Strategies for Preparing Graphene Liquid Cells for Transmission

6 days ago - lowering the liquid.34 (b) Scanning electron microscopy image of two graphene layers enclosing liquid on a holey carbon film. Reprinted w...
2 downloads 0 Views 5MB Size
Download PDF
Mini Review Cite This: Nano Lett. XXXX, XXX, XXX−XXX

pubs.acs.org/NanoLett

Strategies for Preparing Graphene Liquid Cells for Transmission Electron Microscopy Martin Textor†,§ and Niels de Jonge*,†,‡ †

INM, Leibniz Institute for New Materials, D-66123 Saarbrücken, Germany Department of Physics, Saarland University, D-66123 Saarbrücken, Germany

‡

ABSTRACT: A graphene liquid cell for transmission electron microscopy (TEM) uses one or two graphene sheets to separate the liquid from the vacuum in the microscope. In principle, graphene is an excellent material for such an application because it allows the highest possible spatial resolution, provides a flexible covering foil, and effectively protects the liquid from evaporating. Examples in open literature have demonstrated atomic-resolution TEM using small liquid pockets and the coverage of whole biological cells with graphene sheets. A total of three different basic types of liquid cells are discerned: (i) one graphene sheet is used to cover a liquid sample supported by a thin membrane of another material (for example, silicon nitride, SiN), (ii) two graphene sheets pressed together leaving liquid pockets with graphene at both sides, and (iii) a spacer material with liquid pockets covered at both sides by graphene. A total of four different process flows are available for liquid cell assembly, but there is not yet a consensus on the best routes, and a number of variations exist. The key step is the transfer of graphene to a liquid sample, which is complicated by practical issues that arise from imperfections in the graphene sheets, such as cracks. This review provides an overview of these different approaches to assembling graphene liquid cells and discusses the main obstacles and ideas to overcome them with the prospect of developing the nanoscale technology needed for graphene liquid cells so that they become available on a routine basis for electron microscopy in liquid. It also provides guidance in selecting the appropriate type of graphene liquid cell and the best assembly method for a specific experiment. KEYWORDS: Liquid cell, graphene, liquid-phase electron microscopy, TEM, STEM, high resolution

N

membrane for regular atomic-resolution TEM of samples in vacuum.10 Importantly, it has been shown that graphene reduces beam damage by radical scavenging,11 which renders it well-suited for the enclosure of liquid samples as these are particularly sensitive to beam-induced changes.12 Graphene enclosures for TEM or scanning TEM (STEM) have been utilized to study samples for the materials sciences, sometimes achieving atomic resolution, such as nanoconfined water,13 metallic, magnetic, or semiconducting nanoparticles,14,15 and to visualize nanomaterial growth,5 lithiation on silicon nanoparticle electrodes,16 and crystallization processes.17 Studies involving biological samples in a liquid environment included ferritin in saline solution,18 DNA-Au-nanoconjugates in buffer,11,19 and fixated20 or immobilized cells.21 Graphene enclosures have also been used for scanning electron microscopy (SEM) of aqueous suspensions of gold nanoparticles,22 whereby the thin graphene layer is advantageous for

anoscale studies of samples in liquid have been possible on a routine basis using liquid-phase electron microscopy for about a decade.1,2 Most published data are based on a liquid cell with electron transparent windows of SiN providing a reliable liquid enclosure. Materials of high atomic number (Z) are imaged with nanometer resolution, and dynamic processes are also studied with these systems. However, for imaging lowZ materials, the combined thickness of the liquid and the windows is typically thicker than one mean free path length of electron scattering at the electron beam energy used so that phase-contrast TEM cannot be applied and the obtained spatial resolution is limited to the 10 nm range.3 There is, thus, a need for assembling liquid cells with ultrathin windows and containing a liquid column of up to 100 nm thickness, which has been shown in some exceptional studies to provide nanometer resolution.4 Graphene is a promising window material for a TEM liquid cell5 because it is a two-dimensional material consisting of a monolayer or a few layers of a hexagonal carbon lattice. Owing to its high thermal and electrical conductivity,6 mechanical strength,7,8 and chemical inertness,9 graphene has already been established as a support © XXXX American Chemical Society

