When do specific characterization methods become outdated? Would it be when there are easier methods available to get the relevant information about a sample? Would it be when there are methods that are more accurate? Faster? Cheaper? Any of these or some other reason could suffice to stop using an outdated characterization method. X-ray diffraction recently turned 110 years old (!) and as an X-ray physicist I sometimes feel like a kid playing with antiques. Why do I stick to X-ray scattering methods? Wouldn’t there be something fresher and more interesting for me?
X-ray scattering is a timeless method for structural analysis
I am returning, as I am writing this text, from the Nordic Scattering from Soft Matter 2023 (NSSM2023) workshop organized at the Norwegian University of Science and Technology in Trondheim. As the name suggests, the topics of the workshop are related mostly to X-ray and neutron scattering. The first time I attended this annual conference that circulates between the Nordic countries was in 2011, also then in Norway. The difference between the results presented in 2011 and 2023 mainly lies in the research itself, not the methods applied.
For sure, this observation could probably be done in many workshops focusing on a certain methodology. Many established methods do not change because the underlying physics remain unchanged. Scattering means that a beam of photons – or other particles like neutrons or electrons for that matter – scatter in all directions in space when propagating through matter. The scattering pattern that can be recorded is a direct function of the structure of the transmitted matter. That physical fact will not change in a million years.
As a matter of fact, X-ray scattering (often equal to X-ray diffraction, XRD) is the only common laboratory method to directly characterize the atomic scale and nanoscale structure of many bulk samples. This structural analysis includes, for example, characterization of the crystal structure of a material, size and arrangement of nanoparticles, or structural information from colloids.
Combining several methods yields more comprehensive results
Sure, for most of us, a good TEM or SEM image tells much more than an X-ray scattering profile. This is because in microscopy “what you see is what you get” whereas the X-ray scattering pattern is a Fourier transform of the measured structure, which is a rather unintuitive way of representing a 3D structure, and the data obviously requires some analytic tools to be decrypted.
On the other hand, an electron microscopy image only describes what is captured in the image, i.e. the sample surface or a thin section of the sample volume. This is for many samples a clearly sufficient characterization. However, regarding certain materials, PhD supervisors and a lot of reviewers keep asking for XRD or other X-ray scattering data. The purpose is not (only) to increase the cognitive pain of the student. Typically, X-ray scattering methods and electron microscopy are complementary methods, and if you did not get a solid conclusion with only one of the methods, you would employ the other as well.
A sample has a structure, which is either well-defined or poorly defined. In either case, characterizing the sample using multiple methods always gives stronger conclusions than just employing one characterization method. It’s like listening to your spouse with both ears and eyes. You are less likely to miss the subtle but, oh, so important details.
The importance of avoiding unfounded conclusions
We know that would our conclusions not be supported by all the data collected, but there would be a contradiction between the conclusions and some of the data; we should re-evaluate our conclusions. That is according to the scientific method. Perhaps something, like previously reported studies on the topic and our own hypothesis of the outcome, did bias our conclusions in the wrong direction. It is also possible that the conclusions are not strictly wrong but simplified to a degree that they do not cover some of the features seen in the data. Unfortunately, neither can we rule out the possibility of an unsuccessful experiment.
Experimental artifacts imply more experimental efforts to get confident conclusions. Even if the “wrong” data were suggested to be caused by unsuccessful experiments, we experimentalists should have the courage to repeat the experiment to find out if the data is reproducible and why the result is “wrong”. For the interest of time and other resources, repeated experiments are often not feasible in real life, at least not in the scope of the article that is about to be written or within the project that should be finished. That’s why experimental artifacts are so harmful. When they are present, we are either left with uncertainty about which data to trust or in the worst case: wrong conclusions.
Speaking about experimental artifacts, as we know, X-rays transmit through matter and common laboratory X-ray scattering methods are nondestructive except for some photoactive materials. For the same reason, X-ray scattering experiments can be carried out in a range of environments as the requirements for sample holders are not as strict as for example in electron microscopy experiments. Samples can be measured in ambient conditions, elevated temperatures, controlled humidity, magnetic fields, etc. In addition, a significant volume of the sample can be measured to be sure that the data represents the whole sample.
