Metrology can be considered a collection of subdomains, each subdomain representing a specific analysis technique. Think about secondary ion mass spectroscopy (SIMS), a technique for determining the composition of a material (surface) – with applications for organic and inorganic structures. Another subdomain is scanning probes, comprising several concepts for measuring, e.g., topography, adhesion, material hardness and chemical properties, and we have our own scanning spreading resistance microscopy (SSRM) technique, allowing to determine carrier profiles in a semiconductor. In recent years, many of these techniques have largely evolved and have become mature enough to be used in research or production of semiconductor applications. Lab-to-fab transition, and an increasing focus on volume, automation and throughput time are seen as important trends.
Orbi-trap SIMS: a revolutionary improvement in mass resolution
In recent years, progress in SIMS has been remarkable. SIMS allows studying the composition of a material surface by sputtering the surface with a beam of energetic ions. As a result, secondary ions are released and analyzed by means of a mass spectrometer. In 2017, the introduction of a new concept for mass spectrometry, the Orbi-trap, has enabled a significant improvement (10-50x) in mass resolution (>250,000), and in accuracy of mass determination (<2ppm). This way, complex molecules can be uniquely identified. Originally developed for applications in biological and medical research (single cell proteomics), these properties represent a quantum leap for the analysis accuracy and interpretation of SIMS data in semiconductor technology. With Orbi-trap SIMS, it is now possible to analyze, for example, photoresists and self-assembled monolayers, or to make a distinction between two elements with very similar mass (such as arsenic and germanium). The technique can also be used to support the self-focusing SIMS concept, which was developed at imec for analyzing extremely small structures.
Hybrid metrology: 1+ 1 =3
Over the last years, the concept of hybrid metrology has become increasingly important. Following this concept, different metrology techniques are used to measure one and the same structure.
This allows either correlating complementary information (such as structure and functional properties), or eliminating specific uncertainties of the individual techniques. An illustration is the combination of transmission electron microscopy (TEM, an imaging technique) with scanning probes (SPM, functional analysis). Combining TEM information (structure and composition) with the observation of functional properties at the nanometer scale via SPM (or, via SRRM for carriers, or via piezo-force for ferro-electrical properties,...) on the same structure provides a unique approach for generating insight in the functioning of new structures. Last year, imec played a pioneering role in the development of the SPM/TEM hybrid metrology – which was also presented at the IEDM conference.
In addition, our team made a major breakthrough towards more accurate 3D analysis, by combining the atom probe technique (APT, atom probe tomography) with atomic force microscopy (AFM). Prior to an atom probe measurement, a sample is prepared in the form of a sharp tip. From this tip, ions are evaporated, captured by a position sensitive detector and individually analyzed according to their mass. The tip acts as an ion-optical component, and creates an image magnification of > 106x. The result is a 3D analysis of the sample with a (theoretical) spatial resolution < 0.2nm. In practice, the exact value of the magnification (resolution) is unknown, because the detailed shape of the tip changes continuously and cannot be determined in-situ, until recently. Most of the labs and manufacturers explore the integration of TEM in an APT system as an expensive and complex solution to this problem. By showing that the APT tip can be imaged with AFM, imec demonstrated a promising, simpler, more quantitative and cost effective alternative. The further exploration of this concept will be complemented with new complex data algorithms that will be developed in collaboration with Vision Lab, an imec research group at the University of Antwerp.
As a final example of the increasing importance of hybrid metrology, I would like to mention a project in which imec combined a SIMS instrument with in-situ SPM: a world first. This will allow determining composition (SIMS) as well as functional properties (electrical SPM). In the future, the exploration and demonstration of this approach will be a main focus point of our research efforts.
Probing small confined volumes: a challenge for metrology
Another important milestone is the commercialization of the Fourier Transform-scanning spreading resistance microscopy (FFT-SSRM) technique. This novel technique is based on SSRM, an analysis technique that was invented at imec years ago. SSRM is among the few techniques that allows to determine carrier profiles in a semiconductor. But the standard SSRM technology (and its underlying approaches) cannot be applied to small volumes, such as FinFETs and nanowires, due to signal distortion by parasitic resistances. Imec’s FFT-SSRM concept overcomes this issue. In 2017, it was translated into a commercial product and installed at several partners of imec.