Received: April 5, 2018 Revised: May 20, 2018

A

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters

Figure 1. Basic types of graphene liquid cells. For type A, a sample is immobilized on a thin film and covered by a sheet of graphene. An example is the coverage of a mammalian cell for transmission electron microscopy (top middle image adapted with permission from ref 24. Copyright 2017 American Chemical Society). The membrane may contain liquid pockets. Top right image reprinted with permission from ref 27. Copyright 2016 John Wiley and Sons. The liquid sample is enclosed on both sides by graphene in liquid cell type B, and supported by a perforated thin membrane. This type has been used to study growth processes of nanoparticles. The liquid cell is typically placed on a support grid. Middle row middle image reprinted with permission from ref 5. Copyright 2012 AAAS. Also, this type of liquid cell was used to image cells. Middle row right image adapted with permission from ref 20. Copyright 2015 American Chemical Society. Another approach, type C, is to enclose a perforated thin membrane at both sides with graphene. The holes in the membrane serve as liquid pockets. Bottom image reprinted with permission from ref 28. Copyright 2018 American Chemical Society.

Basic Types of Graphene Liquid Cells. Depending on the specific application, different types of graphene liquid cells are used. The following three basic types are distinguished (Figure 1). For type A, the liquid sample is placed on a thin sample support film or membrane of, for example, carbon or silicon nitride (SiN) mounted on a support frame (for example, a microchip or a standard 3 mm grid for TEM); and the sample is then coated with a graphene layer.24 The thin film may have wells in it providing liquid pockets once the graphene is placed on the film.27 For type B, the liquid sample is enclosed at both sides by graphene, and the assembly is supported by a perforated thin film (for example, a lacey carbon film or a holey carbon film with well-defined periodic perforations).5,18,20 The liquid cell is mounted on a support frame (for example, a standard 3 mm grid). For type C, a different approach is to enclose the liquid sample in wells in a perforated thin film and enclose this film at both sides with graphene.28 We review the different protocols to obtain graphene liquid cells of these three basic types. Most published literature involves liquid cells enclosed on both sides with graphene layers of type B even though type A is easier to handle because it requires the preparation of one graphene window only. In many experiments, one would hardly notice the decrease in spatial resolution due to the presence of a thin membrane at one side of the flow cell. For example, a SiN membrane of 50 nm thickness is a sturdy sample support,29 and a spatial resolution of 1.1 nm would still be achieved for STEM of gold nanoparticles in a 50 nm thick water layer using an electron dose of 100 e− Å2 and an electron energy of 200 keV.3 Even though the resolution would improve to 0.7 nm without SiN, this improvement would not be relevant in many experiments because 1.1 nm is already sufficient to examine, for example, the dynamic processes of the nanoparticles of high atomic number (Z). However, the thickness and the orientation (whether the support is located at the top or the bottom of the liquid cell)

elemental analysis. Others studied graphene-coated COS7 cells23 and cancer cells24 with SEM. However, graphene flakes of a typical size of ∼1 mm2 as needed to cover a sample tend to be fragile during handling,25,26 which has hampered the exploitation of the advantageous properties of graphene for TEM. Furthermore, graphene is typically provided on supporting material to increase the rigidity of the graphene sheet for handling, and this material needs to be removed to obtain the bare graphene. Third, graphene sheets are not always wetting so that a water layer tends to “escape” the sample when attempting to cover the sample. Fourthly, liquid-handling procedures are required to enclose the sample, and the selected procedure needs to be compatible with the sample intended for examination. Here, we review a range of possible strategies for graphene transfer from its support material to the specimen as needed to assemble a graphene liquid cell applicable to both samples from materials science and biology. Differences in the applicability of the transfer strategies depend on the type of sample. In particular, when handling delicate samples such as most biological specimens, several experimental requirements need to be considered. (1) The liquid in which the specimen is maintained cannot be exchanged because the structural integrity of the sample depends on the specific properties of the liquid. (2) The specimen is often available in small quantities, such that the final step of graphene transfer cannot involve larger amounts of liquids than a few microliters to milliliters. (3) The temperature of the specimen needs to be constant within a typical range (for example, from room temperature to 37 °C), preventing any heating steps for the preparation of liquid cells. We will not consider other types of thin membranes even though a wealth of possible materials exists of which some have been used to make liquid enclosures for TEM such as graphene oxide applied from suspension for encasing bacteria.21 B

