Measuring Vibrations in the Ultra
Microscopy Hall
THESIS
submitted in partial fulfillment of the requirements for the degree of
BACHELOR OF SCIENCE
in
PHYSICS
Author : Dani¨elle van Klink
Student ID : 1375458
Supervisor : Dr. Milan Allan
2ndcorrector : Dr. Irene Groot
Measuring Vibrations in the Ultra
Microscopy Hall
Dani¨elle van Klink
Huygens-Kamerlingh Onnes Laboratory, Leiden University P.O. Box 9500, 2300 RA Leiden, The Netherlands
July 7, 2016
Abstract
Physicists like to find out how things work on the smallest level. When doing experiments in which we want to achieve atomic
resolution, there are many factors that can influence the experiments, for example: thermal fluctuation and external vibrations. The latter is what we focused on during this project.
At the Leiden Institute of Physics we have built a new low vibration laboratory in order to perform experiments that can achieve atomic resolution. In this report we will describe and compare the vibrations in both the old and new low vibration
laboratory. We have found that the vibrations in the new laboratory are much less and that this new lab is comparable with
some of the most successful scanning tunneling microscopy laboratories in the world.
Contents
1 Introduction 1
2 Theory 3
2.1 Measurement Islands 3
2.1.1 Kamerlingh Onnes Laboratory 3
2.1.2 Ultra Microscopy Hall 4
2.2 Geophone 4 2.3 Vibrations Sources 5 2.4 Linear Spectrum 7 3 Methodology 11 3.1 Calibration 11 3.2 Vibration Measurements 11
4 Results and Discussion 15
4.1 Calibration of the geophone 15
4.2 Vibrations in the Kamelingh Onnes Laboratory 16
4.2.1 The HS-1 Geophone 16
4.2.2 The Guralp CMG-40T 20
4.3 Vibrations in the Ultra Microscopy Hall 24
4.4 Week measurement 27
4.5 Comparing the Different Measurement Halls 32
5 Conclusion and Outlook 37
Appendix A Possible Problems 41
A.1 Gain of the Pre-Amplifier 41
vi CONTENTS
Appendix B Additional measurements 47
B.1 Partly lifted measurement island 47
B.2 Oscillations in frequency space 48
B.3 Week measurement 49
vi
Chapter
1
Introduction
Physicists like to find out how things work on the smallest level. This is why techniques such as electron microscopy, scanning tunnelling mi-croscopy (STM) and many others are widely used. With these techniques atomic resolution can be achieved, but this can only happen under special conditions. When measuring at such small length scales, there are a lot of factors that can influence the experiments, such as thermal fluctuation or external vibrations. The latter is what we focused on during this project.
In order to reduce vibrations, special low vibration laboratories can be build [1]. At the Leiden Institute of Physics (LION) we currently have a special low vibration laboratory, the Kamerlingh Onnes Laboratory (KOL). With the move to a new building we saw the chance to develop an even better low vibration laboratory, the Ultra Microscopy Hall (UMH). In the KOL and the UMH the scientific setups are placed on so-called measure-ment islands, which are heavy blocks of concrete (or other materials) vi-brationally decoupled from the foundation of the building.
The goal of this project is to quantize the vibrations at these two low vibration laboratories and develop a program to measure and visualize vibrations continuously.
Chapter
2
Theory
2.1
Measurement Islands
In this section I will explain the design of the measurement islands in the Kamerlingh Onnes Laboratory and the Ultra Microscopy Hall.
2.1.1
Kamerlingh Onnes Laboratory
Isolation from external vibration in the Kamerlingh Onnes Laboratory is achieved by placing the measurement islands on a different foundation than the building itself, thus not linking the measurement islands to the vibrations of the building. Each measurement island consists of two is-lands and a so-called pit (Fig. 2.1). The measurement setup stands on the two brown islands, which are on a different foundation than the building and they are not connected to the floor where the scientists are walking. The white pit is on the same foundation as the building.
Figure 2.1: Schematic drawing of the measurement island in the KOL. The two brown block are the measurement islands and the white part in the middle is the pit.
4 Theory
Figure 2.2:Schematic drawing of the measurement island in the Ultra Microscopy Hall.
2.1.2
Ultra Microscopy Hall
The islands in the Ultra Microscopy Hall are designed in a different way, as can be seen in Figure 2.2. The basement floor of the measurement hall is a big concrete floor of 2,000,000 kg. On the basement floor, a 27-ton mea-surement island stands on top of four air spring dampers. Such a big mass is needed for two reasons: first, a heavy experimental setup will not influ-ence the measurement island performance and second, the measurement island itself has a low resonance frequency. Figure 2.2 shows a schematic drawing of the measurement island.
2.2
Geophone
Most of the vibration measurements are performed using a geophone (Fig. 2.3).
A geophone consists of a magnet connected to a metal case and a coil on an inertial mass that is connected to the metal case via springs. When the surface where the geophone stands on moves, the coil moves relatively to the magnet. The change in flux induces a current across the coil and we can measure the voltage that this produces. In order to obtain the spec-trum in the frequency domain, the Fourier Transform of the signal can be taken.
