If you know me, you know that I have a tendency to become obsessed with imagery. I usually stick to optical microscopes, but occasionally the people who play with electrons and ions also do something exciting. A recently published paper on ion microscopy has me pretty excited right now, and that’s about the only excuse I need to dig into it.
Ion microscopy is similar to electron microscopy. In a typical electron microscope, you fire a beam of electrons at a sample and examine the angles at which the electrons are scattered. These angles are directly related to the surface of the sample, so with a pair of optics (magnetic lenses, in the case of electrons) you get an image. Because electrons are heavy and energetic, they have a very short wavelength, so smaller features can be imaged.
After the development of the electron microscope, scientists realized that you can do similar things with ions – the nucleus of an atom that has had some or all of its electrons removed. Hence ion microscopy.
All forms of imaging are subject to the laws of physics, which limit the size of the visible features. Once you’ve got rid of all the imperfections in an imaging system, you’re left with noise. Even if you had optics that magically transcended physical limits, the noise in the lighting system will prevent you from seeing very fine features. In other words, nature hates you and your microscope too.
Stop the noise
Since we are talking about imaging using an ion source, we need to discuss the noise of an ion source (they are actually the same for electrons and light). Imagine you have a source of ions that emits an average of 100 ions per second. That average is subject to an expected variation of about ten ions in a one-second period. Simply put, when you create an image by scanning an ion beam past a sample, at some level, you can’t tell whether a darker spot is due to a characteristic of the sample or to the source being slightly less bright at that particular moment. emits ions.
The nice thing about these resources is that averaging wins. Consider our source above – the scale of the fluctuations as the square root of the mean. In 0.1 seconds the average is 10 and the fluctuations are about three. So the noise is almost as large as the signal. After one second the noise is at the 10 percent level and after ten seconds the noise has dropped to 3 percent.
However, resources is not ideal. Typically, imaging a sample changes it, so the longer it takes to acquire an image, the more likely it is that the feature you’re imaging will be altered by the ions you photograph on the sample. And if you are interested in how the sample changes due to some externally applied control, averaging is not your friend.
The short story is that you cannot easily avoid these fluctuations. At least not unless you know exactly when an ion will arrive. And figuring this out is exactly what researchers in Germany have done.
Old tools get a new job
Over the past three decades, researchers have developed so-called cold ion traps. These traps essentially use some lasers and electric fields to trap ions like beads on a string. Once trapped, they sit there, rocking gently back and forth. Normally, researchers use the captured ions for quantum physics experiments, quantum computer gates, or the like. But our intrepid scientists realized that with the proper application of an electric field, they could pull a single ion out of the trap. That ion would leave the trap at a well-defined time and have a well-defined energy.
That situation completely changes the statistics. What is the average number of ions in the beam? It is the number of times the gate field (used to eject ions) is applied per second. What are the fluctuations in the number of ions per second? That’s a trick question – the answer is no. As long as there are ions in the trap, the researchers can eject an ion with every change of the field. There is always one ion and no fluctuations.
There is a subtle point, however: the precise moment at which the ion leaves the trap changes with each ejection, so the fluctuations are non-zero at very high temporal resolution. But as long as the detector averages over a period greater than that small fluctuations, there are no fluctuations. That is exactly what the researchers do.
This would basically mean that an image should be noiseless. But it is not – there are several factors that deteriorate the image. An unavoidable one is that the detector is not perfect. For every 100 ions that fall on the detector, only 96 are detected. This introduces a small amount of unavoidable noise.
Another problem is that sometimes the detector reports an ion when there isn’t one (called a “dark count”). Because the ion source only produces ions within a fixed time window (say, a few moments after the gate field is turned on), the researchers removed the vast majority of dark counts by discarding all detections that occur outside that window. . Thanks to this gate technique, they were able to reduce this noise by a factor of a million.
The biggest problem comes from the environment: vibrations. Vibrations can be eliminated, but that is actually a real technical challenge. The vibrations limit the focus of the ions to about six nanometers in diameter, which is nothing unusual in the world of ion microscopy. And because the trap only emits three ions per second, the amount of signal to work with is still very small.
Indeed, in one of the researchers’ demonstration images, each pixel is obtained by releasing one ion from the trap and whether or not it is detected. Surprisingly, the system works quite well. And to show that this process is much better than a traditional source with the same average number of ions per second, the researchers had their gated source emitted at random times (so that it behaves exactly like a classical ion source). The image obtained from randomly emitted ions was much blurrier and missing many features.
Normally, if the diameter of the ion beam was six nanometers, it is very difficult to see features significantly smaller than six nanometers. But this is where the power of a deterministic source (it only emits an ion when you ask for it) comes into play. It opens up statistical techniques that cannot be easily applied to arbitrary sources. The researchers demonstrated the power of these statistical techniques by showing that they could determine the diameters of holes much more accurately than the native resolution of the ion microscope would allow. If you have an ion source that provides only one ion per pixel, edge detection becomes very noisy. But the use of statistics allows the edge to be solved with higher precision.
Even with the power of statistics, the image results aren’t fantastic. However, that shouldn’t be your takeaway message. This is the first microscope of its kind and the researchers are thinking of improvements. For example, the trap’s design could be changed so that it automatically charges ions at the same time it ejects them, allowing for a higher ion current (the researchers estimate they could exceed 100 ions per second). Second, because these are ultracold ions, their state can be manipulated with lasers. This means you can imagine doing polarized spin microscopy so that the microscope is directly sensitive to the magnetic properties of the sample. Or you can put the ion in an excited state and examine the final state to learn more about the ion-sample interactions.
A really cool idea is that because the ion current is directly under the control of the researcher, things like stroboscopic imaging with picosecond time resolution are possible. Finally, to make this really exciting, you can use the time it takes for the ions to travel from the trap to the detector to correct the lens and get even more detailed images.
In other words, this is a first step on a very interesting journey.
Physical assessment letters2016, DOI: 10.1103/PhysRevLett.117.043001