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters

Figure 2. Transferring graphene to a liquid sample. (a) Different methods for graphene transfer to a liquid sample. (i) Graphene supported by a perforated thin film is lowered on a second graphene sheet containing the sample.19 (ii) A droplet is placed on a stack of two graphene sheets on a perforated thin film, and the liquid is pulled between the graphene sheets.5 (iii) A thin membrane containing the sample is used to “scoop up” a graphene sheet at a liquid surface.24 (iv) A loop is used to manipulate a graphene sheet on a liquid droplet.33 (v) Manual handling is avoided by lowering the liquid.34 (b) Scanning electron microscopy image of two graphene layers enclosing liquid on a holey carbon film. Reprinted with permission from ref 5. Copyright 2012 AAAS. (c) Beaker with water placed below a binocular microscope with the sample held by tweezers ready for “scooping up” the graphene. (d) Microchip with a graphene-covered sample in the process of drying. (e) Top view of liquid droplet with floating graphene in a loop. (f) Transfer of graphene on a liquid droplet in a loop to sample on 3 mm grid. (g) Top view of a graphene-transfer system of custom design. (h) Magnification of the transfer system’s cylinder with a centered graphene sheet supported by poly methyl-methacrylate (PMMA) floating on water.

membrane is removed. A disadvantage is that the perforated membranes and their supporting frames are usually not entirely flat, and so it may be difficult to obtain liquid cells in a reproducible way (Figure 2b). Typically, liquid pockets of varying submicron sizes are found at random positions. For method ii, a stack of graphene sheets on a perforated thin membrane is first assembled in dry state, and a droplet is placed on the top graphene sheet. The liquid is then believed to pull into small pockets between the graphene layers.5 The wetting properties of the graphene for the liquid in use is then a critical factor. For method iii, a substrate (for example, a microchip or a 3 mm grid) containing a thin film on which a sample is immobilized is handled manually with tweezers and brought into contact with graphene floating on liquid from below (“scooped up”)24 (Figure 2c,d). For a small graphene sheet, it is sometimes difficult to correctly center it on the substrate this way. For method iv, better centering of graphene sheets can be accomplished by handling the graphene on a liquid droplet held by a metal18 or nylon loop33 (Figure 2e,f). For method v, manual interaction with the graphene and the substrates during transfer can be omitted entirely by using liquid-handling equipment34 (Figure 2g,h). Starting with a graphene sheet floating on liquid, contact between a floating graphene sheet and a submerged substrate is obtained by lowering the level of the transfer liquid.

needs to be optimized for the particular experiment. For STEM, the SiN membrane should be at the bottom, while it is optimal in TEM to have it at the top. For imaging low-Z materials, the SiN thickness needs to be minimized.3,4 Graphene-Transfer Methods. The most-delicate steps in the preparation of a graphene liquid cell involve the transfer of graphene. The methods for the transfer of graphene to flat and dry sample support substrates have been reviewed extensively.30,31 Most approaches described so far involve a dry contact between the graphene and the substrate, and the processing steps typically involve the dissolving of a support substrate (for example, copper used for graphene growth) followed by steps of drying, stamping, peeling, and annealing at elevated temperatures. For a liquid sample, certain steps cannot be used and the experimental protocol needs additional steps for covering the liquid sample with graphene. For example, a liquid cell was assembled by sandwiching the sample between two perforated thin carbon membranes on regular TEM grids that were coated with graphene and dried in a prior step.5,32 A total of five basic methods exist for transferring graphene onto a liquid sample (Figure 2a). In transfer method i, graphene is first mounted on a perforated thin membrane, and this is used to transfer the graphene.19,32 A liquid droplet is placed on a first graphene sheet, and a second sheet is then placed on top to obtain a liquid cell. After the graphene sheets adhere, the top perforated C