However, this is not so simple, and there is something else that has to be taken into account. The geophone itself has a transfer function, which behaves as a high pass filter (Fig. 2.4). Frequencies lower than its reso-nance frequency will be damped. This will influence the measurements, because the geophone is not as sensitive to noise at low frequencies com-4
2.3 Vibrations Sources 5
Figure 2.3:Schematic drawing of the cross section of a geophone[2]
pared to the frequencies above the resonance frequency. To compensate for this the transfer function of the geophone needs to be used. The for-mula for this is given by[3]:
H(f) = f
2
−f2+f2
0 +2j f f0ζ
(2.1) Here f0is the resonance frequency and ζ is the damping coefficient.
A transfer function can be written as H = Aeiφ, the product of an am-plitude part and a phase part. For the geophone the relevant part is the amplitude of the transfer function. So the following equation will be used:
A(f) = f 2 q (f02− f2)2+4ζ2f2 0f2 (2.2)
In the chapter Methodology I will explain how the calculation of the two constants f0and ζ is performed. Figure 2.4 illustrates the shape of the transfer function for different values of the damping coefficient.
2.3
Vibrations Sources
There are many factors that can cause vibrations in a building. In this section I will talk about the vibrations sources that we have run into during this project.
To start with, the location of the new measurement hall is not perfect, there are a lot of external sources that cause vibrations. One of the sources
6 Theory
Figure 2.4:Transfer function of the geophone following the formula 2.2. For this example we set the resonance frequency to 10Hz and we vary the damping coef-ficient ζ.
of noise are the roads close to the building. The closest road to the building is a road with a maximum speed of 50 km/h and there is also a highway nearby. The effects of the road can be seen in the measurements previously performed by Marcel Hesselberth on the measurement islands of the new Ultra Microscopy Hall [4]. His results are shown in Fig. 2.5. The effect of the traffic is seen at high frequencies above approximately 200 Hz (Fig. 2.5(b)), taking into account that a lot of noise is already absorbed by the spring-damping system.
Secondly, looking closely, there is also a peak at 1 Hz on Fig. 2.5(b). This is assumed to be related to the strong wind present during the mea-surement. The wind is assumed to make the building resonate, therefore causing this 1 Hz peak. This shows that the wind could also be something to monitor when doing experiments.
At last, there are a lot of people working in the building. People walk-ing around can cause many vibrations and there are more factors that cause vibrations in a building. For example the lights, air conditioning, heaters etc can cause vibrations on lower frequencies.
6
2.4 Linear Spectrum 7
(a)Spectrum at 16:23, before rush hour. (b) Spectrum at 18:17, during rush hour.
Figure 2.5:Velocity spectral densities of the vibrations on the measurement island in the UMH on two different times. The effect of the traffic is seen in the high frequencies.The at 1Hz peak is expected to be caused by the wind resonating the building.
2.4
Linear Spectrum
The geophone measures a signal in the time domain. To get the signal in the frequencies domain we use a Discrete Fourier Transform (DFT). We can express the result of the DFT in 4 different ways[5]. These different ways can be seen in table 2.1, assuming the input time series is measured in Volts.
Abbrev. Name Relation Unit
PSD power spectral density V2/Hz
PS power spectrum PS = PSD∗ENBW V2
LSD linear spectral density LSD=√PSD V/√Hz
amplitude spectral density
LS linear spectrum LS=√PS =LSD∗√ENBW V
amplitude spectrum
Table 2.1: Naming convention for DFT outputs. ENBW is the equivalent noise bandwidth.[5]
The power spectral density describes how the power of a time series is distributed with frequency. It is defined as the Fourier transform of the
8 Theory
Figure 2.6:Segmented data stream with a window and without overlap.[5]
autocorrelation function of the time series. We prefer to use the linear spectrum to show the results in this report, since this can be directly re-lated to the parameters of the experiment as can be seen in this following example. Let us have a give sine wave with frequency of 42 Hz, then its linear spectrum is a peak at 42 Hz with the amplitude of the original sine wave. In this report we will show the results in the Linear Spectrum. This can also called the Narrow Band Spectrum.
The equivalent noise bandwidth (ENBW) is used to convert a spectral density to a spectrum and the other way around. The ENBW can easily be determined when computing the DFT. However, even when the spectrum and the spectral density are provided, the ENBW cannot be reproduced without additional information.
When we compute the DFT in the standard way, we get a very noisy signal. Thus we want to find a way to reduce the noise in our Linear Spec-trum. The usual way to do this is by taking the average of M estimates. In order to do this, the properties of the signal must remain stationary during the averaging. This will not work for our signal. We will use the Welch method to calculate the Linear Spectrum via DFT, which is a combination of using a window function and averaging several spectra by overlapping them. We chose this way of computing the DFT since we could control the amount of overlap, so that we did not average out too much of our signal. Let us look more closely at what the Welch method does. When we have a long continuous data stream that is simply split into several non-overlapping segments of N points and we process each segment by a DFT with a window function, we would have the following situation (Fig. 2.6): Near the boundaries of the window function the window is typically very small or zero. This means that a significant portion of the data stream is ignored in this analysis. This can be improved by overlapping samples as can be seen in Figure 2.7.[5] Now the question remains how much the segments should be allowed to overlap. This depends on the window that is used and on the requirements. In this project we used the Hanning window and an overlap of 10% to not average out too much signal.
8
2.4 Linear Spectrum 9
Chapter
3
Methodology
3.1
Calibration
The geophone needs to be calibrated prior to use. To calibrate the geo-phone we need to find its transfer function. The transfer function of a system is the response of the system to different frequencies. We will use this principle to calibrate the geophone by driving it at different frequen-cies and measuring its response. The geophone can be driven by putting in an excitation voltage. By changing the frequency of the signal we send in, we can drive the geophone at a range of frequencies and get the transfer function.
