|Home | About | Journals | Submit | Contact Us | Français|
Along with more prevalent solid-state nanopores, glass or quartz nanopipettes have found applications in resistive-pulse and rectification sensing. Their advantages include the ease of fabrication, small physical size and needle-like geometry, rendering them useful for local measurements in small spaces and delivery of nanoparticles/biomolecules. Carbon nanopipettes fabricated by depositing a thin carbon layer on the inner wall of a quartz pipette provide additional means for detecting electroactive species and fine-tuning the current rectification properties. In this paper, we discuss the fundamentals of resistive-pulse sensing with nanopipettes and our recent studies of current rectification in carbon pipettes.
Nanometre-sized pipettes can be easily fabricated by pulling borosilicate or quartz capillaries. Nanopipettes have been widely employed in analytical chemistry [1,2], nanoelectrochemistry [3,4], bionanotechnology  and biosensing . They also found applications as tips in scanning probe microscopies [3,7–10] due to the small physical size and needle-like geometry. These features render nanopipettes suitable for local measurements in small spaces, e.g. inside biological cells and vesicles. The very small orifice radius (e.g. a<10nm [11,12]) also allows nanopipettes to be used for resistive-pulse detection of single biomolecules  and current rectification sensing [14–17].
In a resistive-pulse experiment, one measures the ion current flowing through a microscopic aperture . The base current (i0) through the orifice of a nanoscale pore or nanopipette is driven by the voltage applied between two reference electrodes (figure 1a). A nanometre-sized particle can enter through the orifice and partially block the current (figure 1b) in a manner conceptually similar to the Coulter counter . The resulting spike in the current versus time curve (resistive-pulse) can be used to detect the particle. A detectable nanoparticle (NP) must be sufficiently small to pass through the orifice, but at the same time large enough to cause measurable change in the recorded ion current (Δi). (Under certain conditions, the particle translocation may also cause an increase in i). Individual blockage events on the millisecond or sub-millisecond time scale can be recorded using a patch clamp amplifier or a similar device. The analysis of the frequency and amplitude of the current pulses can provide information about the concentration of the detected species, as well as their size and identity.
If the NPs/biomolecules are initially present in the filling solution inside the nanopipette, suitable voltage can be applied between two reference electrodes to deliver these species to the external solution (figure 1c). A current spike is expected to occur every time when a particle is ejected from the pipette, thus enabling precise control of the number of delivered particles. This approach can enable localized delivery of single NPs or biomolecules by using a nanopipette as a scanning ion conductance microscope (SICM) tip [7,8].
Another application of nanopipettes is based on ion current rectification (ICR) that is one of the most important and well-known transport features in conical pores . ICR manifests itself in asymmetry of the current–voltage response, i.e. the ion current at one potential polarity (+V) is much higher than at the opposite polarity (−V) for a potential of the same magnitude. Rectification sensing relies on the ion current sensitivity to surface charge due to its electrostatic interaction with mobile dissolved ions [20–22]. For example, smart bio-inspired nanopores responding to various external stimuli, including pH, temperature, light and potential, have been fabricated by modifying the interior surface with suitable stimuli-response materials . Functionalization of the interior surface of the nanochannels by depositing conductive materials (e.g. carbon) [24–26] or embedding electrodes  has also been performed to control and manipulate the ion transport processes. Thus, the electrostatic interactions can be directly controlled by biasing the inner pipette surface, and at the same time they can provide additional means for optical , conductometric  and electrochemical measurements of the transport processes.
In this paper, we summarize the literature on resistive-pulse and rectification sensing with glass and carbon nanopipettes (CNPs). The fundamentals, including fabrication, surface modification and characterization of the nanopipettes and modelling of their behaviour, are surveyed first, followed by a few examples of pipette applications to resistive-pulse sensing and delivery of nanoparticles/biomolecules. Finally, we discuss the possibility of controlling the surface charge by using conductive CNPs and their potential advantages as rectification sensors.
