Surface roughness of as-printed polymers: a comprehensive review

Abstract

Surface roughness is gaining increasing recognition in the processing design methods of additive manufacturing (AM) due to its role in many critical applications. This impact extends not only to various AM product manufacturing but also to indirect applications, such as molding and casting. This review article discusses the role of processing on the surface roughness of AM-printed polymers with limited post-processing by summarizing recent advances. This review offers a benchmark for surface quality improvement of AM processes, considering the surface roughness of polymeric parts. For this purpose, it lists and analyzes the key processes and various printing parameters used to monitor and adjust surface roughness under given constraints. Four AM techniques for manufacturing polymeric parts are compared: fused filament fabrication (FFF), selective laser sintering (SLS), vat photopolymerization (VPP), and material jetting (MJT). A review and discussion of recent studies are presented, along with the most critical process parameters that affect surface roughness for the selected AM techniques. To assist in selecting the most appropriate method of 3D printing, comparable research summaries are presented. The outcome is a detailed survey of current techniques, process parameters, roughness ranges, and their applicability in achieving surface quality improvement in as-printed polymers.

1 Introduction

Additive manufacturing (AM), often known as 3D printing, combines materials to fabricate products from 3D model data, typically in a layer-upon-layer process [1]. AM has continued its exponential growth in many applications because of its attributes, such as mass customization, waste minimization, and on-demand design revisions [2]. However, AM parts cannot precisely replicate 3D CAD models due to the inherent surface roughness and accuracy limitations of the AM process. In-process enhancing the quality of the surface in AM technology is presently one of the most significant challenges of advanced manufacturing. It is a critical element for compatibility with surface coatings, the fatigue resistance of the products, liquid trapping, and the presence of moving particles [3].

Recently, the influence of the primary processing parameters (PPP) on the quality assessment of 3D-printed (3DP) objects has received considerable attention from academia and industry, mainly because optimizing these PPPs provides more fabrication competence based on mass customization, on-demand design revisions, and waste minimization. Enhancing the 3D model, material and process selection, and surface modification can satisfy the performance constraints of the 3DP parts, such as tooling [4], jewelry [5], sensors [6], performance improvement, production, personalization and customization, spare parts, maintenance, repair, art, design, and architecture [7]. Considering the growing applications of personal 3D printers (shortened form as “printer”) and the small-scale market for AM technologies, it is increasingly essential to thoroughly understand the surface morphology created by various 3DP methods. Not surprisingly, most post-processing machines are currently either unavailable or high-priced to most AM users.

Polymers have been the center of attention in fabricating 3D parts because of their cost, availability, ease of production, and appearance options, particularly in the case of entry-level 3DP machines [8, 9]. The Wohlers Report 2021 [10] reported 7.5% growth in AM industry. It shows almost half of AM service providers offer polymer 3D printing, and 29% provide polymers besides other materials, such as metals and ceramics. As a result, over 80% of the AM market is involved with polymeric materials. In addition, polymer powder consumption is mentioned as rising by 43.3% in 2021, overtaking photopolymers as the most commonly used additive manufacturing material. In total, the polymer 3DP market is expected to increase to $24 billion in 2024 and $55 billion in 2030 [11]. Since 2012, 14,150 out of 54,275 (26.1%) publications in the field of additive manufacturing applied to polymer-based techniques (data from Web of Science, 2023). As almost 80% of the AM market is dedicated to polymer-containing materials, this is still growing, and there is significant potential for future research and development of polymeric AM objects.

The polymeric AM parts suffer from poor surface finish and geometric deviation. Among textural appearance attributes, surface roughness is a critical indicator for assessing the quality of a product and the manufacturing process. For many direct and indirect applications, the surface of the printed object must meet specific criteria and properties such as mechanical [12], physical [13], tribological [12], and other quality attributes (QAs) [14].

Surface roughness is a metric relating to the QAs of AM parts because of its impact on the aesthetic appearance and the integrity of the piece in terms of its ability to interface with other components. It influences not only the appearance but also the functional properties of a part [15].

In most circumstances, the finishing process is rarely used to modify the part dimensions except for reducing the surface roughness via sandblasting and polishing or for structural applications [16]. In some cases, various painting and coating methods are used to achieve the required surface finish. These practices are insufficient to form the printing process and must overcome several challenges. For instance, material accumulation in fused filament fabrication (FFF), as a material extrusion (MEX) AM process, occurs along the edges and inside the products, which cannot be resolved by painting or coating [17, 18]. To choose a process based on the specifications of a part, Gordon et al. [19] provided a decision tree as a framework. They suggested the appropriate design modifications considering the desired surface to account for the selected techniques.

Various kinds of polymers are primarily supplied for AM in the shape of filaments, pellets, resin, or powder [20]. Furthermore, composite polymers reinforced by fibers and particles offer a favorable combination for almost all the existing AM methods [20]. While there are many choices of available AM processes to 3D print polymers, the mechanisms of the different AM methods distinctly differ from one another. Polymers are sensitive to printing parameters, mainly changes in temperature. Hence, the printing process and material should be carefully considered according to the end-user applications [11]. Several review papers have previously discussed roughness in AM processes, including material extrusion [18, 21], vat photopolymerization [22, 23], material jetting [24, 25], and selective laser sintering [26, 27] techniques. However, the reviews have not yet observed a thorough study of methods and quality evaluation trends in AM polymer products.

The AM process for polymers presents different challenges in surface quality than conventional manufacturing. A uniform standard for evaluating the roughness and dimensional accuracy of 3D-printed objects does not exist at present. Studying the surface roughness issues and the various PPP techniques for improving surface quality is also lacking in the literature. The roughness of AM technologies differs considerably, as was reported in this study and in the studies that compared 3DP methods [3, 16, 28].

This study focuses on recent advances in investigating the roughness of 3D-printed surfaces. It describes the primary AM processes for polymers and the corresponding PPP in the pre-processing and printing steps. The next section provides an overview of the AM processes for polymers and the research methodology. This is followed by an explanation of roughness measurements and metrics in the next section. For each of the AM methods studied, including FFF, selective laser sintering (SLS), vat photopolymerization (VPP), and material jetting (MJT), the main process parameters as well as surface roughness studies have been presented in separate sections. Furthermore, a discussion on the issues surrounding the setting up of polymer key AM processes. In another section, studies on several AM methodologies and their results were compared. This review study concludes with a summary, as well as a discussion of future trends and capabilities in the later sections.

2 Overview

2.1 AM processes for polymers

This work defines AM techniques by ISO 52900 – 2021 [29] and their generally accepted terms. A wide range of advanced manufacturing techniques is available, from the nanoscale to the macroscale. Part size is primarily determined by the working volume afforded by the system of motion of the machine. This literature review is focused on manufacturing at the meso- and micro-scales, where surface roughness can significantly affect the visible quality of parts [30]. Table 1 lists relative characteristics overview of AM techniques for polymers and their most relevant AM applications, including prototypes, medical devices, and precision mechanisms.

Table 1 Characteristic overview of polymer AM categories. Data extracted from references [9, 32]

In many applications, AM is still in its infancy and requires post-production finishing techniques (PPFTs), which include post-processing and surface finishing [17]. The process can either be used as the primary manufacturing process or as part of a chain of manufacturing processes. Figure 1 classifies finishing as a critical step and quality assurance information flow in a digital thread in additive manufacturing (DTAM). A series of pre- and/or post-processing is available to alter the surface and significantly enhance the appearance of AM parts. However, some of these methods are limited in changing the surface morphology of complex shapes inexpensively and reliably over time [31]. The following sections will provide a detailed description of AM key processes.

Fig. 1

Stepping through the digital thread in additive manufacturing (DTAM)

Thermoplastics and their composites are the primary polymer materials used for AM [33], which can be divided into crystalline and amorphous states. Table 2 lists some of the main polymers and their specifications used in the AM process. The publication share of main AM polymers is shown in Fig. 2. Most of these polymers are mixed and enhanced by manufacturers under various commercial market trademarks, especially resin-based feedstocks. Besides, many other polymeric compounds are used in specific AM processes, such as polydimethylsiloxane, ethylene vinyl acetate (EVA) [34], and commercial digital materials from 3DP machine manufacturers.

Table 2 Characteristic overview of different unfilled AM polymers for general purposes

Fig. 2

Total number of publications corresponding to polymers and their composites in additive manufacturing since 2012 (data from Web of Science, 2023)

2.2 Research method

This literature review focused on the current state of academic investigation with the broadest possible analysis of all recently published articles on surface roughness and 3D printing parameters. The review process was based on the content analysis of 55 articles. This review paper benefited from the preferred reporting items for systematic reviews and meta-analyses extension for scoping reviews method (PRISMA-ScR) for reporting scoping reviews as a general guide [52].

By utilizing the PRISMA-ScR method, a systematic and comprehensive scoping review approach is provided. A clear reporting framework facilitates transparency and replication of the review process, and a reduction of bias in the selection and interpretation of findings is achieved. It should be noted, however, that the method used had a few limitations. The scoping review methodology used may not provide a comprehensive review of all literature on surface roughness for polymer AM, particularly considering the broadness of the topic. A review process may have been limited by the quality of the selected articles and their generalizability for some methods. Thus, the content analysis of the articles may have been influenced by subjective judgments [52, 53]. For instance, available MJT papers in the studied field were considerably fewer than those for FFF (Fig. 3), resulting in more challenges for generalizations of the results. Besides, there were a variety of hand-made and tweaked 3D printers studied in the literature that may affect the review procedure.

Fig. 3

The most widely used polymer AM technologies and the total number of publications since 2012 (data from Scopus and Web of Science, 2023)

The publications were explored on Web of Science and Scopus to be as comprehensive as possible, as these scientific databases have high coverage of reputed high-impact publications.

Figure 3 shows the most common terminology and methods for polymers mentioned in the literature. Based on the number of publications in each AM category, 5 key processes are determined to be studied further. Accordingly, the authors selected the FFF process for MEX, SLA, and DLP processes for VPP, the SLS process for powder bed fusion (PBF), and the MJT category, including the PolyJet process. Other AM methods which generally use polymers in the fabrication of different materials and composites have been excluded, such as binder jetting (BJT) and sheet lamination (SHL). The market report on polymer 3DP machine sales [33] also confirms the same trend and technological share for selected key processes. Other techniques which were not mainly dedicated to polymers or rarely used for research are not included in our study. As AM growth is dynamic on a daily basis, their capabilities are sporadically mentioned.

Various generic keywords, such as “3D*print,” “additive manufacturing,” and “additive tech*,” were employed as criteria in the search section of the title, abstract, and keywords. The following is a formulation of research key strings for AM processes:

(("3d*print*" OR "advanced manuf*" OR "additive manuf*" OR "additive fabric*" OR "additive proces*" OR "additive tech*" OR "additive method*" OR "additive layer* manuf*"))

To examine the role of PPP on roughness, the post-processing of samples should be restricted to 1st level processes, as suggested by the Wohlers report [54]. It secures a minimal impact of post-processing on the roughness of the as-printed samples compared to the slightly post-processed replicas. It is usually less than a 20% deviation, depending on the specific needs of the project or application. Most studies reviewed here limited the PPFTs to a minimum number of steps to minimize the dimensional variations.

