Scanning probe microscopes are notoriously sensitive to many types of external and internal interferences including electrical, mechanical, and acoustic noise. Sometimes noise can even be misinterpreted as real features in the images. Therefore, quantification of frequency and magnitude of any noise is the key to discover the source and eliminate it from the system. While commercial spectrum analyzers are perfect for this task, they are rather expensive and not always available. We present a simple, cost-effective solution in the form of an audio output from the instrument coupled to a smart phone spectrum analyzer application. Specifically, the scanning probe signal, e.g., the tunneling current of a scanning tunneling microscope, is fed to the spectrum analyzer which Fourier transforms the time domain acoustic signal into the frequency domain. When the scanning probe is in contact with the sample, but not scanning, the output is a spectrum containing both the amplitude and frequency of any periodic noise affecting the microscope itself, enabling troubleshooting to begin.
In recent decades, scanning probe microscopes (SPMs) have experienced rapid development and employment after the invention of the scanning tunneling microscope (STM)1 and the atomic force microscope (AFM).2 The ability to measure physical and chemical properties of surfaces at the nanometer scale has been invaluable and has contributed to the widespread use of SPMs in fields such as chemistry, physics, materials science, and even biology.3–5 In general, these techniques rely on the use of a sharp probe tip, which in some cases is terminated by a single atom or molecule at the tip apex,6 in close proximity to a sample surface. Piezoelectric actuators allow for the probe tip to be reproducibly moved very small distances in the x, y, and z directions, thereby building a three-dimensional image of the sample surface.3,6 Various feedback mechanisms are utilized in order to maintain the distance between the probe tip and the sample (often referred to as the z-gap) constant in order to maintain, for example, a constant tunneling current in STM or force in AFM.3,6,7 Given the scale at which these microscopy techniques work, the signals involved are extremely small. For instance, the STM measures tunneling current at the pA level,8 while AFM relies on force feedback as small as the pN scale.9 Therefore, scanning probes often have problems with interference from external noise sources, which can skew results and produce surface “features” that can be misperceived as real.10 Knowledge of the frequency of these external noise sources is crucial for finding their origin in order to eliminate noise.11 Spectrum analyzers are very useful for the measurement of both the magnitude and frequency of noise sources in SPM images,12,13 but, in general, they tend to be relatively expensive.
We describe an apparatus that just requires the use of a smart phone for determining the frequency of noise sources via a fast Fourier transform (FFT) application. Specifically, the SPM output signal, which in our case was the tunneling current from a STM, is connected to an amplified loudspeaker input. Several smart phone applications that can turn the acoustic signal from the speaker into the frequency domain via an FFT plot are available. There are two main advantages to convert the signal from an electrical signal to audio and then back to an electrical signal: (1) it makes home building this device very straightforward as all one needs is an audio speaker and the smart phone FFT application and does not require any electronic boards to be built and (2) the fact that one can measure the frequency of audible noise from various noise sources in the laboratory and either identify or eliminate each device as the source of the noise. As many sources of noise for SPMs are low frequency, we tested the lower limit in terms of frequencies our loudspeaker could produce. This involved connecting the speaker to another smart device with a noise generating application and testing to see whether low frequency signals could be transmitted via the speaker and detected in the spectra generated by the FFT application. We use several STM imaging experiments to demonstrate the capabilities of the setup in identifying the frequency of external noise sources.
II. METHODS AND DISCUSSION
The overall schematic of the noise analyzer setup for our STM experiments is outlined in Fig. 1. As the scanning probe tip passes over a sample surface to which a positive voltage has been applied, electron tunneling takes place from the tip to the sample, producing an electron tunneling current that is measured. This tunneling current can be output to produce an image of the apparent topography of the surface, and in our case, to also produce the acoustic signal necessary for our smart phone spectrum analyzer input. This step was accomplished by fashioning a cable capable of converting the STM signal output to a loudspeaker input, which involved soldering a BNC connector cable and an audio jack cable together. The microscope output signal is the converted acoustic noise and can be picked up by the built-in microphone of the smart phone and converted to an FFT plot in real time. We demonstrate this setup with the “FFT plot” application by ONYX Apps, but there are several other options available that can perform the same function. The spectrum analyzer application measures the magnitude of the signal as the sound pressure level in decibels relative to full scale (dBFS), as well as the frequency in Hz. When analyzing a typical FFT spectrum of the STM tunneling current signal, there are typically several distinct peaks, the frequency of which enables the identification of sources of periodic noise.