The response of a nanopipette-based sensor strongly depends on its size, geometry and chemical properties of the inside wall. The electrical resistance and other sensing properties of a nanopipette are largely determined by its narrow shaft adjacent to the orifice. Unlike solid-state nanopores with the fixed thickness, the effective length of the pipette sensing element is not well defined and depends on the pipette angle at the tip (θ; figure 2a). This underscores the need for careful fabrication and characterization of nanopipettes whose visualization is not straightforward [1,3,4,32].
Nanopipettes can be pulled directly from glass or quartz capillaries using a laser pipette puller, e.g. P-2000 produced by Sutter Instrument Co. A suitable type of capillaries has to be chosen for a specific experiment, considering such factors as the material (e.g. quartz or borosilicate glass), thickness of the capillary wall, and the presence or absence of a filament. Borosilicate glass has a lower melting point and is easier to work with because its properties change gradually with temperature. However, because of the softness of this glass, it is difficult to produce ultra-small nanopipettes with a relatively short taper (which may be required to minimize the pipette resistance). Quartz is preferred in this case, but it is very sensitive to uneven heating that might result in asymmetric pipettes. Using quartz capillaries with a thicker wall (greater than or equal to 0.5mm) can help, but the resulting pipette tip may be less sharp because the RG (=rg/a, i.e. the ratio of the class radius at the tip to that of the orifice) is largely determined by the o.d./i.d. of the capillary. It is recommended to pre-clean the capillaries with piranha solution before pulling. In most cases, having a filament inside the nanopipette helps to completely fill it with solution; otherwise removing the air or bubbles from the narrow shaft may be difficult. The taper angle of the fabricated pipettes is typically close to 10°.
With a P-2000 puller, the pulling process is controlled by adjusting five parameters in the program, which are HEAT, FILAMENT, VELOCITY, DELAY and PULL. The HEAT parameter controls the laser output power, and the larger HEAT value corresponds to the higher softening temperature. FILAMENT specifies the scanning pattern of the laser beam. VELOCITY determines the velocity at which the puller bar moves before the hard pull is executed. DELAY controls the timing of the start of the hard pull relative to the deactivation of the laser. PULL determines the force of the hard pull. Generally speaking, to obtain smaller tips, one can increase the value of HEAT, VELOCITY or PULL and/or decrease the value of FILAMENT or DELAY. To control the length of the taper while maintaining the nanometre size, one can try to limit the value of VELOCITY and increase PULL at the same time. Importantly, the parameter values suitable for a given size and shape may not be the same for different P-2000 pullers; and even for the same instrument, the parameters change with time and have to be adjusted occasionally. For instance, representative pulling parameters were reported for approximately 150nm diameter quartz pipettes: HEAT=760, FILAMENT=3, VELOCITY=16, DELAY=128, PULL=140 . However, these parameters vary significantly for different pullers. A pulling program consisting of more than one line can be used to achieve better results.
The surface pKa of glass or quartz is known to vary within a broad range due to different types of the surface silanol groups . To obtain stable and reproducible responses, the glass nanopipettes or nanopores are often modified with specific functional groups (e.g. –NH2, –CN) via silanization reaction. The chemically modified glass nanopores generally display more stable responses, as the surface pKa can be better controlled and the reactive surface sites can be blocked or eliminated . The glass surface can also be modified with analyte-specific binding molecules (e.g. an antibody or a DNA aptamer) to selectively capture and detect analyte molecules with high selectivity and sensitivity [35,36] and with chemical functionalities that respond to external stimuli (e.g. pH or light) [37,38].
Besides silanization, the glass pipettes can be modified by conductive surface films [24–26]. In this way, several nanosensors, including a sensor for dopamine, were recently fabricated by chemical vapour deposition (CVD) of carbon into quartz nanopipettes followed by etching of a micrometre long portion of the quartz wall [39–41]. The outer surface of the glass nanopipette can be coated with a thin gold film by sputtering, and used to concentrate biomolecules at the metallic tip via dielectrophoretic trapping . Coating the interior surface of the glass nanopipette with Au by photochemical reaction under UV irradiation has also been reported .
