Mitigating print-through effects through an optimized method for CFRP mirror production in Chile
S. Castillo, G. Hamilton, N. Soto, C. Lobos, L. Pedrero, C. Rozas, A. Bayo, P. Mardones, H. Hakobyan, C. García, M.R. Schreiber, W. Brooks
MMitigating print-through effects through an optimizedmethod for CFRP mirror production in Chile
S. Castillo a,b , G. Hamilton c,d , N. Soto b,c , C. Lobos a,b , L. Pedrero b,c , C. Rozas b,c , A. Bayo a,b,d ,P. Mardones b , H. Hakobyan b,c,d , C. Garc´ıa b,c,d , M.R. Schreiber b,c , and W. Brooks c,da Instituto de F´ısica y Astronom´ıa, Facultad de Ciencias, Universidad de Valpara´ıso, Av. GranBreta˜na 1111, 5030 Casilla, Valpara´ıso, Chile b N´ucleo Milenio de Formaci´on Planetaria - NPF, Valpara´ıso, Chile c Universidad T´ecnica Federico Santa Mar´ıa d Centro Cient´ıfico Tecnol´ogico de Valpara´ıso, CCTVal
ABSTRACT
In the manufacturing process of Carbon Fiber Reinforced Polymer (CFRP) mirrors (replicated from a mandrel)the orientation of the unidirectional carbon fiber layers (layup) has a direct influence on different aspects ofthe final product, like its general (large scale) shape and local deformations. In particular, optical methodsused to evaluate the surface’s quality, can reveal the presence of print-through, a very common issue in CFPRmanufacture. In practical terms, the surface’s irregularities induced, among other artifacts, by print-through,produce unwanted scattering effects, which are usually mitigated applying extra layers of different materials tothe surface. Since one of the main goals of CFPR mirrors is to decrease the final weight of the whole mirrorsystem, adding more material goes in the opposite direction of that. For this reason a different layup method isbeing developed with the goal of decreasing print-through and improving sphericity while maintaining mechanicalqualities and without the addition of extra material in the process.
Keywords: layup, cfrp, print-through, PFI, NPF
The production of low cost segmented primaries for four to eight meters class infrared telescopes is one ofthe main technology requirements of the future infrared interferometers such as the Planet Formation Imager(PFI) (Monnier et al., 2018). Developments in the manufacturing process of replicated carbon fiber reinforcedpolymers (CFRP) mirrors have been made by the collaboration of engineers, astronomers and experimentalphysicists from the N´ucleo Milenio de Formaci´on Planetaria (NPF) and the Centro Cient´ıfico Tecnol´ogico deValpara´ıso (CCTVAL).The manufacturing of mirrors using composite materials has become a necessity for the next generation ofoptical elements of big size since it reduces weight, costs and time needed for production, while at the same timeoffering high stiffness and low thermal expansion coefficient (Wei et al., 2017). The method, called replication,consists of copying an optical quality mold surface with a series of layers of composite materials like carbonfiber impregnated with resin. This combination of materials makes it possible to clone the surface quality of thepreviously polished mandrel, but with the advantage of CFRP strength-to-weight ratio, which is better whencompared to other commonly used materials, like glass or ceramic (Schmidt, 2008). Through this replicationmethod, often called replica method, we found two main advantages: Using composite materials as carbon fiberreinforced polymer opens a range of possible uses when compared to traditional materials and secondly, thepresence of a polymer allows the replica to match the surface quality of the mold down to nanometer scales.