This work considered journal articles and conference proceedings to obtain a broader understanding of the topic. Upon eliminating duplicates, the titles and abstracts, availability of full text, and English language were screened before the full-text review. Following this, papers were controlled by their relevance to the present review paper, their originality in providing roughness evaluation for polymers, and their comprehensiveness and uniqueness in terms of the studied parameters and reported roughness metrics. Thus, papers that were out of these criteria were excluded from the study, which resulted in 55 articles separately being exported to Endnote and OriginPro 9.9.5 for in-depth analysis. Figure 4 summarizes the selection procedure used in the current study. The authors have 3D-printed several specimens for each AM category to visualize the surface roughness and texture conditions discussed.

Fig. 4

An overview of the screening procedure employed in this study

3 Roughness measurements and metrics

Additively manufactured surfaces are composed of various spatial frequency components, including profile, form, waviness, and roughness (Fig. 5). Each of these components has different origins and influences the appearance and functional performance of products differently. The waviness may reveal machine vibration, the form is usually produced by the poor performance of the manufacturing system, and the profile can be ascribed to layer-by-layer manufacturing. Roughness, however, is generated by surface irregularities due to printing and material removal errors. The waviness appears as a signal noise because of the planarity of the motion system and any deformations caused by weight or residual stress [55]. As a result of the specific printing process and materials used, there may be other sources of waviness, including defects in the printing process, thermal distortion, poor adhesion between layers, inadequate support structures, and mechanical deformation during post-processing [56, 57].

Fig. 5

General spatial frequency components of additively manufactured surfaces

Specifically, surface roughness is a critical texture component for assessing the quality of manufactured items by investigating the distribution of topographical features on the surface. Different metrics describe surface roughness because different industry sectors refer to various measures. Due to uncertainty in the surface quality of 3D-printed products, using several metrics would also be efficient [58]. For instance, Triantaphyllou et al. [59] concluded that average area roughness (S_a) and area root mean squared height (S_q) are appropriate metrics for measuring area surface roughness, as they were not sensitive to measurement parameters such as sampling length and evaluation length. In contrast, area height distribution skewness (S_sk) was found to sufficiently characterize the upside and downside surfaces from SLM parts.

Surface topography measurements based on data obtained from the 3D scanned images of a sample surface are either reported from a linear measurement, referred to by R, or from an aerial surface measurement indicated by S. The R and S metrics are defined and parameterized in ISO 21920–2 [60] and ISO 25178 [61], respectively. Area ratio, or the overall real surface area over the theoretical area of a smooth surface, can reveal how rough a surface is. Generally, the standard height-based metrics employed to describe surface roughness based on a linear profile can be derived from Eqs. (1)–(10) in Table 3. On the other hand, area roughness parameters are sometimes used to describe the roughness variation on a surface (Table 4).

Table 3 Surface roughness metrics based on linear measurement [60, 62]

Table 4 Area roughness parameters [28, 65, 66]

The roughness parameter R_a is widely used by researchers in AM studies as a straightforward metric to define and measure. Considering height variation as a general concept makes it easier to understand, but R_a is insensitive to wavelength variations [63]. Li et al. [64] revealed that the highest peak-to-valley distance parameter (R_z) was superior to R_q and R_a as standard metrics for measuring surface roughness. Li et al. reached a more significant correlation between R_z, tactile evaluation outcomes, and visual assessment results. Other appearance factors, however, influence sensory judgments, such as surface texture and color of the build material. Therefore, R_z alone is insufficient to comprehensively characterize the differences in human perception and surface QAs among samples.

Extraction of the roughness profile is not a common reproducible method because it depends highly on the operational instrument, shape, rotation, and displacement speed [3], as reported in several studies [3, 65]. While 2D profile measurements based on the stylus, according to ISO 4287, are still popular, there is a growing interest in X-ray computed tomography (CT) scan and contactless 3D optical profilometry, according to ISO 25178–2, to obtain more information without scratching the surface.

Launhardt et al. [66] evaluated four alternative methods for evaluating the surface roughness of Polyamide 12 components produced by SLS. According to them, stylus-based techniques scratch the surface somewhat without substantially altering its roughness. Despite being unable to measure the valley depth, the tactile method was the most reproducible among other studied techniques.

On the other hand, optical methods do not physically alter the surface but are hypersensitive to light reflection and surface transparency, leading to defect detection [67]. The advantages of this method, such as contactless measurement and a comparable R_a and roughness trend to tactile systems in the focus variation mode, make it a viable technique for smooth polymeric surfaces. The focus variation is a vertical scanning method with a shallow depth of focus. It simultaneously allows the measurement of steep flanks, form, and surface roughness [68]. Optical methods could also detect a three-dimensional topography of the surface and its roughness.

The focus variation method suffers from error because of the translucent polymer. The fringe projection and confocal laser scanning microscope represent higher roughness values and more sensitive measurements prone to outliers and faults [66]. The lower wear resistance of polymers in tactile methods and the possibility of the semi-translucent appearance in optical techniques make them more sensitive to method selection in roughness measurement.

Beitz et al. [69] used a confocal laser scanning microscope (CLSM) and X-ray micro-computed tomography (XMT) to measure surface roughness. They reported that roughness resulting from XMT data diverged substantially from those obtained from CLSM data. Due to the inability to level peaks, smaller measuring lengths result in higher arithmetic roughness values. The method also has a smaller sample size, resulting in lower surface roughness along the measurements section. Thus, assessing the roughness metrics of AM polymeric surfaces requires identifying the roughness evaluation method.

Regarding test artifacts for quality surface evaluation, most studies have used the twisted pillar (truncheon) [24, 65, 70], sloped surfaces [70,71,72], standard test artifact [65], or faceted sphere [24, 73] to measure roughness (Fig. 6). The design of an AM artifact should consider adaptability to various AM processes and machine sizes, as well as its ability to perform non-contact and contact measurements, editable geometry, and minimal material, and energy consumption. The twisted pillar is the preliminary design for this purpose which consists of a sequence of square segments rotated 0 to 90° around a central axis with 3° or 5° increments. The design is appropriate for measuring the surface roughness of an angled plane in the range of 0 to 360° [24]. Yet, it cannot meet all the above requirements of the measurement. Understanding the measurement process is an essential step in interpreting the results.

Fig. 6

3DP designs for studying build orientation (wedge angles): a Twisted tower, b tilted surfaces, c faceted sphere, and d Standard Test Artifact (STAR)

4 Fused filament fabrication (FFF)

Even though FFF 3DP has made significant advances so far, the fabricated parts tend to have a poor surface finish, including rough surfaces, voids, and prominent striations [2]. The morphology is relatively rough due to various limiting factors such as phase transformations, fast cooling, and exhaustive energy (Fig. 7). Although FFF is attractive for demanding applications, printed parts deviate from their initial designed geometry, volumetric error, and hardware settings in translating a CAD file to a physical object [74, 75]. To control the shape deviation, the corresponding allowances are approximately estimated before printing [76]. However, the FFF parts still required PPFTs to meet the market [17]. For instance, temperature variation during the layer-by-layer part fabrication procedure undesirably affects the printing quality [2].

Fig. 7

Appearance and accuracy of gradient lattice-based structures in as-printed FFF samples using PLA filament and Prusa i3 mk3s + (scale bar represents 1 cm)

Generally, there are two types of bonding in the FFF process: inter-layer and intra-layer (Fig. 8). The high thermal expansion in polymers can play an important role in the weak bonding among the layers during the build process, leading to staircase (stairstep) formation. As an inherent issue, the formation of staircases has a considerable negative impact on the surface quality of FFF components [18, 75].

Fig. 8

Bonding and stages of neck formation in the FFF process

Among literature reviews on the process parameters of FFF, Turner et al. in 2014 [77] summarized the process design and modeling of FFF. They reviewed the bonding of the raster, the model spread of the deposited raster, and the motor torque and power. Chohan et al. [21] have reported a literature review on pre- and PPFTs to improve surface characteristics of FFF parts. In 2018, Singh et al. gathered results from studies on the effect of pre- and post-processing procedures on FFF patterns to develop biomedical implants from the route of AM and investment casting (IC) [78]. Several AM materials and their mechanical performance have been reviewed by Popescu et al. (2018) [79]. In addition, deliberately structuring the polymeric surfaces using FFF 3D printing has become a popular choice for AM processes. Cuan-Urquizo et al. [17] reviewed the literature on the characterization and projection of the mechanical behavior of FFF products using analytical and computational approaches (2019).

4.1 Process parameters

In FFF, pre-processing includes instructions generated by a slicer software to gain data, slice the design file into layers of 3D pixels (voxel), model construction, optimization of the toolpath for the printing process, and material preparation. Parameters directly affecting the process are categorized based on operation, geometry, material, and machine-specific parameters [80]. Generally, the accuracy of the motion system limits the precision of the part. To minimize the issues due to the design and pre-production phase, each voxel must have accurate position information and print-process parameters in tool pathing. In the following, the main process parameters influencing the surface roughness of FFF polymeric surfaces are discussed.

4.1.1 Filament material

The growing interest in using polymers ranging from rubber-like materials to rigid plastics leads to new applications in vehicle parts, shoe soles, and biomedical applications [81]. While high surface quality and desired roughness are advantages for 3DP parts, other material properties and manufacturing features are usually considered in the selection of the AM method. Figure 9 illustrates a general cumulative performance score based on the comparative scoring of each parameter on a scale between 1 and 9. PLA and ABS are the most well-known feedstock among other materials for FFF. Since PLA filaments can provide better surface quality and biodegradable polymer derived from corn, it is considered more eco-friendly than petroleum-based ABS [82].

Fig. 9

Recommended polymers for the FFF process. Data extracted from references [43, 83]

In most cases, materials are already enhanced for optimum performance. Their parameters are challenging to alter later in the production stage, but choosing the appropriate material to improve surface roughness is crucial. This may prevent the usage of FFF prototypes in some cases where the surface should be smooth. An overview of the major polymers used in FFF is presented in Table 5.

Table 5 Typical FFF filament specifications. Data extracted from references [43, 83]

4.1.2 Print temperature

In FFF 3D printing, print temperature and cooling speed play an important role as it affects the quality of the 3D-printed object. The model temperature should be high enough to melt the outgoing feedstock before extrusion. However, if the temperature is too high, the filament will melt too rapidly, resulting in a porous and brittle object with poor surface quality [21]. Depending on the type of filament being used, the optimal printing temperature will vary. Typically, PLA is printed at 190–220 °C, while ABS is typically printed at 220–250 °C [35,36,37]. The ideal printing temperature should be referred to by the filament manufacturer to ensure the best results. A stable temperature at heaters is crucial in the fluidity of material, resulting in a smoother surface finish. Besides, the surface roughness can be decreased by increasing the model temperature because of the solidification delay. However, plastic adhesion with a base plate is problematic in very high or low printing temperatures [21]. As well as affecting the quality of the printed part, the temperature of the nozzle can also affect the speed of the printing process. In general, a higher temperature will lead to faster printing speeds, but it is crucial to find a balance between speed and quality.

4.1.3 Layer thickness

The height of each deposited stairstep is a notable parameter that can be controlled in the extrusion nozzle tip and/or shift in the Z-axis between consecutive slices accumulated on the bed. Adjusting the height of each stairstep can improve print quality and reduce the printing time. In general, a smaller stairstep height will produce a higher quality print, but the printing process will take longer [84, 85]. This parameter is the most significant challenge in obtaining a high surface finish in a cost-effective production time. The minimum feature size (smallest linewidth) is determined primarily by the nozzle diameter, which affects layer thickness nonlinearly, but geometry and build orientation also play an essential role [86]. Correctly setting up an optimum balance between layer thickness and printing time is highly influential in the pre-processing step. A smaller nozzle tip will generally result in lower layer thickness, better surface quality, and possibly decreased post-processing time. However, it increases the printing cost and time for the as-print parts, leading to more nozzle clocking and quality issues because of the pressure drop [21].

