Introduction

The continuous miniaturization of technologies, including scaffolding for medical implants [1, 2], complex vertical electronics [3, 4], and low-density and high-strength components for automotive and aviation industries [5, 6], challenges traditional fabrication processes [7]. Consequently, a variety of additive manufacturing (AM) techniques has emerged to generate three-dimensional (3D) objects by the successive layer-by-layer addition of material. Many AM processes, including 3D printing technologies, were originally developed for polymeric materials. More recently, AM techniques for metals and metal alloys have enabled the synthesis of morphologies with feature sizes below 10 μm [7]. Several studies are reported that have attempted in situ growth of Ni-based nanostructures that are considered as cost-effective electrocatalysts for oxygen evolution reactions in water splitting applications [8,9,10,11,12,13]. Most studies are, however, conducted in liquid solutions and did not directly observe growth mechanisms at the nanoscale. For pure metals, as opposed to metal oxides or metal–organic composites, laser engineered net shaping and electron beam melting are commonly employed that rely on the local fusion of metal particles. Prominent examples are selective laser melting and selective laser sintering, which use a laser beam as power source to locally sinter powdered materials for the assembly of 3D morphologies. An advantage of these techniques is that no binding agents or suspensions are employed that could cause contamination of the final products. The size of the smallest attainable features generally depends on the diameter of the feedstock particles and is limited by the relatively high thermal conductivity of metals [7].

During in situ transmission electron microscopy (TEM) experiments, Rufner and co-workers discovered the current-assisted growth of one-dimensional metal nanopillars from a reservoir of feedstock particles at the tip of a tungsten needle [14]. Growth of a metal nanopillar was accomplished by supplying an electric current via a scanning tunneling microscopy (STM) tip that was in direct contact with individual nickel nanoparticles. Growth rates as high as 16 nm/s were reported, and proposed mechanisms for mass transport include electromigration and diffusion in a temperature gradient (Ludwig Soret diffusion). However, experiments were conducted under compressive stress between a non-moving STM tip and a flexible amorphous carbon film that was displaced by the growing nanostructure [14].

This study reports the first tip-guided growth of nickel nanostructures that was accomplished with an intentionally retracting nanoindenter in a scanning electron microscope (SEM). Experimental results demonstrate that dielectric breakdown of surface nickel oxides is required to initiate current flow through nanoparticle agglomerates. Finite element modeling was used to identify the amount of Joule heating that has led to partial sintering of feedstock particles and unidirectional growth through solid-state diffusion. Unlike previous reports, this study demonstrates the feasibility of directionally growing metal nanostructures guided by a moving tip. In the future, this capability is posed to enable the fabrication of pure metal nanostructures with complex 3D geometries and unprecedented feature sizes.

Experimental details

Fabrication of nickel micropillars

Nickel micropillars with 1 µm diameter and 1 µm in height were fabricated using photolithography and electroplating on thermally grown SiO2 substrate that was supported by a single-crystalline silicon wafer. Before fabrication, the substrate was cleaned under sonication in acetone, methanol and propanol for 5 min each. Subsequently the substrate was rinsed in deionized water and blow-dried with N2. Afterward, the substrate surface was plasma cleaned by reactive ion etching in O2 for 30 s followed by Ar for 30 s.

Figure 1 summarizes the subsequent fabrication process. A Lesker labline sputtering system (Kurt J Lesker Company, Jefferson Hills, PA) was used to deposit nominally 70 nm of titanium and 220 nm of nickel as seed layers (cf. Figure 1a). Deposition was carried out in 5 mTorr of Ar atmosphere, followed by another plasma cleaning step. Next, a 1-µm-thick positive photoresist layer was deposited (cf. Figure 1b). A periodic pattern was generated on the photoresist layer using a GCA 8500 i-line Stepper (GCA Corporation, San Francisco, CA). As displayed in Fig. 1c, a mask with the desired pattern was placed on top of the photoresist for subsequent exposure to UV light. Areas of the photoresist illuminated by UV light became soluble in CD-26 developer solution. After dissolution of the exposed photoresist for 60 s, the substrate was immediately rinsed in deionized water and blow-dried with N2. The resulting substrate configuration is sketched in Fig. 1e. Metallic nickel with a nominal thickness of 0.958 µm was subsequently electroplated at 60 °C with a current of 6.6 mA (Fig. 1f). The plating solution was Transene sulfumate nickel plating SN-10. After electroplating, the remaining photoresist was removed using Futurrex RR41. Figure 2a and b shows top view SEM images of as-fabricated Ni micropillars. Figure 2c shows a side view of a series of micropillars that exhibited a ‘mushroom-top’ morphology due to excess electroplating deposition.