Further author information: (Send correspondence to S.C)S.C: E-mail: [email protected], Telephone:+56982291703 a r X i v : . [ a s t r o - ph . I M ] D ec evertheless, using composite materials such as CFRP comes with a price, which in this case is a veryknown issue called fiber print-through (FPT) (Hochhalter et al., 2006). In this paper we will be reporting onour experimental advances to optimize a layup method capable of mitigating the FPT (by changing the layuporientation) while keeping the elements that gives the CFRP replica its stiffness. In our experimental setup, the replica method consists mainly of a process called hand-layup, where consecutivelayers of Unidirectional Carbon Fiber Reinforced Polymer (UD-CFRP) are positioned over the mold (mandrel)with specific orientations to achieve the desired mechanical properties. This layers can be made of differentfabrics and different resins, but always a combination of both. Each one of this combinations have differentthermo-mechanical properties and, from the mechanical point of view, the final layup has direct impact in thoseproperties too (Hongkarnjanakul et al., 2013). There are three important concepts when defining the layup forthe specific properties of the material used: symmetry, balance and quasi-isotropy (see Fig. 1).A laminate is said symmetric when plies above a imaginary mid-plane are a mirror of those below the mid-plane (Joyce, 2003). (a) Symmetric laminate which consider an imaginarymid-plane (b) Quasi-isotropy laminate
Figure 1: Layup considerationsOn the other hand, balance is achieved when an equal number of negatively and positively angled layers isemployed, that usually translates to orthogonal pairs of layers (Joyce, 2003).Finally, quasi-isotropy aims to achieve an even distribution of the in-plane forces. Therefore, a quasi-isotropiclaminate has either randomly oriented fiber in all directions, or has fibers oriented in a way that guarantees thatan equal amount of strength is distributed all over the plane of the section. Generally, this can be achieved byusing 4 orientations [0/90/45/-45] (Joyce, 2003).Once the proper layup is decided, a release agent is applied over the whole surface of the mold to avoidadhesion between the resin and the substrate. The hand-layup is then done and both CFRP and mold areenveloped in a series of materials which allow gases and resins to flow properly when cured. A brief summary ofthe process is presented below (and the full sequence of CFRP manufacture is displayed in Fig. 2).To ensure proper vacuum, prior to the curing process, the vacuum level and the sealing quality must bechecked to guarantee that there is no leaking. The last stage, curing, can be done in an autoclave or in an oven(often called out-of-autoclave) where heat is applied. The importance of this steps resides in the fact that theigure 2: CFRP Mirror replication methodpreimpregnated epoxy resin needs to be heated to crosslink (Pham & Marks, 2005), a process through which theresin acquires its crystal atomic structure, achieving its full mechanical properties. Besides that, while the resinis being heated it also becomes more fluid so it can fill any gaps between the layers and the substrate-CFRPinterface. In addition, during this heating phase, gases can flow outside the laminate. The main advantage ofusing an autoclave is the higher pressure applied over the CFRP laminate, producing better quality pieces dueto the smaller amount of bubbles allowed to form inside the laminate.
During the replica process many factors can impact the final surface quality, such as surface contamination (Fig3a), bad layer compaction, the condition of the resin (Fig 3c) and the layup itself (Fig 3d).Surface contamination can be mitigated by using clean rooms during the whole process, decreasing the amountof dust and other materials present in the air.On the other hand, despite having a controlled environment, leakage in the vacuum during the vacuum cyclecan induce the presence of air and end up allowing bubbles between the different layers. These pockets of airwill impact the accuracy of the final surface of the mirror. To expel the bubbles from the system it is necessaryto apply enough pressure so they are forced out of the layers and a proper curing schedule needs to be set toensure enough time for the resin to achieve lower viscosity.Finally, the layup itself plays a major role in the presence of surface aberrations which are mostly visiblewhen layers are placed in orthogonal arrays where residual stress appears as “fiber print-through” (FPT fromnow on) on the surface.
The traditional layup methods typically optimize the stress distribution to maximize the stiffness of the finalproduct. However, for our application, surface quality is as important as mechanical properties, and the classicalmethods do not guarantee the former (see for example the optical tests displayed in Fig. 4). Therefore, newlayup methodologies need to be explored and optimized.The unidirectional CFRP is strong in the direction of the fiber and exerts zero force perpendicular to itsorientation. In order to overcome this weakness the common way to proceed is to place a number of layers indifferent directions. With this approach FPT appears as a consequence (see Fig. 4b). One way to mitigate thiseffect is to apply extra resin over the surface after the first curing cycle. Even so, residual FPT can still bepresent and transferred to the extra resin layer surface. In this case, the amount of residual FPT is related tothe thickness of this final resin layer and is also dependent of the laminate type: woven fibers show higher FPTin comparison to unidirectional fibers. This effect dominates the mid spatial frequencies and can be clearly seenin the surface waviness profiles. a) Surface contamination (b) Mandrel defects(c) Bad curing cycle (d) FPT example
Figure 3: Common surface errors
To analyse this kind of aberration and understand its effects on the surface, visual deflectometry is used in orderto estimate the surface shape and scattering, but this was not yet quantified in our experimental set-up becauseof a lack of instrumentation. A mechanical profilometer can also be used to measure mid spatial frequencies,where waviness can be observed. Finally, on the qualitative side of methodologies, naked eye tests are done suchas focault, ronchi, interferometry and others (See Fig. 4).
