Abstract - LPHE · 2009-06-18 · Abstract The characterization of ultra high energy neutrinos could provide insights into cosmological source evolution and cosmological accelerators

Abstract

The characterization of ultra high energy neutrinos could provide insights intocosmological source evolution and cosmological accelerators which are responsiblefor ultra high energies. Due to the low expected cross section of ultra high en-ergy neutrinos a much larger detection volume (>100 km3) is required than thatof currently operating detectors. Inspired by Askaryan, both radio and acousticdetection techniques could extend the detection volume thanks to longer wave at-tenuation lengths (> 1km in theory,) allowing for a sparser and less expensiveinstrumentation compared to the currently used optical detection technique.

This work presents a prototype of a 4-channel acoustic sensor and its absolutesensitivities and discusses capabilities of a possible vertex reconstruction producedby an interaction of ultra high energy neutrinos.

Contents

1 Introduction 1

2 Askaryan’s thermoacoustic model 5

3 Design of the sensor prototype 73.1 Housing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73.2 Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

4 Experimental setup 114.1 Absolute calibration of the sensor . . . . . . . . . . . . . . . . . . . 134.2 Setup for characterization of the polar orientation of the sensor . . . 15

5 Analysis in the time domain 175.1 Extracted primary signal . . . . . . . . . . . . . . . . . . . . . . . . 175.2 Noise level analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 185.3 Calculation of the S/N ratio . . . . . . . . . . . . . . . . . . . . . . 195.4 Absolute sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

6 Analysis in the frequency domain 236.1 Data tapering windows . . . . . . . . . . . . . . . . . . . . . . . . . 236.2 Absolute sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . 246.3 Frequency reconstruction . . . . . . . . . . . . . . . . . . . . . . . . 25

7 Analysis of the polar orientation of the sensor 277.1 Measurement of sound propagation in the sensor under water . . . . 29

8 Error discussion 31

9 Hardware and software during the experiments 339.1 Detected errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339.2 Recommendations for later data acquisition . . . . . . . . . . . . . 36

10 Summary and outlook 37

A Acknowledgement 43

B Statement of independence 45

C Appendix 47C.1 The proceedings of the 31st ICRC, Lodz 2009 . . . . . . . . . . . . 47C.2 Attached data CD . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

List of Figures

1.1 Astrophysical messengers . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Neutrino flux upper bounds/ sensitivities for IC-9 / IC-80 (at 90%

C.L) and other experiments compared to the Engel, Seckel, Stanev(ESS) GZK flux model . . . . . . . . . . . . . . . . . . . . . . . . . 2

2.1 Askaryan’s thermoacoustic model . . . . . . . . . . . . . . . . . . . 52.2 Simulated pressure pulses in 400m distance to the center of the cas-

cade caused by a neutrino of 1016 eV . . . . . . . . . . . . . . . . . 6

3.1 The housing with polished and porous surfaces . . . . . . . . . . . . 73.2 Housing with electronics . . . . . . . . . . . . . . . . . . . . . . . . 83.3 Piezo Pz26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83.4 The connection scheme of the amplifier . . . . . . . . . . . . . . . . 93.5 The upper layer of the amplifier with its technical scheme . . . . . . 103.6 Gain of the total amplification process as a function of frequency . . 10

4.1 The tank and data acquisition hardware . . . . . . . . . . . . . . . 114.2 Data acquisition scheme . . . . . . . . . . . . . . . . . . . . . . . . 124.3 Used transmitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124.4 The commercial hydrophone SENSORTECH SQ03 . . . . . . . . . 134.5 Total and extracted signal . . . . . . . . . . . . . . . . . . . . . . . 134.6 Shapes of transmitted signals in Volt . . . . . . . . . . . . . . . . . 144.7 Simulation of the transmitted pressure signals . . . . . . . . . . . . 144.8 The experimental environment . . . . . . . . . . . . . . . . . . . . . 15

5.1 Almost possible reflection paths in the tank . . . . . . . . . . . . . 175.2 Total recorded signal with an extracted primary response . . . . . . 175.3 Noise signal polluted with spikes . . . . . . . . . . . . . . . . . . . . 185.4 S/N ratio for the reference hydrophone as a function of frequency

and shape of the transmitted signal . . . . . . . . . . . . . . . . . . 195.5 S/N ratio for the 4-channel sensor as a function of frequency and

shape of the transmitted signal . . . . . . . . . . . . . . . . . . . . 205.6 Absolute sensitivity of the 4-channel sensor as a function of fre-

quency and shape of the transmitted signal . . . . . . . . . . . . . . 215.7 Pressure spectrum of the reference hydrophone as a function of fre-

quency and shape of the transmitted signal . . . . . . . . . . . . . . 215.8 Pressure sensitivity as a function of frequency of transmitted Neu-

trino pulses normalized to S/N = 1 ratio . . . . . . . . . . . . . . . 225.9 Pressure sensitivity as a function of frequency of transmitted Gaus-

sian pulse normalized to S/N = 1 ratio . . . . . . . . . . . . . . . . 22

IV List of Figures

6.1 Different data tapering windows . . . . . . . . . . . . . . . . . . . . 236.2 FFT of the extracted primary signal for the piezo element no.3 and

the reference hydrophone . . . . . . . . . . . . . . . . . . . . . . . . 246.3 Absolute sensitivity in the frequency domain (left) and the detection

threshold (right) for piezo element 3 . . . . . . . . . . . . . . . . . . 256.4 Reconstructed frequencies . . . . . . . . . . . . . . . . . . . . . . . 25