Figure 1
figure 1

Sketch of the processing steps for nickel micropillar fabrication using photolithography and electroplating.

Full size image
Figure 2
figure 2

a Top view SEM micrograph of the as-fabricated nickel micropillar arrays. b Side view SEM micrograph of the nickel micropillars exhibiting the effects of overfilling during nickel deposition sketched in Fig. 1e.

Full size image

Feedstock particles

Nickel nanoparticles with nominal diameter of 300 nm served as feedstock particles and were dispersed by sonication in isopropanol and subsequently drop casted on top of the micropillar substrate.

Current-activated growth of nickel nanostructures

In situ current-activated growth experiments were carried out with a ThermoFisher Quattro S environmental scanning electron microscope (ESEM). No additional sample heating was used during the experiments. Figure 3a and b presents detailed sketches of the experimental setup and the sample geometry, respectively. The substrate was adhered with double-sided copper tape to a pre-tilt SEM sample stub. A second piece of copper tape was used to create a conducting pathway between the nickel seed layer and the sample stub (cf Fig. 3b). Electric bias was supplied to feedstock particle agglomerates with a tungsten carbide cone shaped indenter head (Micro Star Technologies, Huntsville, TX) installed on an InSEM nanoindenter (KLA, Milpitas, CA) that is attached to the ESEM. A Keithley model 6517B electrometer was used to apply a positive DC bias and measure the resulting electric current (see circuit diagram in Fig. 3a). Both bias application and current measurement were controlled by a Python program. The sample stage was connected to ground. One directional movement of the nanoindenter along its main axis was achieved by the displacement and extension control of the actuator through the InView software package. The growth process was recorded through a series of SEM micrographs saved as video files.

Figure 3
figure 3

a Sketch of experimental setup and the respective circuitry. b Sketch view of the as-prepared sample supported by a SEM sample stub made from aluminum.

Full size image

For each experiment, mechanical contact between the nanoindenter head and feedstock particles was established before any bias application. Current–voltage (I–V) scans ranging between − 0.1 V and + 10 V were recorded to ensure and inspect electromechanical contact between the particle/micropillar configurations and the nanoindenter. During in situ experiments, a constant positive DC bias of 6–8 V was applied. A maximum current limit was set to 1 mA to protect the InSEM electronics. Mechanical drift of the indenter head toward the sample stage was monitored during the experiments and manually compensated when applicable.

Results

After electromechanically contacting the nanoindenter head with the nickel particle agglomerates, in situ experiments were carried out either under static (non-moving) conditions or dynamically with a retracting nanoindenter.

Static in situ experiments

Figure 4a and b shows a nickel micropillar with attached Ni nanoparticles before and after contacting with the indenter, respectively. Nickel particles are highlighted by arrows in Fig. 4a. After contact with the nanoindenter was established, the particle labeled ‘1’ in Fig. 4b appears compressed in comparison with Fig. 4a. Periodic intensity fluctuations along the vertical direction observed in Fig. 4b are caused by interferences between the nanoindenter electronics and the scan coils of the ESEM.

Figure 4
figure 4

SEM micrographs of the nickel nanoparticle and micropillar configuration a before and b after contacting with the nanoindenter.

Full size image

Figure 5a shows a plot of recorded current I as a function of applied bias V between − 0.1 V and + 10 V acquired from the configuration shown in Fig. 4b. The first recorded I–V curve exhibits Ohmic behavior above + 4 V. All subsequently acquired I–V curves reveal Ohmic behavior for biases above + 2 V with overall increased currents.

Figure 5
figure 5

a Current–voltage (I–V) curves sequentially recorded form the nanoparticle/nanopillar configuration shown in Fig. 4b. b and c Recorded currents as a function of time for constant applied biases of + 8 V and + 7 V, respectively. During the time interval between 900 and 1193 s in b, no current data were saved due to a software malfunction. However, after 1193 s the current safety limit was triggered.