To produce a CNP, a layer of carbon is deposited inside a pre-pulled quartz capillary by CVD at a moderately high temperature (e.g. 875°C) to avoid softening the nanometre-sized tip [24,26,44]. In most cases, methane was used as a carbon source and argon—as a protector. The Ar flow of 200ccmin−1 was passed through the CVD reaction chamber during heating until the furnace temperature reached 875°C, followed by the mixed flow of CH4 and Ar (3:5 CH4 to Ar ratio). For open CNPs discussed in this paper, the CVD time was about 30min.
Similar to glass nanopipettes (figure 1), the ion current through CNPs can be driven by applying voltage between the internal and external reference electrodes. To avoid a contact between the inner reference and carbon layer, the tip of a plastic pipette was attached to the back of the CNP, sealed with parafilm and filled with the same aqueous solution . The reference (e.g. Ag/AgCl) electrodes are inserted into the plastic receptacle and the external solution. If necessary, the electrical connection to the carbon layer can be made via insulated wire, e.g. enamelled copper wire.
The two main parameters of the pipette geometry are the orifice radius (a) and the pipette angle at the tip (θ; figure 2a), which define the shape of the narrow tapered glass (or quartz) shaft adjacent to its tip and largely determine the pipette resistance. Both parameters can be obtained from transmission electron microscopy (TEM) or scanning electron microscopy (SEM). A representative SEM image of a nanopipette is shown in figure 2b. The magnification in TEM images (figure 2c) is usually higher, and they can provide more detailed information about the pipette inside geometry. The shortcoming of TEM is that the pipette's tip has to be cut before imaging, and, therefore, it cannot be used again after geometry characterization.
Electrochemical methods can also provide information about pipette geometry [1,3,4]. Steady-state voltammetry of ion transfer at the interface between two immiscible electrolyte solutions supported at the nanopipette tip can be used to evaluate the orifice radius from the diffusion limiting current (id) expressed as follows:
where zi, D and c are the charge of the transferred ion i, its diffusion coefficient and bulk concentration in the external solution (e.g. 1,2-dichloroethane; DCE), respectively; and x is a function of RG [45,46]. A representative ion-transfer voltammogram is shown in figure 2d.
θ can be determined in two different ways—from resistance measurements  and by common ion voltammetry . The pipette resistance can be found from current–voltage (i–V) curves recorded in the same aqueous solution used for resistive-pulse experiments. For larger nanopipettes (e.g. a>~100nm; figure 2e), the i–V curves are essentially linear, and the total resistance, R, can be extracted from the slope. i–V curves obtained with smaller pipettes are nonlinear because of current rectification (figure 2f) [14,49–51]; whereas an essentially linear part of such a curve recorded at low applied voltages (e.g. ±20mV; the inset in figure 2f) can be used for determining R. The total pipette resistance comprises two components, i.e. the resistances of the inner and outer solutions:
Rext is entirely determined by the radius and solution conductivity (κ) ,
The resistance of the internal solution can be estimated as :
When a CNP is fabricated by CVD, the carbon layer first deposits on the interior quartz wall, and gradually fills the narrow shaft adjacent to the nanopipette orifice. Depending on the deposition time and other parameters, a carbon film may also appear on the outer pipette wall. The optimal geometry of the deposited carbon depends the intended pipette application. For resistive-pulse and rectification sensing, only a thin carbon layer needs to be deposited on the inner quartz wall with an open path left in the middle of the pipette and no carbon coating on its outer wall. The geometry of a CNP can be characterized by TEM and voltammetry. For instance, figure 3 shows two different types of CNPs, an open pipette (figure 3a) and an almost completely filled pipette (figure 3b) with a nanometre-sized cavity at the orifice (electrochemical nanosampler ). In both images, the outer pipette wall appears to be carbon-free. The details of the inside geometry and the thickness of the carbon layer can be revealed by TEM images taken from different angles and/or three-dimensional TEM tomography .
In a typical resistive-pulse experiment, the nanopipette is filled with aqueous solution and dipped into the same solution containing the analyte species. The resistive-pulse signal usually includes a number of short (e.g. ≤1ms) blips during which the measured current changes by a few picoamperes. A patch clamp amplifier (e.g. Multiclamp 700B by Molecular Devices [30,31]) is a suitable instrument for recoding pA-range currents on the microsecond time scale. It is used in the voltage-clamp mode to apply voltage between the internal and external reference electrodes and to measure the resulting current. The signal can be digitized using an analogue-to-digital converter at a sufficiently high sampling frequency (e.g. up to 500kHz with a Digidata 1550A; Molecular Devices). A low pass filter can be used to decrease the noise, but its bandwidth (typically, 1–30kHz) must be high enough to ensure that shorter current pulses have not been missed or filtered out.
