Signature of a single spin. Adapted from Nature 430, 329–332 (2004).
Magnetic resonance imaging (MRI; Milestone 15) and atomic force microscopy (AFM) have been two of the most widely used imaging techniques of recent decades. Building on previous work in nuclear magnetic resonance, MRI can provide three-dimensional (3D) images of systems containing unpaired nuclear spins (such as the protons in water molecules), and is now used extensively in many areas of fundamental research and medicine. AFM, by contrast, is a member of the large family of scanning probe microscopies and can provide detailed images of a wide range of different surfaces.
In conventional MRI, the sample is placed in a magnetic field gradient, which causes all the unpaired spins in the sample to point in the same direction as the magnetic field. A resonant radiofrequency (RF) pulse is then used to 'flip' the spins so that they point in the opposite direction. Over time, the spins flip back again, emitting an RF signal that contains a wealth of information about the sample.
The basic idea of AFM is to scan a flexible cantilever containing a sharp tip over a surface, and to measure the deflection of the cantilever caused by the forces acting between the surface and the tip. Many different forces — including van der Waals, chemical and magnetic — can be exploited in AFM to produce images of the surface with atomic resolution.
In 1991, John Sidles proposed combining these two techniques to make the magnetic resonance force microscope (MRFM). The first MRFM was demonstrated by Dan Rugar, Costantino Yannoni and Sidles a year later. The resolution that is possible in MRI is limited by the use of coils to detect the RF signal. The MRFM gets around this problem by placing the sample on an oscillating cantilever and detecting the magnetic resonance mechanically (by using a laser to measure changes in the resonant frequency of the cantilever), although a coil is still needed to produce the RF field.
In their initial paper, Rugar et al. reported that they had imaged the spatial distribution of electron spins in their sample with a resolution of 19 
m in one dimension. In the years that followed, performance was improved by using lower temperatures, higher gradients and more sensitive cantilevers. By 1994, nuclear spins — which have much smaller magnetic moments than electrons — had been detected and the resolution had improved to 2.6 m in one dimension; in 2007, it reached 90 nm.
The ultimate goal in MRFM is to detect a single nuclear spin, but Rugar and colleagues passed a major milestone in 2004 when they detected the spin of a single electron.