The setup that is used can be seen in Figure 3.1. A Lock-In Amplifier is used to drive the geophone at frequencies between 0.6 and 10 Hz and to filter out the signal from the noisy background. The cut-off frequency is expected to be around 3 Hz. A Pre-Amplifier is used to amplify the weak signal of the geophone. To find the resonance frequency f0and the damping coefficient ζ of the geophone, we fitted the measured data to Equation 2.2.
3.2
Vibration Measurements
Performing a vibration measurement is quite simple: a geophone is placed on the surface of the object that needs to be measured, and its output signal is read in and analyzed.
For the measurements we used two different instruments: the HS-1 In-dustrial geophone and the Guralp CMG-40T, which is a professional seis-mometer that measures directly the ground acceleration. The advantage
12 Methodology
Figure 3.1:Schematic drawing of the setup used in calibrating the geophone. The Lock-In Amplifier is used to drive the geophone at different frequencies.
Figure 3.2: Schematic drawing of the setup used for measuring vibrations with the HS-1 Industrial geophone.
of the geophone is that it is small and that it was already available from the beginning of my project. The Guralp CMG-40T is a quite big sensor (diam-eter: 168mm, height: 210mm, weigth: 5kg), but it has the advantage that it measures vibration on three axes (X,Y,Z), while with the geophone we get information only about the Z axis. The Guralp instrument was bought during my project, therefore we used two different experimental setups to perform the measurements (one only for the geophone and one for both of them after the Guralp arrived). At the beginning we used the setup that can be seen in Figure 3.2 for measuring only with the HS-1 geophone. The geophone is put on the surface and the Pre-Amplifier amplifies the signal and sends this to the DAQ-card. The oscilloscope is connected for hav-ing live feedback of the signal. Via USB the DAQ can be connected to a computer to get the data.
When the Guralp CMG-40T came in we had to change the setup, be-12
3.2 Vibration Measurements 13
Figure 3.3: Schematic drawing of the setup used for measuring vibrations with the HS-1 Industrial geophone and the Guralp CMG-40T.
cause the Guralp CMG-40T has 3 channels of data where the HS-1 Indus-trial geophone has just 1 channel. This new sensor can record vibrations in 3 direction and it stands on 3 points with adjustable height, so that it can be leveled. To read in the data of more than one channel we got a new record-ing instrument that can receive 6 channels of data at once. This recorder has a build-in amplifier. The setup for this measurement can be seen in Figure 3.3. The data recorded by the DAQ can be stored on a USB-stick or it can be send to a computer via FTP.
Chapter
4
Results and Discussion
4.1
Calibration of the geophone
We started with the calibration the HS-1 Industrial geophone. We did not have the opportunity to calibrate the Guralp CMG-40T.
The calibration of the geophone was done by driving the geophone with an excitation voltage at different frequencies, performing a frequency sweep. A Lock-In Amplifier fed the geophone with a signal of an am-plitude of 10 mV varying the frequency from 0.6 Hz to 10 Hz. A Pre-Amplifier was set to a gain of 50. This Pre-Pre-Amplifier can also introduce a high pass and a low pass filter into the circuit. For the calibration mea-surement the cut-off frequency of the low pass filter was set to 100 Hz. And the cut-off frequency of the high pass filter was set to 0 Hz, so that there is no high pass filter in our circuit. We then measured the output of the geophone. The data was then fitted to Equation 2.2. As a fitting procedure, instead of using the least square method, we minimized the following expression:
(yf it−ydata)2 yf it
(4.1) Where yf it is the amplitude of the fit in point x and ydata the amplitude of the data in point x. We chose this expression so that the points with a lower amplitude would wage more. In this way the slope of the transfer function, which is the important part for us, will be more accurate. The first results of the fit can be seen in Figure 4.1.
This fit clearly does not give use the correct transfer function. With this fit we get a value of 0.21891 Hz for the resonance frequency and a value
16 Results and Discussion
Figure 4.1:Frequency sweep from 0.6 to 10 Hz with the geophone. The data was fitted by Eq. 2.2. The fit gives f0=0.21891Hz and ζ =37.2657
of 37.2657 for the damping coefficient. Both values are rather unrealistic. This bad fit is caused by the offset of the data. A transfer function of a high pass filter will be 0 dB for frequencies above the resonance frequency (Fig. 2.4), however the data seem to approach -10 dB. To fit the data correctly we should take this offset into account, which can be done by multiplying the transfer function with a constant:
A(f) =C∗ f 2 q (f02− f2)2+4ζ2f2 0f2 (4.2)
This offset could be due to some loss of signal in our system. When we use equation 4.2 to fit the data, we get a much better result (Fig. 4.2). For the resonance frequency we get a frequency of 3.1357 Hz and a the damping coefficient is 0.51748. This results in the relation ζ =0.16 f0. For the constant C we found a value of 3.2367.
4.2
Vibrations in the Kamelingh Onnes
Labora-tory
4.2.1
The HS-1 Geophone
After the calibration measurement, we started measuring vibrations in the Kamerlingh Onnes Laboratory. The first measurements were performed 16
4.2 Vibrations in the Kamelingh Onnes Laboratory 17
Figure 4.2:Frequency sweep from 0.6 to 10 Hz with the geophone. The data was fitted by Eq. 4.2. The fit gives f0=3.1357Hz, ζ =0.51748 and C=3.2367
with the setup that can be seen in Figure 3.2 with the gain of the Pre-Amplifier set to 10000, , the cut-off frequency of the low pass filter was set to 1000 Hz, the cut-off frequency of the high pass filter was set to 0 Hz and the sampling rate ( fs) was 2000 Hz.