7.1 Recorded signals of the 4-channel sensor at 0o . . . . . . . . . . . . 277.2 Time of first signal maximum as a function of the polar orientation

of the piezo element 3 . . . . . . . . . . . . . . . . . . . . . . . . . 287.3 Time of first signal maximum as a function of the polar orientation

of the sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287.4 Prototype of an instrument for measurement of sound propagation

in the housing of the sensor under water . . . . . . . . . . . . . . . 29

9.1 Influence of one channel receiving a signal to remained channels . . 339.2 Connection scheme between the input channels of the hardware and

software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349.3 Improved channel response and coupling . . . . . . . . . . . . . . . 349.4 Decreasing offset of the hydrophone . . . . . . . . . . . . . . . . . . 359.5 Damaged surface of the housing . . . . . . . . . . . . . . . . . . . . 35

1 Introduction

Throughout the history of humanity, stars, heavenly bodies, and universe as awhole have fascinated billions of people and continue to fascinate them. Humancuriosity about these unknown objects has created one of the oldest science onthis planet: astronomy. Through the surprising results of Victor F. Hess’ balloonexperiments detecting higher air ionization above 5000m in 1912 [1], the new fieldof astroparticle physics was born, with the goal of studying elementary particles ofastronomic origin. With the indirect discovery of neutrinos by Frederic Reines in1956 neutrino astronomy has been created as a new form of celestial observation[2].

Neutrinos have the attribute of departing straightforward from their sources andtheir trajectories remain in the majority of cases unaffected by forces producedthrough planets or magnetic fields among others. By the detection of extrater-restrial neutrinos and their trajectorial reconstruction, potential sources and cos-mogenic accelerators can be localized in the sky (Fig. 1.1). Up to now only twosources of cosmic neutrinos have been found: the Sun [3] and the supernova SN19876A detected by the Kamiokande II [4] experiment with neutrinos in an energyrange 20 − 40 MeV. So far, no extraterrestrial neutrinos could be detected abovethese energies despite theoretical predictions [5].

Figure 1.1: Astrophysical messengers [6]

2 Introduction

In 2005 the IceCube Neutrino Detector has been established among others atthe South Pole with the goal of neutrino flux characterization (Fig. 1.2) [7]. Thedetector could soon prove the existence of ultra high energy neutrinos (UHE) inan energy range from 108 − 1012 eV which follows from interactions of ultra highenergy cosmic rays with the cosmic microwave background. With around 5000photo multipliers detecting the Cherenkov light produced by nucleons coming froman interaction with a UHE neutrino, it would also be possible to extract informationabout the cosmic rays accelerators producing such high energies among others.

Figure 1.2: Neutrino flux upper bounds/ sensitivities for IC-9 / IC-80 (at 90%C.L) and other experiments compared to the Engel, Seckel, Stanev(ESS) GZK flux model [8, 9]

As neutrinos are elementary particles participating only in weak interaction, theircross section is small in comparison to the cross sections of photons and chargedcosmic rays, but their mean free path is much longer [10]. Due to the difficultyof their detection a large volume of target material is required in order to obtainreasonable results. For example the probability of a cosmogenic neutrino flux inthe range EdN/dE ∼ 10−17 s−1 cm−3sr−1 at E = 1018 eV would be 0.01 - 1 event /year / km3 in ice [8].

As the rate of detection is limited by the attenuation length of light in ice ( 102m)and by the number of affordable photo multipliers, new detection methods are re-quired to expand the detection volume. In 1957 G. Askaryan introduced a thermoa-coustic model describing the sound generation mechanism by high energy particlesin a liquid medium [11]. After a verification of the model in 1979 [12] in water, themodel can also be confirmed for solids such as ice [13].

Apart from the emission of a coherent Cherenkov radio pulse in the range of0.1-1GHz close to the shower axis [14], the interacting neutrino would also producea thermoacoustic disc perpendicular to the shower axis with a frequency rangeof 10-100kHz. While the larger attenuation lengths could be already convincingly

3

demonstrated for the radio emission [15, 16, 17], the South Pole Acoustic Test Setup(SPATS) [18, 19] in 2006/2007 and later the Hydrophone for acoustic detection atSouth Pole (HADES) [20] in 2007/2008 have been installed to investigate theirtheoretical predicted acoustic attenuation lengths in a range of 103 m [21] andthe sources of background noise [22]. As a further goal of the established acousticsetups, the determination of the sound speed as a function of depth can be alreadyconfirmed with great accuracy meeting theoretical expectations [23].

Contrary to salt and water, ice allows for the detection of three distinct attributesaccompanying a neutrino event: Optical Cherenkov light, coherent radio Cherenkovand thermoacoustic emissions. Implementing the acoustic detection method men-tioned above would expand the detection volume up to 100km3 at the same costsand 10− 103 interacting neutrinos could be expected after 10 years [24].

This thesis describes a first approach of the next step on the way to an acoustic-radio-optical hybrid detector: the development of a high sensitive acoustic 4-channel sensor allowing for UHE neutrino spectrum analysis and the ir vertexreconstruction.

Chapter 2 gives a short introduction into the Askaryan thermoacoustic model inliquids and solids. In chapter 3 and 4 the prototype of the 4-channel sensor andthe experimental setup for characterization purposes are presented. Chapter 5 and6 describe the obtained absolute sensitivities and signal-to-noise ratios in the timeand frequency domain relative to a commercial reference hydrophone. Then theanalysis of the polar orientation is shown in chapter 7. After a short discussion oferrors in chapter 8, chapter 9 presents the behavior of hardware and software duringthe experiments. Finally, chapter 10 concludes with an outlook and a summary.

4 Introduction

2 Askaryan’s thermoacoustic model

The acoustic high energy particle detection bases on the thermoacoustic modelwhich was first discussed by Askaryan in 1957 [11]. The model describes the energydeposition of a particle shower produced by a high energy particle such as a neutrinoand their obtained acoustic signals in target medium. The local energy deposition

Figure 2.1: Askaryan’s thermoacoustic model [19]

causes a fast heating of the target medium followed by a thermal volume expansionwhich depends on a media specific expansion coefficient α(T ).