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters These five transfer methods differ substantially in the volume of transfer liquid required. The smallest amount of transfer liquid is required when using droplets (methods i and ii) or a loop (methods i and iv) because only a fraction of a microliter is needed to cover the surface of the used substrates or to support a graphene sheet of a few square millimeters in the case of a loop. Note that these methods, except method ii, are also used for transferring graphene from one liquid surface to another as needed in some steps of the involved protocols prior to assembling the liquid cell. Assembling Graphene Liquid Cells. The enclosure of a liquid sample involves several steps of graphene transfer and sample handling, whereby the aforementioned basic five methods of graphene transfer are used. The sequence of such a protocol depends on the type of sample as well as on the type of graphene liquid cell desired (Figure 1) and the type of support to which the graphene is attached. The preparation of a graphene liquid cell typically involves one of four different process flows (Figure 3). Each process flow involves two groups of processing steps. In the first group of steps, a precursor is made. The liquid cell is then assembled from the precursor in the second group of steps. The following description reflects the smallest number of steps for each precursor type. Sometimes additional steps are used for practical reasons that

involve exchanging one precursor type into another (see Figure 4).

Figure 4. Schematic alternative possibilities for obtaining the precursors needed for liquid cell assembly. Precursor I can be used to produce precursor II by “scooping up” the graphene on the liquid surface with a perforated membrane mounted, for example, on a 3 mm grid. Similarly, precursor III can be used to obtain precursor II via an intermediate wash in acetone to remove PMMA. Precursor III is also used as basis for producing precursor IV via transfer of graphene on a solid support in a transfer solution.24 Precursor I is obtained in a subsequent step involving the transfer of the graphene to a liquid surface. PMMA-coated graphene can be obtained on a cellulose sheet as a starting point of the process flow, from which precursor III is obtained via a water-bath transfer.

The basis of producing graphene liquid cells is most frequently the growth of graphene on a metal foil by chemical vapor deposition (CVD) on a metal catalyst such as platinum, nickel, or copper.36 Graphene grown on a copper foil is also commercially available from several sources. Process flows I−III use graphene on a metal foil (Figure 3). Process Flow I. Only a few steps are involved in process flow I (Figure 3).20 The metal is etched away in an etching bath, resulting in a graphene sheet floating on the etching liquid.30,35,37 The etchant is usually composed of strong acids and harsh chemicals such as sodium persulfate,32 copper sulfate, ferric chloride, or ammonium persulfate, and it should be avoided that these chemicals become trapped between the graphene and the substrate upon transfer. Second, it should be avoided that residues of the solid material used for graphene growth remain on the graphene. Although the release can be carried out on a large volume of etchant or solvent, a contamination with residual material cannot be fully excluded. In the case of sensitive experiments, this workflow may include repeated etching steps followed by dilution steps of the etchant solution or solvent until the concentration of impurities becomes negligible. The graphene is transferred to a water bath to obtain precursor I. Transfer can be done by transfer method iii, whereby the substrate is a glass slide20 or using a loop to hold a liquid droplet as in transfer method iv.18 This procedure, however, is highly delicate as the graphene is easily

Figure 3. Process flows for preparing graphene liquid cells of types A− C. Process flows I−III start with graphene on a metal sheet. In process flow I, a graphene sheet floating on a water surface is obtained after etching away the metal foil,18,20 referred to as precursor I. Next, a sample in liquid is placed on a thin membrane. A liquid cell of type A is then obtained via “scooping up” a graphene sheet at a liquid surface from precursor I.24 This process flow can also be combined with a sample placed on precursor II to obtain liquid cell type B.20 In process flow II, the graphene on metal foil is mounted in dry state on a perforated thin supporting membrane, and the metal is then etched away;5,32,35 the resulting structure is precursor II. A pair of ways exist by which to obtain a liquid cell of type B 19 5. Process flow III involves the coating of the graphene on metal with a layer of PMMA and etching the metal away to obtain precursor III.24,27 Liquid cell A is finally obtained after sample loading and the subsequent removal of the PMMA.23,27 Process flow IV starts from a block of graphene material. This process flow uses dry processing to remove a sheet of graphene from a block of solid graphene to obtain precursor IV,28 which is graphene on a solid piece of support material, and the subsequent steps lead to liquid cell C.28 D