The first measurement we did was measuring the vibrations on the floor, the pit, the island and the setup. To measure on the setup, we put the geophone on the vibration isolation table on which the setup is stand-ing, in our case this is a STM. This vibration isolation table is standing on the pit instead of the islands, since the islands have lost most of their vi-bration isolation properties over the years. Due to sand flowing between the pit and the island, the pit is proved from previous measurements to be slightly better at isolating vibrations than the island. The results of this measurement can be seen in the Figures 4.3, 4.4, 4.5 and 4.6 (note the logarithmic y-axis). The data analyses was done with the scripts that can be found in “Manual python scripts measuring vibrations” [6]. When we compare the floor and the pit (Fig. 4.3), we see that the pit has a lot less vi-brations. We do see that a considerably big peak at approximately 25 Hz is present on both the floor and the pit. Our hypothesis is that this vibration could be due to things like lights or air conditioning.
It is much more interesting to look at Figure 4.4 to compare the mea-surement island, that was originally designed to damp vibrations, and the pit. At a first glance the pit does not look better at isolating vibrations, be-cause above approximately 80 Hz the vibrations on the island are less than
18 Results and Discussion
Figure 4.3:Vibrations measurement on the 15th of April 2016 on the floor and the pit in the KOL.
on the pit. However, looking at the lower frequencies, there is a big peak around 23 Hz on the island. On the pit this peak is hardly present. Since for STM experiments we are more sensitive to noise at frequency below 100 Hz then at higher frequencies, for our purposes the pit is better than the islands. There the 25 Hz peak is present on both the island and the pit, but this is less intense on the pit than on the island.
Now let us look at Figure 4.5 where we compared the vibrations on the pit with the vibrations on the setup. It can be observed that the table dampens vibration below 100 Hz, while it is worse at higher frequencies (Fig. 4.6), as expected from a damping system. However, an unexpectedly big peak at about 3 Hz is measured. In section 4.1 we found that the reso-nance frequency of the geophone is 3.1 Hz. We believe that the vibrations isolation table vibrates a little at or around 3 Hz, thus exciting the geo-phone at or around its resonance frequency giving such a noisy signal. We can see this peak in Figure 4.5 and more clearly on the double log scale in Figure 4.6. Since the geophone is excited at its resonance frequency, it will not measure as accurate as it did with the other measurements, resulting in a very noisy signal for the measurement on the vibration isolation table. To find out how the vibrations on the vibrations isolation table really are, we should measure with a geophone with a hugely different resonance frequency so that we do not have this problem. Unfortunately we did not have the time to do this measurement thoroughly, but a quick mea-surement with the Guralp CMG-40T, which has a resonance frequency of below 0.6 Hz, shows that there are a lot of vibrations around 3 Hz.
18
4.2 Vibrations in the Kamelingh Onnes Laboratory 19
Figure 4.4: Vibrations measurement on the 15th of April 2016 on the island and the pit in the KOL.
Figure 4.5: Vibrations measurement on the 15th of April 2016 on the pit and the vibration isolation table in the KOL.
20 Results and Discussion
Figure 4.6:Vibrations measurement on the 15th of April 2016 on the floor, the pit, the island and the vibration isolation table in the KOL.
4.2.2
The Guralp CMG-40T
The following measurements are performed with the new sensor, the Gu-ralp CMG-40T. Since this sensor records 3 channels of data, the X, Y and Z direction, we had to use a different setup (Fig. 3.3). We measured the vibrations on the measurement island with the geophone and the Guralp CMG-40T in the same place. We let this measurement run for 90 minutes and then we took the average spectrum of all the data, which can be seen in Figure 4.7. For every minute the time data is saved into a new file, so we can also plot the intensity of the vibrations at different frequencies as a function of time (Figures 4.8, 4.9 and 4.10).
Let us start with looking at Figure 4.7. At first we see that the vibrations in the X (north to south) and Y (east to west) direction are higher than in the Z direction. This is expected, because the islands are designed to damp the vibrations in the Z direction.
Secondly, we can compare the signal we get in the Z direction with the Guralp CMG-40T and the HS-1 geophone. At most frequencies the lines (nearly) overlap, but there are also some differences. At 25 Hz there is a peak in the X and Y direction. The geophone also measures this, while the Guralp does not. From approximately 37 Hz on, the geophone signal 20
4.2 Vibrations in the Kamelingh Onnes Laboratory 21
Figure 4.7: Vibration measurement on the 4th of May on measurement island west 1 in the KOL.
Figure 4.8:Intensity plot of the narrow band spectrum versus time in the X direc-tion on the 4th of May on measurement island west 1 in the KOL. Time on the x axis is indicated in hhmm.
22 Results and Discussion
Figure 4.9:Intensity plot of the narrow band spectrum versus time in the Y direc-tion on the 4th of May on measurement island west 1 in the KOL.
looks more similar to that of the X and Y direction of the Guralp CMG-40T than the one of the Z direction. In section 3.2 we described that the Guralp CMG-40T can be leveled, since it records 3 channels of data. The geophone can not be leveled thus might also pick up more vibrations from the X and Y direction resulting in these differences between the Guralp and the geophone. Since the noise amplitude at 25 Hz is roughly one order of magnitude higher in the X and Y direction than in the Z direction, we can infer that the noise originates in the horizontal direction, and the noise measured on the Z direction is mainly due to crosstalk. It is possible that these vibrations are due to lights or air conditioning, but we did not have the chance to test this hypotheses.
