Binnig, Quate, and Gerber developed a scanning probe technique known as atomic force microscopy (AFM) and as scanning force microscopy. Unlike its predecessor, scanning tunneling microscope (STM), AFM also images nonconducting samples, such as biological specimens, in a liquid environment at molecular and even atomic resolution. Currently, structural information about biological materials at the molecular level is obtained from other microscopic techniques, including electron microscopy, electron crystallography, X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, and vibrational spectroscopy. These techniques require extensive sample preparation and unfavorable operating environments, and they are unsuitable for providing real-time functional information. Molecular function is studied by various molecular biological, biochemical, and electrophysiological techniques, but it is difficult to combine both structural and functional studies in one technique. Moreover, these techniques provide incomplete information about the surfaces of biological macromolecules, the very sites of their interactions with other molecules. In contrast, AFM images the surfaces of biological specimens, where most of the regulatory biochemical and other signals are directed. Other microscopic techniques also image surfaces, for example, the scanning electron microscope (SEM), but AFM images living cells and molecules in a liquid environment at comparable and often greater resolution.
1. Principle of Operation
AFM is based on the general physical principle that the interactive force between two bodies is inversely proportional to some power of the distance separating them and on the physicochemical natures of the interacting bodies. A tip that is sharp on the molecular scale is attached to a cantilevered spring: As it is moved across the surface of the specimen, it is deflected by the interactive forces between the atoms of the tip and those of the specimen (see Fig. 1 of Scanning Probe Techniques). Because the spring constant of the commonly used cantilevers (10-1 to 10- N/m) is much smaller than the intermolecular vibrational spring constant of the atoms in the specimen (10 N/m), the cantilever senses exquisitely small forces exerted by the individual sample atoms. The tip’s deflection is a measure of the forces sensed by the cantilever, which are transduced to generate molecular images.
In practice, a microfabricated cantilevered tip is pressed against a sample surface by a small tracking (loading) force. The tip is raster-scanned in the x-y plane over the specimen by moving the sample beneath the tip or by moving the tip over the sample. The sample’s vertical position (z) is also monitored. The three planes of movement are controlled by a piezoelectric xyz scanner, and the information about the three coordinates is used to create the image (Fig. 1). The cantilevered tip is brought sufficiently close to touch the sample (known as "contact" mode), or it oscillates at a finite distance (>a few nanometers) from the surface ("noncontact" mode). A noncontact mode microscope has the advantage that it does not perturb the sample, but the lateral resolution in these microscopes is poor, and hence they are not commonly used for biological imaging.
Figure 1. Schematic illustration of the operating principle of multimodal atomic force microscopes. (a) Schematic of the combined scanning ion conductance and atomic force microscope. A pipette serves as the probe: A laser beam reflecting from a mirror glued to the back of the pipette provides the deflection signal for the topographic image as the pipette is moved across the surface. The electrode has a nanometer-sized hole. Electrodes within the pipette and in the bath measure electrical currents (R. A. Proksch, R. Lai, P. K. Hansma, G. Morse, and G. Stucky (1996) Biophys. J. 71, 2155-2157). (b) A combined light and atomic force microscope. The first focal point is located inside the upper portion of the piezoelectric scanner. After the positions of the lenses are adjusted, the scanning focused spot accurately tracks the cantilever, and the zero-deflection signal from the four-segment photodiode is independent of the position within the scan area. One of the key advantages of this new AFM is that there is optical access to the sample from above and below. Thus, the new AFM can be combined with an optical microscope of high numerical aperture. The other key advantage is that the sample is stationary during scanning and can be large, so that techniques for on-line perturbations and recordings are easily incorporated.
The deflective force is translated into a detectable signal in several ways. The most common is by an optical lever system (Fig. 1). A monochromatic laser beam reflects from the upper face of the cantilever, the angular direction of which changes as the cantilevered tip undergoes deflections. The reflected beams are captured and converted into electrical signals by position-sensitive multisegmented photodetectors. Such an optical lever amplifies the cantilever’s deflection as much as a thousandfold, so deflections even less than a nanometer are measured.
1.1. Modes of Operation
Using an appropriate feedback system, the cantilever’s deflection is kept constant or left to respond freely to the sensed forces. In the constant deflection mode (also called "constant force mode"), the feedback loop changes the height of the sample (to maintain the constant deflection) by adjusting the voltage applied to the z portion of the xyz piezoelectric scanner. The amount of z change corresponds to the sample’s topological height at each point in the x-y raster. Combining the information from the three coordinates generates the 3-D image.
