant focus of research has been the analysis of temperature-dependent forming, which has enabled the demonstration of the glass morphology as a function of temperature and dimensional stability through conventional melting processes of the raw material. This investigation led to the observation that shaping of the powder material is possible above the softening temperature, but mechanically stable not until the deformation temperature has been reached.

The investigations on materials analysis of the thermal deformation and the resulting glass morphology and dimensional accuracy are considered in order to characterise the process limits and phenomena. In addition, the entire range of glass structures, ranging from open-pored to closed-melt configurations, could be analysed for the first time. In general, it has been determined that the characteristic vitrification strands are a consequence of the volume energy density and the process-related line-like energy input. Furthermore, it has been shown that these strands are formed independently of the laser-beam diameter, yet that they vary in shape and size. This indicated that the characteristics of the resultant vitrification strands depend on the HAZ resulting from the setting parameters. Broad HAZs and therefore large beam diameters result in a broad and homogeneous surface morphology and therefore high component densities. Conversely, the smaller laser-beam diameters lead to the formation of small HAZs, consequently resulting in filigree vitrification strands. This, in turn, results in low component densities. However, it should be noted that these beam diameters do not permit the creation of a commercial dense glass micro structure, but they do permit component densities ρ_rel > 90%. In principle, a component density range of ρ_rel = 37.8…94.2% is possible.

This research contributes to the overall understanding of the producible glass morphology and the process limitations of the PBF-LB process. The study provides insights into the effects of varying laser-beam diameter and energy density on the glass morphology, component density, and dimensional accuracy.

6. Outlook

The following investigations will focus on optimising dimensional accuracy through reverse engineering, mechanical and thermal post-processing, or laser-based processing of the individual layers. The utilisation of statistical evaluation methodologies will facilitate the optimisation of process parameters, thereby enhancing the quality of components and investigating their long-term stability.

Furthermore, the resulting PBF-LB manufactured components will be examined with regard to their segregation behaviour, the achievable nanoporosity and bimodal pore distribution, as well as their mechanical stability.

In order to achieve a more comprehensive understanding of the interaction mechanisms between glass powder and laser radiation, further in situ measurement methods must be developed. In particular, it is necessary to quantify the HAZ and its correlation to the resulting glass strands using high-resolution imaging measurement methods. Furthermore, the characteristics of the resulting crystalline and amorphous phases of the emerging glass structure will be investigated.