Introduction

Metal-organic frameworks (MOFs) are porous materials built from metallic ion clusters and organic linker molecules1. Important properties, such as pore size, electrical conductance, or chemical activity, are strongly dependent on the combination of these building blocks2. The properties can be further adjusted by modification of linker molecules or metallic clusters or by incorporating atoms, nanoparticles, or nanostructures into the MOF3,4. These changes can be made in situ5 or post-synthetic6. This freedom in tunability makes them applicable in many fields, for instance, for helium7 or hydrogen8 separation, for catalysis9 and as gas sensors10,11,12 like for methanol13 sensing.

Conductive MOFs are essential for chemiresistive sensing applications14. There are several design strategies to promote the conductance of MOF materials, among others, to tune the charge transport over an extended conjugation pathway, a through-bond pathway, or a through-space pathway15,16,17. Other mechanism to promote charge transport in MOFs without changing the building blocks of a specific MOF is to load the pores with redox-active guest molecules18 or by adding conductive materials12,19,20. Carbon nanotubes (CNTs) have emerged as prominent additives due to their high surface area and high current carrying capacity21,22,23,24. The functionalization of CNTs plays a key role in the in situ synthesis of MOF/CNT composites23. MOF synthesis in the presence of functionalized CNTs enables direct MOF growth on the surface of the tubes25,26,27. In particular, UiO-6628 grows preferentially on carboxyl (-COOH) functionalized carbon nanotubes5,29,30. An explanation for this is the chemical similarity between the UiO-66 linker molecules and the -COOH functionalization of the tubes. UiO-66 is one of the best-suited MOFs for sensing applications due to its remarkable thermal and chemical stability, especially its high stability against moisture. Networks of many multi-walled carbon nanotubes (MWCNTs) with UiO-66 MOF were shown to exhibit chemiresistive ethanol sensing properties that are absent in bare CNTs5. MOF/CNT hybrid structures with other MOFs have shown detection capabilities for substances such as lead31, H2O225, nitride26, humidity27, and others19,23.

Originally, CNT surface functionalization was introduced as a standard procedure to enhance the dispersibility of CNTs in solvents32. Typically, this involves oxidation to introduce carboxyl and hydroxyl groups to the tube surface24, while also removing catalytic metal residues and amorphous carbon32. It should be noted that this process is reversible; defunctionalization of functionalized carbon nanotubes in a solvent leads to the formation of large clusters of entangled tubes33.

Here in this work, electron beam-controlled, UiO-66 MOF patterning on functionalized MWCNTs is presented. Precise local control of MOF growth is crucial for complex device architectures, such as e.g., necessary for miniaturized nanosensor arrays. In previous studies, MOF patterning34 was performed using different methods35,36,37,38,39, e.g., by decorating substrates with surface-modified nanoparticles and utilizing the aligned particles as nucleation sites40,41,42. In contrast to these methods, our method involves inhibition of MOF growth by spatial deactivation of nucleation sites. The mechanism is shown schematically in Fig. 1. First, we deposited -COOH functionalized MWCNTs on a Si chip with SiO2 surface by drop-casting (Fig. 1a). The individual MWCNTs are then locally irradiated by an electron beam, resulting in defunctionalization of the MWCNTs (Fig. 1b). Subsequently, the chip is placed in an UiO-66 synthesis solution containing Zr-salt, linker molecules, and additives (Fig. 1c; synthesis details are provided in the Methods section). Growth can occur in tube areas that were not irradiated. This can be explained by an intact functionalization of the MWCNT that possibly acts as a coordination center for the Zr4+-ions allowing MOF nucleation on the surface of the MWCNT (Fig. 1d)5,29,30. Scanning electron microscope (SEM) imaging was used to quantify the MOF growth. Our method allows to pattern the growth of MOF crystals locally on individual MWCNTs.

Fig. 1: Schematic illustration of the UiO-66 growth inhibition mechanism.
figure 1

a COOH functionalized multi-walled carbon nanotube (MWCNT). The functionalization molecules are shown only on the top site and not around the tube for the sake of clarity (not to scale). b Irradiation and local defunctionalization by an electron beam. c Locally irradiated MWCNT in an UiO-66 synthesis solution, which contains Zr-salt, linker molecules, and additives. The Zr4+-ions bind to the functionalization, seeding the growth of the MOF. d After synthesis. UiO-66 crystals have grown on the MWCNT. Growth is inhibited in the irradiated areas.