Since the quality of internal surfaces does not engage in the appearance of the part in most cases, consistent layer height has been seen as a waste of time [72]. Thus, it can be varied in different areas of the object according to the expected time-quality factor. For most FFF printers, the domain of layer thickness is variable in a certain range, i.e., there is a maximum and minimum value for changing the nozzle diameter. A few machines are limited to a single value, so it is required to set up other parameters to change the surface quality [21].

In several papers [84, 85], the layer thickness is reported as the most significant process parameter influencing surface roughness. However, this influence also depends on the other process parameters. Anitha et al. [87] established a set of experiments that showed that layer thickness significantly impacts the roughness of the FFF part compared to rod width and speed of deposition. Haque et al. [84] attempted to minimize the surface roughness of FFF build features using a numerical approach. After investigating various equations to control FFF parameters such as layer thickness, overlap distance, part orientation, and raster width, they reported that layer thickness has more influence on roughness than other parameters. They observed that increasing the overlap distance between two layers and part orientation leads to lower surface roughness. However, higher layer thickness and raster width increased surface roughness.

4.1.4 Infill density and air gap

The infill density defines the level of incorporated material inside the fabricated object. It might be varied from 0 to 100 percent according to the required balance between material consumption and mechanical properties [88]. Generally, a higher infill density leads to a heavier and stronger part, which increases the cost and the amount of material used in the printing process. Infill density and pattern are significant process parameters influencing surface quality [72]. Support structures must also be designed appropriately to support the geometry. Thus, they should be accurately chosen by considering the design and strength requirement, as well as the build time of the printed part. For instance, surface artifacts such as gaps and porosity have been observed even for 100% infill density under a scanning electron microscope (SEM) [89].

The distances or spaces between two adjacent rasters on the same layer are called the air gap or road gap [90]. Figure 10 depicts the air gap compared to other adjustable FFF process parameters. The default value taken for the air gap is zero, which means the end of the two nearest beads is in touch. There are two types of positive and negative gaps. The positive gap increases the gap to reduce the density and build time of structures, whereas the negative gap means overlapping two roads resulting in a long printing time and dense objects. While both positive and negative air gaps can enhance the surface finish, zero air-gap minimizes dimensional accuracy and part quality [21].

Fig. 10

Controllable parameters in FFF machine

4.1.5 Raster width and angle

The raster width, also called road or contour width and (tool) path width, refers to the width of the melted bead path, which is added to fill interior regions of the FFF-printed samples [21]. According to the filament material, it is regularly 1.2 to 1.5 times the nozzle tip size [80]. As seen in Fig. 10, the contour tool path width surrounding the object is defined as the contour width, slightly smaller than the nozzle diameter [21, 86]. The width of the most minor features in the XY-plane in FFF is about two times the path, much larger than the thickness of the layer in the Z-direction. To reach high mechanical performance, dimensional accuracy, and surface quality, the raster width should be minimized [86, 91]. It has been noticed that a wide contour width enhances geometrical precision and surface quality because heat evolved during extrusion can easily deform thin contours [92].

The raster direction compared to the X-axis of the build platform is known as the raster angle (Fig. 10). This parameter significantly affects the internal layer bonding and object appearance. The main approaches to raster angle are criss-cross (− 45°/ + 45°), cross (0°/90°), and 30/°60°. Because of the variation of CAD models and printing parameters, there is a loose correlation among the studied strategies. Sood et al. [91] used a bacterial foraging algorithm to show that the 0° raster angle is the best option for dimensional accuracy, and 45° results in the best surface appearance. Kumar et al. [93] stated − 45°/ + 45 as the best raster angle for the surface characteristics, including roughness.

4.1.6 Build orientation

The orientation (deposition) angle is a notable and highly flexible process parameter involving surface quality. It corresponds to the CAD model and coordinates machine system (Fig. 11). Wang et al. [94] considered the build orientation the most substantial PPP factor regarding dimensional accuracy. As a result of gravity and residual stresses, overhanging surfaces should be supported at less than 45° from the horizontal plane [19]. Several papers have studied the role of positioning the models in various X- and Y-directions on the surface finish and build times. For instance, Kattethota and Henderson [95] reported that the orientation angle of 0° yielded the best surface finish. Moreover, 0° and 90° were reported as the optimum build angles for balancing the build times, cost, and surface finish. By contrast, orientation angles between 40 and 60° were neither cost-effective nor quality-enhancing due to the maximum support material required for tilting the model [21].

Fig. 11

Orientations commonly used for FFF parts

Since different surface angles result in various surface roughness, the test part should comprise features considering different surface angles. In 1997, Reeves and Cobb [96] introduced a benchmark model called twisted pillar. It can consist of 18 or 31 square blocks, depending on the intervals (step), twisted 5° or 3° compared to the previous square (see Fig. 6a). Durgun and Ertan [97] also confirmed the close relationship between build direction and surface roughness.

Among the defects associated with different build orientations are warping, layer delamination, deformed overhangs, and poor surface quality [98, 99]. In terms of surface roughness, Buj-Corral et al. [99] reported both simulated and experimental amplitude roughness values to rise with build orientation angle, due to the stair-stepping effect. As reviewed by Jiang et al. [98], a change in print orientation also affects support generation, which ultimately affects the surface roughness after support removal. The choice of support structures can have a significant impact on the surface roughness of FFF prints, thereby influencing the post-processing process. Furthermore, the orientation of the part impacts the support contact area, the build time, and the cost of the fabricated part.

4.1.7 Adaptive slicing

Adaptive slicing is termed as a protocol to slice various zones of the part into different thicknesses during building [78]. It contains balancing techniques needed to reach the optimum printing time versus surface finish. Generally, the CAD model is divided into polygons or closed curves by either the CAD software or a slicer before transferring the mesh model (STL file) to the printer. This process is called slicing, and the distance between two sequential horizontal planes is known as a slice [100]. The slicing process and the tessellation of the CAD file are expected to be the significant parameters involved in creating rough surfaces in the procedures of layer fabrication. The containment problem causes the original CAD model to deviate from the designed form when slicing a tessellated CAD model. Aside from the containment issue, the layer deposition causes a problem known as staircase effects [101].

Several researchers [17, 102] have studied several types of stepwise refinement, adaptive slicing, and identifying nonuniform fillet radius at different areas of printed objects. These methods consider automatically slicing algorithms, generating variable tool paths (.gcode), reducing build time, and minimizing surface roughness through varying heights depending on geometry [103]. These specific parameters rely entirely on the shape and dimensions of the designed part [21].

The major categories of slicing methods available for FFF printing are flat-layer, non-planar, and mixed-layer adaptive slicing, respectively. Zhao and Guo [104] listed the most important research studies on non-planar and mixed-layer adaptive slicing. They suggested method planning of mixed-layer adaptive slicing, which discusses the strategy for the process planning of more straightforward adaptive slicing approaches. Table 6 provides an overview of these slicing methods.

Table 6 Adaptive slicing classification. Extracted from references [103, 104]

4.2 Surface roughness studies and discussion

The literature on process parameter optimization is classified according to the resulting properties. It can be based on the surface finish and smoothness, dimensional accuracy, build time, material behavior, dynamic and static (tribo-) mechanical/thermal behaviors, and manufacturing cost. Studies have centered on finding the best combinations of geometry and operation-specific characteristics. For instance, Durgun and Ertan [97] reported that surface roughness significantly impacts the flexural strength of ABS parts manufactured with infill 100% at different orientations and angles. Among the factors responsible for PLA and ABS surface roughness, the researchers identified layer thickness [105,106,107], build orientation [107], printing speed [105, 106], nozzle diameter [105, 106], and temperature [106] as the most critical parameters. Table 7 summarizes some of the significant literature on the essential parameters in the process optimization of FFF on the roughness range. Optimizing these parameters is one of the highly critical tasks for acquiring the desired surface quality and improving superior mechanical properties and material response.

Table 7 Overview of studies on surface roughness in FFF (see text^* below the table for abbreviations)

Table 7 indicates that layer thickness can be considered the most significant FFF parameter that affects surface roughness and surface finish. Based on the results, surface roughness was reported primarily by R_a, S_a, R_z, and R_q. Furthermore, the range of roughness can be significantly different depending on the processing conditions from sub-micrometers to 33.65 µm for R_a in the studied works. Although the reported roughness varies tremendously, it is generally between 0.1 and 1 times the layer thickness, depending on the materials and measurement geometry. Also, build orientation is crucial in determining print quality, as horizontal and vertical surfaces differ. Slicing settings, object orientation, and considerations regarding the design for additive manufacturing (DfAM) guidelines could enhance roughness and resolution while decreasing the number of PPFTs required [118]. These results did not consider more complex features, such as ridges, holes, slots, and posts, coupled with motion and material flow dynamics.

5 Selective laser sintering (SLS)

Parts produced by SLS are expected to have a high surface quality because of the precise nature of the laser, resulting in extensive usage in meeting functional needs. On the other hand, SLS-produced components have generally a greater surface roughness than other polymer AM techniques [119]. Several functional properties are affected by surface roughness, including frictional properties, heat transfer, and fatigue resistance in polymers, as well as the possibility of powder becoming loose, e.g., as a medical implant in the human body [120]. During SLS, the build platform is heated uniformly to exactly below the material melting point; this is a highly effective method to boost the build rate. However, it can cause unwanted “caking” of powder on the outer surface of the part. A preheating lamp to maintain the temperature under the melting temperature reduces thermal stresses, leading to part distortion, shrinkage, and lower dimensional accuracy at the surface. Due to entrapped air, many grainy features, voids, and porosities can be formed inside the packed powder (Fig. 12). Therefore, porosity is an inherent defect in SLS objects that can appear on the surface [3, 121].

Fig. 12

2D laser scanning image of a part made of PA6, showing the typical surface texture of solid part samples printed with an EOS P500 FDR system

SLS typically prints features as fine as 0.1–0.5 mm, making them an excellent choice for printing intricate latticework with thin walls and beams. While SLS parts do not need support due to a powder bed, temperature gradients during printing may deform the part and create very thin surface issues. In addition, because of thicker layers (90–150 µm) in SLS, the technique is more sensitive to the staircase effect. This effect in an SLS object is most prevalent on semi-horizontal surfaces [3]. In a similar method, SLM, since the melt pool is typically more extensive than the laser spot, the scan contour tracks are naturally shifted inwards to account for this issue and, consequently, rougher surfaces [122]. PPFTs usually improve the quality of large surfaces in these cases, but it would be more demanding for more complex and minor features.

There is a direct correlation between the amount of porosity in a part and material properties, such as the shape and size distribution of the powder and part processing conditions. Compared to semi-crystalline thermoplastics, amorphous thermoplastics produce more porous parts, which can either be an advantage or a disadvantage depending on the desired property of the piece. The fundamental problems with sintering are porosity and shrinkage in the parts. Still, they can be overcome with an optimal packing density (for porosity) and a careful choice of the sintering parameters [123, 124].

Low porosity is required if the appearance and mechanical properties are essential. The size distribution, reduced porosity, and enhanced surface finish affect powder flowability and packing density. Using infrared lamps or ambient heating helps prevent nonuniform shrinkage by keeping the polymer above the glass transition temperature, allowing the shrinkage process to be controlled. This requires slow cooling after the build is complete and must be considered when calculating the processing time for each part. On the other hand, semi-crystalline thermoplastics experience a volume reduction during cooling due to crystallization, making amorphous thermoplastics the better choice [123, 124].