Full size image

A constant bias of + 8 V was subsequently applied to the nanoindenter head and resulting current through the particle agglomerate was recorded as a function of time (Fig. 5b). To protect the nanoindenter electronics, a maximal current safety limit of 1 mA was employed. For the duration of 1193 s, the current continuously increased from around 147 µA until it reached the current safety limit. The in situ experiment was subsequently restarted at a reduced bias of + 7 V (Fig. 5c). After restarting the experiment, the current steadily increased over 364 s from around 722 to 938 µA before it abruptly increased to 11.6 mA, which exceeded the current safety limit. Subsequent lowering of the applied bias to + 6 V, + 5 V, + 2 V and + 0.5 V instantaneously resulted in currents above the safety limit of 1 mA.

Additional in situ experiments carried out with similar nanoparticle/micropillar arrangements reproduced the results displayed in Figs. 4 and 5. Figure 6a–c shows experimentally observed currents for three additional in situ experiments as a function of applied bias recorded either immediately before the software controlled current limit was triggered or when a stable current below 1 mA was registered. In all cases, linear relationships between current and applied voltage were observed up to the current safety limit, which was reached for biases around + 2 V in Fig. 6a and b, or + 1 V in Fig. 6c.

Figure 6
figure 6

Recorded current as a function of applied bias after hard dielectric breakdown for three different particle/micropillar configurations.

Full size image

For all nanoparticle/micropillar configurations, in situ current measurements consistently remained below the current safety limit when a bias of + 0.1 V was applied. Figure 7 shows the observed current as a function of elapsed time for one representative experiment. During shaded time intervals labeled I through VI, constant bias was continuously applied by a static indenter head. During intermittent non-shaded intervals in Fig. 7, no bias was applied. For intervals I and III at a bias of + 0.1 V, initial currents of around 650–850 µA have gradually decreased to zero. No current was detected during interval II despite the application of a larger bias of + 0.3 V. During interval IV and an applied bias of + 0.1 V, the observed current initially increased from around 800 µA to roughly 820 µA before it gradually decreased again to zero. At the onset of intervals V and VI, the initial currents were zero and began to gradually increase after some incubation time. During interval V, the current reached a maximum of approximately 360 µA for a bias of + 0.1 V before it decreased again to zero. During interval VI, a bias of + 0.2 V was applied. The recorded current gradually increased to around 300 µA before it began to increase more rapidly and exceed the safety limit.

Figure 7
figure 7

Current recorded as a function of time over a series of time intervals (labeled I through VI) during which the provided constant biases were applied. During intermittent times no bias was applied. Instant current measurements were taken at times marked 1 through 6 with biases listed in the adjacent table. For each of these measurements, the recorded current instantaneously reached the current safety limit of 1 mA.

Full size image

At times labeled 1 through 6 marked by crosses in Fig. 7, biases between + 0.2 V and + 1 V were briefly applied (cf. table embedded in Fig. 7). In all cases, the resulting currents instantaneously exceeded the safety limit of 1 mA.

Dynamic in situ experiments

Dynamic growth experiments were conducted with a constant bias of + 0.1 V applied to the same particle/nanopillar configuration utilized for the experiment presented in Fig. 7. After initiation of the constant bias at t0 = 0 s, the indenter was held statically for a duration of 60 s. For the first 50 s, the observed current was mostly constant with small fluctuations around 660 µA (see Fig. 8). During this holding time, the current temporarily decreased by up to 10% for approximately 15 s. After t1 = 50 s, the current began to gradually increase until it reached a maximum of around 760 µA after t3 = 70 s. The current stayed approximately constant for the remainder of the experiment. After t2 = 60 s, i.e., during the near linear current increase, mechanical retraction of the nanoindenter head was initiated. Between t2 and t3 (stage I), no change of the relative position of the nanoindenter was observed. However, after t3 and for the remainder of the experiment (stage II, t3 < t < tf = 93 s) stepwise displacement of the indenter was recorded (cf. Figure 8). During stage II, growth of an elongated nanostructure was observed between the micropillar and the indenter tip. The total displacement of the nanoindenter was approximately 100 nm, which corresponds to an average growth rate for the nanostructure of 3.12 nm/s. Growth terminated at tf = 93 s when the indenter tip and particle agglomerate lost contact.

Figure 8
figure 8

Current through the particle agglomerate at an applied bias of + 0.1 V (black line) and relative position of the nanoindenter head (red and blue marks) as a function of time. Red dashed line mark stages I (t2 < t < t3) and II (t3 < t < tf) of the in situ growth experiment.

