In situ Monitoring of the Seeding and Growth of Silver Metal–Organic Nanotubes by Liquid-Cell Transmission Electron Microscopy

Abstract

Metal–organic nanotubes (MONTs) are highly ordered one-dimensional crystalline porous frameworks. Despite being nanomaterials, virtually all studies of MONTs rely on characterization of the bulk crystalline material (micron-sized) by single-crystal X-ray diffraction. For MONTs to achieve their raison d’être as tunable one-dimensional nanomaterials, individual tubes or small finite bundles of tubes must be synthesized and characterized. Therefore, to directly observe their formation under a variety of reaction conditions in solution, we employ liquid-cell transmission electron microscopy (LCTEM), which allows the early stages of MONT assembly to be monitored in real time. Notably, changing the metal-to-ligand ratio alters the local concentrations of reactant monomers, resulting in multiple nucleation and growth pathways and diverse morphologies at the nanoscale. These various initial seeds grow to form the same nanocrystalline needle phase. This approach of employing LCTEM to study these nanomaterials is analogous to monitoring typical homogeneous solution phase reactions by NMR for controlled nanomaterial formation.

Graphical Abstract

Metal–organic nanotubes (MONTs) are crystalline one-dimensional frameworks composed of metal ions bridged via coordination bonds with organic ligands.^1–5 Their porous and tubular architecture, with high specific surface area, makes them potential candidates for applications including gas storage, sensing, and separations and as hosts for small molecules, such as organic conductors.^6,7 Like their more widespread three-dimensional cousins, MOFs, the principles of isoreticular design can be leveraged to control size and porosity, but also like MOFs, understanding their atomic structure is predicated on successful single-crystal X-ray diffraction of the bulk material. Despite the value in gaining an improved understanding of MONTs at the nanoscale, either as individual tubes or as small finite bundles of them, very few studies have been conducted that evaluate them on this length scale.^8–10 Methods for monitoring the progress of MONT reactions in solution in real time are highly desirable if they can be coupled with detailed characterization of the purified material in a manner analogous to in situ NMR and in situ IR as tools for monitoring reactant and product concentrations for solution-based organic syntheses.^11,12 At present, time lapse transmission electron microscopy (TEM) of bulk experiments aliquoted onto TEM grids at defined time points are commonly employed to monitor morphological changes and different structures at a given reaction condition.¹³ The ability to visualize morphological changes in real time would enable us to elucidate MONT formation pathways, phase transformations, and kinetics of given reaction conditions.

Liquid-cell transmission electron microscopy (LCTEM) has shown potential for monitoring morphological changes during reactions and self-assembly of nanomaterials in situ.^14,15 LCTEM allows for analysis of the nanoscale outcome of various reactions on a minute-by-minute basis, beginning at the very first instance of the appearance of nanoscale seeds. To date, LCTEM has predominately been used to study the nucleation and growth of metal nanoparticles,^16–18 with limited studies on more beam sensitive, predominately organic materials.¹⁹⁻²³ Recent exceptions include studies on porous materials including metal–organic frameworks (MOFs),^24,25 covalent organic frameworks (COFs),²⁶ and our recently reported copper-based MONT, where one-dimensional growth was initiated by multicomponent mixing and thermal initiation of reactions subsequently observed in the liquid phase by TEM.¹³

In this article, we explore the early stage growth and nanoscale morphology of MONT crystals over a range of reaction conditions by varying metal and ligand ratios and flowing these components into the liquid cell, where they are heated and visualized (Figure S1). We previously reported the synthesis of these MONTs using a ditriazole ligand (L1) that adopts a syn conformation to form a two-pillared motif about the silver ion.⁴ Single-crystal X-ray diffraction previously confirmed the structure of the MONT, giving key intermetallic distances (Figure 1). Here we conduct studies by direct observation and elucidation of growth mechanisms using LCTEM. These studies demonstrate that the process of forming the same bulk material can proceed through multiple pathways depending on the reaction conditions.⁴ This case study of changing reaction conditions allows us to tailor the synthesis of these nanobundles of tubes in a manner analogous to how NMR monitors reaction conditions in solution-based synthetic chemistry.