Full size image

Results and discussion

Electron beam-induced growth inhibition

Figure 2 shows the influence of electron irradiation on the growth of UiO-66 crystals (crystallinity verified with powder X-ray diffraction, PXRD, data shown in Supplementary Note 1) on various MWCNTs. Initially, MWCNTs were found to be around 35 nm thick and individual tubes are clearly distinguishable, as shown in Fig. 2a. After synthesis and without prior irradiation, thick MOF crystals fully encapsulate the MWCNTs, leaving no visible tube area, as shown in Fig. 2b. The -COOH functionalization acts as a seed for UiO-66 MOF growth and the MOF crystals grow preferred on such functionalized MWCNTs5. In contrast to this, Fig. 2c shows an image of an area irradiated with a dose D = 9000 μC cm−2 at an acceleration voltage Vacc = 6 kV. The irradiation prior to synthesis leads to strongly inhibited MOF growth. Only very few MOF crystals have grown on the MWCNTs. Interestingly, the combination of higher electron energies, due to a higher acceleration voltage, and a higher dose results in a less distinct growth inhibition. This can be seen in Fig. 2d. The area was irradiated by electrons accelerated at Vacc = 30 kV and a dose D = 10,000 μC cm−2. Notably, oxidized graphene, which has a similar structure to our functionalized MWCNTs, exibits an electron beam-induced defunctionalization effect at an acceleration voltage of Vacc = 80 kV43. In the case of MWCNTs in solution, UVC irradiation has been found to induce a process known as photodecarboxylation, resulting in the defunctionalization of MWCNTs functionalized with -COOH molecules33. Drawing parallels from these findings, our observation of MOF growth inhibition on MWCNTs by electron irradiation, as seen in Fig. 2c and d, can be attributed to a similar defunctionalization effect. A -COOH defunctionalization is also supported by X-ray photoelectron spectroscopy (XPS) analysis of MWCNTs before and after electron irradiation (XPS data shown in Supplementary Note 1).

Fig. 2: MOF growth inhibition by electron irradiation.
figure 2

a Typical scanning electron microscope image of multi-walled carbon nanotubes (MWCNTs) before MOF synthesis. b, c, and d are images after UiO-66 MOF synthesis. b Non-irradiated MWCNTs, c Vacc = 6 kV and D = 9000 μC cm−2 and d Vacc = 30 kV and D = 10,000 μC cm−2.

Full size image

Impact of the electron dose

To study the dose-dependent growth inhibition in more detail, we systematically irradiated areas with various MWCNTs by different doses, prior to UiO-66 synthesis. Figure 3 shows SEM images of MWCNTs after synthesis, with MWCNTs colored in red and MOF crystals colored in blue (for the uncolored images, see Supplementary Note 2). The areas presented were irradiated with increasing doses at a fixed electron acceleration voltage Vacc = 6 kV. For the lowest dose D = 1000 μC cm−2 (Fig. 3a) minor changes in MOF growth are observed, compared to the non-irradiated case (see Fig. 2b). Most of the MWCNTs are covered by MOF (blue area), and only small parts of the bare MWCNTs (red area) are visible. However, in contrast to the non-irradiated case, individual MOF grains become distinguishable. An irradiation dose of D = 2000 μC cm−2 further inhibits the growth of MOF (Fig. 3b) and the crystal size of the MOF appears to be reduced. MOF crystals are separated from each other, and the MWCNTs can be recognized. The areas irradiated by a further increased dose D = 3000 μC cm−2 (Fig. 3c) follow this trend. Fewer MOF crystals grow on the MWCNTs and large parts of the MWCNTs are visible. MWCNTs in areas irradiated by D = 5000 μC cm−2 (Fig. 3d) show strongly inhibited MOF growth and only a small number of MOF crystals have grown on the MWCNTs.

Fig. 3: Impact of electron dose on MOF growth.
figure 3

After synthesis scanning electron microscope images of (prior to the synthesis) electron irradiated multi-walled carbon nanotubes (MWCNTs). The MOF area is colored in blue, and the bare MWCNT area is colored in red. The irradiation doses are a D = 1000 μC cm−2, b D = 2000 μC cm−2, c D = 3000 μC cm−2, and d D = 5000 μC cm−2 all at Vacc = 6 kV. The higher the irradiation dose the stronger the MOF growth inhibition.

