Using an atom probe microscope to characterize read heads
09/01/2004
 |
The accurate characterization and control of buried interfaces is increasingly important to manufacturers of devices at the nanoscale. With thin films only a few atoms thick, device properties now are dominated by the properties of these interfaces between various films. A new atom probe microscope allows users to characterize in a meaningful manner the behavior of these interfaces. This information speeds up process development and pinpoints failure analysis.
The three forces of new structures, new materials, and ultrathin films have combined to create a sort of "perfect storm" for technologists in the magnetic storage industry. New structures such as TGMR, CPP-GMR, spintronics, and ballistic GMR have device engineers worrying about how to optimize device performance and extend reliability. Exotic materials including ruthenium, platinum, titanium, and nano-oxide layers (NOL) have process engineers and capital equipment manufacturers scurrying to control film properties even as these layers shrink to a few individual atoms. Lastly, "new physics" are being observed as interactions between individual atoms and interface effects increasingly dominate thin-film properties and, therefore, device performance.
The problems surrounding these materials and structures are exacerbated because no analytical techniques can measure atomic interactions and the behavior of films at the atomic scale — especially at buried interfaces.
A new microscope called the local electrode atom probe (LEAP) microscope can image and chemically identify each individual atom in the bulk of a material in a timeframe suited for industrial applications. This atom probe microscope is especially suited for emerging analytical problems in the magnetic storage industry. It is being adopted by leading manufacturers to solve a variety of problems. For example, it can identify composition variation around grain boundaries, and analyze it along NOLs; pinpoint contaminants and quantify roughness in a film stack, and analyze the interdiffusion at interfaces.
Why study buried interfaces?
Interfaces need to be atomically smooth — in a nutshell, this is why the study of buried interfaces is important. The need for atomically smooth interfaces can be studied in the read heads used in disk drives. Since the mid-1990s, the read head of a disk drive has been separated from the write head. While write heads are still based on the principle of induction, read heads are now based on the principle of giant magnetoresistance (GMR). To understand why interfaces are important, it first helps to understand how GMR devices operate.
Principle of GMR operation
 Figure 1. Schematic of a read head structure. |
A GMR read head contains at least two ferromagnetic layers that are able to rotate with respect to each other in response to an external magnetic field. A typical GMR read head will consist of a nonmagnetic spacer between these two ferromagnetic layers (Fig. 1). One of the ferromagnetic layers is adjacent to an antiferromagnetic layer to "pin" the field or hold it constant. Thus, at the most elementary level, there are four layers in all.
Electrons have two spin states; let's call them up and down. If the resistance encountered by these electrons as they travel through a medium (circuit resistance, in other words) is dependent on the electron spin, it is said that spin-dependent scattering is occurring. Spin-dependent scattering does not occur when electrons pass through bulk metal wires, but it does occur in GMR devices.
The circuit/current resistance in the GMR head depends on the orientation of the magnetic moments of the two ferromagnetic layers. This is because the interaction between the magnetic moments affects the electron scattering. When the magnetic moments in the two ferromagnetic layers are in opposite directions, then the electrons traveling through the layers are scattered. Electrons therefore encounter a higher resistance. When the magnetic moments in the two ferromagnetic layers are aligned, then the electrons of one spin state are scattered less strongly than that of the other and they form a low-resistance channel current. Since electrons encounter lower resistance, a stronger signal is thereby detected by the read head.
The presence of an external magnetic field can align these two ferromagnetic layers. When the media (or platter) rotates underneath a read head sensor and a "bit" is present, it causes the magnetic moments to align, thereby reducing the resistance of the sensor head and causing a current to pass through. This current indicates the presence of a bit.
At nanoscales (i.e., when the films in a read head are 10–15 atoms thick), it is easy to see how the roughness, composition, chemistry, and impurities at the interface can affect electron scattering and therefore the resistance and the current that pass through a sensor.