Full size image

Figure 9a and b shows the nickel nanoparticle/micropillar configuration before and after nanostructure growth, respectively. Prior to tip-guided growth the nickel particle labeled ‘1’ exhibited round morphology with a diameter of 478 nm. The particle agglomerate also included two smaller particles that are highlighted in Fig. 9a. After the growth experiment, the morphology of particle 1 has changed to an elongated nanostructure with a blunt tip pointing toward the indenter tip (see Fig. 9b). The dimension of particle 1 has increased to 600 nm in the direction parallel to the indenter axis. Arrows included in Fig. 9b highlight small, elongated metal structures (‘spikes’) with relatively blunt tips. Figure 9c exhibits a similar feature on the tip of the indenter head. The spikes are oriented parallel to the moving direction of the indenter indicated by the red dashed line in Fig. 9c.

Figure 9
figure 9

SEM micrographs of nickel nanoparticles supported by a micropillar a before nanoindenter contact and b after conclusion of an in situ growth experiment during which the indenter head was gradually retracted. c Image of the nanoindenter head after in situ growth displaying particle elongation and two high-aspect-ratio nanostructures (marked by arrows) that align with the moving direction of the indenter.

Full size image

Discussion

The experimental results demonstrate tip-guided growth of metal nanostructures by the application of a bias that induces current flow. A prerequisite for current flow though the particle agglomerate is the formation of electrically conducting pathways.

Dielectric breakdown of nickel oxide layer

After exposure to air, a thin native nickel oxide layer is usually present on the surface of nickel nanoparticles that is typically 2–3 nm thick and prevents electric current flow at room temperature [15,16,17,18]. In the presence of a sufficiently high electric field strength across the surface oxide layer, defects may be created and cause dielectric breakdown [16, 19, 20]. At lower voltages, the I–V curves of the nanoparticle/micropillar configuration displayed in Fig. 4a exhibit the characteristics of a Schottky contact, which is a reflection of the semiconducting property of the nickel oxide film [21, 22]. The current increases with each consecutively acquired I–V curve. The same behavior was previously observed by Bonifacio and co-workers and indicates dielectric breakdown of the native oxide layers covering the nickel particles [20]. The gradual current increase observed under constant bias stress shown in Fig. 4b provides additional evidence for soft dielectric breakdown [16, 19, 20]. The removal of oxygen from inter-particle contacts gives rise to conductive pathways through the formerly insulating oxide film [16]. The continuous current increase observed in Fig. 4c suggests additional soft breakdown events that eventually lead to percolation and the formation of highly conducting pathways—a phenomenon often referred to as hard breakdown [16]. Figure 6 reflects the electrical properties of the nanoparticle/micropillar configuration once a significant ohmic pathway is established between adjacent particles. Partial local removal of the surface oxide is consistent with ‘surface cleaning’ effects proposed by Groza and co-workers for field-assisted sintering technologies [23]. Linear regions of the I–V plots in Fig. 6 reflect ohmic contacts with a total resistance of 80–180 Ω. Control experiments with the nanoindenter head contacted directly to the nickel seed layer instead of the micropillar (cf. Figure 2b) have revealed an ohmic resistance for the experimental setup of 27 Ω. Subtraction of the data obtained from the control experiment from the total resistance measurements hence reveals that individual nanoparticle/micropillar configurations are characterized by ohmic resistances in the range between 50 Ω and 150 Ω. The relatively low resistance of the nickel seed layer and the low porosity of electroplated nickel suggests that both Joule heating and geometric changes observed during in situ experiments are mostly confined to the nanoparticle agglomerates.

Rearrangement of particle agglomerates

The parabolic decrease in the current through the nanoparticle/micropillar configuration observed during intervals I, III and IV in Fig. 7 suggests gradual disconnection between the particle agglomerate and the nanoindenter tip. Since neither the ESEM sample stage nor the nanoindenter was intentionally moved during these time intervals, it is concluded that particles have rearranged leading to shrinkage of the particle agglomerate in the direction of the nanoindeter axis. Shrinkage of the agglomerate is further confirmed by the absence of any current flow during interval II (cf. Figure 7) even for an increased bias. Between intervals II and III, the indenter head was re-attached to the agglomerate. However, small mechanical drift of the indenter tip toward the sample stage of the ESEM was observed. In some cases, continuous forward drifting of the indenter has caused electrical re-connection with the particle agglomerate during times when no bias was applied. Due to manual re-connection and/or the forward drifting of the indenter tip, instantaneous current increase is observed for data points labeled ‘1’ through ‘6’ in Fig. 7. Particle rearrangement during intervals I, III and IV has led to repeating detachment from the indenter tip, which suggests that the rate of shrinkage for this specific particle agglomerate is larger than the rate of mechanical forward drifting of the indenter. Detachment of the particle agglomerate from the indenter tip was not directly observed during in situ SEM imaging because the contact area was blocked from the electron beam due to the indenter geometry.

