Metal-Assisted Transfer Strategy for Construction ... - ACS Publications

Dec 26, 2018 - Scale bar: (a, d, e) 500 nm; (b, c) 200 nm; (f, g) 5 μm; inset of (g) 200 nm. ACS Nano. Article. DOI: 10.1021/acsnano.8b06623. ACS Nan...
0 downloads 0 Views 986KB Size
Subscriber access provided by University of South Dakota

Article

Metal-Assisted Transfer Strategy for Construction of 2D and 3D Nanostructures on Elastic Substrate Wenjie Liu, Qiushun Zou, Chaoqun Zheng, and Chongjun Jin ACS Nano, Just Accepted Manuscript • DOI: 10.1021/acsnano.8b06623 • Publication Date (Web): 26 Dec 2018 Downloaded from http://pubs.acs.org on December 27, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Metal-Assisted Transfer Strategy for Construction of 2D and 3D Nanostructures on Elastic Substrate Wenjie Liu,†,# Qiushun Zou,†,‡,# Chaoqun Zheng,†,‡ Chongjun Jin†,* †State

Key Laboratory of Optoelectronic Materials and Technologies, School of

Materials Science and Engineering, Sun Yat-sen University, Guangzhou 510275, China ‡School

of Electronics and Information Technology, Sun Yat-sen University,

Guangzhou 510275, China *E-mail: [email protected]

Abstract Comparing with conventional rigid devices, the elastic substrates integrated with functional components offer various advantages, such as flexibility, dynamic tunability, and biocompatibility. However, the reliable formations of 2D nanoparticles, nanogaps, and 3D nanostructures on elastic substrates are still challenging. The conventional transfer method plays an important role in the fabrication of microstructures on elastic substrates, however, it could not fabricate structures with feature size less than a few microns. In this article, we have developed a flexible technique based on the "metal-assisted transfer" strategy. The key concept is to introduce a metal film as an assistant layer between nanostructures and silicon substrates to help the fabrication of nanostructures which can’t be successfully transferred in the conventional transfer method. Various 2D nanostructures, that are difficult to achieve on elastic substrates, could be reliably defined using this approach. The desired gap distances and even sub-10-nm metal gaps between adjacent nanoparticles can be controllably achieved. Moreover, 3D nanostructures can be directly assembled from the pre-strained 2D precursors based on the developed technique. Comparing with the previous reports, our fabrication method contains only a one-step transfer process without selective bonding or second transfer process. Significantly, the 3D nanostructures presented here are two orders of magnitude

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 22

smaller than the state-of-the-art mechanically assembled 3D structures in unit cell size. The proposed method may become a mainstream technology for the nano-optics and ultra-compact optoelectronic devices due to its multi-functionalities and superior advantages in achieving tunable nanoparticles as well as 3D nanostructures. KEYWORDS: elastic substrate, metal transfer, sub-10-nm metal gaps, mechanical assemble, 3D nanostructures Within the past few years, components and devices with superior properties have been demonstrated on elastic substrates,1-10 for example, SERS substrates,7 full-color active tuning devices,8 flat zoom lenses,9 stretchable organic transistors,10 etc. Unfortunately, direct manufacture of nano/microstructures on the elastic substrates is incompatible with the standard nanofabrication techniques, for instance, electron beam lithography (EBL) and focused ion beam (FIB) milling. To solve this problem, alternatives have been explored, such as nanostencil lithography (NSL),11,12 liquid polydimethylsiloxane

(PDMS)

transfer,9,13,14

(PMMA)-Hydrogen silsesquioxane (HSQ)-metal

poly

(methyl

transfer8,9

methacrylate)

methods. Aksu et al.

