On the application of components manufactured with stereolithographic 3D printing in high vacuum systems

We report on a method for using stereolithographic (SLA) additive manufacturing to rapidly and cheaply prototype components for use in high-vacuum environments. We demonstrate the primary vacuum contaminant from freshly printed SLA plastics is water with no evidence of polymers out-gassing from the material and thus the vacuum performance can be controlled with simple treatments which do not involve surface sealing. An unbaked vacuum system containing SLA printed parts achieved 1.9e-8 mbar base pressure whilst retaining structural integrity and manufacturing accuracy. Preliminary results indicate that our method can be extended to achieve ultrahigh-vacuum compatibility by baking at higher temperatures. We further report on the effect of atmospheric exposure to components and present evidence to suggest that re-wetting occurs exclusively in the component skin layer, by showing that the bulk mass changes of the material is irreversible on the timescale investigated (<2 weeks).

Additive manufacturing is a rapid fabrication technique with a large amateur and industrial user base.Printer sales are growing at more than 20% per annum [1] and projected sales of 21.5 million units in 2030.Additive manufacturing benefits from wide applicability in smallscale manufacturing, prototyping and refining of new parts due to fast turn around times, capability to fabricate internal and external structures impossible to fabricate using conventional techniques, low per-unit costs, efficient tool-chain from concept to part and wide material choice.These benefits are particular relevant to research & development and small volume-high value manufacturing, two areas where the use of high-/ultrahigh-vacuum equipment is ubiquitous.Several printing technologies are available in the marketplace from large industrial machines to desktop devices, in addition different print media can be used to suit precision and material properties required.
The application of printed parts in high-vacuum systems has been limited, though there are reports discussing the use of fused deposition modeling (FDM) and multi jet fusion (MJF) methods which have been shown to have some vacuum compatibility depending on ultimate surface finish and porosity [2][3][4][5].Stereolithography (SLA) has become a standard method due to the control and precision of manufacturing and is substantially different to it's stablemates since the finished part is closer to a chemically bonded solid body and structures are not defined by lamination layers, thus SLA is a promising candidate for producing vacuum compatible bespoke parts.SLA printers are available at low cost and are accessible to most users readily available printing consumables.Layers of 10 − 20 µm with overall part tolerances of ±0.15 µm are commonplace in the consumer market (e.g.3Dhubs.com),and contract printing services are readily accessible at low unit part cost for one-off prints or small manufacturing runs.
Previous work has explored the application of vacuum sealants, such as "VacSeal" from SPI Supplies Inc. [6], post-manufacturing on acrylonitrile butadiene (ABS) and polycarbonate (PC) FDM prints, to create a physical barrier through which trapped water cannot out-gas [7][8][9].Designed primarily as leak sealants for high-and ultrahigh-vacuum systems, such products typically contain toxic solvents which affect repeatability and present environmental challenges.Crucially, application of a surface coating limits the degree of internal complexity of printed components as any surface left un-coated will result in performance limited out-gassing.Many common vacuum sealants require high temperature curing, 260 -300 °C for VacSeal specifically [6], which are not compatible with many materials used for fabrication.Typical pressures achieved using leak sealant coatings range between 10 −4 − 10 −7 mbar when applied to ABS or PC [7][8][9][10].
In the current work, we investigate the vacuum properties of SLA printed components and show that effect of the bulk water can be mitigated without vacuum sealants, reaching the test vacuum chamber base pressure of 1.9×10 −8 mbar.We define a post-manufacture method to process SLA plastics to make them compatible in high-and ultrahigh-vacuum systems (HV and UHV, respectively), alongside characterisation of their re-wetting properties upon exposure to atmosphere.Following the outlined procedure it is possible to use SLA printed components in HV systems whilst retaining structural integrity and manufacturing tolerances.While there have been few reported usages of SLA components in HV systems, recent usage, such as that of the complex optical components for neutral helium microscopy [11,12], indicate that there is a wider interest in the application of SLA components in HV systems beyond vacuum science itself.
To investigate the out-gassing and re-wetting properties of SLA 3D printed plastic, a regular tetrahedron (nominal 4.80 cm edge length, ∼ 13 cm 3 volume) was chosen as the test sample.The chosen sample is of typical size for a vacuum component one would want to produce using additive manufacturing.Six samples were printed by 3DHubs [13] using Formlabs "Clear Resin" [14,15] with a Formlabs "Form 3" printer costing approximately £ 5 per sample with a lead time of 3 days.While the samples were ordered from 3DHubs, the chosen resin and printer are standard and representative of the most widely available SLA additive manufacturing services and methods.Upon delivery the samples were within the stated dimensional tolerance (±0.15 mm) with smooth, flat faces, and sharp edges with the exception of small aberrations on the side where a support structure attached during fabrication.