∆V = V0α(T )∆T

With ∆T = ε(~r,t)cp

the thermal volume expansion can be expressed as a function of

energy deposition density ε(~r, t) and material properties (cp describes the specificheat capacity of a target medium).

∆V = V0α(T )

cpε(~r, t)

Aside from this the thermal volume expansion results in an acoustic wave extendingperpendicular to particle shower. In liquids such as water this acoustic wave ismathematically characterized by a pressure signal p(~r, t) as a variation of the staticpressure p0

1. The received pressure signal can be described by an inhomogeneous

1In solids such as ice this scalar pressure signal p(~r, t) is replaced by a longitudinal displacementvector ~ul(~r, t). Furthermore the total displacement vector characterizing the sound propaga-tion in solids is defined by the sum of the longitudinal and transversal part [25]

6 Askaryan’s thermoacoustic model

wave equation in fluids.

− 1

v2sound

∂2p(~r, t)

∂t2+∇2p(~r, t) =

α(T )

cp

∂2ε(~r, t)

∂t2

where vsound represents the speed of sound in the target material (for water2:≈1484m/s). Assuming that the instantaneous energy deposition occurs almostwith the speed of light and is much faster than the sound propagation and heatconduction, the excitation term can be simplified to ε(~r, t) = ε(~r)∆t and the inho-mogeneous wave equation can be solved using the Kirchhoff integral:

p(~r, t) =α(T )v2

4πcp

∂

∂R

∫S′

ε(~r′)R

d2r

With S ′ as the total surface of a sphere of radiusR = vsound·t around the observationspot ~r the integral only depends on the spatial distribution of energy depositiondensity. Assuming a homogeneous energy deposition in a cylindrical volume, thepeak frequency can then be calculated as:

fpeak =vsound2 · d

With theoretically estimated values for such a tube (length L = 5− 10m, diameterd =≈ 0.1m for a UHE neutrino at 1018 eV), the peak frequency is fpeak ≈ 20kHz.Contrary to this assumption made by Askaryan [12], Learned chose another modelproposing a line distribution whose width is smeared with a normal distributionfor energy disposition [26]. Together with simulation done by Dedenko [27] all pre-sumptions predicted a bipolar pulse shape (Fig. 2.2)

Figure 2.2: Simulated pressure pulses in 400m distance to the center of the cas-cade caused by a neutrino of 1016 eV

Principle of detectionThe principle of detection is based on piezoceramics converting acoustic (pressure)signals into electric signals. Due to their polarized dipole structure, an imposedpressure upon a piezo crystal causes a dislocation of its dipoles producing a mea-surable difference in the electric potential.

2The speed of sound in water is a function of temperature [48]

3 Design of the sensor prototype

3.1 Housing

The main part of the housing has a cylindrical shape and is made of aluminum.With a diameter of 10cm and a height of 9cm, this part contains 4 slots for piezo-ceramics as shown in Fig. 3.1a. Each of these slots has the same form for the

(a) Housing of the sensor (b) Porous and polished surfaces

Figure 3.1: The housing with polished and porous surfaces

piezoceramics, but its position is different in the housing. The piezoceramics areplaced symmetrically in two planes (Fig. 3.1b) separated by ≈5cm in the verti-cal axis with the objective of a later horizontal and vertical vertex reconstructionwith one sensor module. A special screw which is illustrated in Fig. 3.2 allowsthe optimal fixation of the piezo elements to the housing. The coupling surfacebetween a piezoceramics and housing is polished for two piezoceramics and porousfor the remaining two in order to get the optimal response of the detection elements(Fig. 3.1b). Aside from the fixation modules for the preamplifier the housing willbe equipped with a digitization board and a gyroscope in the newer version madeof steel [28].


Figure 3.2: Housing with electronics

PiezoceramicsThe piezoceramics Pz26 [29] used have a diameter of 10mm and a thickness of2mm. They are made of lead-zirconate-titanate alloy and characterized by a piezo-electric charge coefficient d33 = 330 · 10−12C/N , a coupling factor k33 = 0.68 and apiezoelectric voltage coefficient g33 = 28 · 10−3V m/N . The resonance of the piezo-ceramics is estimated at 200kHz [30]. More details can be found on the data CDattached

Figure 3.3: Piezo Pz26

Electronics 9

3.2 Electronics

Each of the piezo-elements has its own preamplifier with the following characteris-tics [30]:

• The first step of amplification is done by a AD745 Bifet ultra low noiseamplifier mounted in a non-inverted configuration. The feedback capacitorand resistor defines the maximum gain and the input RC network limitsthe gain at low frequency. The amplifier acts as a band pass filter with amaximum gain of 220 where the low cut off frequency is 3.3 kHz and the highcut-off frequency is 330 kHz.

• The second step of amplification uses the AD797 low noise amplifier installedin a bandpass filter configuration with the same maximum gain and cut-offfrequencies as in step one.

• In the last step the signal is converted with AD8138 to be differential andits positive and negative output are sent through a differential twisted paircable. The output bandwidth of the differential amplifier is limited at 30 kHzin order to reduce high frequency noise.

Figure 3.4: The connection scheme of the amplifier


All amplification modules are supplied with +/-5V and are installed on a doublesided circuit with ground plane. In order to prevent external noise the input part isshielded and earthed with a copper foil as shown in Fig. 3.5a. Power consumptionof the whole system is around 2.5W. Figure 3.5 illustrates the top layer of theamplification board with components on both sides. In Fig. 3.6 the gain factor of90db can be seen in the working area (17-30kHz) of the whole amplifier.