Poor surface quality is a common complaint when working with powdered raw materials. The conventional roughness range (referred to by R_a) of PBF is 5–25 µm. As well as the printing parameters, the quality of the surface can also be affected by the 3DP machine itself. Depending on the SLS machine, R_a can range from 10 to 20 µm with a peak-to-valley distance of up to 0.2 mm [3]. Sachdeva et al. reported a range of 8–12 µm for R_a [124]. Mechanical properties, particularly fatigue, can be affected by different surface roughness.

The AM powder-based procedure comprises semi-molten grains adhering to the outer surface, with occasional sharp grooves in between, prone to crack initiation [3]. Figure 13 depicts the three states influencing the surface quality in PBF.

Fig. 13

Typical PBF surface configurations that impact surface quality

The incomplete composition of powder materials may cause “orange peel” surfaces (Fig. 14). Incorrect powder reuse or non-homogeneous mixing is the main reason for this surface issue in the SLS technique [125, 126].

Fig. 14

Orange peel surface texture on PA12 (2200) SLS part

5.1 Process parameters

Several factors influence the performance of SLS systems. Precision and resolution are limited by the combination of powder particle size, layer thickness, and laser spot diameter. The selective melting of powder is done using two main tracks, including the contour and hatching tracks. In contour tracking, the outer layer of the required profile is melted, while the hatching track is used to melt the internal area bounded by the contour track [122].

The SLS as-printed surface roughness values differ considerably based on the preparation method, the equipment utilized, and the position of the sample surface concerning the layer accumulation [3]. The SLS parameters of the process (see Fig. 15) have been widely investigated in engineering fields [32, 127]. They can be classified into three major categories: laser, material, and chamber. The average powder size, layer thickness, and surface orientation all influence surface roughness. The physics of melting and solidification are also important. It is influenced by scan speed, laser power, hatch spacing (distance) or laser scan spacing, material viscosity, surface tension, and thermal boundary conditions like bed temperature. Research has shown that materials that have been degraded for enough time adversely affect the surface quality and increase the viscosity of melted powder [119]. Several factors determine the accuracy of the part and the minimum feature size, including powder sizes, laser spot sizes, feature orientation, aspect ratio, ability to control the melted region, and the resulting solidified geometry of the scanned areas. A thin layer of unfused powder adheres to the part surface due to heat dispersion into the surrounding powder, which must be eliminated during post-processing to achieve the best surface finish [32]. Overall, the main SLS printing factors influencing surface roughness can be classified as design, laser, material, and build chamber.

Fig. 15

A schematic of the SLS 3D printing process

Particle size distribution and particle shape can impact the quality and strength of the manufactured objects. These procedures are continued until the component has been produced entirely. SLS processes encounter many difficulties due to the complexity of the thermal interactions involved, which calls for technology-appropriate design and process planning. Relative stresses, microstructural formation, and surface quality are a few difficulties resulting in part deformation or failure [19]. To avoid thermal deformation, the component stays in the powder bed throughout the slow cooling phase [9, 128].

5.1.1 Design

The position and orientation of the SLS chamber are crucial when using 3D nesting during the design stage. As a result of the lower temperature of the platform, the bottom section of the construction volume is not as warm as the upper area. Due to the high temperature in the top half of the build volume, this area is more likely to experience powder adhesion. Process modeling for powder sintering might aid in predicting essential aspects that must be addressed during the early design phase [16].

The size of geometrical features has been discovered to be a significant determinant for the volume of partly sintered powders adhering to the component surface owing to heat intensities, severely impacting manufacturing precision. SLS design criteria should consider massive hot masses, a well-known phenomenon. For this purpose, Minetola et al. [16] suggested the SLS modulus, a metric developed to detect crucial heat concentrations in the chamber that can affect the dimensional accuracy of the produced part. Generally, an approximate part accuracy of ± 200 µm for small dimensions and ± 0.1–1% for large dimensions, as well as a minimum feature size of 0.5–1 mm, should be considered in the design of SLS parts [32, 121]. In Fig. 16, SLS shows sensitivity to feature size where there is a distinct area of separation using a gradient lattice-based design.

Fig. 16

Accuracy of gradient lattice-based structures in as-printed SLS samples

A cross-section of SLS parts shows how local and global features in the design influence part quality. The microscale portion comprises fused polymer powder particles ranging from 10 to 100 µm in diameter. The laser heats the particles, fusing them together before solidification. However, incomplete fusion can leave pores within the part, reducing strength, durability, and surface inhomogeneity. SLS parts feature a coarse surface texture at the mesoscale due to particle size polydispersity and unfused powder adhesion based on heat diffusion into the surrounding powder. Non-vertical features are stepped based on layer thickness and orientation [28, 129].

The laser spot size and heat dispersion into the powder limit the minimum feature size in the design step. Freeform geometries, interior cavities (with holes for loose powders), and delicate lattice structures are all possible with SLS. Living hinges, latches, and interlocking parts can also be developed. Then, the interior features must be constructed so that loose powder may evacuate when the component is formed.

Because no support structures are required, and cooling warpage is minimal, unsupported walls and horizontal bridges are more flexible than the MEX system. Nevertheless, design and tolerancing must consider temperature gradients during printing shrinkage. Otherwise, very thin features can deform due to temperature variations in the print environment [130]. SLS cannot attain the same surface quality as other polymer AM, such as photopolymerization (SLA). In general, PBF AM processes, including SLS, slice in the Z-direction with constant or adaptive layer thickness, whereas the slicing method in filament-based AM is strictly a limiting factor due to lower dimensional precision because of the staircase effect, the required support structures for overhangs, and poor performance because of the anisotropic design caused by the slicing method [104].

5.1.2 Materials

SLS is versatile as it can process many types of polymers available in powder form, either thermoset or thermoplastic. The powder is vital for the packing density, which is influenced by other parameters, including particle size distribution, particle shape, and spreading system. Bimodal powder distribution can improve the packing density [131, 132], which is given by other parameters, such as particle shape, size width, distribution exponent, and packing mode [133]. Particle shape can affect the powder size distribution as the finer powder can facilitate reaching higher packing densities and improve flow and spreadability. The spreading system is also responsible for optimizing the packing density. A wrong method for the powder on the bed can cause flaws in the packing process and entail artifacts or holes in the part once sintered [134, 135]. The most common materials used in SLS are semi-crystalline polymers of PA12 and PA11 due to their well-defined melting temperature and melt-freeze thermal hysteresis [32]. Compared with amorphous thermoplastics, these polymers result in more favorable processing conditions and improved powder recyclability. Table 8 lists the common polymers used in SLS and their applications.

Table 8 SLS AM polymer powders classified by structure and performance [32, 129, 136]

SLS can fabricate complex internal cavities with 3D lattice structures, but features for draining unfused powder must be included. Powder parameters, such as diameter, morphology, size distribution, crystallinity point, flowability, and melting point, also significantly affect the SLS method and the part quality. As a result, there is an increased demand for powders capable of performing SLS [137].

Powders were described as pre-processing parameters but can also be included as printing parameters. Since lasers and powder are connected, it is necessary to study and determine the power based on the powder that will be melted before picking up a laser. Polymers must be considered in terms of their heating properties to be suitable for SLS. Furthermore, the optical characteristics of powders determine the wavelength of light the materials absorb [138]. Semi-crystalline polymers are preferred because of their processability (e.g., PA12). There is a clear temperature range for these polymers, ranging from the glass transition temperature to the melting temperature, with a hysteresis window between melting and re-crystallization. The powder type, including recycled and virgin, can affect the surface quality. In general, recycled powders increase the instability of the feedstock properties, resulting in a higher surface roughness [119]. High flowability is a critical property of materials in the SLS technique. Powders must be highly flowable to minimize highly jagged and microscopic particles with strong inter-particle forces that lead to agglomeration and surface issues. Goodridge et al. [139] reported 45 and 90 µm as the most favorable range for powder size in SLS 3DP.

In PBF processes, loose powder is typically removed manually. However, it can be more aggressively removed by shockwave cleaning or dry-ice blasting [140], which alters the surface roughness significantly. Low-temperature heat treatment would be beneficial to improve surface quality. It would relieve imposed stress during laser sintering and improve mechanical properties.

Surface modification of powders can increase laser light absorption, flow, and spreadability. Feedstock enhancement could be used to investigate various powder blends with various or multi-modal particle size distributions (PSDs) and create particle-based models for forecasting the deposition parameters (e.g., spreading speed) and the powder bed packing (e.g., inter-particle friction) and [69]. The thermal history of the polymers is also essential as the molecular weight of the polymer can change when it goes through heat cycles, influencing its melting and solidification behavior. The working conditions significantly impact this powder, and the sintered cake cannot be recycled near the component. Fresh powder must be mixed with the remaining powder from the previous 3D printing procedure to closely match the specifications [3].

This technology works with a higher layer thickness (90–150 µm), making it more susceptible to the staircase effect. Because of the sintered grains on the surface, cleaning is difficult [3]. Powder bed systems are exposed to a frictional force with the bed and an inter-particle force that restrains their motion. For this reason, chemical additives can be added to the powder to have a higher spreadability [141, 142]. There is a variation in the laser beam over the whole build plate. The angle of incidence can significantly impact the outcome of the part [143]. Therefore, new machines with two laser beams are currently available. However, having two different power sources leads to a slightly different sintering process. Furthermore, portions of the stage where the two lasers work almost simultaneously cause a point or line of poor or no sintering.

5.1.3 Laser

The mechanical strength and density of SLS objects depend heavily on the laser beam energy density. As the laser rasterizes over the powder bed, its movement strategy and direction are essential. Generally, higher laser beam energy density results in higher densities and greater mechanical strength of the final part, up to a certain point. If the powder particles are over-melted beyond this threshold, the properties of the part may begin to deteriorate [144]. The most often used SLS printers use diode lasers. The laser power (P), the laser scan speed (v), the focus diameter, and the hatch distance (scan line distance) all affect energy density (ED) [145]. The hatch distance includes two parameters: layer thickness (t) and scan space (s). Depending on the laser power (P) and beam movement on the materials, the powder bed obtains the proper heat for fusion. Accordingly, ED is the laser beam input energy per unit area (J/mm²), and it can be calculated using Eq. (17) [146]:

$$ED=\text{P/(}v \, \times \, t \times s\text{)}$$

(17)

In this equation, if the hatch distance (in mm) is more than the effective laser diameter (in mm), the effective laser spot size would be considered [147].

Sintering requires a certain amount of laser power due to laser-powder interaction, which depends on the material and layer thickness. Because the laser beam diameter and material formulation affect absorption, they can raise or lower the sintering/melting temperature. It affects the laser wavelength required for sintering. In addition, pulse durations (for pulsed lasers), geometries, and frequencies can impact surface quality. Shorter pulse durations can lead to more precise melting and less heat buildup, resulting in smoother surfaces with less porosity. There is usually an offset between the laser and the design border because of the light spot that must be considered. During the sintering process, a low-power laser is used to smear the particles that can remain attached along the contour [134, 148].

The dimensional accuracy of the printed product causes excellent process accuracy in SLS. Shrinkage is one of the critical factors influencing accuracy, and the quantity of shrinkage seen hugely depends on materials and laser sources. Laser power and scan length have a substantial impact on X-direction shrinkage, whereas beam speed and laser power have a significant effect on Y-direction shrinkage. In contrast, bed temperature, beam speed, and hatch spacing considerably affect shrinkage in the Z-direction [130].