Full size image

Impact of the electron acceleration voltage

We also analyzed the dose dependence on growth inhibition for irradiation with higher energy electrons using an acceleration voltage of Vacc = 30 kV. To quantify the different impacts of Vacc = 6 kV and Vacc = 30 kV, the MOF covered area AMOF was determined directly from SEM images (see Supplementary Note 2). Similarly, the visible areas of the MWCNTs were determined. From the highest dose values D ≥ 6000 μC cm−2 (strong growth inhibition), the mean area of MWCNTs ACNTmean in the images was approximated. The ratio AMOF/ACNTmean for different irradiation doses D at Vacc = 6 kV is shown in Fig. 4a. The decrease in MOF area is well described by an exponential decrease of the form \({A}_{{{{{{{{\rm{MOF}}}}}}}}}/{A}_{{{{{{{{\rm{CNTmean}}}}}}}}}=A\cdot \exp (-D/g)\) with A6kV = 4 and g6kV = 1785 μC cm−2. This is also the case when using electrons of higher energy accelerated by Vacc = 30 kV, as seen in Fig. 4b. The exponential fit is described with the parameters A30kV = 4.5 and g30kV = 5271 μC cm−2. At this increased acceleration voltage, the growth inhibiting effect is less pronounced in contrast to Vacc = 6 kV. In both cases (Vacc = 6 kV and Vacc = 30 kV) the used acceleration voltages generate electrons with kinetic energies that are several orders of magnitude higher than typical organic binding energies (range of eV). The acceleration voltage has strong impacts on the electron penetration depth in the substrate and the backscattering (Electron scattering simulations can be found in Supplementary Note 2). Figure 4c illustrates the backscattering effect (e.g., Vacc = 6 kV). Electrons from an incoming electron beam (red) penetrate the surface, are absorbed in the substrate, and they lose energy. Secondary electrons are backscattered and return to the surface44 (blue). These electrons contribute to the defunctionalization process. Figure 4d illustrates the scattering for a higher acceleration voltage (e.g., Vacc = 30 kV) than in Fig. 4c. Electrons penetrate deeper into the substrate before scattering, and fewer electrons reach the surface. In this case, there are fewer electrons that contribute to the defunctionalization process.

Fig. 4: Influence of electron acceleration voltage.
figure 4

a Normalized MOF area AMOF over the irradiation dose D at Vacc = 6 kV (dots) and b for Vacc = 30 kV (triangles). The higher the electron irradiation, the more suppressed the MOF growth. Solid lines are exponential fits of the form \({A}_{{{{{{{{\rm{MOF}}}}}}}}}/{A}_{{{{{{{{\rm{CNTmean}}}}}}}}}=A\cdot \exp (-D/g)\) with A6kV = 4, g6kV = 1785 μC cm−2, A30kV = 4.5, and g30kV = 5271 μC cm−2. The error bars denote the standard deviation calculated by the propagation of uncertainty using the individual standard deviations of AMOF and ACNTmean. c Schematic illustration of electron scattering within the substrate, with many electrons scattering near the surface (incident electrons colored in red and scattered electrons colored in blue). d Electron scattering for acceleration voltages higher (e.g., 30 kV) than in (c) (e.g., 6 kV). Electrons penetrate the substrate deeper and fewer electrons scatter near the surface, leading to less growth inhibition.

Full size image

Locally controlled MOF growth

This knowledge was now used to locally deplete the growth of MOF crystals in specific areas of individual MWCNTs. Before MOF synthesis, stripes of s = 200 nm width crossing the MWCNTs were irradiated at an acceleration voltage Vacc = 6 kV. Irradiation with different doses results in MOF-free areas with certain lengths that are different from the stripe width s = 200 nm. Colored SEM images of these MWCNTs can be seen in Fig. 5a–c (MOF colored in blue, bare MWCNT colored in red). Figure 5a shows a MWCNT irradiated by D = 8000 μC cm−2, resulting in a MOF-free area of 470 nm length along the tube, this is larger than the initial stripe width s = 200 nm. Lowering the irradiation dose reduces the lengths of the MOF-free tube area, as seen in Fig. 5b (see Supplementary Note 3). We were able to achieve a MOF-free tube area of 220 nm length along the tube by using D = 6000 μC cm−2. By using D = 4000 μC cm−2, as shown in Fig. 5c, a MOF-free tube area of 130 nm length was obtained although the width of the irradiated stripe was s = 200 nm. Defunctionalization by electron scattering is a statistical process, and lowering the dose leads to a higher probability of remaining functional molecules, resulting in a possibly shorter length of the MOF-free tube area compared to that initially irradiated. The observed enlargement of the MOF-free area along the length of the tube, as shown in Fig. 5a, b, can be explained by defunctionalization outside the irradiated tube length s, due to randomly off-axis backscattered electrons, as shown schematically in Fig. 5d.

Fig. 5: Locally controlled MOF growth.
figure 5

a, b and c show colored scanning electron microscopy images (MOF colored in blue, multi-walled carbon nanotube (MWCNT) colored in red) of locally irradiated, -COOH functionalized MWCNTs after UiO-66 synthesis. s = 200 nm wide stripes were irradiated by a Vacc = 6 kV electron beam, prior to the synthesis. The irradiation doses are a D = 8000 μC cm−2, b D = 6000 μC cm−2, and c D = 4000 μC cm−2. MOF growth in the irradiated MWCNT areas is strongly inhibited. d Schematic side view of the stripe-irradiation process (not to scale) using stripe width s. Some backscattered electrons are off-axis scattered and reach the surface outside the irradiated width s. These electrons participate in the defunctionalization process, which could enlarge or reduce the MOF-free tube area.