(a) (b)

Figure 3.5: The upper layer of the amplifier with its technical scheme

Figure 3.6: Gain of the total amplification process as a function of frequency

4 Experimental setup

Water tankThe whole experimental setup is shown in Fig. 4.1a. The water tank has thedimensions 1.5m x 0.7m x 0.7m and can contain around 70l of water. The innerwalls of the tank are covered with 3.5cm thick Styrofoam in order to damp signalreflections1. The distance between the transmitter and the sensor is chosen to bemore than 80cm in order to simulate an acoustic plain wave reaching the sensor.Two fixation rails are attached on the top of the tank where a sensor module, anabsolute calibrated hydrophone and a transmitter are also connected.

(a) (b)

Figure 4.1: The tank and data acquisition hardware

The external generator and oscilloscope were used for comparison purposes withthe new self-modified acquisition board made by National Instruments (NI). Theazimuthal and polar motors can be additionally moved by the controller boardwith digital display simplifying the offset for the polar orientation. The acquisitionhardware in Fig. 4.2 shows the connection of the hardware and the signal recordingprocess.

1Based on this idea the Styrofoam could be replaced by porous foam as is done in absorbing soundrooms. Regrettably, the absorption and damping of ultrasound has not been investigatedfor water or ice yet. One of the advantages of such a study could lower costs of sensorcharacterizations as well as lower time and effort used for characterization of sensors in largeenvironments [31].


Figure 4.2: Data acquisition scheme

TransmitterThe transmitter is made of a piezo element in a form of a ring covered by epoxy.The characteristics of the azimuthal and polar radiation as well as the dependencyon the transmitted pulse shape have not been established yet. Their importanceand estimations are discussed in following chapters.

Figure 4.3: Used transmitter

Absolute calibrated HydrophoneIn order to define the absolute sensitivity of the sensor prototype, the absolutecalibrated hydrophone SENSOTECH SQ03, is used as a reference (Fig. 4.4). Thehydrophone works in the range of 1Hz to 65kHz and consists of a piezoceram-ics (black part in Fig. 4.4) equipped with an amplifier with a gaining factor of100 (40db). Regarding the description supplied by the manufacturer, the absolutesensitivity of the order of 7.94µ/Pa (−163.3± 0.3 dB re. 1V/Pa) is linear and fre-quency independent. Studies carried out on the SPATS and HADES sensors showa decrease in the absolute sensitivity to −167.5 ± 1 dB re. 1V/Pa correspondingto 38% less sensitivity after 3 years [19, 25].

Absolute calibration of the sensor 13

Figure 4.4: The commercial hydrophone SENSORTECH SQ03

Data acquisition programA data acquisition LabView program interfaced to NI board allows for data record-ing at a maximal sampling rate of 1.25MHz for one open channel. If more channelsare open, sampling rate decreases to 1MHz/number of channels. Furthermore, theprogram enables pulse generation and automatic relative orientation between thesensor module and the transmitter.

4.1 Absolute calibration of the sensor

The comparison method [19, 20, 25] is used, in order to determine the absolutesensitivity of the sensor. In contrast to the reciprocal method [35], this calibrationmethod is faster angular resolutionand easier2, but it requires an absolute pre-calibrated reference to compare.

Figure 4.5: orientation of the hardware

Figure 4.5 shows an experimental arrangement for the comparison method. Unfor-tunately, the large water tank was still too small and it was not possible to carryout the calibration in a bigger environment as the water-proof cables for the sensorwere not completed. Due to the negative influence of micro air bubbles attached tothe housing of the sensor, the sensitivity calibration took place after the sensor wasalready in the water for 24h [33]. Therefore, the primary signal is always pollutedwith reflections leading to the challenge of extracting the biggest part of primarysignal in the time and frequency domain. This is discussed in the next chapter.The sensitivity calibration is done with two different signal shapes:

• Neutrino pulse equals to a damped sinus: f(x) = sin(

2πa

)exp

(−x2

2b2

)2No voltage-pressure-transfer function of the transmitter is required [36].


• Gaussian pulse: g(x) = exp(−x2

2c2

)

Figure 4.6: Shapes of transmitted signals in Volt

Each pulse is first simulated through the signal analysis program AUTOSIGNAL1.73 using the FFT and Wavelet function in order to get optimal values (for thevariables a, b and c above) for the generator. In this way, simulated Gaussianand damped sine pulses are in the range from 13kHz to 85kHz. Additionally, thetheoretical expectations for the Gaussian pulses could be calculated with f = 1√

2πc

[13].The sampling rate for these 3 channels is 330kHz.

Figure 4.7: Simulation of the transmitted pressure signals

3I have used the 30 days trial version [37]

Setup for characterization of the polar orientation of the sensor 15

4.2 Setup for characterization of the polarorientation of the sensor

The experimental composition of the hardware does not differ much from the ab-solute calibration. The three connected channels are replaced by five, excludingthe commercial hydrophone, in order to achieve higher sampling rate (200kHz). Ahigh sampling rate per channel is essential in order to reach centimeter resolutionin aluminum or steel housing. This characterization could not be carried out withthe LabView program. A difference of more than 30 degrees between the manualcontrollers and these in the program is still available after one 360o turn. Thereforethe sensor is steered by the manual controllers in order to find the right angle. Thisway the sensor makes a 360o turn in 10 degrees steps. Due to its dominant peakonly a Gaussian pulse was transmitted for these purposes.

Figure 4.8: The experimental environment


5 Analysis in the time domain

5.1 Extracted primary signal

Due to the small experimental environment the response of the sensor is polluted byreflections. Assuming the law of total reflection for all reflection paths1, the time of

Figure 5.1: Almost all possible reflection paths in the tank

arrival for every shortest path is measured and calculated2(Fig. 5.1). The receivedtime window is not stationary and is a function of frequency of transmitted pulse.For further calculations, a fixed time window is chosen in order to simplify theanalysis. Programming a flexible time window was more difficult than anticipated.Figure 5.2 shows a total recorded signal with extracted primary signal in a timewindow of 0.15ms.