In Figures 4.8, 4.9 and 4.10 we show the color plots of the linear spec-tra as a function of frequency and time, in order to visualize how the vi-brations evolve with time. In these graphs we calculate and plot a linear power spectrum for each minute of the measurement. We can see that at some minutes the noise is suddenly higher (vertical “bright lines”), this is probably due to people walking nearby the measurement setup. We also notice that the peak at 31Hz sometimes suddenly stops for some minutes and then starts again, while the peak at 25 Hz is very strong and always present.
22
4.2 Vibrations in the Kamelingh Onnes Laboratory 23
(a)Measured with geophone.
(b)Measured with the Guralp.
Figure 4.10: Intensity plot of the narrow band spectrum versus time in the Z di-rection measured on the 4th of May with the Guralp CMG-40T and the geophone.
24 Results and Discussion
4.3
Vibrations in the Ultra Microscopy Hall
In this section we show the vibration measurements performed in the Ul-tra Microscopy Hall. The first place where we measured in the UMH was the pump corridor next to the new measurement hall. When we will move to this building, all the vacuum pumps will be placed here. We chose this spot for our first measurement, because the building was still empty and we wanted to learn the noise level before the building starts to be used. Particularly, in the room next to the pump corridor, a potentially vibration-generating piece of equipment will be placed, and we want to be able to see if this will make any difference in the vibration of the pump corridor. The vibration spectra of this measurement can be seen in Figure 4.11 and 4.12. It can immediately be seen that this spectrum is a lot different than in the Kamerlingh Onnes Laboratory. In the pump corridor of the UMH the low frequency (10-50Hz) noise is mostly 1 order of magnitude lower than on the island of the KOL.
Figure 4.11:Intensity plot of the vibrations in the Z direction in the pump corridor of the UMH.
24
4.3 Vibrations in the Ultra Microscopy Hall 25
Figure 4.12: Average narrow band spectrum of the vibrations in the pump corri-dor in the UMH.
The second place where we measured was the measurement island. The measurement island (Fig. 2.2) works with an air spring damping sys-tem. In order for this system to work, the air legs of the island should be filled with air by using a compressor. For this we need electricity, but at the time of the measurement it was not available in this part of the building yet. In Appendix B we will report this measurement.
We then were able to fully lift the island and to perform the vibration measurements on it. The results of the vibrations on the lifted island can be seen in Figures 4.13(a) and 4.14(a).
The first thing we notice is that the vibrations in the X and Y direction increase dramatically after 16:30, but this change is hardly present in the Z direction. Why this happens is not clear, maybe it could be that someone turned something on that caused vibrations in the horizontal plane or that something happened that suddenly decreased the quality of the measure-ment island. It is hard to determine what happened, since we were not present when this change happened. We do know that one of the legs of the measurement island is a bit leaky, this could have changed the damp-ing of the measurement island. When we compare the spectrum in the X direction of a minute before this sudden event and a minute after this event, we can clearly see a difference (Fig 4.13(b)). The vibrations around
26 Results and Discussion
20 Hz and below 10 Hz are much more present. In Figures 4.13(b) and 4.14(b) we also compared the not fully lifted island with the lifted island.
(a)Intensity plot.
(b)Narrow band spectrum 02:12 18-06-2016, 02:12 19-06-2016 and 02:12
20-05-2016
Figure 4.13:Vibrations on the X direction on the measurement island in the UMH.
(a)Intensity plot.
(b)Narrow band spectrum at 02:12 18-06-2016, 02:12 19-06-2016 and 02:12
20-05-2016
Figure 4.14:Vibrations on the Z direction on the measurement island in the UMH. The biggest difference we see is that the vibrations above 40 Hz go down a lot for the lifted island. However, we think that this an artifact of the measurement equipment rather than the island damping these vibra-tions, because in Figure 4.12 we can also see this behavior. For vibrations with a frequency below 40 Hz there is hardly any difference between the 26
4.4 Week measurement 27
partly lifted island and the fully lifted island. In both the X and Z direction the vibrations are the highest at 02:12 19-06-2016, when the measurement island was fully lifted.
Lastly, we see that the measurement island damps vibrations very well. Vibrations with a frequency below 5 Hz are not damped much, but for higher frequencies they are damped significantly.
To really determine how well the measurement island in the Ultra Mi-croscopy Hall damps the vibrations, we should compare the vibrations on the basement floor to the vibrations on the measurement island. In order to be able to compare this, I measured on the basement floor on which the measurement island is located. In section 4.4 I will show more of these results. The comparison of the vibrations can be seen in Figures 4.15(a) and 4.15(b). It is very clear that the measurement island is designed to damp vibrations in the Z direction, these are damped of about 1 order of magnitude. For the X and Y direction the damping is less effective. Note that the measurement islands does not seem to damp vibrations below a frequency of 2 Hz, which is expected since its resonance frequency is a bit higher than 1 Hz, and vibrations below the resonance frequency are not damped.
(a)X direction (b)Z direction
Figure 4.15: Narrow band spectrum of the vibrations on Saturday June 18th on the island and Saturday June 4th on the basement.