The parts were baked in a vacuum oven with UHV base pressure (∼ 10 −10 mbar).Vacuum properties were investigated in a separate vacuum system using a Hiden Analytical residual gas mass spectrometer (Model: HAL/3F RC301 PIC300) [16].The atmospheric exposure time was varied to probe the vacuum properties of the samples.Atmospheric temperature and relative humidity were recorded using a "PICO Humidiprobe" [17].The SLA printed plastic preparation and out-gassing measurement procedures are outlined below.
1. Clean the surface of the part by wiping with isopropyl alcohol and dry immediately.2. Record mass and dimensions of samples.3. Bake parts in a vacuum oven.Increment the temperature by 20 • C every half hour until 120 • C is reached, bake for 48 hours.Allow parts cool slowly by turning off heating element and allowing for natural cooling within the oven.Gradual heating/cooling prevents structural distortion from material and water expansion.4. On removal from oven, visually inspect the samples and measure dimensions.Transfer samples to either the test vacuum chamber or store in an opaque storage container with temperature and humidity monitoring [17]. 5. Planned atmospheric exposure times, in the storage container, of 0, 1, 2, 4 and 7 days such that each sample is exposed for a different length of time.6. Record pump down curves for the samples in the test vacuum chamber.7. Acquire mass spectra using residual gas analyser [16] immediately before venting for next sample.
Use mass spectra to find the partial pressure of water out-gassing from each sample by integrating the spectrum to the total pressure.8. Repeat steps 4-7 for all samples.
The samples were all placed in a vacuum oven and heated to 120 • C over a two hour period.Oven pressure was initially 4 × 10 −5 mbar at room temperature, rising to above the accurate range of an Edwards wide range gauge (WRG) (±30% at 1 × 10 −3 mbar) [18].Bake temperature was sustained for 48 hours after which the chamber gradually cooled over approximately 3 hours.
Once the oven was vented, the 0 day sample was weighed and transferred, using gloves and in atmosphere, to the test chamber in less than 10 minutes.The other samples were also weighed and placed in an opaque, temperature and humidity monitored container for storage.The mean mass lost during baking was 0.29 ± 0.005% from the initial 17.27 ± 0.05 g mean.
All analysis of vacuum properties were conducted in a test chamber (20 L volume with nominal 450 L s −1 pumping capacity) with a Hiden Analytical mass spectrometer used for residual gas analysis connected to the sample area via a UHV gate valve such that the ion source for the mass spectrometer remained on for consistency of measurements.
Pump down curves for each sample were recorded to demonstrate the re-wetting effect caused by atmospheric exposure over time.The partial pressure contribution, or outgassing, of each sample by subtracting the empty chamber pump down curve.The pump down curves for the samples exposed to air for different duration are shown in Figure 1.
Figure 1 shows a strong correlation between exposure time and re-wetting of the samples, with the majority of the re-wetting occurring in the first 24 hours of exposure.Taking a vertical slice through the figure at ∼ 21 hours elapsed time yields the out-gassing solely due to the sample itself, shown in Figure 2.
Figure 2 shows that the re-wetting is very well described by a single exponential, showing that almost half the total re-wetting occurs within the first 24 hours of exposure, making a matter of hours the critical time period for exposure to achieve high-vacuum in a time comparable to an empty chamber.
A base pressure of 1.9 × 10 −8 mbar was achieved with the 0 day sample in the testing vacuum chamber, which has never been baked, and 9.9 × 10 −10 mbar in a baked using sample transfer in air, demonstrating that SLA is fundamentally compatible with HV and UHV systems.
Figure 3 presents mass spectra taken for each sample after 21 hours, corresponding to data points in Figure 2 FIG. 1. Out-gassing curves for samples exposed to atmosphere for between 0-7 days.An empty chamber pumping curve was subtracted from the samples' pumping curves to estimate partial pressure contribution of the samples to the vacuum.Vertical slice (dashed line) through the pumping curves is shown in Figure 2 to illustrate the dependence of base pressure on exposure time.and the dashed line in Figure 1.The mass spectrum of each sample was normalised to the respective out-gassing partial pressure after 21 hours pumping down to distinguish between chemical composition of vacuum contaminants after varying exposure to air.Mass spectra have been cropped at 50 amu because there was no significant detection of heavier species, indicating that the SLA plastic itself, and any cracking products, do not evolve into the vacuum.