Figure 5.2: Total recorded signal with an extracted primary response

1This assumption is chosen with the goal of simplicity. The absorption coefficient has to beconsidered for correct calculation

2The propagation of sound in water was estimated at 1487± 1.5m/s from the collected data


5.2 Noise level analysis

The noise level is calculated by two different methods. The first local methodchooses the noise during the recording of the sensor response. The recorded noiseis exactly in time with when the signal is sent, but before the first response of thesensor.(See Fig.5.2 in previous section) Then the effective value of the noise levelis calculated with:

NoiseRMS =

√√√√ N∑i=0

xi2

N

The advantage of this method lies in the locale determination of the signal-to-noiseratio (S/N). It shows the possibility of a pulse detection during the local noise level.During the examination of the piezoceramics and the hydrophone, the local noisewas not constant and sometimes even oscillations could be noticed indicating someunsolved problems in the NI board or cables. With the external oscilloscope thesepollutions could not be seen.

The second method determines the global noise level. The noise was recordedfor few seconds in 24h. As it can be seen in Fig. 5.3 the noise level is distorted bypeaks with unknown origin. A possible explanation for this may be found in the

Figure 5.3: Noise signal polluted with spikes

appearance of micro bubble which produce a tri-polar pulse[33, 34]. After removalof spikes the noise level is calculated with the same method for every piezo elementand the hydrophone as above and only the global noise level is used for furthercalculations.

Piezo 1 Piezo 2 Piezo 3 Piezo 4 SQ03Noise level: 3.2mV 3.4mV 3.6mV 3.2mV 0.89mV

One should note the significant reduction in noise level, by a factor of 4, for theSQ03.

Calculation of the S/N ratio 19

5.3 Calculation of the S/N ratio

There are three different definitions of the signal-to-noise ratios3 (S/N ratio):

• The correct definition of the S/N ratio uses the effective values of noise andsignal. The effective values are calculated with the same formula for RMS asin the chapter before.

S/NRMS =signalRMS

noiseRMS

• The second definition uses the Peak-to-Peak difference to calculate S/N. Peak-to-Peak value in this case means the difference between the maximum andminimum values of the signal.

S/Npp =Differencepeak−to−peak

RMSnoise

• The Zero-to-Peak calculates the values between the average and a maximumvalue.

S/Nzp =Differencezero−to−peak

RMSnoise

As the two last definitions give higher values for S/N and they have been introducedby manufactures in order to improve the characteristics of their products, only thefirst definition is used for further analysis and also because of the different responsesbetween the hydrophone and the sensor [38]. Figure 5.4 presents the S/N ratio ofthe reference hydrophone as a function of frequency.

(a) Neutrino pulse (b) Gaussian pulse

Figure 5.4: S/N ratio for the reference hydrophone as a function of frequencyand shape of the transmitted signal

It is also worth noting the increasing S/N ratio in the range up to 40kHz. After thispoint, the values seem to be independent of frequency, but this assumption has to be

3The theoretical definition is S/N = Psignal

Pnoise=(

Asignal

Anoise

)2

(corresponding to S/N(dB) =

20 log10

(Asignal

Anoise

))


handled with care due to the unknown radiation characteristics of the transmitterat this time. Regarding another analysis for the same hydrophone SENSORTECHSQ03 [19, 20] the S/N ratio there is almost independent of frequency. This wouldindicate a better response of the transmitter to the signal sent in the range above40kHz. In general, it is important to know the radiation characteristics of thetransmitter in order to correctly determine the S/N ratio.


Figure 5.5: S/N ratio for the 4-channel sensor as a function of frequency andshape of the transmitted signal

As illustrated in Fig. 5.5, the S/N ratio for the piezoceramics has its best values inthe range 17-30kHz which corresponds to the working area of the amplifier cuttingfrequencies above and below this range. The figures also show also the importanceof the state of the surface coupling piezo elements to the housing. The S/N ratiofor a polished surface (piezo elements 1 & 3) has up to 2 times better resultsthan for a porous one (piezo elements 2 & 4). Additionally, the shape of sentsignals plays a role during the determination. The response to a neutrino pulseis up to 3 times better than to a Gaussian one by the same transmitted voltageamplitude. This could indicate either a dependency of the piezoceramics used intransmitter for different signal shapes or a dependency of the piezoceramics used insensor for different pressure shapes. A composition of both reasons is also possible.Nevertheless, this fact shows the importance that the sensor should be calibratedthrough a theoretically estimated neutrino pulse4.

5.4 Absolute sensitivity

The sensitivity is calculated with the following formula:

Ssensor =Msensor

MSQ03

· SSQ03

where Spiezo indicates the absolute sensitivity of the sensor (its 4 piezo elements) andSSQ03 the absolute sensitivity of the reference hydrophone (SSQ03 = 7.94µV/mPa).

4This corresponds to a bipolar pressure signal which could be described by the second derivativeof the error function

Absolute sensitivity 21

For Msensor and MSQ03 the relation between the S/N ratios of the reference hy-drophone and sensor is chosen due to the experimental uncertainties with thehardware. Instead the S/N ratios the effective values and Peak-to-Peak differ-ences could be compared as well. Figure 5.6 illustrates the received sensitivity of


Figure 5.6: Absolute sensitivity of the 4-channel sensor as a function of frequencyand shape of the transmitted signal

our sensor. In the working area of the amplifier the sensitivity can be estimated to:0.2mV/mPa for the neutrino pulse and 0.1mV/mPa for the Gaussian pulse. Withan average noise level of 3.2mV a neutrino producing a pressure pulse of 10mPacould be already detected. Of course, this assumption neglects the backgroundnoise in the south pole, but it demonstrates the potential of the sensor in its cur-rent form and that it is in developing state. As presented in Fig. 5.7 the spectrumof transmitted pressure pulses varies also with shape.