4.4
Week measurement
To learn more about the difference in vibrations over time, we measured vibrations for a whole week on the basement floor of the UMH. We had technical problems in the new building for a flood that happened a week
28 Results and Discussion Figure 4.16: Intensity gr aph of the vibrations in the Z dir ection on the basement floor in the UMH. T ime span of this measur ement is 31-05-2016 till 07-06-2016. 28
4.4 Week measurement 29
Figure 4.17: Narrow Band Spectrum of the vibrations in the Z direction on three different times: 03:00, 12:00 and 21:00. Note the difference in intensity for the lower frequencies, the peak below 1 Hz and the peaks at approximately 40 Hz.
before the measurement, therefore the external conditions were not perfect and this might have influenced our measurement. The data of this week is organized into one image per direction, Figures B.5, B.6 and 4.16. Note that there has been a loss of data between 15:53 05-2016 and 06:10 06-07-2016, this can results in sudden jumps in the graphs that I will show in this section. This section will only focus on vibrations in the Z direction.
The first thing we notice is the variation in intensity in the low fre-quency vibrations over time. These variations indicate the difference be-tween day and night. At day time there is more activity in and around the building, such as people walking and cars driving outside. When we take 3 narrow band spectra at different times, we can see this difference rather clearly (Fig. 4.17). When we look at this graph, two other things stand out: first, there are peaks at approximately 40 Hz at 03:00 and 21:00, but there is not a corresponding peak at 12:00 and second, there is a big peak below 1Hz for all times.
Let us first examine the peaks at approximately 40 Hz more closely. We can see that these vibrations are mostly present during the night, but they are not present during the weekend. We can add the intensities of all the frequencies between 40 and 41 Hz to see the evolution of the
vi-30 Results and Discussion
Figure 4.18: Change in intensity of the vibrations in the Z direction between 40 and 41 Hz. Note the shape of the peaks.
brations. These results can be imaged with plotting it against time, as we did in Figure 4.18. As a first thing we see that they indeed vary over time, in the weekend the vibrations are very low. Secondly we notice the pe-culiar shape of the peak which can been seen the best in Figure 4.18. The vibrations seem to suddenly start and then slowly decrease.
Now let us look at the peak below 1 Hz of Figure 4.17. When we look closely at this peak, we see this is a peak between 0.3 and 0.6 Hz. When we zoom in to the intensity plots, we can see this more clearly (Figure 4.19). This peak can also be found in the X and Y direction. Even though such a low frequency noise will probably not influence our STM, it is a very intense vibration which we would like to know more about. In Figure 4.20 we plotted the change in intensity between 0.3 and 0.6 Hz against the time. This shows a rather irregular pattern. This vibration could be due to the wind I talked about earlier (Section 2.3). But since we did not monitor the wind, we can not be sure about this. One might think that the fast changes in the vibrations are due to traffic or people in the building, since we saw that the vibrations at low frequency got more intense during the day time. But when we look at the moments that the fast changes in this vibration happen, we see that these do not coincide with the increase or decrease in traffic.
When we go back to Figures B.6 and 4.16 there is an other interesting vibration. It is better seen when we zoom in, as we did in Figures B.4 and 4.21(a). On Figure B.4 there can be seen a “dashed” line at 33 Hz and this 30
4.4 Week measurement 31
Figure 4.19:Intensity plot of the vibrations between 0 and 10 Hz in the Z-direction on 02-06-2016. Note the high intensity around 0.5 Hz. The black dashed lines show the area where the frequency band change was computed.
Figure 4.20: Change in intensity of the vibrations in the Z direction between 0.3 and 0.6 Hz over a week. The vertical lines show midnight.
32 Results and Discussion
“dashed” line can also be seen at 25 Hz in Figure 4.21(a). The length of one “dash” is approximately 2 hours. In the X direction this “dashed” line is also present at 25 Hz, but not as strong as in the Y and Z direction. For this frequency band we can again plot the change in intensity against the time and we get something very interesting (Fig. 4.21(b)). It is very clear that the “dashed” lines in all 3 directions are related, so therse vibrations are very likely to originate from the same source.
(a)Intensity plot, Z direction. (b)Comparing X, Y and Z
Figure 4.21: (a) Intensity plot, note the yellow “dashed” line. The black dashed lines show the area where the frequency band change was computed. (b) Change in intensity of the vibrations between 24.5 and 25 Hz for the X and Z direction and between 33.33 and 33.67 Hz in the Y direction. The vibrations in the Z direction are divide by 4 to scale it to the intensity of the X and Y direction.
Another interesting thing are the oscillations in frequency space in some minutes, this for example can be seen in Figure 4.19. More about these os-cillations can be seen in Appendix B.
4.5
Comparing the Different Measurement Halls
We would like to compare the vibrations between the Kamerlingh Onnes Laboratory and the Ultra Microscopy Hall. First let us start with looking at the vibrations on the floor (Figures 4.22(a), 4.22(b) and 4.23). In these fig-ures we can clearly see the difference between a building in which many people work, the Kamerlingh Onnes Laboratory and a building in which is mostly empty, the Ultra Microscopy Hall. I should note that the mea-surement in the KOL building are done on the lab floor on which people constantly walk an the measurement in the UMH is done on the basement 32
4.5 Comparing the Different Measurement Halls 33
(a)X direction. (b)Y direction.