reported an advanced NSL method to improve the resolution limit.12 However, besides the complex fabrication process of the stencil, the nanostructures with sub-10-nm gap size or triangles with sharp edges are still unattainable. Recently, transfer strategies provide a promising way to peel the patterned films or structures from a temporary substrate, and thus attract the attention of scientists from diverse research backgrounds.8,9,13-18 The principal advantage of the strategies is that, the nanostructures on elastic substrates can be completely identical with the original ones on rigid substrates which are fabricated using the well-developed approach. Liquid PDMS were commonly adopted in the transfer process9,13,14 but at the cost of burying the isolated nanostructures inside of the elastic substrates. PMMA, HSQ, and metal have to be integrated together to accomplish transfer process.8,9 However, HSQ and PMMA could not be subsequently removed in this technique, because the appropriate etching solutions of HSQ/PMMA, such as HF/acetone, can dissolve the PDMS substrate/make it swelling, thus finally destroy the nanostructures. The above two transfer methods are well developed for the fabrication of nanostructures on the elastic substrates, but have limitations in some applications due to the obvious shortcomings: the metal nanostructures are either buried in the substrate or sandwiched between other non-metal materials. These shortcomings block their use in

ACS Paragon Plus Environment

Page 3 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

the applications, such as sensing and SERS, which required the plasmonic "hot spots" exposed to air. More significantly, these methods could not be used to construct tiny metal gaps19 or assemble advanced complex architectures.20-23 To solve these issues, the most simple and elegant way is to directly transfer the metal structures onto the elastic solid, for example, PDMS slab2 or tape.15 Here, we also investigated the characteristics of the conventional transfer process by directly peeling the metal structures from a silicon substrate using a PDMS slab. The detailed description of the fabrication process is illustrated in the figure S1-S3 of the supplementary file. The results indicate that the yields are independent of the shapes of the structures or PDMS peeling directions, however, depends critically on the dimension of the structures. As the structural dimension decreased to about 10 μm, the yields reduced dramatically to 0%. Thus, this conventional transfer method could not fabricate structures with less than a few microns on PDMS substrates. In this article, we have developed a flexible technique based on the "metal-assisted transfer" concept to overcome the limit of the conventional transfer method in the transfer of patterns with feature size less than 10 μm. The key concept is to introduce a metal film as an assistant layer between nanostructures and silicon substrates to help the transfer process. Various nanostructures that are not feasible to achieve on elastic substrates, could be reliably defined using this approach. Furthermore, tuned metal gaps and even sub-10-nm metal gaps between adjacent nanoparticles can be obtained. Especially, we explored the possibility in mechanical assembling. We demonstrate that the 3D nano-microstructures can be directly assembled from the pre-strained 2D precursors based on the developed transfer technique. It is worth mentioning that, the fabricated 3D nanostructures are two orders of magnitude smaller when compared to state-of-the-art mechanically assembled 3D structures in unit cell size, showing great potential in the miniaturization of advanced functional materials.

Results and Discussion We have demonstrated a "metal-assisted transfer" strategy to the construction of 2D nanoparticles as well as 3D nano-microstructures on elastic PDMS substrates. The key concept of this technique is to introduce a metal film as an assistant layer between nanostructures and the silicon substrate to help the transfer process. As we

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

experimentally demonstrated in the supplementary file, metal structures (with metal A) with larger dimensions (such as a layer of metal film) were much easier to be peeled off from the silicon substrate. Thus, taking this advantage, the structures (with material B) that initially formed on the metal film A can be transferred together with the metal film layer to the elastic substrate. As long as the etching solution of the metal A has no influence on the material B and elastic substrate, we could fabricate such material B-based nanostructures on this elastic substrate. For now, we have explored some material schemes that are suitable for this approach. For example, we could fabricate Au or Pt nano/microstructures using Ag-assisted transfer layer, and, Pt or Al nano/microstructures using Au-assisted transfer layer. Furthermore, polymer structures such as PMMA could also be fabricated using Au-assisted transfer layer. In this work, we choose Ag film as an assistant layer to fabricate Au nano/microstructures on the PDMS substrate. The typical fabrication procedure is schematically presented in figure 1a. We take the disk array as an example. The Au disk array with 50 nm thickness was formed on a silicon substrate with 120 nm-thick silver film (named an Ag-coated silicon substrate) through EBL, Au evaporation, and lift-off process. Next, we bonded the Si/Ag/nanostructures on the PDMS slab and then separated the silicon from the PDMS slab. Benefit from the weak adhesion between the whole layer of Ag film and silicon, the Au nanostructures and Ag film clung tightly to the PDMS slab. After etching the Ag layer using phosphorous acid, Au nanostructures were left behind on the PDMS substrate. Figure S4(a-d) in the supplementary information illustrate four photographs of the samples during the practical fabrication process. More importantly, this basic method can be explored to flexibly tune the gap distances (Figure 1b) and directly assemble 3D nanostructures from the 2D precursors (Figure 1c). The details of these two applications will be discussed later.