The interdiffusion of adjacent layers of even 2–3 atomic layers on either side can mean that up to 40% of a specific layer is now a different material. Ferromagnetic layers may come to resemble the properties of their nonmagnetic spacers and vice versa. Similarly, the presence of impurities at interfaces can be detrimental. Therefore, the accurate characterization of buried interfaces is a critical component of device development and manufacturing.
Characterization of buried interfaces
Cobalt iron-copper (CoFe-Cu) stacks are commonly used by the magnetic storage industry to manufacture read heads. The atom probe has been used to analyze reliability/lifetime problems on test structures for read heads.
To analyze a sample for use with the atom probe, a specimen is prepared from the sample (e.g., the wafer) that has to be analyzed. This specimen is inserted into the atom probe. Data is collected from this specimen. Atom probe data exists as a database file that contains the location and identity of almost all atoms in an analyzed specimen. Data sets generated by the local electrode version of the atom probe are typically a cube of 50 to 100nm. (For details, see "How does the atom probe microscope work?").
 Figure 2. Subset of a larger 3D data set showing a 24??24nm 2D view of a GMR test structure. |
Once the data has been collected, it can be visualized and mathematically analyzed using a variety of software programs. Visualization software can be used to view the specimen in 3D, render it in different orientations, take a thin slice through the data, or display only specific elements or atoms from a region of the specimen. Numerical algorithms can be applied onto the data set to extract valuable quantitative information about the specimen.
Figure 2 shows a 2D view of a multilayer test structure used for read heads. It consists of alternating layers of CoFe-Cu. CoFe atoms are shown in blue, Cu atoms in red. The height of the data displayed is 24nm. Each layer is about 2.5nm (~12 atoms thick).
Several questions commonly asked by process and equipment engineers about these and other multilayer stacks are:
- What is the difference in interface roughness when copper films are deposited on cobalt films, and what is the interface roughness when the order of these films is reversed, i.e., when cobalt is deposited onto copper?
- Which interfaces in the stack are the first to diffuse (get destroyed) when the structure is subjected to accelerated lifetime tests?
- What is the average composition of the individual layers upon deposition and after accelerated lifetime tests?
As explained earlier, the answers to these questions are used to improve manufacturing and deposition processes. The improved processes will result in devices that have better properties such as MR response. The atom probe is the technique that can be used to address these questions.
 Figure 3. Sharp interfaces between CoFe and Cu in the as-deposited stack. |
Analysis of multilayers. These questions can be answered by analyzing various subsets of the data shown in Fig. 2. Figure 3a shows a thin slice of the as-deposited CoFe-Cu films from Fig. 2. Figure 3b plots the composition of the CoFe and Cu films with depth. The growth direction is from right to left. Figures 4a and 4b show the thin slice and composition, respectively, but only after the film was subjected to an accelerated lifetime test.
Interface roughness. Cu films deposited on CoFe films have a sharper interface than if CoFe films are deposited on Cu films. This is evident by observing both Figs. 3a and 3b. In Fig. 3a, more Cu (red) atoms are observed in the CoFe (blue) films when CoFe is deposited on Cu than the other way around. Figure 3b shows the concentration gradient from 0 to 100% with distance shown in angstroms. In Fig. 3b, the slope of the Cu (red) line is sharper when it is to the right of the CoFe (blue) lines — when Cu is deposited on CoFe. The concentration gradient of CoFe on Cu is more diffused.
 Figure 4. Heavy intermixing between the CoFe and Cu layers when the structure is annealed. |
Interface that is most diffused. Figure 4a shows that the Cu and CoFe layers have become intermixed upon annealing. Figure 4b shows the composition gradient from 0 to 100% with distance in angstroms. In this composition gradient, all the CoFe films are seen to diffuse the most; the CoFe peaks have spread out the most, whereas the Cu peaks are spread out over less distance. Also, the first CoFe film from the right has diffused the most. Specimens annealed for different times can be analyzed in the atom probe to show which layer is the first to break down.