The current–voltage (i–V) responses of the nanopipettes in rectification sensing experiments can also be measured with a patch clamp amplifier. However, these experiments do not require high acquisition frequency and, therefore, can also be performed with a low-current potentiostat.
As an NP enters the pipette, the displacement of the electrolyte solution from the narrow shaft results in the increased resistance of the inner solution (the effect on the resistance of the outer solution is negligible). The question here is what relative change in ion current (Δi/i0) and translocation time (τ, i.e. the half-width of the resistive-pulse) can be expected when such a particle enters a pipette with the radius a and angle θ? The extensive literature on mathematical modelling and numerical simulations of particle translocation through conical and cylindrical nanopores has recently been reviewed . An approximate mathematic model developed for conical pores [55,56] was later adapted for a spherical particle translocating through a nanopipette  to calculate the shape and amplitude of the current pulse (figure 4). Similar to conical pores, the pulses are expected to be asymmetrical with the fast initial decrease in current followed by a slow relaxation (tail). This shape reflects the sharp increase in the pipette resistance that occurs when the particle enters the tapered narrow shaft and subsequent slower decrease caused by the further NP progress into the pipette.
The amplitude of the current pulse depends on the a, θ and rp values. While the radius of the transferable particle has to be smaller than that of the pipette, the larger the rp/a ratio the larger the change in pipette resistance and hence Δi/i0 (figure 4a). For a given rp/a, the larger the θ the higher the resistive-pulse signal, as shown in figure 4b. For the same orifice radius, a larger taper angle corresponds to a lower pipette resistance (equation (2.4)) and a larger relative change in resistance; and, thus, a more significant current blockage. This data can be used as a guideline for selecting pipettes with a suitable geometry to generate detectable current pulses with a sufficiently high signal-to-noise ratio for an expected range of rp values in a given NP sample.
Figure 4c shows an experimental current transient fitted to the aforementioned approximate theory. The effective mobility of the NP found from the fit was very low, suggesting a significant effect of the electroosmotic flow, whose direction was opposite to that of the electrophoresis, resulting in the diminished value of the effective mobility .
The surface charge is known to play a key role in the ionic transport processes through nanochannels. Originating from the surface group deprotonation and/or ion adsorption, the excess surface charges attract the counter-ions and repel the co-ions in the solution layer adjacent to the pipette wall. Consequently, the counter-ions are the main charge carriers in nanopipettes and other types of nanochannels, and this selective ion transport exhibits a number of interesting features, such as ICR , concentration polarization  and memory effects .
The ICR is believed to stem from the asymmetric geometry and/or surface charge distribution inside a nanochannel . In a tapered nanopipette with a negative net surface charge, at the negative bias (defined as the potential difference between the internal and external references), both the applied voltage and the field generated by negative surface charges would drive the ingress of the cations into the pipette and the egress of anions to the external solution. However, at a positive bias, the applied voltage would drive the cation egress from the pipette, which is impeded by the negative surface charge. The combined effects of the applied voltage and excess surface charge thus lead to the high and low ion currents at the negative and positive bias, respectively. Besides the surface charge  and geometry , the ICR was shown to depend on other factors, including electrolyte concentration [49,62], pressure , applied voltage and scan rate [62,64]. Such dependences are potentially useful for designing rectification sensors.
Although the effect of surface charge density on ICR is fundamentally important for rectification-based sensing, direct measurement of this parameter in nanopipette experiments is challenging. The correlation between the ICR response and surface charge density was investigated by finite-element simulations based on solving the Poisson and Nernst–Planck and Navier–Stokes equations (PNP–NS model) numerically . Such simulations offer fundamental insights in the physical origins of the ICR features , and help elucidate the ionic concentrations, fluxes and potential distribution inside the nanopore. If the pore geometry is known, the surface charge can be determined by fitting the experimental current responses to the simulated curves [51,65]. The ICR was found to be very sensitive to the surface charge density and distribution , so that a 1e−1nm−2 (=160mCm−2) change in the former can induce a nanoampere change in the ion current . Such a high sensitivity suggests that by measuring ICR at conductive nanopipettes (e.g. CNPs), one may be able to detect low concentrations of redox species.