ACS Paragon Plus Environment

Page 4 of 22

Page 5 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Figure 1. (a) Schematic flow-charts showing the basic procedure of metal-assisted transfer strategy. EBL stands for electron beam lithography. A large dimension of Ag film was performed as an assistant layer to transfer Au nanodisks to the PDMS slab which can’t be achieved based on the conventional transfer method. (b) The simplified schematic diagram to obtain sub-10-nm gaps. The PDMS slab was initially stretched to a certain pre-strain, and bonded to Ag/silicon with Au nanostructures, then released after transfer process. (c) The simplified fabrication process to assemble 3D structures. The PDMS slab was initially stretched biaxially, and bonded to Ag/silicon with Au nanostructures, then released after transfer process.

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Commonly used plasmonic nanostructures fabricated by metal-assisted transfer strategy, such as disk array (a), bowtie (b), disk dimer (c) and square dimer (d), a series of disks with diameter varied from 100 nm to 1000 nm (e), bull's eye structure with 1-μm-wide circular grooves spaced every 2 μm (f), and (g) zoomed-in figure of the ETPM structure. The functional ribbon is 40 μm in length, and the functional nanostructure is 50 nm in gap width, as shown in the inset of (g). Scale bar: (a, d, e) 500 nm; (b, c) 200 nm; (f, g) 5 μm; Inset of (g) 200 nm. Structures with various geometries and dimensions can be facilely prepared on PDMS substrate, as shown from the scanning electron microscopy (SEM, Zeiss AURIGA) images in figure 2 and figure S5. Figure 2a-d demonstrate the commonly used gold plasmonic nanostructures such as disk array (200 nm in diameter), bowtie (150 nm in height), disk dimer (200 nm in diameter) and square dimer (450 nm in length, 150 nm in width). Also, a series of disks with diameter varied from 100 nm to 1000 nm were presented in figure 2e. We also fabricated micro-sized structures to verify the versatility of the proposed method. Figure 2f presents the bull's eye structure with 1-μm-wide circular grooves spaced every 2 μm. The full structure in figure 2f is imaged by a CMOS camera as displayed in figure S5 d, where the maximum circular diameter is 64 μm. It is worth mentioning that, an electronic tunable plasmonic metamaterial (ETPM) structure combined with nano, micro, and

ACS Paragon Plus Environment

Page 6 of 22

Page 7 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

even millimeter size could be perfectly obtained on PDMS substrate. The full CMOS image of the ETPM is shown in figure S5 e, where the plasmonic functional region with 40 μm in length locates between two electrodes. The whole length of the structure is 1 mm, and the functional nanostructures are 50 nm in gap width, as shown in figure 2g and the inset. Such structures with both nano to millimeter-sized portions were only reported on rigid substrates,24 now we are capable of achieving them on the elastic substrates. Benefit from the elasticity of the substrate, a larger reversibly tuning of the frequency range and tuning depth could be expected.

Figure 3. (a) Parameters of the bowtie structure (b-f) SEM images of bowtie nanoantennas on the released PDMS substrate at the pre-strain of 0% (b), 20% (c), 40% (d), 50% (e) and 60% (f). (g) The bowtie gap distances as a function of pre-strains. The solid lines are the linear fitted curve and the error bars represent the standard deviations of 20 bowties measurements. With the developed approach, tuned metal gaps and even sub-10-nm metal gaps between adjacent nanoparticles can be obtained. As is known that, sub-10-nm gaps between two closely spaced metallic nanostructures enable strong localized electromagnetic field enhancement, thus is highly desirable in various research areas, such as SERS, single photon emitting and quantum tunneling, etc. However, the reliable formation of metallic nanogaps, especially on an elastic substrate, is still challenging. In this part, we will demonstrate that the desired gap distance can be easily obtained just by exerting the PDMS slab to a certain pre-strain before the transfer process. Here we take a bowtie as an example to illustrate the tuning process.