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters

Process Flow II. The metal foil (for example, a copper foil) with graphene is directly placed on a sample support frame in process flow II using, for example, a 3 mm TEM grid made of gold. The copper is then etched away, resulting in a support substrate with graphene in the etching liquid.30,32,35 The materials of the TEM grid should obviously be chosen such that it does not etch. The sample is then washed and dried. The result is precursor of type II consisting of graphene on a sample support frame. The precursor of type II is also for sale by several suppliers. The simplest procedure is to use precursor II to obtain a liquid cell of type B, for which two identical precursors are needed. A droplet containing the sample is placed on the graphene sheet of the lower precursor, and the second precursor II is then placed upside-down on the liquid and pressed down.19 When the graphene contacts the membrane of the sample substrate, van der Waals forces lead to a maximization of the contact area between both membranes, thereby trapping small amounts of liquid into micrometer-sized pockets. Finally, the top perforated membrane is removed by carefully lifting the supporting frame (for example, a 3 mm grid), thereby leaving the graphene behind. This last step, however, is not always reproducible and introduces additional strain. Alternatively, the frame is folded such that the liquid droplet becomes enclosed between two graphene sheets.40 Because the substrates are in a dry state at some point during the process, this process flow allows for pretreatment by plasma cleaning to render the graphene hydrophilic41 or for annealing at high temperature.42 A drawback is that two graphenecovered perforated thin membranes, such as a holey carbon foil, placed onto each other do not necessarily make full contact if they are not perfectly flat, which is rarely the case. As a result, this approach may yield less areas in which the graphene layers encase the liquid. A variation of this procedure is to first bring the graphene sheets into contact and then to place a droplet on the stack (right branch in process flow II in Figure 3). The droplet is then supposed to be pulled into small liquid pockets between the graphene sheets.5 Process Flow III. A sacrificial support layer is applied over the graphene for process flow III (Figure 3). Typically, poly methyl-methacrylate (PMMA) is applied to the graphene on metal foil by spin-coating. The reason for applying PMMA is that it allows easier handling and reduces mechanical strain of the graphene when transferring to liquid surfaces as compared to graphene alone. The copper substrate is then etched away.30 Next, the graphene on PMMA is transferred to a water surface so that one obtains precursor type III. The sample is then immobilized on a thin membrane.23 The thin membrane may contain liquid pockets obtained, for example, by etching cylindrical holes into a SiN membrane.27 Graphene covered by PMMA at the liquid surface is scooped up with the sample substrate, and the sample is then dried. Hereafter, the PMMA is removed in acetone to obtain a liquid cell of type A. The advantage of this process flow is that graphene handling has become much easier ever since. However, success is uncertain in practice because the liquid sample might get in contact with the acetone via cracks in the graphene sheets. Also, there is a risk of substrate contamination with residual PMMA if polymer removal is insufficient. Improvements of the PMMA removal step include prevention of graphene wrinkles by drop-casting an additional layer of PMMA before removal,43 using acetic acid instead of acetone

