Topic #2:  Spin-Polarized Scanning Tunneling Microscopy

Scanning tunneling microscopy (STM), a technique for imaging the structure of a surface with atomic-scale resolution, was invented in 1981 by researchers at IBM Zürich Research Laboratories.  The importance and potential of the technique was quickly realized; as a result, Gerd Binnig and Heinrich Rohrer were awarded the Nobel Prize in 1986.  Since its invention, STM has become more and more popular in research labs to investigate surface structures as the ultimate high-resolution imaging tool.  In addition, the STM is now being used not only for imaging, but also for manipulation, of the atoms on a surface.

The STM produces data consisting of height information (z) as a function of lateral position (x,y) on the surface.  Correct calibration of the data can give quite accurate measurements.  Thus, while a picture is worth a thousand words, STM “images” are even more meaningful, as they are related to the true “structure” of the surface.  For example, single atomic-height steps can easily be measured in STM images; if a single step is 2 Å in height, then a double step will measure 4 Å in height.  In fact, it is tempting to interpret STM data as topographic information.  However, since the tunneling in STM is based on quantum mechanics, an STM image is not simply a topograph.  While interpreting STM data as topographic is a good starting point, in fact STM images are really maps of the local density of states (LDOS) of the surface.  As such, two atoms whose nuclei lie within the same horizontal plane may not appear in the image to be located at the same topographic (z) height since their LDOS are probably going to be different.  This can in fact comprise a method of chemical specificity, although chemical identification of atomic features in STM images has proven to be in general extremely difficult.  Additional information about the surface under investigation is usually required to make a chemical assignment.  Still, this information can often be deduced.

The height profile measured by the STM is given by contours of constant tunneling current; this is the constant current mode of STM.  While the tunneling current is sensitive to the LDOS of the surface, it is also dependent upon the LDOS of the tip.  This is because electrons occupying states (therefore filled states) on one side of the tunneling gap (barrier) will tunnel into empty states on the other side of the gap.  The total tunneling current is a convolution integral of filled and empty states of sample and tip.  In normal STM imaging, the LDOS of the tip may not be of great interest and is often assumed to be a constant.  However, in spin-polarized STM (SP-STM), the LDOS of the tip is very important.  The reason for that is explained below, but first is given a brief description of what SP-STM is and does.

SP-STM has been explored since about 1990.  The most significant body of SP-STM work has come from the group led by Prof. Roland Wiesendanger of Hamburg, Germany.  More recently, other groups have begun to have success with the technique, including that of Ohio University NIRT project director Art Smith.  But now let’s discuss the method.

In SP-STM, both the surface and the tip are spin-polarized – that is, the number of spin-↑ and spin-↓ states are unequal for both tip and surface.  Along a given quantization axis, one may take the difference between spin-­ and spin-↓ LDOS to arrive at the magnetization LDOS (MLDOS).  Then along the same quantization axis, one may similarly compute the MLDOS for the sample surface as a function of the position.  Now, if the MLDOS is (+), we can think of it as net spin-↑, whereas if it is (-), net spin-↓.  In SP-STM, it turns out that the tunneling current equation has an additional term inside the convolution integral which is the vector dot product of the MLDOS of the sample and the MLDOS of the tip.  For example, if the net spins of sample and tip are parallel, the tunneling current will be increased; whereas, if the net spins of sample and tip are anti-parallel, the tunneling current will be decreased.  Because of this term, the tunneling current is sensitive to the spin directions on the sample surface.  This allows SP-STM images to show spin-polarization.  Since STM has spatial resolution down to the atomic scale, this method is ideal for the study of nanoscale magnetism (nanomagnetism).

The method of SP-STM requires of course that the tip have a net spin-polarization, or a magnetization.  This can be achieved using ferromagnetic (FM) or antiferromagnetic (aFM) tips.  However, the most success has been achieved through the use of FM or aFM tip coatings [see e.g. A. Kubetzka et al., Phys. Rev. Lett. 88(5), 057201 (2002)].  Thin layers of magnetic material are deposited onto the apex of the tip through the use of thermal evaporators.  Various types of tip coatings have been applied with success.  In fact, it is possible that the control of the spin-polarization of the tip is one of the major roadblocks to increased success with SP-STM.  Wiesendanger et al. have reported their methods of tip preparation, in which they can controllably prepare tips having spin vector parallel to the sample surface or perpendicular to the sample surface.

SP-STM has been applied so far to a variety of different types of magnetic surfaces, including those which are FM and also those which are aFM.  In the case of FM surfaces, it is common for the surface to form domains having their MLDOS in different directions.  Moreover, as a function of the position crossing between domains, the spins usually undergo a gradual transition by means of spin rotations – these regions of spin gradient are referred to as domain walls (DWs).  Wiesendanger’s group has studied domains and DWs in various metallic systems such as Fe/W, making use of in-plane and out-of-plane polarized tips to resolve the in-plane and out-of-plane components of the surface magnetization [see e.g. M. Bode et al., Phys. Rev. Lett. 89(23), 237205 (2002)]. 

In the case of aFM surfaces, it is possible that the spin direction within one atomic row will be opposite to the spin directions of the neighboring (adjacent) atomic rows.  Resolving such ultra-short period (atomic-scale) magnetization reversals was reported first in a metallic system [see Heinze et al., Science 288, 1805 (2000); also D. Wortmann et al., Phys. Rev. Lett. 86(18), 4132 (2001).] and more recently in the case of a transition metal nitride system [Mn₃N₂(010)] by Smith’s group [see Yang et al., Phys. Rev. Lett. 89, 226101 (2002); also A. R. Smith et al., Surface Science 561, 154 (2004).].  Shown here is a SP-STM image obtained by our group on the Mn₃N₂(010) surface showing the spin resolution.  Each of the rows shown is chemically equivalent, and with a non-magnetic tip (not shown here) appear the same.  But with a magnetic tip (as shown here), the rows alternate in height (which appears as an alternation in brightness).

Yang et al. showed that this SP-STM atomic-scale image contains both the magnetic component and the non-magnetic component of the surface LDOS.  Moreover, it was shown how to separate these two components.  This was performed by averaging the horizontal line profiles (in x-direction) along the vertical (y) direction.  This averaged total line profile is then shifted by half the magnetic period along the x-direction.  The shifted profile is then subtracted from (added to) the original profile and then divided by 2 to produce the magnetic (non-magnetic) component.