The similarity in the distribution of species present in the normalised mass spectra in Figure 3 indicates that, although the pump down speed slows as exposure time increases, re-wetting is skin deep on the time scales investigated.If the re-wetting of baked parts was a bulk process, one would expect water to permeate deep into the sample such that it cannot be quickly removed by heating, hence returning to the previously observed base pressures ∼ 10 −5 mbar.Baking effectively removes water from the bulk of the material and prevents water from reabsorbing into it.Therefore, although the majority of re-wetting occurs in the first week of exposure, water can only saturate the surface which functions as a finite volume of water when out-gassing into vacuum.In contrast to the unbaked plastic whose bulk water content acts as an infinite volume of water which cannot out-gas on a practical time-scale of days to weeks, hence the base pressure of only 4 × 10 −7 mbar achieved in the first instance after almost 2 weeks pumping down, in comparison to 1.9 × 10 −8 mbar for a baked sample.Note that the pump down time to achieve such pressures with baked parts depends on time exposed to air, as discussed previously.Taking the chemical composition of FormLabs "Clear Resin" as 75% poly(methyl methacrylate) (PMMA) and 25% hydroxypropyl methacrylate (HPMA) by mass according to its safety data sheet [15], the total water weight, referred to as weight in weight (w/w), pre-baking can be approximated.PMMA is reported to hold up to 2% w/w water, with HPMA holding 0.2% [19][20][21].Averaging the water w/w values yields an overall max water content ∼ 1.6% for FormLabs "Clear Resin".If one takes the 0.29% average mass loss during baking to be solely water based on mass spectra in Figure 4, then 0.29/1.6≈ 20% of the maximum initial water mass was degassed due to baking.
Figure 4 presents normalised mass spectra for an unbaked and baked sample with the empty chamber below, 21h after evacuation.It is clear that water is the principle species introduced from the untreated plastic, and it can been seen that the vacuum properties of a baked sample are indistinguishable from the empty chamber to pressures lower than 1.9 × 10 −8 mbar.
A further experiment was conducted to evaluate the ultrahigh-vacuum compatibility of SLA Clear Resin by baking out the vacuum oven with the sample inside to give a true UHV platform in the 10 × −10 mbar range such that the achievable pressure with no atmospheric exposure of the part could be established.The same baking procedure presented earlier was followed except the maximum temperature was increased to 170 • C. The pressure achieved with the part in situ was indistinguishable to that without the part installed, validating the potential for using Formlabs Clear Resin in UHV systems where a moderate temperature bake may be used.However it FIG.3. Mass spectra of samples exposed to atmosphere for 0-14 days (right side y-axis labels) were normalised to their respective base pressures achieved (Figure 1) to show that the vacuum properties of all baked samples are consistent provided pump down times approximately equal to the time the samples were exposed to the atmosphere.All spectra have the same logarithmic y-axis range.
must be noted that upon removal from the chamber, the surface of the sample developed shallow cracks which can only be attributed to the higher temperature used causing water vapour to escape destructively.We recommend that care must be taken to heat gently and a maximum bake temperature of 150 • C is used for the material.
In the current work we have demonstrated that Form-Labs Clear Resin SLA is an appropriate material for SLA additive manufacturing applied to cheap, rapid prototyping of small vacuum components with complex internal geometries, without application of external vacuum sealants post-manufacturing.A simple baking protocol has been designed and tested that achieves high-and ultrahigh-vacuum compatibility with SLA printed Formlabs Clear Resin, we anticipate that the result reflects the general performance of SLA prepared samples.
We demonstrate that the surface of SLA plastics rewets quickly under room temperature atmospheric conditions.We recommend that SLA components be kept in a dehumidified or vacuum environment for short term storage post-baking to minimise vacuum chamber pump down times.Additionally, evidence is presented to suggest that re-wetting is exclusively a surface process, making the change in bulk water content due to baking permanent over the time scale investigated.Further investigation into > 2 week atmospheric exposures is needed to explore the rate and extent to which the bulk of the plastic can reabsorb water to determine whether baking SLA plastics permanently changes their mass composition, and therefore induces permanent HV and UHV compatibility.
FIG. 4. Mass spectra, normalised as in Figure 3, for an unbaked sample (top), baked sample with 0 day exposure time (middle) and an empty chamber (bottom).All spectra have the same logarithmic y-axis range.The spectra for an empty chamber and baked sample are near identical, both at the base pressure of 1.9 × 10 −8 mbar.It should be noted that the electron multiplier saturated during the unbaked measurement, taken at 4 × 10 −7 mbar, providing lower bounds for the 2, 18 and 28 amu peaks.Some limitations have been found for higher temperature treatment of SLA printed plastics at 170 • C which reduce the treatment time and potentially further improve base pressure by extracting more water in a timely manner.Surface cracking of the plastic was observed which we attribute to the water desorption occurring explosively, we can't conclusively show that the ultimate temperature, heating and cooling rates and postmanufacturing curing processes are the dominant contributor to that process, however there is clearly further capacity to investigate and improve all these aspects of the process for more demanding vacuum requirements.Additionally, high-temperature and high-strength variants of SLA plastics are also available which may be more resistant to cracking from rapid water degassing during baking.It is likely that dehydration of components will cease to be the dominant factor and that other out gassing products will be the ultimate limitation to ultimate vacuum performance.

FIG. 2 .
FIG. 2. Data points taken from a vertical slice through Figure 1 at t = 21 hours.A single exponential with time constant −0.70 accurately describes the re-wetting rate under atmospheric conditions with R 2 = 0.988.