High-speed atomic force microscopy (HS-AFM) is a form of atomic force microscopy designed to record successive nanoscale surface images rapidly enough to follow dynamic processes. A sharp probe attached to a miniature cantilever scans a specimen while a control system maintains a prescribed probe–sample interaction. By combining small cantilevers, fast scanners, sensitive detection, and high-bandwidth control, HS-AFM can produce image sequences rather than only static surface maps. It is especially important for observing individual proteins and other biological assemblies in liquid, often without fluorescent labels.
Operating principle
Like conventional atomic force microscopy, HS-AFM measures interactions between a sharp tip and a surface. The instrument converts changes in cantilever motion into a topographic image as the tip moves through a raster pattern. Unlike an optical microscope, it does not form its image by collecting light from the specimen; spatial resolution depends strongly on tip geometry, probe–sample interactions, and mechanical stability.
Biological HS-AFM commonly uses amplitude-modulation, or tapping-mode, operation. The cantilever oscillates near a mechanical resonance, and its tip interacts intermittently with the sample. These interactions alter the oscillation amplitude. A feedback controller adjusts the vertical position of the scanner to keep that amplitude close to a selected setpoint. The required vertical displacement provides an estimate of surface height.
Intermittent contact reduces lateral forces compared with continuous-contact scanning, but it does not eliminate disturbance. The tip still transfers force and energy to the specimen. Imaging conditions must therefore balance reliable surface tracking against deformation, displacement, or damage.
What makes imaging fast
Several response times constrain acquisition speed:
- Cantilever response: the probe must react quickly to changes in surface height.
- Amplitude detection: the electronics must estimate oscillation amplitude with little delay.
- Scanner response: the positioning system must move rapidly without excessive vibration.
- Feedback response: vertical correction must follow surface features without instability.
- Data handling: acquisition and processing must keep pace with the scanning hardware.
Improving only one component is insufficient. A rapidly moving scanner paired with a slow cantilever, for example, may produce tracking errors rather than useful high-speed images.
Instrument design
Miniature cantilevers and detection
HS-AFM commonly uses cantilevers much smaller than those employed in conventional AFM. Their low effective mass permits high resonance frequencies while retaining a relatively low spring constant. This combination allows rapid response without necessarily requiring a stiff probe that would strongly deform soft specimens.
A useful approximate timescale for the amplitude response of a driven cantilever is
where is the quality factor and is the resonance frequency. A higher resonance frequency and, within practical limits, a lower quality factor shorten the response time. This expression describes the oscillator response, not the bandwidth of the complete microscope.
Operation in water or aqueous buffer introduces fluid damping. Although damping lowers the quality factor, it can support fast amplitude response. Liquid operation also allows many biomolecules to remain active under suitable chemical conditions.
Cantilever deflection is often measured with an optical beam-deflection system: a focused laser beam reflects from the cantilever onto a position-sensitive detector. Very small cantilevers require correspondingly precise optical alignment and focusing.
Fast scanners and control electronics
The sample or probe is positioned using actuators based on piezoelectricity. Fast scanners are designed to combine adequate travel with high mechanical resonance frequencies. Structural stiffness, low moving mass, vibration damping, and compensation for scanner dynamics help suppress image distortions.
The vertical feedback loop is particularly important. If its response is too slow, the tip may lose contact with descending features or press excessively against rising ones. Controllers may include measures to reduce this asymmetric tracking behavior and prevent prolonged loss of surface tracking.
Specialized amplitude detectors, low-latency electronics, and carefully tuned control parameters reduce delays. Computational correction can compensate for some predictable scanner behavior, but it cannot recover arbitrary molecular events that were not adequately sampled.
Imaging speed and spatial resolution
There is no single frame-rate threshold that universally defines HS-AFM. The term describes a substantial increase in imaging speed relative to conventional AFM while retaining useful spatial resolution and controlled probe–sample interaction.
Biological experiments commonly operate at several to tens of frames per second over small scan areas. Higher rates are possible in specialized configurations, often with reduced field of view, fewer scan lines, or other compromises. Image speed is meaningful only when reported alongside scan dimensions, pixel count, specimen properties, and image quality.
For a raster image, the frame time can be approximated by
where is the number of image lines and is the rate of completed image lines. Retrace and other overheads can increase the actual acquisition time.
The required feedback bandwidth also depends on scan velocity and feature size. A feature with lateral extent , encountered at tip velocity , generates height changes on a characteristic timescale of approximately . Small features scanned rapidly therefore place especially demanding requirements on the instrument.
Vertical sensitivity can reach subnanometre scales, while lateral resolution in molecular imaging is commonly on the order of a few nanometres under favorable conditions. These values are not universal specifications. Tip sharpness, specimen motion, surface attachment, noise, and feedback errors all affect the achieved resolution. Despite the name “atomic force microscopy,” high-speed movies of biological molecules do not ordinarily resolve individual atoms.