Average composition. Eighteen percent of the as-deposited stack is >90 Cu. After anneal, only 3.6% is >90% Cu. Similarly, 24% of the as-deposited film is >90% CoFe, while only 12.6% of the stack is >90% CoFe after anneal. These compositions have been calculated by counting the number of Cu atoms and CoFe atoms within a user-specified volume in the specimen. The specimen is sampled at several locations and the results are averaged.
Conclusion
The ultimate objective of most process development and failure analysis activity is to correlate device properties with the microstructure that has resulted from specific processing conditions. Since the atom probe provides the location and identity of individual atoms, critical device problems faced by the industry can finally be addressed.
Acknowledgments
The authors wish to thank Dr. Peter Ladwig, formerly of the U. of Wisconsin at Madison, for conducting the experimental work. LEAP is a registered trademark of Imago Scientific Instruments.
Sanjay Tripathi is COO of Imago Scientific Instruments Corp., 6300 Enterprise Ln., Madison, WI 53719; ph 608/274-6880, e-mail [email protected].
Tye T. Gribb is VP of R&D at Imago Scientific Instruments.
Thomas F. Kelly is chairman, founder, and CTO at Imago Scientific Instruments.
Robert M. Ulfig is applications lab manager at Imago Scientific Instruments.
How does the atom probe microscope work?
Time-of-flight mass spectrometry identifies individual atoms by their elements or isotopes, while point-projection microscopy identifies where atoms were originally located in the specimen (i.e., in 3D). The atom probe microscope uses the principles of both time-of-flight mass spectroscopy and point-projection microscopy to identify individual elements and to locate them within the bulk of a material.
As in transmission electron microscopy (TEM), the user creates a specimen. Instead of the thin foil familiar to users of TEMs, the atom probe specimen is a small pointed tip (microtip) with a ~100nm radius of curvature (item #1 in the figure). This microtip can be carved into a wafer using commonly available techniques such as the focused ion beam. Several other techniques can be used to make the tips either before or after the film is deposited, and with or without the use of a focused ion-beam tool.
The specimen is then inserted into a cryogenically cooled, UHV analysis chamber. The analysis chamber is cryogenically cooled to freeze out atomic motion. It is at ultrahigh vacuum to allow individual atoms to be identified without interference from the environment.
A positive voltage is applied to the specimen via the voltage pulser (item #2 in the figure). The positive voltage attracts electrons and results in the creation of positive ions. These ions are repelled from the specimen and pulled toward a position-sensitive detector.
 Schematic showing the principles of an atom probe. |
Under an identical voltage differential, lighter ions (those from lighter elements, shown in green) will reach the detector quicker than their heavier counterparts (depicted in red). To accurately measure the time of departure of the ion, the positive voltage is typically pulsed. The arrival time of the ion is recorded by a set of timing devices on the single particle detector. This detector operates under the principle of a delay line detector. The time-of-flight of the ion identifies the mass-to-charge of the ion, which provides the elemental identity, so the atomic species is obtained. The numbers of atoms of various elements that hit the detector are recorded and a mass spectrum is extrapolated.
The location of the atom in the specimen is determined from the ion's hit position on the delay line detector (item #3 in the figure). The x and y coordinates of the hit relate to the ion's original position on the specimen. It is a fundamental rule of physics that neither potential lines nor atomic flight paths can intersect or cross each other; therefore, an ion that hits the detector to the right of another ion must have been located to its right on the specimen. The depth, or z dimension, is provided by the sequence (time) of the ion hit on the detector. Placing the detector at a distance away from the specimen magnifies the tip of the specimen. The specimen has a curvature (r) on the order of nanometers. The detector (d) is millimeters away from the specimen. This configuration magnifies the specimen by a million X (r/d multiplied by a constant).
In due course, atoms from the surface ionize, exposing another layer of atoms under them. This process of field ionization continues until the specimen has been fully analyzed, and provides a 3D image of the entire specimen. In this technique, the specimen is the primary optic, so there are fairly stringent considerations for the conductivity, interface strength, and geometry of the specimen.