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 22

The simplified fabrication procedure is demonstrated in figure 1b and the triangle parameters of the bowtie structure are shown in figure 3a. The PDMS slab was fastened to a home-build tool (shown in our previous report in Ref. [25]) and exerted to a certain pre-strain. Then, the bowtie was transferred to the pre-strained PDMS slab according to the proposed method and followed by pre-strain releasing. The percentage of the pre-strain is defined as R=(L-L0)/L0,

where L and L0 are the

lengths of the PDMS after and before stretching. The applied pre-strains were chosen to be 0%, 20%, 40%, 50%, 60%, and the measured gap distance were 108, 67, 34, 19, and sub-10 nm, respectively. Figure 3b-f show the SEM images of the bowtie structure on the PDMS substrate after releasing the pre-strains to 0%. A linear relationship between the gap distances and the pre-strain was observed, as summarized in figure 3g. This means that, for these nanoparticles, the gap distance in nanoscale region can be linearly controlled by manipulating the PDMS length in macroscopic scale. Using this strategy, structures with different size and geometry can be obtained on the PDMS substrate with a sub-10 nm and zero gap distance, as further demonstrated in figure S6. For the control of the nanogap distance, based on the fitted curve in figure 3g, the 70 μm change of the PDMS total length results in 1 nm change of the nanogap distance. In our experimental measurement, the precision of the PDMS length is 0.1 mm, we estimated that, in our experimental condition, the distance of the nanogap can be tuned accurately in 1.5 nm. However, because the pattern contains tens of bowties, the original gaps of the bowties are not the same due to the fabrication tolerance. Therefore, in our final experimental results of figure 3g, the error bar is ±5 nm.

In figure 4, dark-field scattering spectroscopy was adopted to determine the optical response of one gold bowtie antenna in a backscattering geometry. Structure parameters of the bowtie are the same as that in figure 3a. The gold bowtie antenna with an initial gap distance of 108 nm was fabricated on a pre-stretched PDMS substrate with 60% pre-strain. Dark-field spectra were recorded from one bowtie as gradually release the strain from 60% to 40%, 20%, 10% and 0%. In this case, the

ACS Paragon Plus Environment

Page 9 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

trend of the measured gap distance is reversed compared to the previous measurement (figure 3) at fullly released condition. As we demonstrated in figure 3, a linear relationship was proved between the gap distances after PDMS release and the pre-strain, thus we can deduce that in this PDMS releasing process, the gap distance is also linearly dependent on the gradually decreased strain. Therefore, given the initial and final gap distance of 108 nm and 5 nm, we can calculate that the gap distances are 73, 39, 22 nm at the strain of 40%, 20%, and 10%, respectively. Figure 4a shows the evolution of the measured normalized back-scattering spectra under the polarization of illumination parallel to the long axis of the bowtie as reducing the strain. A clear red-shift ranging from 810 nm to 927 nm was observed as reducing the strain from 60% to 0%. The peak wavelength and strain as a function of estimated gap distance were summarized in figure 4c, where an exponential redshift trend was clearly observed. At the strain of 60% and 40%, one main scattering peak was presented and as we further reduced the strain to 20%, the original resonance mode split into two modes and moved remarkably toward the longer wavelength. The two resonance modes changed with decreasing the gap distance can be well explained by the hybridization model, which is similar as the previous reports.12,26 Using the calculated gap distance, we simulated the corresponding scattering spectra with the incident light polarization parallel to the long axis of the bowtie (figure 4b). The spectra shapes and movement trends are similar to the experimental results, confirming the reliability of the proposed method.

Figure 4. (a) Normalized dark-field back-scattering spectra of single bowtie nanoantenna as decreasing the strain. (b) Numerically simulated back-scattering spectra of the bowtie nanoantenna at different calculated gap distances. (c) The

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

measured scattering peak wavelength and the applied strain as a function of calculated gap distance. The error bars represent the standard deviations of the measured scattering peak wavelength of 5 bowties. The red curve represents exponential fits to the data.