Figure 1.

Reaction of triazole ligand L1 and AgNO₃ to form the MONT. Crystal structure of the MONT evaluated herein. X-ray diffraction structures were obtained from the bulk scale synthesis of the respective MONTs and analyzed on micron-sized crystals. Key intermetallic distances (Ag···Ag) are shown.⁴ For LCTEM studies presented herein, various solutions of L1 (in N-methyl-2-pyrrolidone, NMP) and AgNO₃ (in H₂O) were prepare separately outside of the TEM and then mixed together in the liquid cell by flowing the reactants into the liquid-cell TEM holder (Figure S1). Reactions were performed during TEM with heating at 85 °C.

RESULTS

Growth of MONT Bundles in LCTEM.

It is well known that the electron beam perturbs the dynamics and chemistry in LCTEM studies. Therefore, it is key to know the effect of the electron beam to prevent the destruction of early stage nucleation and growth of MONT.^13,25 We identified the structural damage to the MONTs as the cumulative doses exceeded 70 e⁻Å⁻² (see Movie S1, Figures S1, S2, Supporting Information, Section I). To initiate MONT growth within the electron microscope, two solutions (10 mg of AgNO₃ dispersed in 3 mL of DI water and 14 mg of L1 dispersed in 3 mL of NMP) were flowed separately into the LCTEM holder to mix inside the cell, giving a 1:1 ratio of AgNO₃:L1. The two solutions were allowed to mix and were then heated to 85 °C (this procedure of mixing and heating in the cell was repeated for the other ratios used in subsequent experiments, vide infra). This ratio of metal to ligand gave particles in the size range of ~50 to ~200 nm (Figure 2), which is in contrast to MONT crystals that typically form highly anisotropic elongated morphologies (Movie S2).^4,8,13 To limit electron beam exposure and to image morphology transformations over extended time periods, we recorded snapshots using 30 s pulses with an electron flux of 0.36 e⁻ Å⁻² s⁻¹ (Figure 2A). Such stroboscopic imaging reveals the growth of small nanotubes (<200 nm in length, Figure 2, green arrows) on the surface of the initially formed primary particles after 5 min of reaction (Figure 2A, Movie S3). After 10 min, fully grown rod- and sheet-like structures (~1 μm) along with the primary particles are present (Figure 2A). At increased concentrations of AgNO₃, the size of the initially formed primary particles is smaller, and growth of elongated morphologies is prominent, appearing within 1 min of initiation of the reaction via heating and mixing of components (Figure 2B, Figure S3, Movie S4, Movie S5). Notably, at a 6:1 ratio of AgNO₃:L1, anisotropic growth is very rapid, with micron long structures produced within 30 s of initiation of the reaction (Figure 2C, Movie S6).

Figure 2.

LCTEM snapshots of growth of MONT bundles at various reaction conditions acquired with an electron flux of 0.36 e⁻ Å⁻² s⁻¹. Varying ratios of AgNO₃ in DI water and L1 in NMP were flowed into the liquid cell separately at a rate of 1 μL min⁻¹ and allowed to mix in the cell, followed by heating at 85 °C. Red arrows represent the primary particles, and green arrows represent the nanotubes. (A) 1:1 ratio of AgNO₃:L1 (Movie S2, Movie S3), (B) 3:1 ratio of AgNO₃:L1 (Movie S4), (C) 6:1 ratio of AgNO₃:L1 (Movie S6). Scale bars are 500 nm.