An example of an STM image with both real and externally derived noise features is shown in Fig. 2(a). In this experiment, 1,1,1-trifluoroiodopropane molecules were deposited on a flat Cu(111) surface and imaged by STM. The large bright islands represent real features, i.e., semiordered, one molecule thick islands on the Cu(111) surface. The diagonal stripes [highlighted by white arrows in Fig. 2(a)] imply the existence of alternating ridges and valleys on the Cu surface, but we know that this Cu surface is flat based on its (111) facetted surface and the fact that we have previously imaged the same system without these stripes present in a noise free environment. In order to characterize this particular noise feature and verify the corresponding features in the FFT spectrum, we first attempted to find the frequency of the noise from the image itself. This was done by using the known scan speed (in nm/s) and image size (in nm) and determining the period of the noise in a single linescan from the STM image. In the linescan, each peak and trough are one waveform, and hence, the number of waveforms per line can be calculated. This then allowed us to calculate the frequency of the noise, which in this experiment was 60 Hz. This frequency, as well as the first harmonic at 120 Hz, was clearly visible in the smart phone FFT spectrum [Fig. 2(b)], verifying that the FT plot can detect noise sources in the image via the electron tunneling current signal.
In order to test the ability of our setup to detect noise of unknown frequencies, we introduced acoustic noise of known frequencies beside the microscope via a signal generator. The STM itself was housed in a custom build “quiet room” with the control electronics outside the enclosure so that the only way these frequencies would be detected by our setup was via mechanical coupling of the generated noise to the ∼1 nm STM tip-sample junction and subsequent coupling to the tunneling current signal. In Fig. 3, the FT spectra of raw STM data measured via point mode in which the STM tip remains fixed are displayed. As can be seen in the figure, the intentional addition of different frequencies can be seen in the raw STM images [Figs. 3(b)–3(d)] and in the FT spectra [Figs. 3(e)–3(g)]. The previously discussed method of obtaining the characteristic noise frequency based on first calculating the period of noise in a single linescan and determining the number of waveforms (one peak and one trough) per linescan was once again utilized in conjunction with known scanning parameters and the STM images taken at constant current [insets in Figs. 3(e)–3(g)]. These calculations led to frequencies indicated by orange arrows. Our apparatus could detect the fundamental driving frequency (green arrows) when the original noise signal was greater than 60 Hz, and in all cases, it could detect the first, second, and sometimes third harmonic frequencies (denoted by white arrows). It can be seen from the data in Fig. 3 that the frequency of noise detected in the images (whether it is the same as the introduced noise or a higher harmonic) depends on the frequency of the noise source itself, as different frequencies will couple to the STM in different ways, and this will differ for every instrument.
We have demonstrated the capability of our simple home-built setup to identify the frequency of periodic noise present in our STM images as a result of external noise sources. This provides a fast and simple way to improve the quality of SPM images by identification of the frequency of the external interference which is crucial for identifying its origin. The primary limitation of this setup, as evidenced by our experiments adding noise with known frequency, is that the apparatus can detect the fundamental frequency of the external source only if it is above ∼60 Hz. This is not a major issue as below this limit, harmonic frequencies (for example, 120 and 180 Hz) can still be used to identify the external noise source. This apparatus can serve as a cost-effective alternative to using expensive commercially available spectrum analyzers in the trouble shooting of external sources of noise in many scanning probe techniques.
The project was supported by the U.S. National Science Foundation (NSF) under Grant No. CHE-1764270.
The data that support the findings of this study are available from the corresponding author upon reasonable request.
E. Charles H. Sykes is the John Wade Professor of Chemistry at Tufts University. Charles received B.S. and M.S. degrees with first class honors from Oxford University in 1998 before moving to Cambridge University for a Ph.D. under the supervision of Professor Richard Lambert. His thesis work explored the structure and reactivity of model gold/titania catalysts. He then relocated to the USA for postdoctoral fellowships with Professor Paul Weiss at Penn State and was the first to directly image and control the placement of catalytically important subsurface hydrogen in palladium. In 2005 Charles began his independent career as an Assistant Professor of Chemistry at Tufts University. He was tenured and promoted to Associate Professor in 2010 and then promoted to Full Professor in 2013.
Sykes has been named a Beckman Young Investigator, Research Corporation Cottrell Scholar, IUPAC young observer, and the Usen Family Career Development Professor. He is also the recipient of a 2009 NSF CAREER award, a 2011 Camille Dreyfus Teacher-Scholar Award, and the 2012 AVS Peter Mark Memorial Award. Charles received the Young Talented Scientist Award at Chirality 2014 and was named a Fellow of the Royal Society of Chemistry in 2015. In 2018, Charles was named a Fellow of the AVS and, together with Maria Flytzani-Stephanopoulos, was awarded the 2019 ACS Catalysis Lectureship for the Advancement of Catalytic Sciences. He is the author of over 150 peer-reviewed publications and has given over 160 invited talks at conferences and universities.