Figure 5. (a) Top-view SEM image of a bowtie array on Si/Ag substrate before transferring to a PDMS slab. (b) Side-view SEM image of the bowtie array after releasing the pre-strain (scale bar: 200 nm). (c) SERS spectra of the MGITC-modified bowtie arrays under the pre-strain and released conditions. For reference, the Raman spectrum of the MGITC-modified PDMS was also recorded. The excitation laser wavelength is 785 nm. To demonstrate the utility of this method, we fabricated a bowtie array for SERS detection. Raman scattering from the malachite green isothiocyanate (MGITC) molecules was examined on the two samples: pre-stretched (55% pre-strain) PDMS substrate with bowtie array and bare PDMS substrate for reference. The bowtie array before (corresponding to a gap distance of 108 nm on a silicon substrate shown in figure 5a) and after PDMS releasing (corresponding to a gap distance of about 16 nm, as shown in figure 5b) were both measured. The samples were dipped into MGITC aqueous solution (10-5 M) for 4 hours, then washed with deionized water and dried with pure nitrogen gas. The SERS measurements were carried out on a Raman spectrometer (Renishaw inVia) with an excitation laser wavelength of 785 nm. As shown in figure 5c, comparing with the bare PDMS and bowtie array under pre-strain, the released bowtie array presents an obvious enhancement on Raman intensity. We performed the simulations using finite-difference time-domain (FDTD) method on the local electric field intensity distribution of the bowtie arrays (figure S7). By comparing the electric field intensity before (figure S7a) and after (figure S7b) releasing the pre-strain, we can confirm that such an enhancement on Raman intensity

ACS Paragon Plus Environment

Page 10 of 22

Page 11 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

arises mainly from the electric field enhancement between the reduced bowtie gaps.

Another important application of the proposed method is to mechanically assemble 3D nano/microstructures directly from the transferred 2D precursors. Formation of 3D nano/microstructures remains broad and significant interest in advanced materials due to their wide range of applications, such as nano/micro-electromechanical systems (NEMS/MEMS), metamaterials, biomedical devices, plasmonics, optoelectronic components, etc. The demand for such advanced materials and devices has fueled the research on upgraded fabrication technologies, including strain-induced bending or folding,27-31,32-35 FIB folding,36,37 improved top-down lithography,38 etc. Although methods have allowed relatively reliable formation of 3D structures, they are limited in terms of fabrication simplicity or geometric diversity. Recently, a series of reports have described a cogent technique for mechanically assembling complex 3D microstructures with silicon, metal, polymer, or inorganic semiconductors.21-23,

39,40

Diverse microstructures were successfully assembled using this method. However, the dimensions of the fabricated structures are still above hundreds of micrometers and the fabrication process is relatively complex. Here, inspired by this mechanical buckling approach, we now can achieve sub-10-micro-sized or even sub-micro-sized 3D architectures on the elastic substrate through our “metal-assisted transfer” technique. Comparing with the previous reports, our fabrication method only contains a one-step transfer process without selective bonding or second transfer process. More significantly, the 3D nanostructures presented here are two orders of magnitude smaller than the state-of-the-art mechanically assembled 3D structures in unit cell size. Figure 1c briefly illustrated the fabrication procedure. Specially designed 2D precursor was first fabricated on the Ag-coated silicon substrate, and then a PDMS slab with pre-strain was bonded to the Si/Ag/nanostructures. Unlike described above, in this case, a constant external pressure of 3.1 kPa was applied to bond the metal nanostructures with PDMS slab more tightly, thus were able to tolerate a stronger compressive force during the metal deformation process. Next, the Si removal and Ag