Full size image

Our technique shows how to nanopattern the growth of MOF crystals on the scale of an individual MWCNT. This allows to restrict MOF growth to specific tube areas. Possible applications would be more complex devices, such as MOF/CNT transistors with MOF dielectrics or MOF/CNT nanosensor arrays.

Because this process is explained by the removal of functional molecules on the MWCNT, other defunctionalization methods, such as EUV or UVC irradiation, should show similar results. The described working principle also indicates the possibility of expanding this process to other MOF species, as well as to other functionalized materials.

Conclusion

In conclusion, the impact of electron irradiation on the growth of UiO-66 MOF on functionalized MWCNTs was studied. In non-irradiated areas, MWCNTs were found to be fully encapsulated by the MOF. However, in irradiated areas, the growth of MOF crystals was inhibited, with stronger inhibition at higher irradiation doses. We attribute this to the defunctionalization of the MWCNTs by backscattered electrons. Lower acceleration voltages result in stronger growth inhibition than higher ones. Electrons accelerated by lower voltages penetrate the substrate less deeply and more electrons are backscattered to the surface. With this process, we are able to locally deplete the growth of MOF crystals on individual MWCNTs. This process opens the way to nanofabricate complex MOF/MWCNT devices based on a single nanotube as e.g., needed in applications as nanosensors.

Methods

Chemicals

Zirconium (IV) chloride (>99.5%, ZrCl4, Sigma Aldrich), N,N-dimethylformamide (99.8%, DMF, Sigma Aldrich), formic acid (>98%, HCOOH, Merck), terephthalic acid (99%, C8H6O4 (H2BDC), Honeywell), denatured ethanol (96%, C2H5OH, Sigma Aldrich), nitric acid (70%, HNO3, Sigma Aldrich), acetone (99.8%, C3H6O, Carl Roth), isopropanol (99.8%, 2-propanol, C3H8O, Carl Roth) were used without purification.

Carbon nanotube preparation

Multi-walled carbon nanotubes (Baytubes C 70 P, Bayer) with diameters around 35 nm and length of several microns were oxidized for 6 hours in boiling nitric acid. This step adds a carboxy functionalization to the surface of the tube. An ethanol solution with 0.001 m% oxidized MWCNTs was prepared. Around 10–20 μL of the MWCNT/ethanol solution was deposited by drop-casting on a 4 × 4 mm2 chip. Standard N++/As doped <100> silicon wafer with 340 nm thermally grown silicon dioxide, provided by Si-Mat, were used as chip material. Arrays of gold/chrome markers served for orientation (the gold/chrome markers are shown in Supplementary Note 2). Before drop-casting, the chips were ultrasonically cleaned (Elma Elmasonic S 30 H) for 10 min in acetone, washed in isopropanol, then cleaned by oxygen plasma for 5 min and 200 W (TePla Plasma Prozessor 100-E) and subsequently ultrasonically cleaned in acetone for 5 min and washed again in isopropanol.

Electron beam lithography

A scanning electron microscope (SEM) LEO 1530 with a Raith Elphy Nanolithography system was used to irradiate squares of 5000 μm2 with doses up to 10,000 μC cm−2 and lines of 200 nm width and doses up to 10,500 μC cm−2 at acceleration voltages of either 6 kV or 30 kV. The squares were at least 100 μm and the stripes 4.8 μm apart from each other.

The same system was used for SEM imaging.

Synthesis of UiO-66/MWCNT composite

The UiO-66/MWCNT composites were synthesized according to a synthesis method developed in the Behrens group5. UiO-66/MWCNT was synthesized in a 100 mL Pyrex glass vessel by sequentially adding 0.1003 g ZrCl4 (1 Eq), 0.031 mL deionized water (4 Eq), 0.812 mL formic acid (50 Eq) and 0.0715 g H2BDC (1 Eq) in 25 mL DMF (750 Eq). The synthesis solution was ultrasonicated for 5 min to ensure that the Zr-salt and the linker were completely dissolved. The SiO2/Si chip with MWCNTs on it was carefully added to the solution and placed top-up at the bottom of the glass vessel. The glass vessel was sealed and heated at 120 ∘C for 1 h. The chip was carefully removed from the solution and washed 3 times with ethanol and dried in air for 1 h. For the powder X-ray diffraction measurements (see Supplementary Note 1) the precipitate which results during the synthesis was separated via centrifugation, washed with DMF and ethanol, and dried under vacuum.