By careful examination of the acquired MONT growth time series, anisotropic morphologies are observed to grow by several mechanisms (Figure 3). Several individual rod-like and sheet-like morphologies appear over time and grow gradually from solution resembling classical monomer attachment, as has been observed in MOF chemistry (Figure 3A, Movie S7).^13,24 In addition, rapid growth of MONTs from the surface of the initially formed particles suggests heterogeneous nucleation and growth, similar to classical zeolite nucleation described by the secondary building unit (SBU) model (Figure 3B, Movie S8).^8,27–29 Moreover, coalescence of particles leads to instantaneous formation of anisotropic morphologies, captured by LCTEM (Figure 3C, Figure S4, Movie S9). Length evolution of individual MONT bundles reveals MONT growth via two different mechanisms (Figure 3D). At a 3:1 ratio of AgNO₃:L1, a gradual increase in MONT bundle length by continuous growth of initial seed is predominant. By contrast, sudden increases in length at specific time points indicate the coalescence of two or more particles in 6:1 AgNO₃:L1 (Figure 3D). MOFs and MOF-like particles predominantly grow through the addition of smaller clusters (~1 nm) in solution, but not through coalescence.^24,30,31 Thus, these results are in direct contrast to 3D MOFs and are more akin to oriented-attachment growth of crystals driven by interparticle interactions typically attached to one end of the growing crystal.³² In all reaction conditions, the growth of the MONT bundles plateaus, suggesting the effect of spatial confinement within the LCTEM environment. Crucially, LCTEM elucidates the difference in the early stages of MONT growth at different ratios of metal ion and ligand that cannot be tracked by traditional TEM by taking aliquots from bulk reaction mixtures, as the processes are too fast and continuous to be captured. To quantify the growth kinetics of MONT bundles under various reaction conditions, an image analysis routine was applied to individual frames of the collected LCTEM data, and change in average size with time is measured (Figure 3E, Figure S5, Figure S6). Growth rate (G) can be related to the activation energy (ΔE) and temperature (T) as G = G₀ e^−ΔE/kT where G₀ is a constant depending on the growth conditions and k is Boltzmann’s constant.³³ Several factors such as precursor concentration and activation energy affect the size and growth rate of MONT bundles. In addition, with LCTEM effects such as confinement, electron beam intensity, and liquid thickness, radiolysis reactions cannot be ignored.³⁴ Several reports describe the formation of electric fields within the cell,³⁵ increases in local supersaturation,³⁵ changes in pH,^36,37 and reduction of solvated species by the electron beam,³⁸ all capable of potentially reducing the free energy barrier for the nucleation and growth events.³⁵ In the first 20 s of the reaction, MONTs grow linearly with time with almost the same growth rate across the measured concentration gradient. It is likely that in the very early stage of the assembly process the activation energy for all the reaction conditions are fixed for a given temperature, and a constant growth rate is achieved. As the reaction proceeds, changes in precursor concentrations (i.e., depletion or availability) at various reaction conditions likely change the growth rate. In 1:1 AgNO3:L1, the initial growth rate of 1.27 nm s⁻¹ drops gradually after 20 s, and we observe a nonlinear increase in average size of MONTs with projected size 〈S〉 ~ t^1/2, consistent with a diffusion-controlled growth mechanism (Figure 3E). This represents a drop in the precursor concentration most likely via secondary nucleation events. At a 3:1 ratio of AgNO₃:L1, a growth rate of 1.25 nm s⁻¹ was consistent over the entire measurement range with size changing almost linearly with time (i.e., 〈S〉 ~ t, Figure 3E). This is consistent with a constant replenishment of precursors as the reaction proceeds, commonly seen in other MOF and MOF-like materials.^13,24,26 However, this trend deviates from the square-root behavior of lengths of individual MONT bundles (Figure 3D), showing the stronger spatial constraint for growth of anisotropic morphologies and the strong influence of primary particles on size measurements in the LCTEM environment. Further increase in the concentration of AgNO₃ (i.e., at a 6:1 ratio of AgNO₃:L1) led to growth rate increases from 1.51 nm s⁻¹ to 7.68 nm s⁻¹ after 20 s of reaction (Figure 3E). This contradicts the reaction-limited monomer–monomer attachment growth mechanism. Instead MONT bundles are formed by coalescence of ensemble particles, which is distinct from how 3D MOFs are typically formed.^39–41

Figure 3.