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

etching procedure were followed. The PDMS slab is uniaxially or biaxially pre-strained thereby to assemble structures with one (figure 6) or two directions (figure 7) deformation. The dimensions of planar precursors of figure 6a-d are shown in figure S8. The precursors are various planar gold chains with uniform thickness of 50 nm. The left and middle columns in figure 6 show the SEM images of the whole structures and the zoomed-in images marked by the dashed box, respectively. The right column presents the simulated structural geometries and stress distribution of a unit cell using finite element method (FEM). Each part of the precursor experiences adhesion force that brought by the PDMS. And with releasing the pre-strain, the thickness of the PDMS increases along z-axis, thus provides an upward trigger force on the precursors, meanwhile, compressive forces are induced and accumulated along x-axis. A threshold reaches when the adhesion force is equal to the upward trigger force. Above the threshold, the induced forces could lift the most weakly bonded structures out of the PDMS substrate, and simultaneously induce out-of-plane spatial deformation. In figure 6a, the "golden bridges" including six unit cells were formed by the 2D precursors with squares (3 μm×3 μm) and lines (10 μm in length, 200 nm in width), as shown in figure S8a. The squares, with much larger areas comparing with the nano-lines, were expected to tolerate much stronger compressive force during the deformation process, thus could serve as fasteners. The final golden bridges were formed after a pre-strain of 30% was released. As shown in the simulated stress distribution image, the stresses are mainly distributed in the bridge (nano-line), and the squares suffer from much less stress as expected, which confirms that this 3D structure is very stable. Figure 6(b-c) present 3D square dimers assembled from the precursors with nanoscale intervals (figure S8(b-c)). The pre-strains for figure 6b and figure 6c were 40% and 50%, respectively. In these structures, the small squares were designed as the connecting piece as well as the fastener to support the adjacent lifted structures. The stress distribution images exhibit that the stresses are mainly distributed at the bending parts, however, the top of the square dimers are under low stress which is different from the golden bridge. The rectangle precursors of figure 6b

ACS Paragon Plus Environment

Page 12 of 22

Page 13 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

possess a higher length-to-thickness ratio (2 μm : 50 nm) comparing with that of figure 6c (1 μm : 50 nm), thus the assembled 3D structures present obvious bending morphology. 3D period triangle dimers (figure 6d, pre-strain of 40%) can also be flexibly assembled, even though only tip contacts are formed at the top. The tip contacts of the triangle dimers suffer from higher proportion of stresses comparing with the top of the square dimers, which is due to the much smaller contact area. We can learn some underlying mechanics of mechanical assembling from previous reports.21,40 Although the configuration and size of the structures are different, we believe some of the underlying mechanics are similar.

Figure 6. Mechanically assembled 3D microstructures after releasing the uniaxial deformation of the PDMS substrate. (a) Golden bridges. (b-c) 3D square dimers. (d) 3D triangle dimers. Each figure contains an SEM image of the whole structure (left column), the corresponding zoomed-in image marked by red dashed box (middle column), and simulated structure with stress distribution (right column). The scale bars in the first column and the second column are 5 µm and 1µm. The pre-strains for (a), (b), (c), and (d) were 30%, 40%, 50% and 40%, respectively.

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7. SEM images of the mechanically assembled structures after releasing the biaxial deformation of the PDMS substrate. (a) 3D bridge array, (b) tent array, (c) microfluidic channels, and (d) pyramid array. Each figure contains an SEM image of the structure with the pre-strain of 0% (left column), the pre-strain of 40% (middle column), and the corresponding zoomed-in image marked by red dashed box (right column). With the biaxial pre-stretch of PDMS substrates, 3D structures such as bridge array (figure 7a), tent array (b), microfluidic channels (c), and pyramid array (d) could be flexibly fabricated after releasing the pre-strains. In figure 7a-d, each figure contains an SEM image of the structure with the pre-strain of 0% (left column), the pre-strain of 40% (middle column), and the corresponding zoomed-in image marked by red dashed box (right column). Here we take the tent array (figure 7b) as an example to

ACS Paragon Plus Environment

Page 14 of 22

Page 15 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

evaluate the induced force. At the threshold, the induced force is equal to the adhesion force between PDMS and Au. We could assume the work of adhesion is the same for the squares and nano-lines since their adhesions are both between Au and PDMS. This means that, the work of adhesions for the weak (in the middle of nano-line) and strong bonding regions (square) are the same. As we learned from previous paper41 that, due to the viscoelastic behavior of the PDMS, the work of adhesion between PDMS and metal varies with separation speed. In our situation, the separation speed is very slow (