5.1.4 Build chamber

A typical mainstream SLS system has a medium build size of 340 × 340 × 600 (mm). When a product exceeds the build platform volume, it must be split into multiple pieces, designing separation planes, and considering subsequent assembly procedures affecting the surface quality [149]. The part size is constrained on the upper end, restricted by the build volume of the printer and the ability of the optics system to scan the entire area. This volume typically ranges between 10 and 1000 L (0.01–1 m³). However, it is possible to reach a smaller size in some designs as 0.005 m³. Due to the minimal residual stresses, the entire build volume may be utilized by stacking pieces on top of one another without the requirement for support to connect the parts to the build platform. The size of small parts is constrained by the minimal feature size, the related precision and resolution, and the assembly of the components [149]. The powder particle and laser spot size restrict the precision of less prominent features. All these variables, coupled with the powder material and thermal boundary conditions, affect solidification kinetics and, hence, dimensional accuracy. In terms of depth and length, the larger-the-better rule applies to the part size to obtain the desired object, whereas roughness and geometric precision are determined by the nominal-the-better rule [127].

Several build chamber parameters, such as layer thickness, roller speed, heating–cooling rates, build size, and powder and feed bed temperatures, impact the SLS process. There are several gradients of temperature in the build chamber to be considered [150]. Besides the Z-axis, where the part suffers a decrease in temperature for the higher layers while the stage moves downward, the platform has some increased temperature along the diagonals, and the center can be caused by the presence of heaters or mechanical components underneath. The blade shape used for powder application also affects the surface quality. Beitz et al. [69] demonstrated that a flat bottom form was more advantageous than sharp or slightly rounded edges. Due to the larger horizontal contact zone between the powder bed and blade, the powder material is compressed evenly, resulting in a more dense and consistent powder layer.

5.2 Surface roughness studies and discussion

SLS allows the production of components with high levels of complexity, almost no geometrical constraints, and no need for a tool or a mold. The resulting parts have an enhanced surface roughness due to optimizing the processing settings, the build orientation, and the powder characteristics [66]. Although roughness cannot be removed entirely, several researchers have attempted to decrease the deficiency by modifying the printing process parameters and operating at different parameter levels.

In many cases, PPFTs primarily affect the surface roughness obtained through SLS. Nevertheless, several studies have attempted to optimize the parameters of the process. Sachdeva et al. [124] investigated and used response surface methodology (RSM) to optimize the SLS process parameters for roughness (R_a, R_z, and R_q). Beitz et al. [66] reported that the CLSM measurement yielded roughness ranges of R_a: ~ 24–31 µm and R_z: ~ 157–181 µm, while the XMT method produced R_a: ~ 22–27 µm and R_z: ~ 128–148 µm for PA12 powder. Most research has focused on PA12 as the feedstock material to determine optimal parameters for processing based on roughness conditions. However, limited research [151, 152] has been conducted on other materials. As listed in Table 9, different process parameters were considered in the method.

Table 9 Overview of studies on surface roughness in SLS (see text^* below the table for abbreviations)

In agreement with the reviewed papers in Table 9, the average particle size, layer thickness, and surface orientation can substantially influence surface roughness among a wide range of variables. Material viscosity, surface tension, and thermal boundary conditions such as bed temperature play a role in melting and solidification. Heat dispersion into the surrounding powder generates a thin layer of unfused powder to cling to the component surface to achieve the most delicate surface quality. The powder size and the laser spot restrict the part accuracy of less prominent features. These parameters, coupled with the powder material and temperature boundary conditions in the system, influence solidification dynamics, which can also affect dimensional accuracy. Shrinkage during cooling can cause additional losses in part accuracy for bigger features. Concerning laser power, a low energy density can result in loose powder particles that are unable to melt and fuse together sufficiently, resulting in parts that are weak and porous. In contrast, if the energy density is too high, the powder may melt and fuse excessively, resulting in rough surfaces, distortions, and even cracks in the finished parts [119, 127].

While the SLS parts generally show higher surface roughness (10 µm < R_a < 20 µm) compared to FFF (1 µm < R_a < 10 µm), SLS offers many benefits compared to other polymer-based AM techniques [158]. To begin with, no support structures or foundations are needed during the SLS process, since the unused powders support the components. Therefore, there is no roughness caused by support removal. Even though SLS uses unused powders to support the components being printed, this is usually insufficient to prevent all types of deformations. As a result, post-processing steps such as stress relieving and annealing may be required to reduce residual stresses and deformations [119, 157].

Additives such as initiators, binders, and catalysts are not required, which implies that the components are more likely to be utilized in the medical field since additives may cause toxicity. Moreover, although SLS resolution is not as excellent as other AM methods, such as SLA, the mechanical properties of SLS components are usually superior, making the surface of these components more stable over time. The resolution issue may be addressed by improving the laser system. In theory, SLS technology is not material-restricted, and most powders may be utilized in SLS, provided that the laser wavelength and power meet the sintering material requirements. However, this kind of laser/point-based technology (e.g., SLA or SLS) has a common drawback of low processing speed because of the methodology of “point → line → face (slice) → body” [157].

6 Vat photopolymerization (VPP)

Vat photopolymerization produces parts with a resolution close to 100 nm [159], leading to superior surface quality. While the design choice can bring some advantages and some disadvantages in terms of texture, the process selection is still critical. In a top-down VPP setup, the build plate dips in the resin to create a new layer, and generally, a recoating system makes the printed surface smoother. Despite this mechanism and a consequent reduction of the necessary supports, the recoating procedure may cause some convex undulations on the resin surface and on the printed layer afterward. It is due to tensions, especially with highly viscous resins or really thin layers. Thus, surface bubbles can remain trapped inside the part but also can groove the surface [160, 161]. Using scraper blades can attenuate these issues by adjusting the layer thickness of the new resin on top of the part [162].

Alternatively, the bottom-up approach, where the light source is placed below the vat, and the build platform move stepwise upwards, resulting in accurate details but a more corrugated surface [163]. Moreover, confined layers due to the space in between the transparent window at the bottom of the vat and the build plate allow the system to achieve better Z resolution. However, the layer, once cured, attaches to the glass window, and the detachment step might cause it to lose material and have defects. To reduce detachment forces during this phase, the vat is usually coated with a polydimethylsiloxane (PDMS) layer or an anti-adhesive membrane made of PTFE [161, 164] with an additional protective layer of fluorinated ethylene propylene (FEP) to have good anti-sticking effect and durability [165]. The latter seems more resistant to degradation and all the problems this may cause to the quality of microfeatures [166] but has a more complex tightening method that can lead to deformed features and light refraction when applied incorrectly [161]. These approaches constitute a layer-wise method and generate staircases along vertical surfaces.

VPP defects generally occur due to non-optimized print parameters, insufficient supports, improper model generation, and contamination in the build platform and resin material. For each layer to be cured entirely through its thickness and uniformly recoated, the print parameters, such as scan speed, power source, and recoating process, must be optimized [167]. Another common problem with VPP resins is their tendency to turn yellow quickly. This is mainly because of overexposure to ultraviolet light, which also causes clear prints to appear matte yellow (Fig. 17).

Fig. 17

Appearance and accuracy of gradient lattice-based structures in as-printed SLA samples using PrimaCreator Value resin. After 12 h, a yellowish color began to appear

6.1 Process parameters

6.1.1 Design

The design role in PPP regards creating the model system on a CAD system and handling the photosensitive resin independently of the technology used. CAD format is tessellated into the STL data, and this is the phase where decisions about surface modeling are made. For instance, essentially flat and/or thick sections are prone to warp because of shrinkage. The STL file approximates three-dimensional surfaces with triangular facets, which may result in errors in dimension, form, and surface [168]. Incorrect conversion of a solid model into an STL file can cause missing or distorted features. In stereolithographic technology, the STL file is sliced into several horizontal layers and then commonly saved in CLI file format [169]. Similar to the FFF method, many parameters such as print direction, layer thickness, the inclination of the part, hatch spacing, fill spacing, hatch over-cure, border over-cure, and fill cure depth are essential to improve surface quality without resorting to a large number of facets and a long build time [168, 170]. Considering the presence of supports is crucial because if the support density is insufficient, the part can also shift or detach entirely from inadequate supports [171]. The supports cause unavoidable staircase effects on bent or sloped surfaces, but they can be reduced by adjusting the printing parameters mentioned above [172]. In this case, the adaptive slicing method, as a typical practice in FFF, is acquiring more attention to improve the surface quality and surface roughness in VPP methods [96, 173].

6.1.2 Materials

The most popular materials for VPP are photosensitive resins, usually containing acrylates, methacrylates, vinyl, and epoxies monomers/oligomers. Acrylates and methacrylates monomers/oligomers are a subcategory of the vinyl group, and the presence of the carboxylic group (-COOH) in the vinyl position confers them high photo speed as they react quickly when exposed to UV radiation. Moreover, they behave differently in terms of radical formation. Acrylates tend to form secondary radical ends, whereas methacrylates tend to form tertiary radical ends. This difference in radical end formation makes methacrylates more stable and less reactive than acrylates. However, both of them undergo significant shrinkage with associated stress that might result in warping or curling [174, 175]. Moreover, they suffer from the inhibiting influence of oxygen, and this facilitates the formation of a sticky surface appearance due to oligomer formation [176]. They also have low viscosity and critical energy, increased light sensitivity, relatively high dependence on humidity and temperature fluctuations, and controllable mechanical properties [177].

Vinyl monomers appear in both radical and cationic polymerizations, and a mix with acrylates or epoxies enhances their respective characteristics. They provide relatively low thermal resistance and low glass transition temperature, and they tend to exhibit brittleness, low elongation, toughness, and impact resistance [178]. Epoxy monomers instead have an oxirane functional group, which is a three-member ring formed between oxygen and two carbon atoms, and when they react, these rings open, resulting in vacancies for other chemical bonds. This opening is known to have an influence on the volume change because the bonds remain the same and, as a result, epoxy resins typically present much smaller shrinkage and much less tendency to warp and curl. Furthermore, the products have high structural stability, higher mechanical performance, insensitivity to oxygen, and lower shrinkage stresses compared to radical polymerization of (meth)acrylates and vinyl monomers [175, 179]. All these monomers can be also added as additives with the function of chain transfer agents, specifically addition-fragmentation chain transfer (AFCT) agents, in a poly-functional way, with the intent of having, for instance, lower shrinkage stress, higher cross-linking density, or tougher polymers [180, 181]. Besides monomers/oligomers and additives, the photocurable resin generally presents reactive diluents, UV stabilizers/blockers, and photoinitiators.

Also, adding particles as reinforcement in a resin can result in reduced or absent curing or in an accumulation of the particles, leading to a nonuniformity or degradation of the support. Studies about the size of the particles [182] and their interaction [183] were found essential to obtain the best outcome. Therefore, in VPP, the cure kinetics of the polymerization process related to the resin viscosity, light intensity, chemical functionality, illumination time, and the additives in the formulations play a crucial role in determining the final surface finishing and appearance of the prints [179]. In fact, the choice of light absorbers, the photoinitiator, and the monomers and oligomers can reduce the staircase effects, improve the resolution of printed objects, and produce optically more transparent layers and surfaces. Kowsari et al. [184] evaluated the influences of polymer formulation on the printing resolution and surface quality. In particular, by trying different formulations of (meth)acrylates-based monomers and oligomers with some of the most common photoinitiators, they found that dimethacrylate-based resins can improve the surface finishing by reducing the staircase effects and removing jagged edges. Moreover, enhancing the expected reflectance at the same wavelength is possible with different photopolymer formulations [177]. The selection of photocurable resins is usually according to the properties needed, such as quality of finish [185], durability [186], flexibility [187], transparency [188], bio-compatibility [185, 187], and cost [189]. They need to be stored in dark rooms to avoid the photopolymerization process initiates.