As previously explained, due to the nature of the CFRP layers, the surface of the replica can present differenttypes of aberrations, among them the FPT . With the goal of improving the surface quality of the mirrors, oneor more extra resin layers are applied on the reflective face obtained after first curing cycle. Since this resin is aviscous liquid and can be easily spread over the surface, it is possible to replicate truthfully the quality of the a) Focault test (b) Interferometry test
Figure 4: Optical testsmandrel’s surface, like its roughness and waviness. Unfortunately, our experience shows that one single layer ofresin is not enough to eliminate the FPT, as it was observed in our experiments.The surface quality measurements of the replicas before the first layer, after the first layer and after the secondlayer can be compared and analysed to help understand the efficiency of extra resin layer in FPT mitigation.Even though our experience shows dramatic improvements after the second layer,there are problems that areintrinsic to the resin that suggest that extra-resin layering alone is not the definitive solution.One of the known problems has to do with the glass transition temperature (Tg) of the resin, which limitsthe range of applications that a CFRP mirror could have depending on how high the Tg is. This reason alonecould lead to exclude space-based and direct sunlight applications.Besides, the epoxy resin suffers, like any other material, mechanical deformations, micro crackings and possiblefractures due to humidity that can infiltrate in the micro crackings creating water deposits causing drasticdegradation of this organic polymer.At last, but not less important, the average density of this resin is 1.4g/cm , similar to the carbon fiberdensity, but without the mechanical properties of the latter, therefore, arbitrary extra-layering could lead to anincrease in fragility making the replicas more prone to failures while increasing the total weight of the system. Allthese considerations lead to the strong conclusion that the thickness of the extra resin layer is a key parameterto be considered and optimized.As a final aspect to be considered, the CTE mismatch between resin and carbon fiber could become an issuesince it induces stress on the piece (Ahmed et al., 2012). Another method proposed in the literature as a solution to the FPT problem is to sputter a series of differentmetals on the mandrel’s surface (Steeves et al., 2014). In this framework, the surface of the mandrel has to bepreviously treated to prevent the adhesion of the metal to the glass while still allowing for exact surface cloningthat can be later glued to an already prepared CFRP replica. Sputtering is a method for depositing a thin layerof atoms from a chosen material onto the surface of choice (Behrisch, 1981). In this method, an electromagneticfield takes atoms of the material that will be deposited and orient these atoms gradually depositing them on thesurface of choice.An advantage of this metallic multi-layer sputtering deposition is that these layers are effective at decreasingthe FPT while also working as a reflective skin. So, in principle, this is a very promising double purpose methodproducing a final product that is closer to the goal, an already reflective mirror with better surface quality.However, the cost-effective scalability (in mirror size) of the process is far from trivial and the experiments shownn ”Design, fabrication and testing of active carbon shell mirrors for space telescope applications” (Steeves et al.,2014) were made only in 15cm hexagonal samples.
As previously mentioned, the traditional layup method optimizes the layers orientation with the goal of obtaininga mechanically resistant final piece, but it does not yield the optic quality needed. This work’s proposition is toreorient a number of layers from the beginning to the end of the layup maintaining the basic considerations ofbalance, symmetry and quasi-isotropy, but allowing the mitigation of print-through thanks to the repetition oflayers in the same direction.Figure 5: Representation of the effects of the micro tubes that compose CFRP in a traditional layupIf each layer is a pattern of unidirectional carbon fibers, we found that, in between different orientations, whenmore layers are stacked together in the same direction, the fibers are allowed to arrange themselves more efficientlydecreasing the undulating surface profile. In the traditional method every layer has a different orientation asseen in Fig. 5 and the result is a surface with visible undulations coming from the inner layers. Our new methodimproves the initial replica’s surface quality producing a final replica with better reflectivity since it decreasesthe FPT and the scattering effects, therefore having better roughness and waviness as seen in 1 and 2.Figure 6: Representation of the effects of the micro tubes that compose CFRP in proposed layupExtra resin layers are still going be necessary to suppress the remaining mostly-unidirectional FPT, but, withthis layup method, its thickness can be reduced minimising the probability of mechanical failures due to themismatch of the materials properties between the resin and the CFRP layers as mentioned in subsection 4.2.1In Table 1 we present our results specifying, for each replica, its ID, mandrel, diameter, number of layers andlayup used. Three of the replicas presented in the table were made with our proposed method (IDs 170-172-173),while the other four were made with traditional and non-traditional methods.able 1: Tests madeID Mandrel Diameter(cm) −
147 CX200PYR 19 8 [0/10/20/30/40/50/60/70/80]157 CX200PYR 19 8 [0/90/45/ − − /45/0/ − /0 /45/ − − /45/0/ − Analyzing these changes in layup orientation we have observed improvements both in roughness and waviness,measuring around four times better quality using our method, mostly in 19 cm spherical CFRP replicated mirrors.Regardless, further testing is needed in terms of scalability of the replica size. In Fig. 7 and Fig. 8 we presentthe results for Rq and Wq for replicas with the traditional method versus the proposed method. (a) Rq Replica 157 (b) Rq Replica 170
Figure 7: Rq Replica 157 and 170 a) Wq Replica 157 (b) Wq Replica 170
Figure 8: Wq Replica 157 and 170As it can clearly be seen in Table 2, replicas made with our proposed method have better roughness, andeven better waviness, on the first curing cycle. This could lead to thinner extra layers to mitigate FPT, yieldinga lighter final piece when compared to CFRP mirrors made with the traditional layup.Table 2: ResultsID Rq Wq128 17 nm 152 nm147 235 nm 2896 nm157 539 nm 17577 nm159 169 nm 10225 nm170 12 nm 51 nm172 15 nm 41 nm173 133 nm 658 nmFrom Table 2 we can also see the improvement even when compared to a 24 layers CFRP mirror (ID 128),meaning that with less layers we can achieve better initial surface quality, specially at mid spatial frequencies.