In the variable deflection mode ("constant height mode"), the feedback loop is open so that the cantilever deflection is proportional to the change in the tip-sample interaction, that is the force sensed by the cantilever. The surface image is constructed from the deflection information. It is called the constant height mode because the z component of the piezoelectric scanner does not change appreciably. This is usually unsuitable for a sample with large surface corrugation (e.g., cells), because the force fluctuations, and thus the cantilever’s deflections, are enormous and often result in disengagement of the tip from the sample.
"Error mode" imaging relies on the imperfection in the feedback loop to operate in the constant-deflection mode. The error signal is amplified to yield contour information in the z plane. In the error mode, the feedback loop gathers high-frequency information that is normally not acquired in the constant deflection mode. This high-frequency information provides details of sharp contour changes (edges) in the sample. Measurements of actual height in error mode imaging are not accurate, however, in contrast to the other modes of operation. The main advantage of error mode operation is that imaging occurs without exerting high forces on the sample.
In "tapping mode" imaging, the cantilever is oscillated at very high frequency, normally near its resonance frequency, as it scans the sample. As the tip approaches the sample surface, its oscillatory amplitude decreases because of energy loss when the tip "taps" the surface. The amplitude of the cantilever oscillation is detected and used by the feedback system to adjust the tip-sample distance for constant amplitude. This ensures a much shorter tip-sample contact time, and smaller lateral forces are exerted on the cantilever. In this way, this mode has been successfully used for imaging delicate and individual macromolecules. The disadvantage of this mode of operation is that the vertical imaging force can be large, thus increasing the possibility of sample damage.
1.2. Sample Preparation
AFM is used to image specimens in aqueous, semiaqueous, or dry conditions. The imaging condition is normally chosen to maintain the specimen in as near a lifelike condition as possible. Where resolution takes priority over the physiological condition, however, the investigator is not as constrained. Imaging conditions also influence the choice of substrate, the stability of the specimen with respect to the tip interaction, and the preservation of the specimen with respect to its physiological or biochemical functions. At present, investigators rely on an empirical approach to find a suitable method, and probably will for some time to come. When it "works", the search stops for the "ideal support" or buffer.
The physicochemical characteristics of the sample determine or suggest ways under which it can be imaged. Problems encountered are as simple as getting the sample to attach to the support. Techniques for sample support include drying down of samples and adsorption to specially prepared surfaces. For example, imaging of plasmid DNA is vastly improved in to both resolution and consistency, under propanol, which increases the humidity and produces a more hydrated condition. This allows reducing the tip-tracking force exerted on the sample to <1 nN , thus decreasing sample deformation.
The interaction of the sample with the support determines the magnitude of the imaging force. If the sample does not adhere tightly to the support, low tracking forces must be used for imaging, or the tip literally sweeps the sample from the support. Originally, graphite (hydrophobic and uncharged), mica (hydrophilic and negatively charged), and glass (usually negatively-charged) were the supports most routinely used. These supports can also be modified chemically to adjust their hydropathy, charge density, and polarity. Today the repertoire has expanded greatly, and examples include gold treated with a variety of agents for DNA imaging. Gold supports maintained under potential (voltage) control have been used for DNA imaging with the scanning transmission microscope, and they may also prove useful for AFM.
The specimen support can also be modified or coated chemically so that it acts as a ligand for the sample and thus orients the specimen in a defined way. It is also possible to use artificial systems to generate constraints where there were none before. Examples include imaging isolated cholera toxin molecules incorporated into synthetic phospholipid bilayers, followed by covalent cross-linking or imaging the vaccinia virus protrusion out of living cells held by a suction pipette.
1.3. Forces in AFM
Interactive forces that deflect the cantilevered-tip are attractive or repulsive, and they vary depending on the mode of operation and the conditions used for imaging.
In contact mode imaging, the tip is deflected mainly by the repulsive forces from the overlapping electron orbitals of the atoms of the tip and sample. The dominant attractive force is a van der Waals interaction due primarily to the nonlocalized dipole-dipole interactions among atoms of the tip and specimen. When imaging in air, (attractive) surface tension is also present because of adsorbed water layers. For imaging in fluids, electrostatic interactions between charges on the specimen and the tip (occurring either naturally or induced by to polarization), osmotic pressure due to charge movements and rearrangements, and structural forces due to hydration, solvation, or adhesion enable a reduction in the net imaging force although both the meniscus and surface tension forces are abolished.
