Nanoscale magnetic resonance imagingC. L. Degena, M. Poggioa,b, H. J. Mamina, C. T. Rettnera, and D. Rugara,1aIBM Research Division, Almaden Research Center, 650 Harry Road, San Jose, CA 95120; andbCenter for Probing the Nanoscale, Stanford University, 476Lomita Mall, Stanford, CA 94305Communicated by Stuart S. P. Parkin, IBM Almaden Research Center, San Jose, CA, December 1, 2008 (received for review August 21, 2008)We have combined ultrasensitive magnetic resonance force mi-croscopy (MRFM) with 3D image reconstruction to achieve mag-netic resonance imaging (MRI) with resolution <10 nm. The imagereconstruction converts measured magnetic force data into a 3Dmap of nuclear spin density, taking advantage of the uniquecharacteristics of the “resonant slice” that is projected outwardfrom a nanoscale magnetic tip. The basic principles are demon-strated by imaging the1H spin density within individual tobaccomosaic virus particles sitting on a nanometer-thick layer of ad-sorbed hydrocarbons. This result, which represents a 100 million-fold improvement in volume resolution over conventional MRI,demonstrates the potential of MRFM as a tool for 3D, elementallyselective imaging on the nanometer scale.MRFM 兩 MRI 兩 nuclear magnetic resonance 兩 molecular structure imagingMagnetic resonance imaging (MRI) is well-k nown in med-icine and in the neurosciences as a powerful tool foracquiring 3D morphological and functional infor mation withresolution in the millimeter-to-submillimeter range (1, 2). Un-fortunately, despite considerable effort, attempts to push thespatial resolution of conventional MRI into the realm of high-resolution microscopy have been stymied by fundamental limi-t ations, especially detection sensitivity (3, 4). Consequently, thehighest resolution MRI microscopes today remain limited tovoxel volumes ⬎40␮m3(5–8). The central issue is that MRI isbased on the manipulation and detection of nuclear magnetism,and nuclear magnetism is a relatively weak physical effect. Itappears that c onventional coil-based inductive detection tech-n iques simply cannot provide adequate signal-to-noise ratio fordetecting voxel volumes below the micrometer size. This sensi-tivity constraint is unfortunate because MRI has much to offerthe world of microscopy w ith its unique contrast modalities, itselement al selectiv ity, and its avoidance of radiation damage.Despite the many challenges, there is strong motivation toextend MRI to finer resolution, especially if the nanometer scalecan be reached. At the nanometer scale, one might hope todirectly and nondestructively image the 3D structure of individ-ual macromolecules and molecular complexes (9). Such a pow-erful molecular imaging capabilit y could be of particular interestto str uctural biologists trying to unravel the structure andinteractions of proteins, especially for those proteins that cannotbe crystallized for X-ray analysis, or are too large for conven-tional NMR spectroscopy. Nanoscale MRI, with its capacity fortr ue 3D, subsurface imaging, its potential for generating contrastby selective isotopic labeling and its nondestructive nature,would be a welcome complement to the characteristics ofelectron microscopy. The key to pushing MRI to the nanoscaleis detection sensitivity.Recently, a sign ificant breakthrough in magnetic resonancedetection sensitivity has been achieved by using magnetic reso-nance force microscopy (MRFM) (9–13), resulting in single spindetection for electrons (14) and substantial progress in nuclearspin detection (15–24). Despite the great prog ress in nuclear spinMRFM, only one previous nanoscale imaging experiment hasbeen demonstrated, and it was limited to 90-nm resolution in 2dimensions for19F nuclei in an inorgan ic test sample (25). Here,we report that MRFM can perform 3D MRI of1H nuclear spins(protons) in a biological specimen [tobacco mosaic virus (TMV)particles] with a spatial resolution down to 4 nm. This capabilityis enabled by several key technical advances, including thegeneration of magnetic field gradients as high as 4 million Tesla(T) per meter, detailed understanding of the MRFM point-spread function, and application of an image-reconstructiontechn ique capable of converting magnetic force measurementsinto a 3D map of proton density.PrinciplesMRFM is based on mechan ical measurement of ultrasmall(attonew ton) magnetic forces between nuclear spins in a sampleand a nearby magnetic tip. Basic elements of our MRFMapparatus are shown in Fig. 1. The test sample consists ofindividual TMV particles that are deposited onto the flat end ofan ultrasensitive silic on cantilever. The end of the cantilever ispositioned close to a 200-nm-diameter magnetic tip that pro-duces a strong and very inhomogeneous magnetic field. Themagnetic tip sits on a copper ‘‘microwire’’ that serves to effi-ciently generate a radiofrequency (rf) magnetic field that excitesNMR (26). Frequency modulation of the rf field induces periodicinversions of the1H spins in the sample, resulting in a periodic forcethat drives the mechanical resonance of the cantilever. Monitoringthe cantilever oscillation amplitude while mechanically scanning themagnetic tip with respect to the sample in 3 dimensions providesdata that allow the reconstruction of the1H density. The imagingis performed in vacuum and at low temperature (T ⫽ 300 mK).The TMV particles are deposited onto the cantilever insolution and then air dried. (See supporting information (SI)Appendix for preparation details.) As shown in Fig. 1B, thesample c onsists of both whole vir us and smaller fragments. TheTMV particles, which have a rod-like geometry with diameter of18 nm and lengths up to 300 nm (27, 28), were chosen as testobjects because they are physically robust and have a size suitablefor evaluating our imaging resolution. They also serve to dem-onstrate that MRFM is capable of imaging native biologicalspecimens. Approximately 95% of the virus mass consists ofprotein, resulting in a1H density estimated to be␳⫽ 4 ⫻ 1028spins per m3. In the future, rapid f reezing techniques, such asused in cryoelectron microsc opy, c ould be used to better pre-serve the structural integrity of fully hydrated biological samples(29, 30).NMR will only occur if the1H spins in the sample are at thec orrect field for satisfying the Larmor resonance condition:B0(r) ⫽␻0/␥' Bres, where␻0is the rf field frequency, and␥⫽2␲⫻ 42.57 MHz/T is the proton gyromagnetic