Solid-state nanopores have been used for DNA sequencing. Detecting and sequencing DNA with nanopipettes may be more challenging because of the geometry of its sensing element is less well defined. Karhanek et al.  were first to use nanopipettes for resistive-pulse detection and identification of single DNA molecules (24-base single-stranded thiol-modified) labelled with 10nm AuNPs. They stressed several advantages of the nanopipette over the nanopore, including its mechanical strength and ease of fabrication. Another advantage—the small physical size and needle-like shape enabling precise positioning and experiments in small spaces—was utilized by Gong et al. , who developed an integrated nanopipette-microfluidic device in which the pipette could be positioned at any point within a microfluidic channel. In this way, it was possible to detect and discriminate between DNA molecules of varying lengths travelling through the microfluidic channel.
Resistive-pulse detection of the folding state of double-stranded DNA was demonstrated by the Keyser group . They studied the translocation of λ-DNA which was driven by an electrophoretic force through the nanocapillary and showed that the folding state of single λ-DNA molecules can be seen from the changes in ionic current. It was concluded that nanocapillaries are a promising alternative to solid-state nanopores for label-free single molecule analysis.
The Radenovic group  reported the use of small (≥10nm diameter) quartz nanopipettes prepared by SEM-induced shrinking of larger nanocapillaries (the technique recently developed in the same laboratory ) for resistive-pulse sensing of single dsDNA molecules. A typical problem in nanopipette fabrication with a laser puller is that the pipette with a smaller orifice radius usually has a longer tapper, resulting in the higher solution resistance and lower signal. In addition to an order of magnitude decrease in the orifice radius, SEM-induced shrinking of a quartz nanopipette caused significant shortening of its narrow shaft. In this way, the estimated sensing length of a nanocapillary was decreased to approximately 32nm, and the signal-to-noise ratio for DNA translocation was comparable to or even better than that obtained with standard nanopores in silicon nitride membranes , as shown in figure 5.
Rudzevich et al.  counted 100nm diameter synthetic liposomes and evaluated their size and velocity distributions by resistive-pulse measurements with a 160nm diameter borosilicate pipette (figure 6a,b). The current pulses were a few milliseconds long and quite similar to those in the recordings obtained with 80nm SiO2 NPs. Compared with light scattering and laser doppler velocimetry, which can simultaneously measure particle size and velocity, the resistive-pulse based method can work with smaller particles, and the sample volume can be quite small (less than 1µl) due to the microscopic size of the pipette tip.
Chen et al.  investigated the translocation dynamics of small unilamellar phospholipid vesicles (SUVs) through significantly smaller pipettes with the orifice diameter ranging from approximately 14 to 72nm. No translocation was observed with 14nm and 31nm pipettes whose orifice diameter was much smaller than that of the SUVs (approx. 50–60nm diameter; figure 6c,d). Several second long translocations were observed when the diameter of the nanopipette orifice was similar (figure 6e) or slightly larger (figure 6f) than those of the SUVs. These events were by orders of magnitude longer than those reported in ref. , suggesting that the physical interaction between the nanopipette wall and the vesicles plays an important role in the translocation process. These results are consistent with the earlier report by the White group, showing resistive-pulses for the pressure-driven translocation of 367±79nm radius liposomes through a smaller (208nm radius) nanopore opening . The translocation time significantly increased with decreasing temperature and became several seconds long close to the lipid bilayer transition temperature.
The Gyurcsányi group  used resistive-pulse sensing with quartz nanopipettes for sizing and quantitation of polymeric NPs and polioviruses. This approach was validated using the suspensions of monodisperse latex nanoparticles as well as mixtures of closely sized NP dispersions. The authors focused on refining the sensing theory to obtain precise sizing information from resistive-pulse data and attained the resolution similar to that of SEM. The nanopipette-based resistive-pulse sensing resolved the mixtures of NPs with narrow size distributions (tested down to 13nm), where dynamic light scattering and NP tracking analysis provided a single peak distribution. The mean diameter of the poliovirus particles was 26±2nm from resistive-pulse experiments and 24.3±3.3nm from TEM images, as opposed to 32.6±4.5nm obtained by dynamic light scattering.