Analyzing extremely small structures (such as nanowires, 3 to 4nm in diameter), 3D geometries and confined volumes (small structures embedded in a complex environment such as tungsten or oxide) presents one of the main challenges for metrology in the years to come.
Even for high-resolution techniques such as TEM, a complex sample preparation (into lamella, < 10nm) and special measurement procedures (TEM tomography) are required to access the area that is relevant (e.g., the channel underneath the metal gate in a gate-all-around device). In this context, imec continues the development of its scalpel AFM technique. In scalpel AFM, the tip of the AFM probe is also used to slice layers on an atomic scale, after which a local functional measurement is performed.
Conceptually, metrology on nanostructures remains a major challenge. A nanowire device, for example, contains about 500 dopant atoms at most. In the source/drain extension, maybe three are left. What is the statistical relevance of measuring such a single device (i.e., counting three atoms)? What is the meaning of ‘concentration’ at the nanometer scale? If we want to continue performing quantitative analyses, we will have to introduce novel metrics, such as the average distance to the next atom. In addition, we will have to take into account the stochastic variability in the properties of the structures (e.g. dopant fluctuations).
Trend: array-based metrology, sensitivity and statistical relevance
As a result of these statistical limitations, the development of array-based metrology is emerging as a major trend. Instead of deploying expensive tools (with very high resolution) for measuring e.g. single devices, we start doing measurements on arrays of devices, with relative small resolution (by using broad beams).
Because of the underlying physics of the measurement technique, the information can be restricted to the structure of relevance, and the signal coming from the environment can be suppressed. By simultaneously measuring multiple devices in parallel, the quality of the signal can be improved, and statistically relevant data are obtained. For techniques such as Raman spectroscopy, micro-four point probe, RBS (Rutherford backscattering spectroscopy) and SIMS, imec has developed solutions that allow to measure and interpret effects such as CD variation, strain and composition variations in array structures.
On the longer term, I expect metrology to further evolve into this direction. A large number of techniques that are now being developed, such as Orbi-trap SIMS, TEM tomography or atom probe/AFM, will be massively deployed in the research and production of semiconductor applications. TEM and AFM, for example, which today are still perceived as single device analysis concepts, are expected to evolve into array based metrology as well. These analysis techniques will increasingly be used in-line; inline TEM, inline SIMS, inline XPS (x-ray photoelectron spectroscopy) will become a commodity.
Metrology as a profit center
On the downside, funding all these developments remains a difficult issue for the semiconductor industry. On the one hand, costs increase dramatically because of the growing complexity of semiconductor developments and because of the need for ever higher resolution. For example, enhancing the resolution from 1nm to 0.5nm is a huge step for the metrology developer. On the other hand, since a few years, the market for metrology tends towards monopolization. For techniques such as SIMS or atom probe, there is only one market player. Hence, by lack of competition, progress is slow and expensive. Nevertheless, it is important that the industry continues to invest in metrology.
Metrology is not a cost factor, but a profit center for industry. Some years ago, VLSI Research published a study showing that, for each dollar invested in metrology, there was a return of 10 dollar. Being able to quickly and accurately analyze semiconductor devices can accelerate technology development and yield enhancement. Advanced equipment and gathering fundamental metrology knowledge are cornerstones of successful technology development and high-volume production. For imec, gaining fundamental insight is key to the added value we bring to our partners. Therefore, we continuously invest in a high-performing metrology group with strong scientific foundations. Also, special attention must be paid to design for testability. In the coming years, providing test arrays needed for measuring array structures, and test structures that are metrology friendly will be key for metrology to achieving optimum results.
Wilfried Vandervorst received his M.Sc. degree (electronic engineering) in 1977 from the KU Leuven, Belgium and the PhD degree in Applied Physics in 1983 from the same University. In 1983-1984 he worked at Bell Northern Research, Ottawa, Canada as a consultant in the field of materials characterization. In 1984 he joined imec where he became Director of the department dealing with materials characterization. Since 1990 he is also holding an appointment as a Professor at the KU Leuven (Physics Department) where he is teaching a course on materials characterization and supervising PhD students. In 2001, after an international peer review, he was elected as an imec Fellow for his outstanding scientific achievements related to semiconductor metrology, and in 2013 as Senior imec Fellow. He is engaged in advanced research on metrology and material (interactions) for semiconductor technology. He has co-authored more than 600 papers in peer-reviewed journals, gave more than 150 invited presentations and is co-inventor of more than 60 patents.
1 February 2018