Moreover, when the resin is poured into the vat, it may contain air bubbles that reduce the achievable resolution and cause surface artifacts in the final object. Therefore, removing all the impurities and air from the photopolymer is necessary. It is possible to do so by shaking the resin manually or with the help of some machines, such as roller grills or shaking machines [161]. When the resin is not fully polymeric but with a mole percentage of other compounds, another step that is part of the pre-processing is the mixing process, which is requested to be as uniform as possible.

Polymers are not the only materials that can be used in photopolymerization processes. Diptanshu et al. [182] assessed how introducing fine ceramic powder can improve the density and reduce the porosity of the prints. The interaction between photopolymer and photons greatly influences the surface quality of the part. As well as laser power and uncontrolled photon flux, a nonuniform surface relief during the material solidification generates features much larger than the voxel size. Ambient factors, such as vibrations and jittering of the laser and its scattering while impinging the surface, may cause voxel displacements and fluctuations of printed voxels, inducing a weaker photon flux and defects in the layer so formed [184, 190]. During the printing process, contaminants, such as partially cured regions or external particles, can cause voids and deficiencies in the build since the resin cannot be recoated evenly. A damaged resin tank or dirty optics may also result in improperly cured regions, resulting in internal voids or inclusions. Last, voids within the resin caused by trapped air or not uniformly recoating the next layer will result in voids in the printed part [177]. Photopolymerization can also be affected by oxygen inhibition due to a different air in the room (with an inert gas), high-intensity irradiance lamps, or physical barriers [191, 192].

6.1.3 Printer setup

The setup of the printer and the influence of polymer mass and viscosity can also affect the surface finish, which is divided into two major categories:

• The bottom-up approach: the material is cured through a window, and a membrane of PTFE is placed in the bottom of the vat with a light source. In this setup, the build plate is raised every time to let the new resin occupy the volume underneath, and a “peel” step is necessary to detach the cured resin from the bottom of the vat. The “peel” step is time-consuming because the resin needs extra time to recover the initial state before starting the new layer [193].

• The top-downtop-down approach: a light source above the vat cures the material, and the build plate is submerged. Instead of “peel” steps, this setup employs continuous light exposure to cure the resin. It enables the achievement of high resolutions and printing speeds for this approach. The surface is traversed with a scraper to recoat and minimize eventual surface tensions [162] or can provide a dynamic characterization of the shape of the surface of the resin and adjust the light intensity accordingly [194].

Regarding the exposure strategies, in the bottom-up approach, there are no micro-fluctuations or contamination of the resin during the process since the bottom of the vat flattens each layer, but the detaching step might cause a corrugated surface with accurate details in the surface of the products [163]. However, in the top-down approach, there is not as much stress on the printed part during the printing process as in the bottom-up approach since the subsequent layer is not being sheared off after each layer is cured. Nonetheless, the overall distortions of the surface due to the motion of the stage normally require a recoating procedure that still can cause undulations on the printed layer. Moreover, longer printing times and slightly better resolution and quality of the printing drive the choice for the bottom-up approach. Other setups are related to different components that make the technology unique and solve some of the drawbacks. For instance, an oxygen-permeable window in CLIP technology solves the peeling step issue by preventing the resin from attaching to the window. At the same time, it controls the curing of the resin letting it have sufficient time to flow underneath the build plate and completely homogeneously the curing of the subsequent layer. In this way, the platform can move almost continuously upward [191, 192].

DPP technology uses an LCD unit to project the sliced 2D images, which has great potential resolution/cost-wise but has overall limited optical efficiency and lower resolution. In addition, working with an electric field that blocks the passage of light has a low switching speed (within 20 ms), and this may cause a few liquid crystals to remain trapped, resulting in weak light leakage and lower resolution [169, 179]. Hot lithography has the advantage of having a heating element able to control the viscosity of the resin and the temperature of the process. A higher temperature increases the reactivity of the monomers and the polymerization rate and the efficiency of the process [195].

Two-photon photopolymerization employs a femtosecond compact laser beam as a light source that absorbs two photons simultaneously. The photons are absorbed by the photoinitiators at the same frequency (degenerate process) in the near-infrared range (NIR). To trigger the third-order nonlinear two-photon absorption process, light sources with very high photon density are required. Since the light intensity in the laser beam respects a Gaussian distribution, in TPP, the square value is considered, and the region where polymerization occurs is lower than in normal conditions. In this way, a pulsed femtosecond laser with an acoustic-optical modulator (AOM) to disperse the beam into zero- and first-order diffractions or two light beams (stimulated emission depletion or STED) are the most preferred setups, because they can modulate the intensity and better control over the cross-linking process. Therefore, the resolution of the process and the connected feature size of the voxel improve until reaching details of 100 nm [32]. The concept is described in more detail in a study by Malinauskas et al. [258].

6.1.4 Curing system

In most cases of VPP, the substance is sensitive to light (but not necessarily), i.e., it will absorb the energy of the photons [177]. This energy induces either phase transitions (evaporation, melting, plasma) or chemical reaction in the material. As many materials can absorb light energy, VPP is a very versatile method for selection. However, there are significant variations in the specifications of the main VPP branches, such as the light source (UV or laser), feature size (nm to m), applicable materials, sensitivity to processing parameters, and the possibility of multi-material printing and commercialization capability.

In most VPP technologies, the object is created by irradiation of the photosensitive resin with a UV or laser beam source that enables the cross-linking of the polymer chains. It is scattered and absorbed when UV radiation or a laser beam hits the surface. This creates a threshold for the penetration of the light in the resin and, therefore, a cure depth for the photopolymerization process. The cure depth is given by Eq. (18) [169]

$${C}_{d}={D}_{p} \mathrm{ln}\left[{E}_{max}/{E}_{c}\right]$$

(18)

where the maximum exposure energy is represented by E_max, the critical energy required from the resin to begin the reaction is denoted by E_c, and D_p is the depth of penetration defined by

$${D}_{p}=1/\left(2.3 \varepsilon \left[I\right]\right)$$

(19)

where [I] is the concentration of the photoinitiator and ε is the molar extinction coefficient.

The energy dose considerably affects surface roughness and dimensional accuracy through the curing process [196]. In SLA, the projection system is the key factor that can sensibly modify the appearance of the part. A galvanometer scanning technology provides the steered beam through a movable mirror system to the resin surface and determines the precision grade. The exposure energy at the surface is determined by the steered beam as follows:

$$E=\sqrt{2/\pi }(P/{W}_{o}{V}_{s})$$

(20)

where P denotes the laser power, W₀ is the beam radius, and V_s represents the scan speed [169]. In line with the equations above, it is possible to see how much the entire process can undergo the projection system.

The power, velocity, and hatching of the laser spot determine the depth and size of the feature. Independently from the build orientation, there is a low anisotropy set by the cure depth according to the layer thickness. In fact, the layer thickness is lower than the cure depth, which means that an additional beam dose passes through to the previous layer to facilitate the adhesion and the bonds between subsequent layers [161, 197]. Furthermore, the width of cured resin (L_w), as determined by the center of the laser beam, influences the resolution and dimensional accuracy. When the laser beam passes over the same points while hatching the layer, some points might be over-cure or under-cure. For a successful solidification, it is essential to define a hatch spacing related to the width of the cured resin [169]. Nonuniform laser power density is another important parameter that can be caused by either acceleration or deceleration of the galvanometer at the end of the hatching and along the boundaries. This can be overcome by either modulating the laser power as a function of scanning speed during acceleration or deceleration or keeping the same scanning speed while switching on and off throughout the job [198].

6.2 Surface roughness studies and discussion

The parameters affecting the surface quality can be identified by further examination of the most critical technologies in VPP. Nowadays, the research in surface quality for the Vat photopolymerization technique is extensive, mainly because it combines an affordable price with high performance in terms of printing precision and time. There has been considerable research on SLA and DLP, the oldest and most commercialized technologies. Still, new trends are emerging due to their faster print speeds, such as CLIP [32, 191], and higher and higher resolution, like TPP [194]. Instead of surface roughness, recent significant research results are presented, focusing on SLA and DLP in Table 10.

Table 10 Overview of studies on surface roughness in VPP (SLA and DLP) (see text^* below the table for abbreviations)

According to Table 10, the quality of VPP printing is primarily determined by the photopolymerization process, exposure strategies, and projection systems. Generally, the projection system plays a prominent role in the resolution and printing quality outcome. The microstructure heavily depends on the technology used, and even if all the parameters are optimized, there will always be an error that is impossible to entirely remove. For instance, the light permanently distorts how it impinges on the resins. However, the sum of errors can also outperform other optimizations.

In the same VPP technique, the most predominant parameters influencing the roughness are the build orientation and the layer thickness. Indeed, they are directly connected with the staircase effects. In the top-down approach, once immersed in the resin, the part goes down with a depth based on the layer thickness [183]. Reeves and Cobb [207] found a mathematical model for the approximation of the surface roughness considering the layer thickness and the plane orientation (Fig. 18). It consists in

$${R}_{a}=\left({L}_{t}\left(\mathrm{tan}\varphi \mathrm{sin}\theta +\mathrm{cos}\theta \right)/4\right)+K$$

(21)

where R_a is the roughness average, L_t is the layer thickness, φ is the layer profile angle, θ the surface angle, and K is the composition roughness (up-facing or down-facing roughness).

Fig. 18

Adapted from Reeves and Cobb [96]

Surface roughness for a stepped plane.

As discussed above, different systems can offer various methods to get very high resolution, but many other parameters must be considered. Regarding the resin material, the interplay with different photoinitiators, monomers, or oligomers [184], the addition of additives such as stabilizers or light absorbers [32], the addition of ceramic powder [182], and the resulting viscosity [183] in the resin can influence the cross-linking process on the resultant resolution and quality of the features fabricated. On the other side, volumetric shrinkage throughout the curing course is the primary source of errors and the primary parameter of geometrical error [169]. Local temperature variations [208], UV intensity, and exposure time are critical factors that can affect surface roughness and cause under-cured or over-cured parts leading to geometrical errors.

Although various researchers have examined the level of mechanical anisotropy for SLA-printed parts [197, 209], Shanmugasundaram et al. [202] and Hague et al. [210] reported that SLA-printed components can be considered isotropic. The capability to produce isotropic components with SLA is a significant advantage over other AM techniques, such as FFF.

7 Material jetting (MJT)

PolyJet technology typically produces high-quality parts with a smooth surface. However, dimensional accuracy for large pieces is usually not similar to other processes, such as VPP, since it uses large droplets, and the accuracy decreases. The accuracy for medium-small size parts is comparable to the other AM technologies [134].

The amount of surface roughness depends on several geometrical and process parameters. MJT produces full-density parts by overlapping adjacent droplets and curing or solidifying them on the spot [8, 211]. A defect can be caused by a clogged nozzle or, in rare instances, by errors in the jetting toolpath. Most AM jetting systems used by professional AM companies are well calibrated and deposit uniform layers without accumulating errors and porosities. However, local variations in topography and layer thickness must be considered when mixing multiple materials and at the interface between the part and the support (Fig. 19). This is caused by the formation of a mini pool and droplets similar to that seen in welding [212, 213].