Historical development
Atomic force microscopy was introduced in 1986. Its ability to image insulating materials and operate in liquid made it attractive for biological research, but early imaging was generally too slow to capture many molecular motions directly.
During the 1990s and early 2000s, improvements in cantilever miniaturization, scanner mechanics, and feedback electronics established practical routes to faster imaging. A major milestone was the demonstration in 2001 by Toshio Ando and colleagues of an instrument designed for high-speed observation of biological macromolecules.
Subsequent development shifted the technique from proof-of-principle imaging toward direct observation of molecular function. Work on motor proteins, membrane-associated assemblies, and conformationally changing complexes established HS-AFM as a method for studying molecular dynamics rather than merely obtaining structural snapshots.
Different instrument designs have since emphasized different goals, including gentle imaging of fragile biomolecules, larger scan areas, higher frame rates, or rapid characterization of comparatively robust materials. These goals do not necessarily require identical operating conditions.
Applications
Protein motion and molecular machines
One of the technique’s central applications is visualizing changes in protein shape, position, and assembly state. Repeated images can reveal conformational transitions, association and dissociation, directional motion, and changes within an individual molecular complex.
Studies of motor proteins have visualized stepping and changes in molecular organization during motion along filaments. Such observations help connect structural rearrangements with the use of adenosine triphosphate as a chemical energy source.
HS-AFM has also been used to examine rotary molecular machines, including the component of ATP synthase. These measurements provide access to the sequence and timing of visible structural states, although identifying the chemical state associated with each image generally requires additional experimental evidence.
Membranes and supramolecular assemblies
Supported lipid membranes provide relatively flat surfaces for observing membrane proteins and their interactions. Applications include protein clustering, rearrangement of membrane-associated complexes, and stimulus-dependent changes in molecular organization.
The relationship between a supported membrane and a natural cell membrane requires care: the support can constrain motion, alter accessibility, or modify mechanical behavior. Observations therefore describe the experimental preparation rather than automatically reproducing the full cellular environment.
Other targets include protein filaments, self-assembling complexes, nucleic-acid-associated proteins, and molecular aggregates. HS-AFM can follow growth, fragmentation, binding, or remodeling when those processes occur within the accessible spatial and temporal range.
Surface and materials processes
Beyond biology, rapid AFM imaging can monitor processes such as adsorption, dissolution, crystal growth, and changes in polymers or thin films. In these settings, the aim may be to observe an evolving interface rather than a moving molecule. Requirements for force control, scan area, and speed can differ substantially from those of biological imaging.
Sample preparation and interpretation
Most molecular HS-AFM experiments require specimens to remain close to a supporting surface, often mica or a chemically modified substrate. Immobilization must be strong enough to keep the target within the field of view but weak or selective enough to preserve the behavior under investigation.
This creates an important experimental trade-off. Strong attachment can suppress motion or bias molecular orientation, whereas weak attachment can allow molecules to diffuse away or be displaced by the tip. Buffer composition, pH, ionic conditions, and surface chemistry can all influence attachment and activity.
A recorded movie is a sequence of surface-height measurements, not a direct measurement of every molecular coordinate. Interpretation may require comparison with biochemical assays, known structures from X-ray crystallography or cryo-electron microscopy, and results from molecular dynamics simulations.
Measured dwell times and transition rates also require statistical care. Events faster than the acquisition interval may be missed, and distinct molecular states may look similar from a particular orientation. Frame rate alone is therefore not sufficient to establish the temporal resolution of a specific measurement.
Limitations and sources of artifact
Probe-induced disturbance. Repeated tapping can alter reaction rates, deform soft structures, or displace loosely attached molecules. Low average interaction forces do not guarantee that brief force peaks are negligible.
Tip-shape effects. The finite tip geometry broadens lateral features and can merge neighboring structures. Changes in tip condition may resemble changes in the specimen.
Sequential acquisition. A raster image is not captured everywhere at once. Different positions correspond to different times, so motion during a frame can produce distorted shapes or apparent intermediate states.
Restricted access. The method primarily measures an accessible surface. It does not directly reveal the internal structure of a thick specimen, and recessed features may be inaccessible to the tip.
Surface dependence. Adsorption can change molecular conformation, orientation, mobility, or interactions. Surface-bound behavior cannot always be generalized to freely diffusing molecules.
Coupled performance limits. Increasing scan area, spatial sampling, or imaging speed generally increases demands elsewhere in the system. The highest frame rate is not necessarily the condition that yields the most reliable biological information.
Quantitative work consequently depends on scanner and cantilever calibration, assessment of measurement uncertainty, checks for tip artifacts, and comparison across imaging conditions. Apparent changes in height or motion must be distinguished from feedback errors, drift, and changes in probe condition.