Figure 4.22: Vibrations on the floor of the KOL and the basement floor of the UMH
floor on which people do not walk that much. In the X and Y direction all the vibrations above 10 Hz are lower in the UMH compared to the KOL. For the Z direction the difference is smaller, but it is still clear that the basement of the UMH has less vibrations.
Secondly, let us look at the vibrations on the measurement islands (Fig-ures 4.24(a), 4.24(b) and 4.25). It is very clear that the new measurement islands in the Ultra Microscopy Hall are much better than the ones in the Kamerlingh Onnes Laboratory. We would like to quantify the quality of the measurement island with a number. We can do this by comparing the two different measurement islands. When we divide the intensity of the vibrations in the KOL by the intensity of the vibrations in the UMH and take the average of this we get a number to quantify the difference between the two islands. This is a very robust way of comparing the vi-brations, but does give us a clear way to express the difference. We can do this for the 3 different directions we measured in.
For the X direction we get a value of 39, which indicates that the vi-brations on the measurement island in the UMH are 39 times less than the vibrations on the measurement island in the KOL. When we do the same thing for the Y direction we get a value of 44 for the whole frequency range. We see that the vibrations below 2 Hz are worse in the UMH in the Y and Z direction, so we can also look at the difference in two different regions: for the vibrations with a frequency below 2 Hz and the vibrations with a frequency above 2 Hz. Below 2 Hz we get a value of 0.6 for the Y direction, meaning that the vibrations are worse in the UMH and above 2 Hz we get a value of 46. The only direction that still remains is the Z
di-34 Results and Discussion
Figure 4.23: Vibrations on the floor of the KOL and the basement floor of the UMH in the Z direction.
rection, vibrations in this direction could influence our measurements the most, so this is the most interesting direction for us. Again we can split this spectrum into the same two parts. For vibrations below 2 Hz we get a value of 0.5 and for vibrations above 2 Hz we get a value of 70.
We should note that this value is not only due to the measurement island. In the Kamerlingh Onnes Laboratory there are more vibrations on the normal floor than in the Ultra Microscopy Hall, so this also influences this value for the difference.
34
4.5 Comparing the Different Measurement Halls 35
(a)X direction. (b)Y direction.
Figure 4.24:Vibrations on the measurement island in the KOL and in the UMH
Figure 4.25: Vibrations on the measurement island in the KOL and in the UMH in the Z direction.
Chapter
5
Conclusion and Outlook
In this report we have seen that the vibration isolation islands and the pits in the Kamerlingh Onnes Laboratory damp vibrations rather well, but at low frequencies the damping could be improved mainly for the vibrations at 25 Hz, which hardly get damped by the pit.
On the floor of the Ultra Microscopy Hall vibrations are lower than on the floor in the Kamerlingh Onnes Laboratory. We do not see any big peaks in the UMH as we did see in the KOL, such as the intense peak at 25 Hz. When we look at the measurement islands in the UMH, we see that these damp vibrations in the Z direction very well, but hardly damp any vibrations in the X and Y direction.
The week measurement has shown us that there are a lot of different vibrations that change over time. There are vibrations around 41 Hz that are present mostly at night, but are not seen in the weekend. And there are vibrations at 25 (X and Z direction) and 33 Hz (Y direction) that are present for two hours, disappear for two hours and then reappear again for two hour. This pattern is present for the entire week.
At last we compared the vibrations on a measurement island in the KOL and a measurement island in the UMH. Vibrations in the horizontal direction are approximately 40 times less in the UMH than in the KOL and in the vertical direction the vibrations are 70 even times less. We can also compare the vibrations in the Ultra Microscopy Hall at LION with the vi-brations of famous low vivi-brations laboratories in the world (Fig. 5.1). This figure shows us that the vibration isolation system in the Ultra Microscopy Hall is one of the best in the world.
During this project we have found many interesting vibrations. With further research we could try to correlate sources to these vibrations. Es-pecially the vibrations at 25 Hz in the UMH would be interesting to find an
38 Conclusion and Outlook
Figure 5.1:Linear Spectrum Density of the vibrations at Cornell, Berkeley, Urbana and LION
explanation for, since this is a rather intense vibration. Another thing that could be researched is how the vibrations in the UMH change when more people move into the building. As a last thing acoustic noise could also influence the measurements, so this could also be something interesting to investigate.
38
References
[1] E. L ¨ortscher, D. Widmer, and B. Gotsmann, Next-generation nanotech-nology laboratories with simultaneous reduction of all relevant disturbances., Nanoscale 5, 10542 (2013).
[2] J. E. Hoffman, A Search for Alternative Electronic Order in the High Tem-perature Superconductor Bi2Sr2CaCu2O8+δ by Scanning Tunneling Mi-croscopy, PhD thesis, 2003.
[3] M. Hesselberth, Microscopie in beweging, Eureka .
[4] M. Hesselberth, Trillingsmeting meeteiland met luchtdempers, page 1 (2016).
[5] G. Heinzel, a. R ¨udiger, R. Schilling, and T. Hannover, Spectrum and spectral density estimation by the Discrete Fourier transform (DFT), includ-ing a comprehensive list of window functions and some new flat-top, Max Plank Institute , 1 (2002).