Intervals V and VI exhibited a slightly different behavior (cf. Figure 7). The gradual increase in current with time indicates broader electrical contact between the agglomerate and the indenter. Both unintentional mechanical forward drifting of the indenter and local Joule heating at the contact area causes neck formation and growth between particles through surface diffusion, which increases the observed electric current. Subsequent decreases in current are once again caused by gradual detachment from the nanoindenter head due to shrinking dimensions of the agglomerate.

Temperature distribution

Particle rearrangement and inter-particle neck formation require temperature-activated mass transport. In the absence of external heating, only Joule heating owing to the observed currents contributed to the required temperature raise. However, the direct temperature measurement during in situ SEM experiments was not feasible. Instead, finite element modeling using the COMSOL Multiphysics program packet was employed to estimate the local temperature distribution during the nanostructure growth process. A simplified three-dimensional model displayed in Fig. 10a and c was used to approximate the more complex experimental geometry. The model was comprised of a large cylinder representing the micropillar and a smaller cylinder resembling the growing nanostructure. An immobile solid was considered with no external current or heat source. A bias of + 0.1 V was applied to the surface of the indenter that is in contact with the small cylinder resembling the nanostructure. The bottom surface of the nickel film (see Fig. 10a) was set to ground potential. The AC/DC and the heat transfer modules of the COMSOL software package were used to calculate temperature distributions that arise from Joule heating while heat dissipation was considered to occur by conduction and radiation.

Figure 10
figure 10

a and c Model utilized to represent the geometry of the in situ SEM experiment for finite element modeling of the temperature distribution. b and d Resulting temperature distributions with or without the boundary condition that both substrate and nanoindenter holder remain at room temperature, respectively.

Full size image

Consistent with the observations in Figs. 5a and 6, the electric current density J scales linearly with the applied bias V following Ohm’s law:

$${\textbf{J}} = \sigma {\textbf{E}} = - \sigma \nabla V.$$
(1)

σ is the electrical conductivity and E is the electric field strength. Heat transfer was modeled following Eq. 2.

$$\rho C_{{\text{p}}} \left( {\frac{\partial T}{{\partial t}}} \right) + \nabla \cdot \left( {{\textbf{q}} + {\textbf{q}}_{{\text{r}}} } \right) = Q$$
(2)

\(\rho\) is the physical density of nickel, \({\text{C}}_{\text{p}}\) is the specific heat capacity, \({\text{q}}\) and qr are the heat fluxes by conduction and radiation, respectively, and Q = −J E k ΔT is an external heat source resulting from Joule heating. The radiative heat flux qr is determined by the Stefan Boltzmann equation

$${\textbf{ q}}_{r} = \varepsilon \sigma_{SB} \left( {T_{{{\text{amb}}}}^{4} - T^{4} } \right).$$
(3)

\(\varepsilon\) is the surface emissivity, σSB = 5.67 × 10–8 Wm−2 K−4 is the Stefan–Boltzmann constant, and Tamb is the ambient temperature. Figure 9b shows the results of the finite element model calculation under the assumption that both the substrate and the nanoindenter holder are ideal heat sinks and, hence, remain at room temperature. Under this condition, the model predicts the temperature of the growing nanostructure to be around 450 K. Figure 9d displays the temperature distribution in the absence of any boundary condition, resulting in a temperature as high as 1100 K for the growing nanostructure.

Tip-guided growth of nickel nanostructure

In situ experiments displayed in Figs. 7 and 8 have demonstrated partial sintering of nanoparticle agglomerates. Figures 8 and 9 provide evidence for unidirectional nickel nanostructure growth that is guided by the retracting nanoindenter head. The stable current around 660 µA prior to movement of the indenter (see Fig. 8) indicates a stable electrical contact between the indenter head and the particle agglomerate. Although the actuator of the nanoindenter was engaged for mechanical retraction during stage I, no displacement of the indenter head was observed. Once the retraction force of the actuator was increased (see stage II in Fig. 8), displacement and nanostructure growth were observed. The recorded current through the nanoparticle agglomerate increased before growth initiation owing the increasing effective cross section of the conductive pathway. However, once nanostructure growth was initiated the current remained approximately constant. From this observation, it is concluded that the steadily increasing current during stage I is compensated by an increasing ohmic resistance during stage II because of the uniaxially growing nanostructure.