Figure 5.7: Pressure spectrum of the reference hydrophone as a function of fre-quency and shape of the transmitted signal


The final plots (Fig. 5.8 & Fig. 5.9) show the pressure sensitivity as a functionof frequency which are normalized to S/N = 1 ratio.

Figure 5.8: Pressure sensitivity as a function of frequency of transmitted Neu-trino pulses normalized to S/N = 1 ratio

Figure 5.9: Pressure sensitivity as a function of frequency of transmitted Gaus-sian pulse normalized to S/N = 1 ratio

These results in the time domain already show good estimations for the absolutesensitivity of the 4-channel sensor, but in order to determine the absolute sensitiv-ity, a calibration in a large environment (Geneva Lake) is required as well as theanalysis of the data in the frequency domain.

6 Analysis in the frequency domain

Due to a time constraint caused by a problem with the data acquisition hardware,the analysis in the frequency domain is carried out with the program AUTOSIG-NAL 1.7. The program has a large data base of functions analyzing signals. Allthe following results are based on the Fourier analysis with the program. Thisanalysis is done only in the working area of the amplifier (transmitted pulses arein the range of 17-30kHz). Beyond this area the amplifier sharply cuts all otherfrequencies. The main aim of this section is to show the possible analysis of anextracted signal in the frequency domain.

6.1 Data tapering windows

In the time domain, the primary signal is analyzed in a time window with a begin-ning and an end. In the frequency domain, which is based on a Fast Fourier trans-formation (FFT), both these limits are not mathematically defined 1. Therefore theextracted signal is convoluted with a tampering window in order to correctly calcu-late the FFT of the signal. In our case the data tampering window requires a widemaximum so that the signal is not too highly convoluted [39, 40, 41]. Figure 6.1

Figure 6.1: Different data tapering windows

shows two examples of different data tapering windows: the blue line correspondsto the extracted window of our signal and its FFT, the green line is a tampering

1In practice these two limits produce high frequencies. This effect is called the Leck-Effect [39].


window in the form of a Gaussian function and the purple data tampering windowis a tampered cosine with a roll-off-factor of 0.2. The roll-off-factor regulates thebandwidth of the filter [40]) Due to its desired shape, the tampered cosine is chosenfor later calculations.

6.2 Absolute sensitivity

The Fourier transformation of the 4-channel sensor and hydrophone is supple-mented by a Fourier analysis of the local noise (see previous section) in order toclean the signal. Then the absolute sensitivity is calculated by dividing the mag-nitude spectra of the four piezo elements and the reference hydrophone.

Due to the low sampling rate the absolute sensitivity could not be determinedfor all piezo elements of the prototype sensor except for piezo element 3. Theirresults were inadequate and not reproducible because of insufficient statistics. Thenumber of data is reduced by a factor of two caused by the FFT. The followingresults are based on the analysis of the piezo 3 whose results could confirm theresults in the time domain.

The plots below demonstrate the FFT of an 25kHz transmitted neutrino pulseafter the noise subtraction. It is remarkable that only few points cover the rangeof the transmitted signal considering the spectrum of the piezo elements and thehydrophone. The calculation of the absolute sensitivity can only confirm the rangeof the results in the time domain.

(a) (b)

Figure 6.2: FFT of the extracted primary signal for the piezo element no.3 andthe reference hydrophone

Frequency reconstruction 25

(a) (b)

Figure 6.3: Absolute sensitivity in the frequency domain (left) and the detectionthreshold (right) for piezo element 3

6.3 Frequency reconstruction

During the analysis of more than 60 different data tampering functions offered byAUTOSIGNAL 1.7 It was noticed that the use of such a filter has to be donecarefully. The tampering windows do affect the maximum height of the magnitudeof our signal and the maximum value for the peak frequency. As these factors arevery important for neutrino characterization in ice, this would be one of the reasonsagainst the use of such filters. Nevertheless, this section shows a possible frequencyreconstruction of transmitted pulses. Figure 6.4 shows a strong linear correlationbetween the transmitted pulses and the responses of the piezos2 in the working areaof the amplifier. A part of the correlation is unworthy the presentation because ofthe large influence of the amplifier. This strong correlation indicates a step in theright direction for neutrino energy spectrum reconstructions.

Figure 6.4: Reconstructed frequencies

2Piezo 2 was excluded due to its inadequate response


7 Analysis of the polar orientationof the sensor

After recording the data each signal is analyzed with the program AUTOSIGNALDue to the low sampling rate the main maximum is only determined but 4 pointsproducing the difficulty of finding the time value for this maximum (See section3). Using the interpolation method of AUTOSIGNAL, this maximum is estimatedand its time and angle value is registered.

Figure 7.1: Recorded signals of the 4-channel sensor at 0o

Figure 7.2 demonstrates the received result for one piezo element. Noteworthyis a good angular response from 90o to 270o with its peak value at 180o. The restof the range has only a flat response due to the dominance of reflections.

28 Analysis of the polar orientation of the sensor

Figure 7.2: Time of first signal maximum as a function of the polar orientationof the piezo element 3

Figure 7.3 shows the arrival time of first signal maximum for all piezo elements inthe sensor and confirms the polar orientation of the piezoceramics in the housing ofthe sensor. Each of the piezo elements has a good coupling in the time range 0.62-0.655ms. Apart from this range, the response of the piezoceramics saturates dueto their orientation towards the transmitter and the dominance of the reflectionsin this areas.