lost, and it is difficult to verify the complete removal of the copper during etching. A graphene liquid cell of type A is then obtained as follows. Starting from precursor I, a sample is first immobilized on a thin membrane (for example, a carbon foil on a 3 mm grid or a SiN membrane supported by a silicon microchip). This sample substrate is next immersed in the water bath and positioned below the graphene sheet swimming at the water surface. The sample substrate is finally lifted up carefully such that the sample becomes covered with graphene, and the liquid cell type A is obtained.20,24 This procedure is known in the biological electron microscopy community for the application of thin carbon foils on thin sections.38 It is also possible to combine process flow I with precursor II (second branch in flow I of Figure 3), a graphene sheet on a perforated membrane obtained in process flow II. The sample is loaded on the graphene sheet that is supported on a perforated substrate, and this sheet is then used to “scoop up” the floating graphene sheet. The sample can consist, for example, of cell cultures on a graphene-coated TEM grid.20 Workflow I involves immobilizing the sample on the substrate before the graphene transfer. Because the transfer is performed in liquid, this step requires sufficiently strong binding of the specimen to the substrate, for example, by electrostatic or covalent interactions in the case of biomolecules or by an immobilization or fixation as in the case of cells cultured directly on the substrate.20 Furthermore, it is required that the transfer of the sample in liquid phase is possible into a relatively large volume of transfer liquid, which will be pure water typically. This would work for fixed cells, or immobilized nanoparticles but not necessarily for protein complexes that may unfold if the right saline conditions are not present. The transfer can, in principle, be accomplished in saline solutions as well but then salt crystals are likely to appear when drying the outside of the liquid cell that may prevent TEM. Alternatively, a transfer is possible by using a transfer liquid that already contains the sample (for example, for the encapsulation of a sodium persulfate solution),37 although remainders of the sample will then be found at both of the dry sides of the liquid cell as well. If a particular experiment requires low sample volumes, then the graphene transfer can also be accomplished by using small liquid droplets via method iv using a loop (Figure 2). A second benefit of using a loop is that it is easier in practice to position the graphene on the sample using a small droplet because its confined geometry substantially reduces liquid convection. Some groups have used a variation of the above process steps to handle a sample that was incompatible with transfer into a large transfer volume. The graphene cell was first assembled using water or saline solution but without a sample. In the next step, a small droplet of sample-containing liquid was pipetted onto the substrate, whereby the sample was transferred to the enclosed liquid via diffusion. This has been shown to work for a suspension of influenza viruses in buffer in a liquid cell.20 However, one should question how much of the sample remained at the dry side of the liquid cell. Another way was to flip the membrane containing the sample so that the liquid droplet faced downward.18,39 The membrane was then lowered onto the floating graphene, until the sample touched the graphene. Finally, the graphene was picked up from the liquid. Here, the question arises how much mixture of the sample liquid in the water bath took place considering cracks in the graphene and the limited size of graphene flakes. E

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters

Figure 5. Examples of results obtained with the three different types of graphene liquid cells. (a) Liquid cell type B was used to study Pt nanocrystal growth at atomic resolution. The image series at different time points show the shape evolution of the Pt nanocrystal by straightening of the twin boundary (red dotted line) and evolution toward a hexagonal shape. reprinted with permission from ref 5. Copyright 2012 AAAS. (b) A tunneling nanotube connecting two cancer cells imaged using liquid cell type A.24 (c) Quantum-dot protein labels reveal the positions of individual HER2 receptors imaged in the region indicated with the dashed square in panel b. Panels b and c are adapted with permission from ref 24. Copyright 2017 American Chemical Society. Pairs of labels are indicated within dashed circles. (d) Tracking of tungsten nanocrystals in water enclosed in liquid cell type C. Trajectories of individual nanocrystals are overlaid with time (in seconds) represented by a color chart in which blue is t = 0 s and each color block is a 60 s increment; the movement paths for two individual nanocrystals are shown. Reprinted with permission from ref 28. Copyright 2018 American Chemical Society.

for increased cleanliness,44 and holding the substrate in place at the acetone surface instead of soaking the substrate in acetone or exposing it to acetone vapor to avoid the tearing of a holey carbon support film.31 Process Flow IV. Liquid cell type C is prepared from a graphene sheet gained via a dry transfer technique from a solid block of graphene containing a few layers of graphene at the surface (Figure 3).28 The process flow is different from I−III because it does not involve any etching and avoids transfer of graphene via a liquid bath. A graphene sheet is first transferred to a solid support substrate (for example, silicon oxide using a dry transfer process).28,45 Other types of solid supports are sodium chloride crystals or mica surfaces. The resulting structure is precursor type IV used to obtain liquid cells of type C. The sample is enclosed in liquid pockets in a perforated thin spacer (for example, consisting of a thin membrane of hexagonal boron nitride (hBN) with etched holes).28 A sheet of graphene coated with PMMA is prepared for sample coverage (right branch in Figure 3). Then, a liquid droplet with the sample is placed on the substrate, and the liquid cell is closed by positioning the graphene−PMMA sheet on top. The liquid cell is subsequently separated from the solid support and transferred to a perforated membrane mounted on a support (for example, a holey carbon film on a 3 mm TEM grid). The PMMA is removed in the final step. The benefit of hBN is a better adherence of the graphene to the substrate compared to using a SiN membrane, such that closed liquid pockets of defined dimensions are obtained in a reproducible manner. Nevertheless, some liquid pockets may be open due to cracks in the graphene. Alternative Ways to Obtain Precursors. Several precursors can be exchanged into other precursor types as needed for graphene liquid cell assembly (Figure 4). These exchange steps are typically applied to facilitate the handling of the graphene sheet and, thus, to enhance the success rate of the experiments. Precursor II is obtained by “scooping up” a graphene sheet floating on a water surface (precursor I) with a

