Pattern transfer of large-scale thin membranes with controllable self-delamination interface for integrated functional systems

Direct transfer of pre-patterned device-grade nano-to-microscale materials highly benefits many existing and potential, high performance, heterogeneously integrated functional systems over conventional lithography-based microfabrication. We present, in combined theory and experiment, a self-delamination-driven pattern transfer of a single crystalline silicon thin membrane via well-controlled interfacial design in liquid media. This pattern transfer allows the usage of an intermediate or mediator substrate where both front and back sides of a thin membrane are capable of being integrated with standard lithographical processing, thereby achieving deterministic assembly of the thin membrane into a multi-functional system. Implementations of these capabilities are demonstrated in broad variety of applications ranging from electronics to microelectromechanical systems, wetting and filtration, and metamaterials.


REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): The authors should be congratulated on an excellent paper, showing a novel fabrication method for membrane transfers. It is a timely and interesting paper. Aside from typos and some grammatical awkwardness, the paper should be published.
I would only make the following three comments: 1. it would be helpful to discuss general approaches which make this method more widely applicable, i.e. to different materials. In particular, a discussion on the theory would be helpful, as would a discussion on the theoretical approaches that could be used to predict the behaviour of the system/approach for various material pairs.
Our response : We highly appreciate the reviewer's positive evaluation. We also sincerely thank him/her for the comment on the process applicability to wide range of materials. In fact, Si membrane/Si substrate pair is one of the hardest cases to self-delaminate as the Si/Si pair has very high interfacial adhesion energy . In addition, Si is rigid/brittle with easy fracture at very small deformation as well as Si/Si pair has flawless contact interface. Even for these challenging conditions, our approach can be widely applicable as long as the thin membrane material is not dissolved to medium liquid such as water/solvents and has decent sacrificial material pair. As a preliminary example, we also showed that pattern transfer of polyimide thin membrane/Si substrate pair with a PMMA sacrificial layer in Fig 7a. Certainly, we believe that this technique can further be extended into other device-grade materials such as an epitaxially grown GaAs membrane/Si or GaAs substrate pair with a AlGaAs sacrificial layer [R1] Our modification to the manuscript: To clarify this point, we have modified and included the following sentence in Page 2 "The theoretical model is established to understand the transfer mechanism based on self-delamination in the liquid media and provides a quantitative guide to experimental demonstrations in great agreement. It is worthwhile to note that the theoretical model certainly ensures the versatility and robustness of this method to be readily extended for other membrane materials while Si membranes are primarily utilized in this work." [R1] Park, Sang-Il, et al. "Printed assemblies of inorganic light-emitting diodes for deformable and semitransparent displays." science 325.5943 (2009): 977-981.
2. a discussion of van der Waals interactions and their impact on the feasibility of the approach. In particular, the authors may want to discuss the large interaction forces computed using many body approaches in Nature communications 11 (1), 1-8 (2020).
Our response: We appreciate the reviewer for this valuable suggestion and sharing the important reference with us. The increasing of van der Waals interaction force will lead to the increasing of interfacial adhesion energy between film and substrate. As shown in Supplementary Fig. 2, the increasing of leads to the increasing of required mechanical peeling strength F/b, which will make the self-delamination more difficult. For example, when the wettability of film and substrate is = 20° and = 7.65°, and surface tension is = 24mN/m, the peeling strength F/b becomes larger than 0 when is larger than 46mN/m for all different porosity of film. Therefore, this self-delamination approach cannot be used when is larger than 46mN/m. Our modification to the manuscript: To clarify this point, we have modified and included the following sentence in Page 4 "Similarly, Supplementary Fig. 2a and 2b show the effect of wettability of substrate and interfacial adhesion energy between thin film and substrate on the required ⁄ , respectively. The required peeling strength ⁄ increases with the incressing of interfacial adhesion energy and ⁄ becomes larger than 0 for all the porosity of film when is beyond a critical value such as the large van der Waals interaction force [27]. In this scenario, applying an external mechanical peeling force is required to assist the delamination at the interface between film and substrate." [27] Hauseux, P., Nguyen, T. T., Ambrosetti, A., Ruiz, K. S., Bordas, S. P., & Tkatchenko, A. (2020). From quantum to continuum mechanics in the delamination of atomically-thin layers from substrates. Nature communications, 11(1), 1-8.
3. What would be the computational and modelling approaches required to generalise the approach proposed here, and how could this be done in a consistent manner, regardless of the material pair.
Our response : We appreciate the reviewer for this valuable comment. In our previous work (Yue Zhang, Qingchang Liu, and Baoxing Xu. Extreme Mechanics Letters 16 (2017): 33-40), we have modeled the peeling of thin film from substrate by an applied mechanical peeling force in a liquid medium using molecular dynamics (MD) simulations for different material pair. The selfdelamination approach proposed here is similar to that mechanical peeling method except that there is no external mechanical force, so the computational modeling of self-delamination approach can also be done using MD simulation in a consistent manner.