Appendix
A
Possible Problems
A.1
Gain of the Pre-Amplifier
For the first measurement the Pre-Amplifier was set to a gain of 100, the cut-off frequency of the low pass filter was set to 1000 Hz, the cut-off fre-quency of the high pass filter was set to 0 Hz and the sampling rate ( fs) was 2000 Hz. We measured for one minute on the floor, one minute on the island and one minute on the pit. In Figure A.1 you can see the results of this measurement. It seem as the floor vibrates less than the pit and is-land. Another thing to note is that the vibrations on the pit and the island seem very constant with frequency. This led to the thought that we were doing something wrong with our measurement. Our hypothesis was that we didn’t amplify the signal enough to measure vibrations on the pit and the island.
To test this we did a measurement on the island while varying the gain of the Pre-Amplifier. We chose to measure at a gain of 100, 1000, 2000 and 10000. As can be seen in Figure A.2, the gain was indeed the problem we had in our measurement. A gain of 1,000 and 2,000 is still not enough to amplification for the signal, but a gain of 10000 is. For the rest of the experiments on the pit and the island we will use a gain of 10000.
42 Possible Problems
Figure A.1:Vibration measurement in the KOL on the floor, the pit and the island. Note that the floor seems to be vibrating less than the pit and the island.
Figure A.2:Vibration measurement on the island in the KOL with different values for the gain.
42
A.2 Powering the recording instrument 43
A.2
Powering the recording instrument
In order to record the data from the Guralp CMG-40T, the recording in-strument should be powered with 12 Volts. At the beginning we used a battery to power this, but the battery only has power to power it for a maximum of 36 hours. Since we wanted to measure for a longer time, we started powering the recording instrument with a power supply. Having connected the device to the power net, we might have some more elec-tronic noise than we have with the battery. To find the difference between the two ways of powering the recording instrument, we did a measure-ment on the same location first powering it with the battery and then with the power supply. The results of these measurements can be seen in Fig-ures A.3, A.4 and A.5.
The first thing that we noticed when looking at these figures is the bright vertical line in Figure A.3(a). This indicates the time at which we changed from the battery to the power supply. The bright vertical line is due to the starting up of the X channel of the Guralp. The first minute this channel will give a very intense signal and does not measure any vibra-tions.
Secondly, we hardly notice any difference between the signal from the battery and the power supply. The average narrow band spectrum for the measurement with the battery is almost the same as that for the measure-ment with the power supply. There is one clear difference in the X and Y direction. During the measurement with the power supply there is a peak around 5 Hz which is not present in the measurement with the battery. In the Z direction there is no difference between the measurement with the battery and with the power supply.
44 Possible Problems
(a)Intensity plot (b)Average spectrum with the battery and the power supply
Figure A.3: Vibrations in the X direction on the floor measured by powering the recording instrument with a battery(09:08-09:39) and a power supply(09:40-10:11).
(a)Intensity plot (b)Average spectrum with the battery and the power supply
Figure A.4: Vibrations in the Y direction on the floor measured by powering the recording instrument with a battery(09:08-09:39) and a power supply(09:40-10:11).
44
A.2 Powering the recording instrument 45
(a)Intensity plot (b)Average spectrum with the battery and the power supply
Figure A.5: Vibrations in the Z direction on the floor measured by powering the recording instrument with a battery(09:08-09:39) and a power supply(09:40-10:11).
Appendix
B
Additional measurements
B.1
Partly lifted measurement island
Here we report the first measurement we did on the island, which unfor-tunately was not fully lifted, therefore not at the best of its performance. The results can be seen in Figure B.1. We immediately see that the vibra-tions are much lower than in the pump corridor. Note that at frequencies higher than 20 Hz, the signal of the geophone saturates to a straight line. The saturation of the geophone is due to the geophone not standing stable enough on the rough surface of the measurement island. When the geo-phone is not stable, it will also be moved by the vibrations in the X and the Y direction resulting in this spectrum.
To put the geophone down more stable, the Fine Mechanical Depart-ment (FMD) made a platter that stands on adjustable points so that we can level the geophone and place it more stable on the rough surface of the concrete measurement island.
48 Additional measurements
Figure B.1: Average narrow band spectrum of the vibrations on the partly lifted measurement island in the UMH.
B.2
Oscillations in frequency space
We have plotted the time signal and the narrow band spectrum of a minute with oscillations in frequency space (Fig. B.3) and one without (Fig. B.3). Let us first look at the two narrow band spectra. The oscillation are only present for frequencies below 20 Hz. Oscillations like this are mostly caused by a discontinuity in the time signal, which however isn’t present in Figure B.3(b). But we do see a lot more noise in this time signal, this could be due to external influences.
(a)Narrow Band Spectrum (b)Time signal
Figure B.2: Time signal and the narrow band spectrum of 12:50 at 02-06-2016. This is a minute without oscillations in frequency space.
48
B.3 Week measurement 49
(a)Narrow Band Spectrum (b)Time signal
Figure B.3: Time signal and the narrow band spectrum of 14:12 at 02-06-2016. This is a minute with oscillations in frequency space.
B.3
Week measurement
50 Additional measurements
Figure B.4: Intensity plot of the vibrations between 30 and 40 Hz in the Y-direction on 04-06-2016. Note the “dashed” line at 33 Hz.
50
B.3 Week measurement 51 Figure B.5: Intensity graph of the vibrati ons in the X dir ection on the basement floor in the UMH. T ime span of this measur ement is 31-05-2016 till 07-06-2016.
52 Additional measurements Figure B.6: Intensity graph of the vibration s in the Y dir ection on the basement floor in the UMH. T ime span of this measur ement is 31-05-2016 till 07-06-2016. 52