Figure 7.3: Time of first signal maximum as a function of the polar orientationof the sensor

Still unknown remains the propagation of the signal in the housing of the sensor.Further studies of the expected interferences in the housing should be done inorder to achieve deeper knowledge about the response of piezoelements to thesignal propagated in a metal housing. A possible method is shown in the previouschapter.

Measurement of sound propagation in the sensor under water 29

7.1 Measurement of sound propagation in the sensorunder water

The measurement is based on a vacuum-insulated container opened only from oneside. In order to use this instrument to different forms of housing of sensor, thisside ends with an elastic ring allowing for an adoption to different forms. Insidethe glass container, a piezo element can be found attached to an adjusting screwoptimizing the coupling of the piezoceramics to the housing.

Figure 7.4: Prototype of an instrument for a measurement of sound propagationin the housing of the sensor under water

The instrument would measure the arrival time of the signal from the housingof the sensor after the sensor is hit by an acoustic signal propagating in water. Inorder to complete this measurement, the responses of the piezoceramics used insidethe sensor should be recorded in order to trigger the time of the piezoceramics usedin the vacuum-insulated container. Replacing this instrument on the sensor wouldgive a better insight in characteristics of sound propagation in the housing of ansensor.

30 Analysis of the polar orientation of the sensor

8 Error discussion

Statistical errorsAll calculated errors are statistical. For the S/N-ratios the statistical error rangevaries from 10% up to 25% for the polished surface. The piezo elements coupledwith a porous surface show by a factor 3 smaller statistical errors. This fact indi-cates a higher certainty of their results compared to the piezoceramics coupled withthe polished surface with higher S/N-ratios but less certainties in these results.

For the calculations of the absolute sensitivity the statistical errors are evenhigher after a Gaussian propagation of errors. The dominant errors still present inthe final graphs are neglected in order to simplify the reading of the results. Dueto low statistical values in the frequency domain the uncertainties in the frequencyspectrum, especially the distribution of the magnitudes of the Fourier-coefficients,could not be calculated.

Systematical errorsThe systematical errors are not considered in all the graphs. Due to the lack oftime and complexity of the errors only the possible sources for these errors arementioned. One of the biggest systematical errors can be found in the choice of thelength of the window for the time and frequency domain. The systematical errormade by a variation of this length can be well seen in the frequency domain but itsextraction was not possible at this time. Aside from the possible unknown errorsin data acquisition, another source of errors can be found in the orientation of thehydrophone and 4-channel senor towards the transmitter which was not alwaysoptimal.

In spite of an established Styrofoam base under the tank the seismic vibrationscould not be excluded. Even clapping and stepping in the surroundings of theexperiment produce visible signals, which could affect the experimental results.

Angular resolutionDue to the frictional effects, a global error of ±3o is assumed for all angles. Thetime error has a statistical nature and is calculated from events recorded, severaltimes.

32 Error discussion

9 Hardware and software during theexperiments

The experimental hardware and software were used for the first time in this instal-lation. Therefore my experiments were accompanied with a lot of bugs and errorsin the data acquisition process. This section presents only problems which are stillunsolved, together with recommendations for faster data acquisition and analysis.

9.1 Detected errors

Error during the data acquisitionOne of the biggest errors which costs me more than 150 hours of data acquisitionand analysis could be found in the modified NI data taking board. The problemwas found in the connection between the coaxial cables and the NI board.

Figure 9.1: Influence of one channel receiving a signal to remained channels

The channels have influenced each other during data acquisition (Fig. 9.1). Inorder to find the influence characteristics the response of one piezo element wassent only through one channel, but with all channels in the program activated.Figure shows that the input of the hardware did not correspond to the input in theprogram. Dome channels, such as CH8 and CH9, did not correspond to them inthe program and all of the channels were influenced by each other. The mix of thechannels could be fixed directly in the program, but only channel 6 could not stillbe activated. The problem with this influence was solved by closing every channelnot in use with a resistance of 50Ω because the influence was caused by lowerimpedance of the air as compared to the impedance of the hardware. Additionallythe noise of every channel could be reduced. Figures 9.3a and 9.3b demonstratethe improved connection between the hardware and software.

34 Hardware and software during the experiments

Figure 9.2: Connection scheme between the input channels of the hardware andsoftware

(a) Improved channel response (b) Improved channel coupling

Figure 9.3: Improved channel response and coupling

Earthing of the hydrophoneAfter having fixed the error above, I have noticed that the offset of the referencehydrophone varied with time. After having activated the hydrophone for a durationof 48h, there was a difference of about 4 Volt. I have realized that during the datarecord this offset was decreasing, but apart from that it was increasing. Afterconsulting with the reference oscilloscope showing an offset of nearly 0 Volt, I havereasoned that the commercial hydrophone is not earthed enough. Unfortunately Icould not find the earthing problem in the hardware and so I was forced to earth thehydrophone and check it with the oscilloscope before each data recording. Figuresbelow show the earthing problem of the hydrophone during the noise recording.

Earthing of the prototype sensorNot only the hydrophone, but also the 4-channel sensor showed some problems withthe earthing. After 2 months in the tank, degradation of the aluminum housingcould be noticed. Figure 9.5 presents the damaged surface of the housing due toan electrolysis caused by inadequate earthing. As these damaged areas providepossibilities for creation of bubbles, which in turn affect quality of the response ofpiezoceramics, this problem should be considered and solved for the new prototype.

Detected errors 35

(a) (b)

Figure 9.4: Decreasing offset of the hydrophone

Figure 9.5: Damaged surface of the housing

Azimuthal and horizontal motorsThe motors used for the triangulation purposes are not strong enough to hold thehousing of the sensor. Therefore only the horizontal motor could be used, but ithas also problems with turning caused by the sensor weight and friction in themechanics.