Fig. 19

aDeposited droplets in a layer-by-layer structure and b porosities in the structure of parts printed by Stratasys PolyJet J55 printer

A variety of methods are available to address and mitigate porosity during the material jetting process, such as using appropriate printing parameters [8, 211], optimizing material properties [214], increasing the number of printed layers to fill in any gaps or voids, as well as modifying the printing pattern [215]. The use of appropriate printing parameters such as droplet size, spacing, and temperature; optimization of material properties such as viscosity, surface tension, and curing time; and modification of the printing pattern such as zigzag instead of straight lines [216] can reduce void formation by ensuring that the material is deposited uniformly and consistently.

While polymer jetting improves the quality and tailor ability of mechanical properties, in general, the properties (particularly toughness and elongation) do not match those of photopolymer parts made by SLA. Jetting compatibility requires resin formulations, compromising the selection of optimal photocuring chemistries [134].

7.1 Process parameters

MJT performance is determined by the fundamentals of the process, such as placement and fusion of consecutive droplets, as well as droplet impact, spreading, and curing. Since the MJT system contains many functions, as described, the printing QAs undergo a higher range of parameters that need to be considered. For instance, part quality in MJT depends on the substrate, material velocity, dynamic viscosity, nozzle distance, total pressure, surface tension, density, the diameter of the nozzle, position accuracy of the XYZ motion stages, platform movements, etc. [134]. This section discusses these parameters in four main categories: materials, designs, surface finish settings, and printheads and rollers.

7.1.1 Design

Design plays a crucial role in achieving greater accuracy and preventing printing errors. Due to the layer-by-layer setup of the technology, most of the aspects related to the design of the model described in the section for FFF also have an impact here. However, the layer thickness, orientation, hatch spacing, and speed are more relevant for this technology. Droplet size, spacing, and how they impact the build platform affect the line surface finishing. The less space between droplets results in defects in the reduction of line edges and improvement of the resolution [217]. Accordingly, layer thickness is the most crucial parameter, as seen for most other AM technologies. Generally, a lower layer thickness improves the QAs of the prints. At the same time, a lower space between lines causes a decrease in the eventual ripple along the top and bottom of a layer, resulting in different surface roughness based on the design orientations [24, 70, 217].

Design orientation, including the building location on the build plate, also affects the surface roughness. According to the study by Yang et al. [218], orientation induces different roughness values in the parts. XZ build orientation seems to give the highest surface roughness value, while XY has the lowest, with the top and the bottom surface reaching high smoothness [218, 219]. Similar to FFF, leveling the build platform prevents distortions and failures and reduces surface defects. An infill density of less than 100% is not regularly achievable for AM parts made from photo resins since support is needed for overhanging features during the printing process to avoid collapse [64]. Figure 20 illustrates how build orientation (wedge angle) can affect the surface texture of samples printed under similar conditions.

Fig. 20

MJT-printed samples at different wedge angles: a 0°, b 45°, and c 90° (scale bar represents 1 mm)

7.1.2 Materials

In most cases, the MJT process uses polymers and plastics, such as Tango and Vero commercial resin. Each of these series has different ingredients. For instance, the FullCure 870 VeroBlack digital material comprises acrylic monomers, epoxy acrylate, urethane acrylate oligomers, and photoinitiators. As an example of support material, FullCure 705 contains acrylic monomers, polyethylene glycol, propane, glycerol, and a photoinitiator [24]. Due to the small and expensive range of materials that can be printed, MJT is limited in its material availability. In particular, waxes and photopolymers are the only commercially available polymers.

MJT “inks” are photo resistors mixed (blended) with waxes and some photopolymers. In modern systems, six liquids (CMYK-W + Support) can be handled simultaneously in separate containers. The printhead can combine multiple materials in a single part and blend pairs and trios of selected base resins to create hybrid properties and colors. They are also called digital materials, defined as composite materials developed for AM (mainly PolyJet 3D printing) with predetermined mechanical and visual properties [220]. Additionally, jetting of multiple materials facilitates support removal and allows high-detail visual representations and functional prototypes to be generated in full color.

Materials are divided into two main categories of base resins and support materials. The term “resin” refers to solids and highly viscous materials, but more commonly to liquids that harden in response to an agent (e.g., heat, setting agents, or light). Base resins carry the main functional characteristics and colors, each with unique characteristics. Thus, a machine can use them as a palette to receive hybrid materials. They are the basis for inks and can be used without mixing or combined to reveal new properties. Resins are a wide variety of different natural and synthetic materials. They consist of long monomer chains, forming cross-link bonds during the curing process. According to their chemical compounds, resins can be silicones, epoxies, acrylics, alkyds, etc. They can be divided into several categories: strong engineering, rigid general-purpose material, biocompatible, transparent rigid, castable, strong and tough, flexible, simulated (digital) polypropylene, simulated (digital) ABS, and composite resins [221, 222].

Material jetting requires supporting structures, and some resins are optimized for easy removal. They come in various types with different solubility parameters, which affect the range of the chemicals needed to fully dissolve them. Besides, support materials used on the surface will determine whether the surface is glossy or matte, which affects the surface roughness. The supporting resins can become soft during the printing process, which allows them to be removed manually, and the final touches are applied using a water jet. Wax-like support material is generally more rigid than gel-like material and, therefore, cannot be easily removed, resulting in a reduction in surface quality [223].

One of the most challenging aspects of this type of technology is the ejecting of the photocurable resin. For this reason, the materials in play assume a crucial role in the process.

In order to be jetted, resins need to have the correct viscosity according to their composition. For instance, Cheng et al. [224] heated the resin to 70 °C, while Jabari et al. [225] mixed and sonicated the resin with a graphene dispersion to obtain an appropriate flowability. There are many methods to gain the correct viscosity for various resins. Still, a resin generally needs to be as liquid as possible to avoid suspending particles that might cause artifacts or reduced accuracy [134]. Particles in the resin lead to imbalances during the drop creation, with sizes that may differ from each other. In fact, this occurs because there are disturbances during the ejection that may change the extrusion setup with an earlier breakup of the drop. Moreover, the droplet spreading and roller performance at each layer deposition can assume different physiognomy [226].

As with binder jetting, another critical parameter to consider is the impact of the droplet on the surface. According to Zhou et al. [227], three forces describe the droplet impact dynamics: inertia, surface tension, and viscous force. At the build plate level, the inertia force is converted into surface energy, and it needs to overcome the surface tension to obtain a flat shape. It means that the inertial force needs to be greater than the surface tension, which is helped by the viscous force. Moreover, the potential difference imparted by the piezoelectric system strengthens the surface tension and increases the impact angle or the height of the ejection [227, 228]. On the other hand, if the height is excessive, head and rear vortexes on the droplet can be formed because the ejection and the drop may impact the surface with different angles and energies [228].

7.1.3 Printhead and roller

Material jetting can print very tiny liquid droplets that reach resolutions around 1600 dpi and 16 μm [229]. The resolution depends on the ejection system but is also related to the droplet size. However, decreasing the specimen thickness increases the magnitude of distortions in the photopolymerization process [219].

Two printing modes are available in PolyJet technology: high speed and high quality with a low layer thickness. If the ejected droplet size is not big enough to cover the spacing between two subsequent ones, the printhead must pass over the same point one more time. According to the offset to fulfill, up to four rounds of jetting may be needed [230]. It is common for MJT printers to jet photocurable resin from different nozzles simultaneously; however, the build platform function may differ (Fig. 21). A roller module leveling mechanism is required to remove the excess resin and flatten the surface to reach the desired layer thickness. After this step, the droplet is immediately cured [223].

Fig. 21

An illustrative photo of the rotary build tray of the PolyJet J55 printer

The hatch spacing in MJT, like other AM techniques such as SLS and BJT, contributes to the texture and morphology of the surface [134, 231]. It comprehends the pulse width and the frequency with which droplets are ejected, and the ejection system series of rounds passes over the same point to create an entire surface. Printing speed is connected with the hatch spacing because it is also based on the sweep speed at which the hatch arises [134].

The quality of the droplet is another aspect to consider. Bussmann et al. [232] showed that impact velocity and angle are essential in terms of quality printing and surface roughness. The effect of rough cured drops may avoid the impact of fingering (perturbed leading edges) and splash (refer to Fig. 20).

Droplet impact, spreading, and curing limit the accuracy of parts and the motion system, which is often a gantry operated by a motor. Droplet spreading and curing restrict the wall thickness, also known as the minimum in-plane feature size (X, Y). As a result of droplet spreading, the wall thickness is much thicker than the layer thickness (Z) [233]. The minimum feature size increases with the aspect ratio because of the precision of layer-layer registration. Regarding the maximum feature size, it is generally constrained by the build volume of the printer. The foundations of MJT do not restrict feature width; however, few printhead components contain blocking features.

Droplets size and printing speed are the key points for a correct saturation rate and quality printing. Droplets of smaller size can give a better resolution of the printing, but if they are too small, they do not spread effectively. High printing speed can cause a loss of accuracy and resolution because the droplet does not have time to spread effectively [217]. Therefore, adjusting the printing speed makes it possible to favor better material spreading. Accordingly, every droplet indeed has a spreading radius time-dependent, r(t), given by

$$r\left(t\right)={\left(a+bt\right)}^{n}$$

(22)

where t is time and a, b, and n are constants that determine the growth rate, initial size, and shape of the droplet. These constants can also be related to other factors, such as the viscosity of the liquid, the surface tension between the liquid and the surface, and the contact angle between the liquid and the surface as in other hydrodynamic theories [234].

Imperfections in finishing are a common problem on the surface of MJT parts, which frequently have a rough or ribbed surface finish caused by overlapping material layers [235]. Miyanaji et al. [236] reported that the three most common jetting techniques for polymeric droplets are single, overlapping, and overlaying droplets. Overlapping droplets give better control for fine sizes.

7.1.4 Surface finish setting

Surface roughness values vary depending on the surface finish settings, and a correlation between surface roughness and finish settings can be complicated to establish [8]. In the case of a glossy finish, surfaces will not be covered by support materials except overhanging structures and the bottom of the part. Support materials will be covered when a matte finish is selected [223]. A glossy surface finish produces a lower surface roughness level, resulting in longer fatigue life. Generally, printing parts with a glossy finish setting is recommended to achieve higher surface quality [237].

Many current studies are trying to reach full-color printing by improving resolution and surface finish related to multi-color materials. Some studies have been carried out by Udroiu et al. [24] on matte and glossy surfaces by setting different process parameters. In particular, they found a correlation with the surface roughness ${R}_{a}$ given by

$${R}_{a}=t/4 (\left|\mathrm{cot}\varphi \mathrm{sin}\theta +\mathrm{cos}\theta \right|\bullet {K}_{1}), {K}_{1}<1$$

(23)

where t is the layer thickness, $\varphi$ is the droplet contact angle, $\theta$ is the plan orientation, and ${K}_{1}$ is the correction coefficient based on the PPFTs used. Some other studies instead have been focused on trying to improve the tuning of colors without jagged shifts between adjacent layers [238].