perforated thin membrane mounted on a support frame. Manipulating a graphene sheet in the etching step is facilitated by using a PMMA coating to obtain precursor II. Residual PMMA possibly can be removed if needed.46 A convenient way to obtain precursor I is via an intermediate precursor IV. In this case, the solid support (sodium chloride crystal or a mica sheet) facilitates releasing the graphene on a water surface.24 Precursor III is initially placed on a solid support via a transfer solution. If the solid support is a salt crystal,24 then the transfer solution should be a salt solution of high concentration to prevent dissolution of the crystal. If needed, the graphene on the solid support can be dried at a temperature of ∼50 °C. The PMMA is then removed in acetone so that a solid support with graphene is obtained resulting in precursor IV, which can be stored for later usage. Precursor I is obtained by carefully positioning precursor IV at a water surface so that the graphene gets released. If the solid support is a salt crystal, the volume of the water should be selected sufficiently large so that the dissolved salt is diluted to such a level that the outside of the liquid cell does not contain salt crystals after drying later in the process. This route was used for the coverage of cells in liquid cell type A.24 A practical alternative route exists to obtain precursor III. Instead of using graphene on a metal foil, or a solid graphene block as starting material, it is also possible obtain so-called “trivial transfer graphene” commercially. This material consists of graphene coated with PMMA on a sponge support. It is easily transferred to a cellulose sheet, which can be stored, cut into the right shape, and used as starting material. Precursor of type III is obtained simply by releasing the graphene with PMMA on a water surface, whereby the cellulose detaches and sinks into the water bath. Comparison of the Different Strategies. Graphene liquid cells have been used for a range of studies in materials science and biology. A total of three examples are shown in Figure 5. The highest spatial resolution has been achieved with graphene liquid cell type B. Atomically resolved time-lapse F

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters TEM image series were recorded of Pt nanocrystals.5 Figure 5a displays five image frames in which the structure of a Pt nanocrystal evolves. It is possible to analyze larger samples than nanoscale liquid droplets containing nanoparticles using liquid cell type A. An example at the other end of the electron microscopy length scale is shown in Figure 5b,c. A tunneling nanotube was analyzed connecting two breast cancer cells grown on a SiN membrane.24 HER2 growth factor receptors were specifically labeled with quantum-dot nanoparticles to determine the positions of individual proteins. Several dimeric proteins representing the signaling active state are indicated within dashed circles. Nanoparticles were enclosed in liquid pockets of liquid cell type C and their dynamic movement analyzed (Figure 5d). With this type of liquid cell, it is possible to analyze the dynamics of freely floating nanoparticles and also to performing electron energy loss spectroscopy.28 The above examples give an impression of what is possible with graphene liquid cells. There is not (yet) one general graphene liquid cell that fits all types of experiments but different types of experiments demand different approaches. To facilitate the choice for a specific process flow and graphene liquid cell type, we summarize the advantages and disadvantages associated with each strategy in Table 1. The study of the