The interplay of local forces determines the stability of the specimen and the resolution. Theoretically the force should be <10-10 N for nonperturbed biological imaging. The sensitivity of AFM is sufficient to record small interactive forces, including the breaking of hydrogen bonds. Imaging in contact mode under liquid, but with a net attractive rather than repulsive force, increases the resolution significantly and has produced true atomic resolution, even with an imaging force of 10-11 N. As explained below, however, successful imaging of cells, membranes, and isolated proteins has been obtained with forces as large as 10 N.
In principle, any movement of the tip caused by its interaction with the sample in the x, y, or z directions provides information about the specimen’s topography. To date, most information has been obtained from z deflections. Improvements in hardware and software have, however, allowed recording movements in the zy or zx planes and measurement of lateral forces for image generation. The contribution of lateral forces to image contrast generation can be substantial.
1.4. Resolution in AFM
1.4.1. Spatial Resolution
The limit of spatial resolution for AFM is not well defined because, unlike conventional microscopies, the images are formed by reconstructing the contours of interacting forces between the specimen and tip. The operating resolution in AFM is defined as the minimum size of two adjacent features that can be distinguished clearly. By selecting a small scan size and suitable operating conditions, one can distinguish two structures that are less than a nanometer apart. Image processing tools used to define resolution in X-ray crystallography and electron microscopy studies may not give correct results for AFM. The operating resolution can be divided into three categories:
1. Instrumental resolution: The lateral resolution is about 1 A and is determined by the limitations of the hardware. The vertical resolution is 0.1 A, and hence molecular perturbations on a sample surface can be imaged.
2. Target resolution: The lateral resolution achieved depends on the characteristics of the tip, the operating environment, and the nature of the specimen. For crystalline solid specimens and many inorganic materials, atomic resolution of 1 to 2 A has been achieved.
3. Resolution in biological specimens: The nature of the biological samples and their preparation play a key role in determining the resolution limits. For the surface of a living cell, the resolution is relatively poor (~10 nm) but greater than that by light microscopy and comparable to that by scanning electron microscopy. In a biological specimen whose the density of particles is high and mobility is limited (e.g., proteins in a membrane), the resolution is comparable to that of a crystalline specimen.
1.4.2. Temporal Resolution
Temporal resolution is limited by the maximum speed at which a specimen can be scanned and still have the tip accurately track surface features. Preliminary studies suggest that the scan speed should be <2.2 um/s for 1 nm spatial resolution on soft and deformable biological materials imaged in aqueous solution. Thus, membrane macromolecules whose dimensions are 10 nm x 10 nm (such as channels and receptors) divided into 10 x 10 pixels (with pixel size ~1 nm) require about 45 to 50 ms to image. However, if only a single line is scanned, the image can be repeated every 4 to 5 ms. Measuring at a single point, rather than scanning, increases the temporal resolution significantly, and hence it is possible to obtain spatial information at very short time intervals. The temporal resolution also depends on the mode of operation (constant deflection or constant height mode), operating environment (solvents, pH, viscosity, elasticity), and the nature of the interactions between tip and sample. In the constant-height mode, the scan speed is limited by the speed with which the deflection of the cantilever changes in reacting to surface features. In the constant deflection mode, the scan speed is limited by how the speed with which the piezo scanner changes its z component. There is ultimately a limit to the temporal resolution imposed by the low-pass filter used to eliminate sampling noise. These filters typically have a cutoff frequency of ~15 kHz , corresponding to a time resolution of 67 us.
Temporal resolution also depends on the material being imaged. Individual molecules at molecular resolution require faster scan speeds than cells at lower resolution. Molecular movements of biological macromolecules can be correlated with their lateral diffusion constant. Lipids in a bilayer have a typical diffusion constant of 10 cm /s , corresponding to a mean velocity of ~2 |um/s. The proteins embedded in natural biomembranes have diffusion constants many orders of magnitude lower (e.g., the acetylcholine receptor in myoblast patches has a diffusion constant <3.0 x 1012 2 cm /s. Thus it is quite possible to obtain images at molecular resolution of proteins and other macromolecules that are properly anchored in a lipid bilayer or immobilized on a substrate.
1.5. Identity of Imaged Structure
Although AFM provides molecular-resolution surface information for crystalline and amorphous materials, it is often difficult to define the nature of individual components, especially if the specimen contains a heterogeneous population of structures. This is the case with most biological systems, except in favorable systems like membranes that contain a crystalline patch of similar protein molecules. For mixed macromolecules, it is essential to compare the information obtained from AFM with that from alternative or complementary techniques, such as structural probes of electron microscopy and X-ray crystallography, biochemical and immunological binding assays, pharmacological labeling, and electrophysiological measurements.