MONT growth analysis. (A–C) LCTEM snapshots of MONT bundles growing via various mechanisms (see corresponding Movies S7, S8, and S9). Red arrows represent the primary particles, and green arrows represent the nanotubes. (D) Length evolution of individual MONT bundles grown at 3:1 (left) and 6:1 (right) ratio of AgNO₃:L1. Colored arrows point to MONTs. (E) Average size of MONT bundles as a function of time and their corresponding linear fit into 〈S〉 ~ Gt. Purple region in 1:1 and 6:1 AgNO₃:L1 represents the deviation from the initial linear growth of MONT bundles after ~20 s of reaction.

post-mortem Analysis of LCTEM Reactions.

By prying open the liquid cells after growth experiments, we can analyze the materials by standard, dry-state TEM (Figure 4, Figure S7). Assembly of multiple anisotropic MONT crystals forming 3D MONT bundles exhibits selected area electron diffraction (SAED) spots between ~7 and ~14 Å depending on the orientation of the crystal (Figure 1). Initially formed primary particles reveal a weak diffraction from the MONT pores and strong diffraction at 3.77 Å exhibited by the aggregation and short-range clustering of coordinated silver centers, triazoles, nitrate anions, and π–π stacked phenyl rings of the organic ligands (Figure 4).⁸ This distinct diffraction spot at 3.77 Å is also predominant in bulk syntheses, where they form and then disappear as elongated morphologies grow from those seeds (see Figure 5B). The elongated structures formed within the liquid cell show diffraction spots consistent with MONT bundles (Figure 4B,C,D). However, they have a distinct morphology in terms of size, shape, and aspect ratio with the varying concentrations of AgNO₃. Notably, the aspect ratio of MONT bundles broadens and increases with AgNO₃ concentration (Figure 4E). Control of the aspect ratio is critical for 1D materials, and this demonstrates that reactant ratios can be employed as a tool for adjusting this critical parameter. EDS analysis on these elongated structures (i.e., MONT bundles) also reveals a uniform distribution of silver atoms (Figure 4F,G, Figure S7).

Figure 4.

post-mortem analysis after LCTEM. (A–D) HAADF-STEM images and SAED of MONT bundles grown within a liquid-cell. (E) Aspect ratio of MONT bundles measured during post-mortem analysis for various ratios of AgNO₃:L1. (F, G) EDS spectrum and silver mapping illustrate the formation of micron-sized MONTs within the liquid cell.

Figure 5.

TEM analysis of MONT bundles grown in bulk synthesis at various ratios (shown in bottom right corner) of AgNO₃:L1.

MONT Growth under Standard Bulk Synthesis Conditions.

MONT formation in bulk synthesis was analyzed by TEM and AFM (Figure 5, Figures S8, S9, S10). Across ratios of AgNO₃:L1, immediately after the reactants are mixed, the solution turns turbid, suggesting the immediate formation of primary particles or MONT nanocrystals directly from solution. Precisely, at a 1:1 ratio of AgNO₃:L1, primary particles are formed within 2 min of reaction (Figure 5A), and TEM after 30 min of reaction reveals anisotropic sheet-like morphologies where lattice spacing represents MONT crystals. Smaller primary particles on the surface of the anisotropic MONT bundles suggest they undergo ripening to form sheet-like structures (Figure 5B,C). These elongated anisotropic structures are observed immediately at 3:1 and 6:1 ratios of AgNO₃:L1 (Figure 5D,G). Sharp edges and corners indicate faceting, which increases in time (Figure 5D,H), and the sheets bend to release surface tension between growing layers (Figure 5H).⁴² We also observe long-range grain boundaries that suggest the assembly of MONT bundles takes place in stages (Figure S9). These observations further support the conclusions drawn from LCTEM data analysis that relatively isotropic primary particles are formed immediately after the reactants mix, which then grow in a preferred orientation to form rod-like or sheet-like structures. These anisotropic structures further aggregate to form bundles of MONTs in all three dimensions.⁸

MONT Growth Mechanisms.