Results from finite element modeling shown in Fig. 10 suggest that the local temperature of the growing nanostructure remained between 450 K (with boundary condition) and 1100 K (without boundary condition). It is therefore concluded that nanostructure growth is facilitated by solid-state diffusion under tensile stress since the estimated temperature range is significantly below the melting point for bulk nickel (Tm = 1725 K). The observed growth rate during stage II was roughly 3.1 nms−1, which is approximately 5 × smaller than that previously reported by Rufner and co-workers for in situ TEM experiments [14]. The growing nanostructure in the previous study was constrained under compressive stress between a non-moving tip and a relatively flexible amorphous carbon film that supported the nickel nanoparticle agglomerate. Such configuration is considered less space constraint for the growing nanostructure. In this study, however, the growing nanostructure is subject to tensile stress between the rigid substrate and the slowly receding indenter tip. Force measurements were not taken in this study.

Potential mechanisms for nanostructure growth, as previously discussed by Rufner and co-workers [14], include surface and volume diffusion activated by Joule heating, Ludwig–Soret diffusion in a temperature gradient [24] and electromigration [25]. The electron wind force determines that the mass transport direction is the same as electron flow direction, which coincides with the observation of this study. Indenter retraction provided space needed for mass transport. The observation of small metal filaments with relatively high aspect ratios in Figs. 9b and c indicates electroplasticity [26,27,28]. Current densities at the contact area between the indenter tip and the particle agglomerate are estimated to be as high as 1010 A/m2. Such current densities were previously reported to be sufficient to cause electroplasticity under the influence of electron wind force [26,27,28].

To evaluate mass transport mechanisms during the in situ tip-guided growth the flux of nickel, JNi is expressed by the experimentally observed growth rate v using the following equation:

$$J = \frac{m}{A \cdot t} = \frac{\rho \Delta V}{{A \cdot t}} = \frac{\rho Al}{{At}} = v \cdot \rho .$$
(4)

m is mass, A and l are the cross-sectional area and the length of the growing nanostructure, respectively, and t is the elapsed time. According to Fick’s first law, flux can also be expressed as

$$J = - D\frac{\partial C}{{\partial x}} = D\frac{\rho }{l},$$
(5)

with D representing the respective diffusion coefficient and the concentration gradient (\(\partial C/\partial x\)). Assuming an elongation of the nanostructure by l = 100 nm (see Fig. 8), the diffusion coefficient calculated from a combination of Eqs. 4 and 5 equals approximately D = 3.1 × 10–6 m2s−1. A comparison to previously reported lattice diffusion coefficients for nickel [29, 30] suggests that the temperature of the growing nanostructure was in the range of 1250–1300 K. Assuming exclusively nickel surface diffusion as the governing mechanism results in an estimated temperature range between 650 and 724 K. The derived temperature ranges are in excellent agreement with those predicted by finite element modeling. The observed shrinkage of particle agglomerates may have been accommodated by particle coarsening and Ostwald ripening. Both together with inter-particle neck formation and growth required to establish conductive pathways through the agglomerates are governed by surface diffusion, which is consistent with the predicted temperatures. It is, however, expected that multiple diffusion mechanisms are activated at the same time, which is reflected by the relatively wide predicted temperature range.

The major challenge for tip-guided growth of metal nanostructures in the SEM compared to more conventional additive manufacturing techniques is attainable growth rates and maximal attainable sizes for parts manufacturing. However, the experimental results of this study highlight the general feasibility of tip-guided growth modes that can find applications in future manufacturing protocols.

Conclusions

This study reports tip-guided growth of nickel nanostructures that is activated by Joule heating resulting from the application of an electric bias to the utilized nanoindenter head. While previous results have shown nanostructure growth at the tip of an immobile STM tip, this study demonstrates directional growth that follows a moving tip. Nickel nanostructure growth was accomplished by a combination of electroplasticity and Joule heating. A combination of finite element modeling and calculation of diffusion flux derived from experimentally observed growth rates confirms solid-state nickel self-diffusion. The ability to grow nanostructures guided by a moving tip may in the future enable the design and fabrication of complex three-dimensionally geometries with feature sizes that cannot be obtainable by other techniques.