Nanopipette-based resistive-pulse sensors were used for selective detection of antibodies to peanut allergens (IgY) . Peanut allergens are glycoproteins that elicit immune responses elevating IgE antibody levels in the human body. Ara h 2–2 peptide-modified gold nanoparticles (AuNP) were employed to capture IgY offline from solution for detection using a nanopipette-based sensor. Resistive-pulse recordings obtained with AuNP, AuNP-peptide allergen and AuNP-peptide-IgY particles (figure 7) were used to demonstrate the capacity of nanopipettes for label-free detection of antibodies. The current pulses produced by these nanoparticles occurred at different translocation voltages (positive for AuNP and AuNP-peptide versus negative for AuNP-peptide-IgY) and exhibited opposite signs of Δi, as shown in figure 7. These major differences enabled selective resistive-pulse sensing of antibodies with nanopipettes. If this behaviour is common to other protein-modified nanoparticles, the developed sensing platform can be useful for detecting other types of antibodies and protein biomarkers.
In most reported resistive-pulse experiments, selective detection of biomolecules was achieved by functionalizing the nanopore (e.g. by immobilizing antibodies on its surface ). By contrast, in  the analyte (IgY) was selectively captured offline using peptide-modified AuNPs. After the analyte capture, the particles could be washed to remove potential interferences and avoid complex mixtures . In this way, selective resistive-pulse sensing of biomolecules can be attained using simple glass or quartz pipettes without laborious surface modification procedures while also avoiding some experimental issues that hinder the experiments in biological media, e.g. clogging and non-specific adsorption of proteins.
A conceptually similar strategy was employed to develop a resistive-pulse sensor for a cancer biomarker—vascular endothelial growth factor C (VEGF-C) . For VEGF-C detection, monoclonal primary antihuman VEGF-C antibodies were covalently immobilized onto carboxylate-functionalized AuNPs . After VEGF-C capture, AuNP-antibody (AuNP-mAb) and AuNP-antibody-VEGF-C particles coexisted in a dispersion. Both AuNP-mAb and AuNP-aAb-VEGF-C particles produced current blockages in nanopipettes with a wide range of radii. Careful selection of the pipettes with well-characterized geometry was essential for selective detection of VEGF-C because of relatively small differences in the current pulses produced by the two kinds of particles. Differentiating between AuNP-mAb-VEGF-C and AuNP-mAb bioconjugates was difficult because of similar sizes and zeta-potentials of these particles. Nevertheless, by choosing the right size of the nanopipette, it was possible to selectively detect either AuNP-mAb (with a<~60nm; the orifice was too small for the translocation of antigen-conjugated NPs) or AuNP-antibody-VEGF-C (with a>~140nm at which the small pulses produced by AuNP-mAb were obscured by the noise) .
In rectification sensors, the detection of the analyte is typically based on its effect on the surface charge density and, therefore, the extent of the ICR. Imidazole-modified glass nanopipettes were employed for Co2+ sensing, as this ion can bind to the negatively charged surface and reduce the effective charge density . Interestingly, decreasing the solution pH resulted in the release of adsorbed Co2+ and regeneration of the binding sites on the pipette surface. These results suggest the possibility of developing reversible nanopore sensors based on recognition elements with intermediate binding affinities. The Baker group also showed that a specific DNA sequence can be detected by the dendrimers-modified nanopipettes in which stable electrostatic DNA-dendrimers complexes can be formed on the surface . Compared with the poly-T and single-base mismatch DNA sequence, the perfect complement sequence caused drastic changes in the i–V response indicative of high selectivity of the nanopipette sensor. On the minus side, the nanopipette sensors could only detect high concentrations of the DNA analytes (micromolar), and additional electrophoretic or convection assisted analyte transport may be needed to improve the detection limit.