Classical nucleation theory (CNT) explains the rate of nucleation J_n = A exp(–Bα³/σ²), where A is the prefactor, which depends on kinetics, B depends on temperature and volume, α is the interfacial energy, and σ is the supersaturation of the solution. The growth rate increases linearly with time, resulting in a nonlinear change in area and/or volume of the crystal growth.⁴³ Here, an inherent assumption is that the pathway is through direct formation of nuclei with an ordered crystalline structure identical to that of the bulk crystal. However, the various concentrations of AgNO₃ applied in our experiments reveal that MONT assembly can take place via several paths deviating from the classical pathway to yield multiple indirect pathways that coexist within the same reaction medium, similar to observations seen in the past for MOF-5 and CaCO₃ formation.^15,44 From our analysis, we see that in the beginning of the reaction less supersaturated reactant monomers mix, and the interfacial surface energy of precursor ions and their density fluctuations create unstable amorphous clusters that immediately lead to the aggregation and clustering of coordinated silver centers, triazoles, nitrate anions, and π–π stacked phenyl rings of the organic ligands forming primary particles. Over time a steady-state supersaturation is achieved, which enhances the growth of anisotropic MONT on the surface of the primary particles by heterogeneous nucleation.^45–47 Some anisotropic MONT bundles are formed by oriented attachment/coalescence of ensemble particles. In the reaction mixture, MONT bundles are also formed by classical monomer attachment by continuous transformation of the supersaturated solution into crystalline MONTs (Figure 6).

Figure 6.

Schematic depiction of MONT nucleation and growth at various ratios of ligand to AgNO₃.

CONCLUSION

In summary, we have shown multiple growth pathways exist for silver MONT formation from the solution phase. Moreover, we demonstrated that differences in pathways can be visualized, and kinetics can be measured using LCTEM by studying the reaction across metal to ligand ratios (AgNO₃:L1). Local concentrations of reactants, precursor ions, and amorphous clusters can dictate the formation of thermodynamically stable MONT crystals and/or kinetically driven assembly within the same reaction medium. We observed that at low concentrations of AgNO₃ surface energy minimization is attained by the aggregation and short-range clustering of precursor ions that form primary particles immediately. As the supersaturation increases, anisotropic MONT bundles are formed by heterogeneous nucleation from the primary particles. Excess AgNO₃ (6:1 ratio) results in coalescence and oriented attachment of ensemble particles to generate bundled fibers formed as MONT crystals. This growth mechanism of coalescence is not generally observed with MOFs and may be characteristic of these one-dimensional materials. In this reaction mixture, MONT bundles also form by classical pathways, which include continuous transformation of the supersaturated solution into crystalline MONTs. More broadly, the ability to monitor reactions that yield solid crystalline materials in real time is critical, and we demonstrated that LCTEM is a viable reaction development tool for these crystalline porous one-dimensional materials. We believe that studying nanomaterials by LCTEM will become a central tool in reaction development and design.

METHODS

Bulk MONT Reactions.

(A) 1:1 AgNO₃:L1 reaction: AgNO₃ (10 mg, 0.059 mmol) was dispersed in 3 mL of deionized water, and separately, L1 (14 mg, 0.059 mmol) was dispersed in 3 mL of N-methyl-2-pyrrolidone (NMP). (B) 3:1 AgNO₃:L1 reaction: AgNO₃ (30 mg, 0.177 mmol) was dispersed in 3 mL of deionized water, and separately, L1 (14 mg, 0.059 mmol) was dispersed in 3 mL of NMP. (C) 6:1 AgNO₃:L1 reaction: AgNO₃ (60 mg, 0.353 mmol) was dispersed in 3 mL of deionized water, and separately, L1 (14 mg, 0.059 mmol) was dispersed in 3 mL of NMP. In each case all the reactants were dispersed by ultrasonication for 3 min and heated to 85 °C for 15 min. The hot reactants were then mixed together and further heated at 85 °C. Aliquots were taken for transmission electron microscopy and atomic force microscopy analysis after 2 min, 30 min, 6 h, and 24 h of reaction.

Transmission Electron Microscopy.