when the sample cannot be immobilized but is dispersed in the liquid. An important experimental requirement in certain studies is the capability of observing processes via in situ electron microscopy. The literature contains a few reports on studying dynamics at the nanoscale using liquid cell type B.5,19 Also liquid cell type C can be used for imaging movements of nanoparticles in liquid.28 However, the other process flows seem to be less favorable for this type of experiment because only a few reports are available.27 Crucial for establishing new technology and science is the ability to reproduce results. The coverage of liquid samples by graphene via process flow I is typically successful in over 50% of the experimental trials for fixed cells,24 but this has shown to be much more difficult with certain other sample types, such as a liquid droplet containing gold nanoparticles. It is usually possible to obtain liquid pockets via enclosure at both sides with graphene using process flow II, but the locations and sizes of the pockets are unpredictable, and the dynamic motion of nanoparticles is rarely observed. Also, for process flow III, there are open questions considering the remaining liquid pockets because the liquid may be exposed to acetone during the PMMA-removal step after sample loading. Process flow IV promises a better control of the liquid pockets, although it also requires processing after sample loading.28 In principle, preformed liquid pockets would allow the control of the liquid volume in each pocket, which would not only enhance the reproducibility of dynamic experiments but also facilitate the quantitative analysis of the sample. Main Disadvantages and Ideas for Improvement. According to our experience thus far, the main experimental difficulty is to obtain liquid pockets containing a sufficient amount of liquid in a reproducible way. The substrate typically contains small pockets with liquid, but it seems impossible to cover the entire surface with a closed layer of graphene, and the location and dimensions of these pockets are not predictable. We have been able to image structures of nanomaterials and biological samples in liquid, but we have not been able to observe movements of nanoparticles because is readily possible with other types of liquid cells not based on graphene.47 From the fact that only a very few publications show dynamical events for graphene liquid cells, we conclude that others must have experienced the same issues. When placed in the vacuum of the electron microscope, nearly all graphene liquid cells suffer from some level of liquid evaporation via cracks in the graphene and also via the gaps between the graphene flakes. A small degree of evaporation is usually acceptable for imaging experiments that do not involve dynamics. The liquid typically transforms into a gel with the remaining salt of the sample. However, leaking is a serious problem for dynamic experiments because these require a liquid pocket that contains a sufficient amount of liquid for nanoscale objects to be able to move. The graphene sheet is usually nonwetting for water. This is useful for handling because a graphene sheet floats on water. However, when a pocket is formed, the liquid is repelled from the covering graphene sheet(s). Most of the liquid disappears when the sample is exposed to vacuum. Possible remedies include four options. (1) A precise tuning of the viscosity, buffer, and salt conditions for each particular experiment is essential for obtaining liquid pockets of sufficiently large dimensions.32 (2) The coating of one side of the graphene could render the corresponding side hydrophilic.48 (3) The usage of large graphene sheets free of cracks to reduce leaking.

Table 1. Specific Advantages of Each Type of Process Flow for Assembling a Graphene Liquid Cell

structure of nanoparticles in liquid is accomplished with all types at a high spatial resolution for high-Z materials. Low-Z materials are best imaged in flow cell type B because this cell provides the thinnest liquid layer and the thinnest windows. However, after the optimization of the thickness of the supporting membrane (for example, 10−20 nm SiN),3 liquid cell type A would also work, and similarly, liquid cell type C would work if the spacer membrane would be kept sufficiently thin (100 nm) samples, however, is best done with liquid cell type A prepared via process flow I. For example, flattened mammalian cells several tens of micrometers width are easily covered by a graphene sheet if the sample is immobilized on a SiN membrane.24 It is a more-delicate procedure to enclose a cell on both sides with graphene, but this is possible via process flow I, thus obtaining liquid cell type B.20 Accomplishing this via process flow II would be practically impossible, and process flow IV would not work because the wells of liquid cell type C are too small. A further question is whether it is allowable to expose the sample to a larger volume of liquid. An option is to use a loop for transfer, but this problem is entirely avoided using methods II and IV, in which the sample is loaded from a droplet, so that, at most, a few microliters are used. This is also advantageous G

DOI: 10.1021/acs.nanolett.8b01366 Nano Lett. XXXX, XXX, XXX−XXX

Mini Review

Nano Letters

needed for high-resolution TEM and STEM. It can be expected that continued technological innovations such as coated- and crack-free graphene sheets, as well as predefined liquid pockets, will overcome those drawbacks in the near future so that the full advantages of the graphene liquid cell for TEM become available to the broad electron microscopy community on a routine basis.

(4) Predefined liquid pockets can be used in a support material that promotes graphene adherence.28 For imaging larger structures, the wetting issue is not a problem. The graphene covers the sample as a foil following the contours of the supporting substrate and the larger objects immobilized on it. This works well to study coated nanoparticles, protein complexes directly at the supporting substrate, and fixed cells in liquid.24 One should be cautious when interpreting the experiment and make sure that the observed movements are indeed movements in liquid and not electron-beam effects on a dried sample under or on graphene. In particular, with the extremely high electron dose needed for atomic resolution,3 electronbeam effects should be taken into account.49,50 It is, therefore, recommended to measure the liquid thickness, which can be done with electron energy loss spectroscopy for thin liquids (