The pH and temperature gated nanopore sensors were prepared by attaching ‘smart’ homopolymer poly[2–(dimethylamino) ethyl methacrylate] to the inner wall of the conical nanochannel . The pKa of the polymer coating was about 7.0–7.5, and the sign and density of the surface charge could be changed by varying the solution pH and, thus, influencing the rectifying behaviour of the pipette. Additionally, the polymer film could be reversibly switched between the swollen state at lower temperatures (<40°C) and the collapsed state at higher temperatures (>50°C). These conformation transitions resulted in the high ionic conductivity ‘on’ and low conductivity ‘off’ states of the glass nanopores at different temperatures and pH.
Although the current rectification in glass nanopipettes was reported back in 1997 , only a few applications of this phenomenon to sensing have been developed to date. Further progress in this area requires better understanding of the quantitative relationship between the surface charge and the extent of rectification in i–V curves.
The geometry of nanopipettes make them especially useful for precise delivery of ions, molecules and particles to surfaces and into small volumes (e.g. inside biological cells) [1,4,5]. The delivery of various nano-objects (e.g. NPs, vesicles and large biomolecules) can be controlled by monitoring the resistive-pulse signal to count the dispensed species (figure 1c). Thus, the delivery of single NPs from nanopipettes to external solution was demonstrated . By using a pipette with a suitable aperture radius, current pulses with the amplitude much larger than the noise level were obtained for the ejection of 10nm AuNPs as well as much larger AuNP-mAb-VEGF-C particles. Similar to resistive-pulse sensing experiments , the pipette radius suitable for the NP delivery was approximately 1.5–3 times that of the translocating particle.
To locally deliver NPs, one has to bring a nanopipette close (i.e. within a few pipette radii) to the sample surface and precisely position it at the right distance above the desired spot. The precise pipette positioning can be attained by using it as a SICM tip. In the non-modulated DC feedback mode of SICM, the ion current flowing through the pipette (i) is measured as a function of the distance between the nanopipette orifice and the sample surface at a constant potential value applied between the internal and external reference electrodes . When the pipette approaches the substrate impermeable to ions, i decreases because of increasing ohmic resistance. By fitting the current versus distance curve to the theory, one can establish the distance scale and precisely position the SICM tip in a close proximity of the substrate . The shape of the current–distance curve depends on the pipette geometry (i.e. a, θ and RG values). The theory was developed in  for RG=10, which is much larger than a typical value for a quartz or glass nanopipette. The following equation is valid for a wide range of RG values :
where L is the distance between the pipette orifice and the substrate normalized by a, and i is normalized by the ion current value at infinite L.
McKelvey et al.  combined resistive-pulse recording with electrochemical collision experiments to attain controlled delivery of nanoparticles to electrochemical interfaces along with particle size measurement. A 700nm radius pipette containing 241nm radius polystyrene particles was positioned above a 405nm radius Pt disc ultramicroelectrode (UME) in solution containing redox mediator (ferrocenemethanol; FcMeOH) for pressure-controlled delivery of single nanoparticles (figure 8a). Two signals were collected for each detected particle: a resistive-pulse spike at the moment of the particle passing through the orifice and the change in the redox current at the UME when the delivered particle collided with its surface (figure 8b). This approach should be useful for electrocatalytic studies at single NPs, provided that significantly smaller particles can be delivered.
Ivanov et al.  used nanopipettes for label-free detection and delivery of single DNA molecules. Unlike traditional resistive-pulse experiments, where a constant DC voltage is applied between two reference electrodes, alternating positive and negative potential pulses were employed in  to minimize clogging of the nanopipette and attain controlled delivery of individual DNA molecules. The time between the application of the negative potential and the detection of the first translocation event was found to be controllable and strongly dependent on the magnitude and duration of the applied pulses. The time-resolved delivery of individual oligonucleotides from solutions with concentrations as low as 3pM was demonstrated. An additional means for monitoring the delivery process was fluorescent imaging of dispensed DNA that was labelled with YOYO-1 fluorescent dye.