7.2 Surface roughness studies and discussion

Compared to other AM techniques, research concerning the surface roughness of MJT technology has been very limited. This contrasts with the fact that MJT appears more relevant to aesthetic applications and artistic purposes. Most of the research in MJT has focused on the relationship between a few pre-processing parameters and the surface finish. However, several publications [70, 239] have used mathematical models to estimate the roughness characteristics of parts printed with PolyJet technologies. Table 11 summarizes significant research related to surface roughness in MJT.

Table 11 Overview of studies on surface roughness in MJT (see text^* below the table for abbreviations)

In agreement with Table 11, positioning the part along the build tray substantially impacts the surface roughness in MJT parts. There was generally less roughness on horizontal surfaces than on vertical planes. On horizontal surfaces, roughness is determined by the droplet spreading and interaction of successive droplets for line and plane formation. In contrast, on vertical surfaces, it is controlled by the interaction of consecutive layers, resulting in stair-stepping equal to the layer thickness of the printed part. The spreading of droplets leads to extremely thin layers, resulting in smooth horizontal and vertical surfaces at the mesoscale [233]. As Udroiu et al. [243] demonstrated, the build type (matte or glossy) substantially affected surface roughness. Although MJT with R_a generally less than 10 µm can be considered to be between FFF (~ 1 µm < R_a <  ~ 35 µm) and VPP (R_a <  ~ 5 µm) in terms of surface roughness, the machines show much less variability compared to FFF and more variation compared to VPP methods. Again, and similar to FFF, SLS, and VPP methods, R_a and S_a were the most commonly reported roughness parameters.

8 Comparative studies and discussion

Table 12 summarizes recent comparative studies on the significant polymer additive manufacturing processes.

Table 12 Overview of comparative studies on surface roughness in FFF, SLS, VPP (SLA and DLP), and MJT (see text^* below the table for abbreviations)

Most benchmarks developed for AM were intended to measure the implementation of a single technology and a limited number of parameters, as discussed in the previous section. The comparison of various AM methods has been the focus of several studies. For instance, Mou and Koc [118] compared three AM technologies, FFF, SLA, and MJT, on four machines in terms of their surface roughness, edge sharpness, and dimensional accuracy. According to their results, FFF produced a rough surface and irregular dimensional accuracy, SLA manufactured smoother surfaces but resulted in the distortion of thin features (< 1 mm), and MJT fabricates surfaces with comparable surface roughness and dimensional accuracy. Sillani et al. [28] reported the trend of surface roughness on the bottom and top surfaces of MJT and SLS seems to be about identical. Minetola et al. [16] evaluated three polymer-based 3DP machines by analyzing their dimensional accuracy using ISO IT grades. They reported a thinner layer gives a greater definition of the features geometry and higher dimensional accuracy. Li et al. [64] compared FFF, SLA, and MJT based on cost, sustainability, and surface roughness quality factors. They reported MJT and SLA as the best and moderate AM methods in tactile and visual assessments, respectively. However, unique SLA materials were considerably more valued in the hedonic sensation category. Results indicated the lowest overall ranking for FFF but with the capability of manufacturing with the lowest environmental problems and costs, confirming its sustainability.

In terms of dimensional accuracy, for instance, Minetola et al. [16] reported that despite the increased layer thickness (0.21 mm against 0.10 mm), the Arburg Freeformer machine outperformed the Prusa i3 for more comprehensive ISO ranges of the primary size. However, layer thickness was the most essential element for improved dimensional accuracy for smaller feature sizes. Roach et al. [2] used inkjet printing for PEGDA/PI material over the PEI substrate manufactured by the FFF method. They reported that direct-ink-writing (DIW) surface modification process for FFF substrate reduces the surface roughness, resulting in improved conductivity for electronics and radio frequency (RF) applications. Nazir and Jeng [252] introduced high-speed additive manufacturing by merging PBF, MJT, and sintering technology without coupling 3DP with subtractive methods. They showed that while the MJF process was substantially quicker than the SLS method, the SLS PA12 parts showed 15% lower R_a when compared to the high-speed MJF.

9 Summary

The 3D-layered nature of AM processes and partially melted particles influence the definition of the component surface. Accordingly, controlling PPPs can dramatically affect the 3D features on a rough surface for AM components. The deliberate surface modification based on surface texture metrics in 3DP products is more demanding than ever with 3D measurement and characterization development, which can comprehensively reflect the surface topography.

Fabricating a part using layer-by-layer deposition in which the produced part exhibits a staircase effect causes the surface to become rougher. It is possible to reduce this problem by being aware of the regular surface roughness of the parts in advance or predicting the roughness values during pre-processing. Accordingly, the PPPs based on process parameter optimization have been discussed to list the parameters that have the most critical influence on the roughness of as-printed polymers. This comparative review emphasized the growing interest in understanding AM system restrictions and discrepancies so that a better selection of 3DP technology can be made based on project constraints. This study summarized the significant advances in additive manufacturing, including the incorporation of AM design decisions to assist in identifying candidate solutions, as well as information regarding roughness considerations for the selected processes.

As mentioned in the previous section, the best possible surface and optimum roughness according to an application can be obtained by adjusting the fabrication parameter. The other choice is to investigate the optimum mix of PPFTs that can be applied to any AM objects. However, it increases the cost, time, and complexity of the process. Production settings may be tweaked to favor speed above surface quality if this combination proves effective.

A variety of strategies have been employed by various groups to achieve this objective. Most studies have focused on the top surface of the parts because many factors contribute to the surface roughness distribution of a 3DP object, such as layer height. It is revealed from the literature review that the workflow of surface modification in the pre-processing step heavily depends on the complexity of the design and the desired quality-time–cost balance. Surface modification techniques are currently not standardized and depend on factors such as geometry and intended application. As a result, the following AM workflow in Fig. 22 can lead to optimum surface modification by altering roughness before and during processing.

Fig. 22

Suggested 3DP workflow to enhance as-printed surfaces

Along with the literature review, the Taguchi method, full factorial method, response surface method (RSM), and analysis of variance (ANOVA) were the most used methods for optimizing the surface roughness of 3D printers. In the case of RSM, it is generally a time-consuming method depending on the orthogonal matrix used [44]. Thus, it has been less widely used than the Taguchi method to date. The combination of specific optimization parameters may result in parts with no known surface roughness, which would have to undergo fundamental design changes. Current capabilities are limited without developing a new surface modification workflow that considers the PPFTs requirements during the pre-processing phase. Frequently, DfAM necessitates the redesign of parts initially planned for conventional methods such as machining. Post-processing must be integrated into the design process at an earlier stage, and the role of each component must be considered. Specifically, post-processing using ultraprecision technology is gaining increasing attention as it provides high-quality parts with improved surface finish and dimensional accuracy [253]. The importance of having precise geometries and smooth surfaces is particularly important for polymer optics and devices, where optimal performance depends on precise geometries and smooth surfaces.

According to the studied literature, the appropriate AM technique selection determines manufacturing efficiency, accuracy, and model size. Thus, it determines whether the actual and nominal roughness is conforming and will allow AM to be better integrated with roughness requirements. For instance, while FFF technology has the advantage of being efficient and cost-effective for small-scale production, accuracy may be limited when the production of large and complex parts is required. The recorded R_a results were generally less than 5 and 10 µm for the studied VPP and MJT methods, respectively, 10 to 20 µm for SLS, and between 1 and over 30 µm for FFF. Accordingly, SLA and DLP technologies, offer high accuracy and resolution for larger and more complex parts with more stable roughness results at various printing processes, considering SLS is at the opposite end of the roughness spectrum. While MJT can fabricate ultimately smooth parts in the sub-micrometer ranges similar to VPP, their 3D-printed surface can be as rough as FFF, depending on the selected processing method. However, both VPP and MJT AM categories can be time-consuming and expensive, limiting their suitability for competing with mass production in the present form.

Selection of the proper AM technique will also require an in-depth examination of the surface measurement techniques. The measuring method results in significant uncertainty in roughness evaluation due to PPP. It is reported that the stylus in contact-based profile measurement scratches and physically smooths the surface, leading to slight compliance with the actual topography. However, it is time efficient, more reproducible, and provides comparable results even when the tip radius varies [3]. Accordingly, tactile roughness measurement and R_a were frequently reported as the main method and roughness metrics in the reviewed articles. However, R_z performed better than R_a since it accurately represents both tactile and visual roughness. Nevertheless, observers are influenced by appearance attributes such as color, texture, glossiness, and translucency when evaluating roughness and surface finish [254]. In particular, polymeric parts represent most of these appearance attributes.

Considering the results, mainstream 3DP technologies differ significantly in terms of surface roughness. The FFF method makes extending applications across various applications challenging due to its poor surface quality. However, FFF objects seem more appropriate for analyzing application-based purposes. SLA is considered a low-cost desktop device that directly competes with FFF because of its higher print resolution and reduced surface roughness. The SLS technique is still being developed for this purpose, and MJF is emerging as a promising technique. In general, the surface roughness of the SLA and MJT was reported to be better than the SLS components. Compared to other polymer AM techniques, their inherent smooth surface finish and great dimensional accuracy imply that painting and coating are rarely required.

Combining these data confirms that the reviewed 3DP methods cannot produce ready-to-use end products and that PPFT is necessary. As a significant finding, it is revealed that there is a demand for further investigation on the appearance of 3D-printed structures, dealing explicitly with their QAs issues. Optimization of AM should not be conducted to achieve complete control over roughness. It is primarily due to the lack of reproducibility in AM technology and the role mainly played by post-processing. Further research is required to establish the links between different PPPs and the quality of the surfaces of AM-made components, as discussed in the next section.

10 Future trends and capabilities

The importance of surface roughness in the final product is recognized by both conventional and advanced manufacturing methods, especially for critical and small products. To increase the applicability of AM processes, it is necessary to assess the surface finish of as-printed polymers and to provide guidance on AM process windows and limitations [8]. 3D and 2.5D printing with multi-materials and multi-colors will be key to the future development of AM technology [255]. The techniques mentioned in this study can also provide insight into other advanced materials, such as nanoparticle and their suspensions with functional properties [256,257,258], surface treatment [259, 260], and liquid metals to use in AM technology. Furthermore, fiber reinforcement and composition can be incorporated into almost all AM methods [261]. The development of eco-friendly materials, the use of polymers, durability, and sustainability are also major concerns [262]. As an emerging trend in advanced manufacturing, the combination of several AM technologies presents new challenges in terms of surface finish.

There is a growing interest in 4D additive manufacturing, which is a relatively new research area. Smart materials can be developed more quickly by developing multi-material 4D printing [263, 264]. A 4D-printed part can thus be carefully controlled in terms of surface texture and topography as a microstructure to achieve more complex geometrical transformations. Therefore, monitoring the surface roughness of smart materials is an essential step. As with 3D multi-material printing, it can present similar challenges, such as limited material choice, printing resolution, slow mechanical performance, and dimensional accuracy [263]. It will be necessary to implement multi-material additive manufacturing in a variety of applications as part of multidisciplinary research and development [262].

The lack of aesthetically appropriate materials for AM necessitates further investigation. An understanding of the induced anisotropic arrangements and their impact on the build platform (chamber) and product properties may be improved by using sophisticated algorithms and numerical techniques [211]. The fundamental material application procedure, as well as the actual applied stresses, heat, and weathering agents, requires further investigation. Aside from the roughness value, the roughness distribution is also of critical importance to the use of AM components in the future. Therefore, there are various capabilities available to examine how printing factors impact other mechanical performance factors, such as compressive strength and tensile strength. As a result of these findings, AM products are likely to be applied to parts for automotive, aerospace, and jewelry applications that require high-dimensional accuracy and proper surface characteristics.