Our modification to the manuscript: To clarify this point, we have modified and included the following sentence in Page 5 "Si membranes with different porosity of 0.04, 0.2, 0.4, and 0.7 are prepared as shown in Supplementary Fig. 3 and the experimental results also qualitatively shows that a higher porosity Si membrane is more favorable for peeling off from a Si substrate in an acetone medium. The computational modeling of the self-delamination of a film on substrates with a broad variety of materials is similar to that of peeling a film from substrates under an applied mechanical force where molecular dynamics (MD) simulations can be employed [25]." The authors present an innovative fabrication technique that can achieve direct transfer of prepatterned nano-to-microscale materials. The fabrication process is driven by self-delamination of single crystalline silicon thin membrane in liquid media. Theoretical model is established to help understand the key factors that affect the self-delamination and how to control the transfer process. This direct pattern transfer technique possesses some unique advantages compared with other methods. It can be used to generate complex 3D Si structures. This method also enables flip and transfer of patterned membrane, which allows for multiple lithographical process on both front and back sides. Then the versatility of this fabrication method is demonstrated through several applications, including hybrid microassembly of LED circuit and re-entrant structures fabrication with omniphobicity and other advanced functionalities.
The idea of utilizing thin film self-delamination to achieve pattern transfer is innovative. It provides a convenient method to achieve direct transfer of large area without physical damage. Designs in this manuscript for multifunctional applications are likely to be of broad interest, but there are a few points that need further clarification Comments: 1.The authors provide multiple examples with patterns on different scales. How the performance/robustness of the technique varies in these scenarios? One factor particularly is the thickness of the thin films. Since the pattern transfer process happens in liquid, the surface tension or environmental perturbation may cause potential fractures, especially for patterns on nano scale. The examples provided in the manuscripts are mostly over 100μm. Can the authors comment on this?
Our response : We appreciate the reviewer's valuable comments on the robustness of the pattern transfer method. The key factor to maintain the integrity of the patterned thin film is to ensure the lateral stiffness of the film. When a thin film is too thin so the lateral stiffness of the thin film is too small, it might cause wrinkles on the transferred film depending on the drying methods [R2,R3]. This could be the drawback of the pattern transfer method but at the same time it could be an engineering opportunity toward the self-assembled surface profile.
On the other hand, the defined inner patterns like Supplementary Fig. 7a and Fig 5d are more susceptible to the surface tension as it has higher aspect ratio. The structures might lose their original shape during transfer and drying process. To prevent this, we introduced physical guide to hold the flexible structures to hold in place during the transfer and drying process. It is quite true that the inner structure become more susceptible to fracture in lower scale during wet transfer. Possible breakthrough is to have more physical guides or to pattern a thin film into the inner structure after transfer like Fig. 3a instead of before transfer.  Table 1 provides the values of parameters (G, contact angle, surface tension) which are calculated using the harmonic mean equation [28][29][30]." 3. The key step of the fabrication technique is the controlled interfacial adhesion in liquid environments. The authors provided the theoretical model that explains the process in detail and matches the experimental results well, which can be used as guidance for other extensible applications. Eq. (1) suggests high wettability and low film-substrate adhesion energy is favorable, but such requirements may also limit the application for this technique, especially for Mode I (2), where the pre-patterned Si membrane will be process more than once and multiple materials are involved. The complexity makes it more difficult to guarantee the delamination happens only in the film-substrate interface. The authors can provide more in-depth discussion on possible extensions to other materials/liquid environments for further applications of the technique.
Our response : We appreciate the reviewer's concern on the possibly difficulty in delamination as the complexity involves. The theoretical model that we have provided assumed the flawless contact between thin film and substrate. In our perspective, the new material and etched-profile on top of thin-membrane surface introduces surface roughness which reduces the interaction between film-substrate rather than increasing the adhesion. Even in the worst case where the new material deposition increases the film-substrate adhesion, surface roughness can be manipulated on certain part of wafer to modulate the self-delamination. As an example of the further applications of this technique, we believe that this technique can be extended into other devicegrade materials over silicon such as an epitaxially grown GaAs membrane/Si or GaAs substrate pair with a AlGaAs sacrificial layer [R1]. However, we think doing further experiments to prove these hypothesis and claim may be out of scope of this manuscript.
Our modification to the manuscript: To clarify this point, we have modified and included the following sentence in Page 2 "The theoretical model is established to understand the transfer mechanism based on self-delamination in the liquid media and provides a quantitative guide to experimental demonstrations in great agreement. It is worthwhile to note that the theoretical model certainly ensures the versatility and robustness of this method to be readily extended for other membrane materials while Si membranes are primarily utilized in this work." 4. The manuscript could have been better if more detailed analysis of this new technique can be provided. The applications of the fabrication methods take relatively large portion of the manuscript while some of them are not directly related to pattern transfer. In the example application provided in Fig. (7), stretch-induced switchable wettability is achieved by utilizing special patterned membranes, which is a well-studied topic. It is not clear how the selfdelamination driven pattern transfer method contributes to the application. What are the unique advantages of this method compared with others in this application?
Our response : We appreciate the reviewer's valuable question on the application stated in Fig  7b. The uniqueness of this application is that it utilizes mechanical metamaterial 'Auxetics' to achieve switchable wettability. Although the stretch-induced switchable wettability is widely studied, the switchable wettability demonstration using the unique auxetic structures like our one seems to be quite rare or none, to our knowledge. Thus, we think that our unique pattern transfer method enabled the auxetic structure for stretch-induced switchable wettability that has not been shown before. Please forgive us if we missed any significant work on wetting control using auxetics and guide us with any related work for our further revising. Finally, this application as an example of Mode 2 uses polyimide thin film rather than silicon membrane. This means the expandability of pattern transfer material choice to the other material over silicon. That is another reason why we have Fig. 7a in this work.