TEM of bulk synthesized MONTs was performed using a JEOL 1230 operated at 120 kV. LCTEM and post-mortem analyses were performed using a JEOL ARM300F GrandARM TEM with a Gatan OneView-IS camera and K3-IS camera operated at 300 kV. Prior to post-mortem analysis, LCTEM chips were opened carefully and washed gently with distilled water to remove excess reactants and solvent. Bottom chips were then placed on the standard TEM holder and analyzed by energy dispersive X-ray spectroscopy (EDS) and diffraction. EDS mapping was employed for elemental analysis of MONT grown in bulk and within the liquid cell. Images were collected in HAADF-STEM mode with a probe semiconvergence angle of 10 mrad and at a camera length of 20 cm. A beam current of 0.3 nA and pixel dwell times between 1 and 5 μs were used.

Liquid-Cell Transmission Electron Microscopy.

A Protochips Poseidon Select Heating holder was used for all LCTEM studies. The liquid-cell chips (with 50 nm silicon nitride thickness) were cleaned in acetone and methanol to remove the photoresist layer prior to use. Subsequently, the top and bottom chips were plasma cleaned for 5 min to induce hydrophilicity. The chips were assembled in the tip of the holder, and a leak check was performed. After inserting the empty liquid-cell holder into the column of the microscope, reactants (two separate solutions made as noted above for bulk syntheses) were flowed separately through the inlets separately using a syringe pump at a rate of 1 μL min⁻¹ to meet inside the liquid cell (Figure S1). After approximately 20 min, we observed wetting of the liquid cell, and the flow of reactants was continued over the entire data acquisition phase (i.e., during MONT growth). BF-TEM images were acquired before and after the flow of reactants to allow for liquid thickness measurements. Once the wetting of the chips was observed, the cell was heated to 85 °C using a Protochips temperature controller at a rate of 1 °C s⁻¹. During heating, the chips were not exposed to the electron beam. Once the temperature reached 85 °C, movies of the reactions were captured using screen recorder software (Camtasia Studio 7 Recorder screen capture software, TechSmith Corporation, USA). NB: All TEM alignments, beam settings, dose settings, and calibrations were conducted prior to the liquid-cell experiments using standard gold nanoparticles or a diffraction grating waffle TEM sample.

LCTEM Electron Flux Measurements.

Beam current (pA cm⁻²), measured using the small screen in the microscope, was calibrated using a Faraday cup (nA). Calibration data were obtained for various electron beam parameters, such as spot size, condenser aperture, and emission current, in the linear regime over the range of beam currents used for LCTEM experiments. At the start of each experiment, the calibrated beam current from the small screen was measured for appropriate spot size, condenser lens system, and emission current used for that specific experiment. The beam diameter (Ų) was measured for the appropriate magnification. This is used along with calibrated beam current (nA) to calculate the electron flux (e⁻ Å⁻² s⁻¹). These measurements were conducted before the LCTEM experiments using a single tilt holder or blank LCTEM chips without a liquid sample. The settings were left unaltered for subsequent LCTEM experiments.

The effect of the electron beam and its interactions with the reactant molecules cannot be avoided.^26,37 Hence, to prevent and minimize the effect of the electron beam during in situ growth experiments, we performed several trial and error experiments at various electron flux conditions. At an electron flux of ~1 e⁻ Å⁻² s⁻¹ (with a beam current of 1.52 nA), dewetting of the chips was predominant upon irradiation with the electron beam. Immediately after electron beam exposure, the reactant mixture receded away from the viewing area (exposed area). By reducing the electron flux to 0.36 e⁻ Å⁻² s⁻¹ (with a beam current of 0.63 nA), dewetting of the liquid cell was prevented. Following this, we irradiated the fully grown MONT nanocrystals with the electron beam and measured the cumulative electron dose. We observed beam-induced structural damage of MONTs as the cumulative dose exceeds 70 e⁻ Å⁻² (see Figure 2, Movie S1).

Atomic Force Microscopy (AFM).

AFM measurements of bulk synthesized MONT bundles were performed using a Bruker Dimension FastScan atomic force microscope operated in tapping mode using silicon tips. The images shown here represent the topological height profile (Figure S8).