An alternative strategy for controlling the delivery of ions made use of a nanopipette with a carbon ring surrounding its orifice . The electroactive ionic species delivered from the pipette to external solution were electrochemically detected at the ring, and fluorescein was utilized to optically map ion enrichment/depletion in the nanopipette tip. An important finding was that the concentration polarization depends on surface charge, and delivery can be controlled by modifying the charge density and polarity. The negative charge on the inner wall of a bare quartz nanopipette enhanced delivery of the anions and diminished the flux of cations, while the positive charge on polyethylenimine-modified nanopipettes produced the opposite effects. Understanding this behaviour is important for nanopipette delivery in biological applications.
The usual way to control the electrostatic interactions and electrokinetic transport processes is to modify the interior surface of a nanopore (or a nanopipette), for example, by covalent attachment of charged molecules. Although useful for selective ion transport control, chemical functionalization is laborious and does not allow one to easily vary the surface charge density and polarity. An alternative approach is the field effect modulation of surface charge by attaching a gate electrode to the nanochannel and insulating it from the solution . The surface charge distribution and the resulting transport features, including ionic rectification and osmotic flow inside the nanopore, can be controlled by the potential of the gate electrode. On the other hand, if the gate electrode is exposed to the solution, it can serve as a working electrode to enable the quantitation of the electroactive analytes in addition to ion current-based resistive-pulse or rectification sensing. A CNP produced by coating the inside of the pulled quartz capillary with a nanometre-thick carbon layer can be employed for a wide range of electrochemical and sensing applications [24–26,40]. The extent of ICR in a CNP depends on the potential applied to carbon, which determines the surface charge and electrical double-layer at the pipette wall (figure 9). Thus, a CNP can work as a tunable resistive-pulse or rectification sensor whose properties can be adjusted to detect a specific analyte. The ability to vary the charge on the carbon surface can be useful for changing the translocation time of charged analytes, e.g. slowing down the translocation of DNA to facilitate its sequencing. Another interesting possibility is to control the electroosmotic flow inside the pipette tapered shaft and study its effects on the ion current, including the intriguing ‘electroosmotic flow separation’ that can cause the increase in i when the SICM probe approaches an insulating surface .
Because of a very large (many orders of magnitude) difference between the surface areas of the nanopipette orifice and its inner conductive wall exposed to the solution, applying voltage between two reference electrodes and simultaneously biasing the carbon surface is not straightforward. It was found recently that the charge distribution on the carbon wall and, consequently, the rectifying behaviour of the nanopipette, can be influenced by adding very low concentrations of redox species to the solution . In figure 10, the extent of rectification of the ion current through the 50nm radius CNP changes significantly upon adding sub-micromole concentrations of K3Fe(CN)6 to the 10mM KCl solution. While the 100nM concentration of ferricyanide is too low to be detected amperometrically, it was found to change significantly the open circuit potential of the carbon layer (−370mV versus Ag/AgCl reference in 10mM KCl versus −320mV in 100nM K3Fe(CN)6+10mM KCl). The ion current at positive voltages (the high conductivity state) decreased markedly with increasing Fe(CN)63− concentration (figure 10).
Besides the possibility of the rectification control without external biasing of carbon surface, this approach is potentially useful for sensitive detection of electroactive species by measuring the ionic conductance of the CNP.
The number of applications of glass nanopipettes to resistive-pulse and current rectification sensing is increasing steadily because of their versatility, robustness and ease of fabrication. In most articles published to date, the pipettes were used for experiments in the bulk solution rather than for measurements in small spaces or localized delivery of single nano-objects. The needle-like geometry of nanopipettes renders them most useful for such applications, and more of them are likely to be reported in the near future.
A major obstacle to the progress in nanopipette-based sensors is the lack of quantitative description of the effects of the surface charge and morphology on the ion current. The improved understanding of the relationship between the surface properties and transport processes in nanocapillaries is needed to enable the development of tuneable resistive-pulse or rectification sensors for specific analytes. Recent progress in preparation and characterization of glass pipettes coated with conductive films, such as CNPs, may also lead to advances in these areas.
We declare we have no competing interests.
The authors gratefully acknowledge the support from the National Science Foundation (CHE-1300158 and CHE-1416116; M.V.M.) and start-up funding from California State University Los Angeles (Y.W.).