With our experiments we have shown that non-traditional layup procedures can help mitigate critical aspects ofCFRP replicas such as FPT, giving us an insight of how crucial this consideration could be on the next generationof lightweight astronomical mirrors. As future work we will systematically characterize the relation between thenumber of CFRP layers in the same direction and the amount of mitigated FPT.The knowledge of how these parameters are related will allow us to determine the optimum setup for a specificreplica considering the total number of layers, diameter, the curing technique and others.
A Azimutal profile for CFRP mirrors
In this appendix we present the azimutal profiles obtained for every replica previously discussed both in 5.1 and6 that were not included in the body of the paper to improve readability. All the profiles here presented wereobtained with a portable surface roughness tester Mitutoyo SJ-410. a) Rq Replica 128 (b) Wq Replica 128
Figure 9: Replica 128 Rq and Wq (a) Rq Replica 147 (b) Wq Replica 147
Figure 10: Replica 147 Rq and Wq (a) Rq Replica 159 (b) Wq Replica 159
Figure 11: Replica 159 Rq and Wq a) Rq Replica 172 (b) Wq Replica 172
Figure 12: Replica 172 Rq and Wq (a) Rq Replica 173 (b) Wq Replica 173
Figure 13: Replica 173 Rq and Wq
CKNOWLEDGMENTS
All the authors acknowledge financial support from Iniciativa Cient´ıfica Milenio v´ıa N´ucleo Milenio de Formaci´onPlanetaria. A.B acknowledges support from FONDECYT grant 1190748, A.B. an N.S. acknowledges supportfrom ESO Comit´e-Mixto and A.B. from QUIMAL funding agencies. M.S., S.C and C.L acknowledge supportfrom the ALMA-CONICYT fund. G.H and N.S acknowledge support from the Programa de Incentivo a laIniciaci´on Cient´ıfica (PIIC) from USM.
References
Ahmed A., Tavakol B., Das R., Joven R., Roozbehjavan P., Minaie B., 2012, in SAMPE International SymposiumProceedings.Behrisch R., 1981, Sputtering by Particle Bombardment I. ”” Vol. 47, Springer-Verlag, doi:10.1007/3-540-10521-2Hochhalter J. D., Massarello J. J., Maji A. K., Fuierer P. A., 2006, in Sasian J. M., Turner M. G., eds, Societyof Photo-Optical Instrumentation Engineers (SPIE) Conference Series Vol. 6289, Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series. p. 628902, doi:10.1117/12.681042Hongkarnjanakul N., Bouvet C., Rivallant S., 2013, Composite Structures, 106, 549Joyce P., 2003, United States Naval Academy, 1Monnier J. D., et al., 2018, Experimental Astronomy, 46, 517Pham H. Q., Marks M. J., 2005, Epoxy Resins. American Cancer Soci-ety, p. 228 (https://onlinelibrary.wiley.com/doi/pdf/10.1002/14356007.a09 547.pub2),doi:https://doi.org/10.1002/14356007.a09˙547.pub2, https://onlinelibrary.wiley.com/doi/abs/10.1002/14356007.a09_547.pub2https://onlinelibrary.wiley.com/doi/abs/10.1002/14356007.a09_547.pub2