Both details and averages are needed for a complete picture
Having data that represents the average structure of the sample is nice, but it can also turn into a problem, even when no experimental artifacts are present. With electron microscopy methods, you often get details with high resolution, but you might not get a reliable average or the 3D structure (without significant effort). X-ray scattering represents the average of the 3D structure, but you might lack relevant information about the details (without significant effort). This becomes truly problematic in the case of polydisperse samples or samples with multiple structural phases.
Let’s assume that you have two different structures present. The X-ray scattering data is then a superposition of scattering from both structures. However, without additional information, you cannot tell the scattering contributions apart. Hence, X-ray scattering and electron microscopy methods are often complementary when characterizing atomic to nanoscale structures. The complementary nature of the characterization methods is one of the reasons why X-ray scattering instruments are a part of the Aalto Nanomicroscopy Center. We aim at listening to our samples using both our ears and eyes, so to say.
Principles remain, but progress is continuous
I also want to return to the topic of the old-aged X-ray scattering method. I recorded my first diffractogram in the summer of 2010, which means that I have been active in the field for barely 10 percent of the age of the discipline. Lessons learned back in 2010 are still valid. The principles and type of information extracted by X-ray scattering are the same. However, the facility I work in has changed significantly. The experimental part of my work is way easier than 13 years ago. What changed?
Progress mostly occurs in small steps. In recent years, the combined progress of multiple parts related to laboratory X-ray instruments has led to a dramatic change. General automatization, detector technology, novel sources, improved X-ray optics, etc. have all contributed to improved efficiency of lab work. To illustrate: If the experimental output is a product of four factors (e.g. output = A x B x C x D) and the factors A, B, C, and D are increased by 10 percent the output increases by almost 50 percent!
This seems to be happening in the field of X-ray scattering, with the exception that certain individual elements of the experimental setup have developed with giant leaps, and the combination of all partial steps of progress has enabled at least seemingly new ways to work. As said, the foundations are the same but now the path from scientific question to experimental confirmation or rejection has decreased almost by a magnitude with respect to certain experiments.
Both academic institutions and businesses have a role in method development
Academic research beyond the purpose of educating clever people is a must in developing characterization methods further. On NSSM2023, I spoke with many of the facilitators of such a development: the companies that make the instruments. These companies exist in a kind of symbiosis with the academic world as they sell their products to universities and institutions. The universities and institutions develop methods and train talented people to be employed by these very companies. More than that, a brief comparison of the exhibitors at the workshop revealed that three out of five of the exhibitors were spin-offs from universities or research institutes.
All the development in the field of X-ray scattering makes this old-time method timeless. The modern advanced methods rely on the same physical principles as 100 years ago and mastering the modern X-ray scattering methods is sometimes just a few steps from understanding the basics of the methods established almost a century ago.
One hundred years ago there were no electronic X-ray detectors. A photographic plate would do the job. Not to mention computation, a pen and paper (+ a clever mind) would have been the analytic tools. In academic research, it can in principle still be sufficient to use whatever old X-ray scattering instrument to record data from which relevant information is extracted using any old software or just pen and paper. The conclusions can be unaffected. More than that – they should be unaffected if no mistakes were done in the process.
Despite advances in technology, the importance of humans asking the right questions remains
Let’s put ancient methods aside. One apparent problem with experimental methods established a long time ago is the opposite of being outdated – they are too advanced. This seems paradoxical. Instead of being nice and understandable tools in materials characterization, they resemble independent scientific disciplines – and they are – with their specific vocabulary/terminology that is gibberish for an outsider (read “anyone who is not from that specific discipline). This is a bit frightening, since there is so much more to learn within the discipline than anyone can master. You will always lack some knowledge that would be useful for you (sad!). On the other hand, if there is some information you need, someone probably did work it out for you already and it is published in a nice article (happy!).
The good news is that we have advanced far from pen and paper – tools that are still good to master – and the X-ray scattering instruments and facilities in combination with user-friendly analysis software make life easy. However, the work for the experimentalist is still to choose for which tasks and how these methods are applied. The questions I ask before experiments: What is the information I should extract from the experiment? What kind of experiment could tell apart a true hypothesis from a false hypothesis? To me, it seems that we should still be asking these questions as the advances in instrumentation decrease the efforts required to obtain good characterization data. This task of asking good questions cannot be outsourced to any instrument or artificial intelligence as far as the aim is to expand the boundaries of human knowledge.
Ville Liljeström reserves the right to publish (including redistributing to a third party for publishing) part of the text or the text in its entirety – modified or unmodified – on any other platform or in any media.