36 Hardware and software during the experiments

9.2 Recommendations for later data acquisition

Characterization of the transmitterThe radiation characteristics and stability of the acoustic sound generation shouldbe determined for the transmitter in order to guarantee the reproduction of results.This would also allow extraction of a transfer-function[36] between the input signaland the output pressure and it would enable further verification of the sensitivity ofthe 4-channel sensor, as well as produce better results for frequency reconstructionin large environment.

Acquisition softwareThe recorded data could be more applicable for analysis with ROOT or anothersoftware. The convertion of the data from self-made ’.waf’-file to a ’.txt’-file took alot of time and it was very susceptible to human errors. It was very easy to mergethe wrong header file with some other ’.waf’-files. Besides, it was tedious to modifyevery .txt-data file in order to make it applicable for ROOT. Each comment of thefile header should begin with a # and instead of a .txt file the recorded data couldbe directly converted into a .dat-file, for instance. Furthermore the program AU-TOSIGNAL 1.7 enables the reading of binary data. In general this data processingcould be done in a more automatic manner saving time and decreasing possibilitiesfor errors.

The embedded generator could be improved by enlarging the data bank of pos-sible mathematical functions (e.g. error function). Additionally the possibilityof writing a small C-code, especially a for-loop, could expand the possibilities ofcalibration methods using for example a sine burst method, simplifying the dataacquisition.

Furthermore, the measured angle and the steering of the motors by the programnever corresponded to real values. As the angle value was always hidden in the fileheader and could not be simply extracted, I was forced to do the polar orientationmanually.

Finally, a higher sampling rate could supplement the improvement of the soft-ware and hardware.

AUTOSIGNAL 1.7The program AUTOSIGNAL 1.7 was a very powerful tool during signal analysis.Due to its very large data bank in signal processing it helped to analyze faster thequality of recorded signals. It allowed to make decision faster improving the dataacquisition. Despite the high acquisition costs (650 Euros) and the restriction ofonly Windows as operating platform, I would recommend the 30-days trial versionof this program that I have used for my analysis.

10 Summary and outlook

In the meantime, a second 4-channel sensor prototype is built of steel. It will beequipped with a new electronics improving the amplification factor by 10db withremaining noise level. A digitization board and a gyroscope will be implementedinto the sensor in order to exclude signal attenuation along transmission cables withlengths of a few kilometers. The new mobile data acquisition system will allow fortesting the device at long distances in large environment. A wide sequence of testsshowing the functionality of the 4-channel sensor at high temperatures and highpressures is planned.

The development of acoustic neutrino detection methods should be continued inorder to scale the sensor sensitivity at sub-mPa levels. The characterization of thenoise in open media such as water and ice is indispensable, especially at the SouthPole, which could complicate the applicability of the technique.

In comparison with the already established optical detection method, the acous-tic method is less expensive. With the same costs a much larger volume could beinstrumented with a resulting higher probability of neutrino detection. Comparingthe 4-channel sensor with the sensors from SPATZ and HADES experiments, theprototype sensor can already achieve lower levels of sensitivities of its counterparts[18, 32, 20].

Considering my objectives at the beginning of my work, I am unsatisfied with myreceived results for the absolute sensitivities. During the project I realized that myobjective changed to the practical applicability of the new hardware and softwarein order to characterize a prototype sensor. I wished to concentrate only on the de-termination of the absolute sensitivity in the frequency domain, but unfortunatelyit was not possible due to small experimental environment as well as unexpectedproblems with software and hardware. Regarding this new goal of this project Ifulfilled my expectations.

38 Summary and outlook

Bibliography

[1] V. F. Hess, Uber Beobachtungen der durchdringenden Strahlung bei siebenFreiballonfahrten, Physik. Zeitschriften XIII (1912) 1084-1091

[2] F. Reines et al., The Neutrino, Nature 178 (1956) 446-449

[3] Y. Fukuda et al., Measurements of the solar neutrino flux from Super-Kamiokande’s first 300 days, Phys. Rev. Lett. 81, 1998

[4] R. M. Bionta et al. Observation of a neutrino burst in coincidence withsupernova 1987A in the Large Magellanic Cloud, Phys. Rev. Lett, 58:1494,1987

[5] G. Sigl, Ultra high energy neutrino fluxes and their constraints, Phys. Rev.D, 66:063004, 2002

[6] S. Cohen, C. Terranova, L. Demirors, M. Ribordy, IceCube, a neutrinotelescope in the Antarctic ice, 2008

[7] R. Abbasi et al., The IceCube Data Acquisition System: Signal Capture,Digitization, and Timestamping, Nucl. Instr. Meth.A601:294-316,200

[8] R. Engel, D. Seckel & T. Stanev, Neutrinos from propagation of ultra-highenergy protons, Phys. Rev. D 64 (2001) 093010.

[9] L. Demirors f. t. Icecube Collaboration, Searches for a Diffuse Flux ofExtraterrestrial Muon Neutrinos with the IceCube Observatory, Int. J. Mod.Phys. D 2009

[10] Raj Ghandi et al., Neutrino interactions at ultrahigh energies, Phys. Rev.D 58(9):093009, September 1998

[11] G. Askaryan, Sov. J. Atom. Energy 3, 921 1957

[12] G. Askaryan, Acoustic detection of high energy particle showers in water,Nucl. Inst. and Meth., 164:267 278, 1979

[13] J. Stegmaier, Optimierung akustischer Detektoren zum Nachweis hochener-getischer kosmischer Teilchen, Diploma Thesis, University of Mainz, May2004

[14] E. Zas, F. Halzen & T. Stanev, Phys. Rev. D45:362, 1992

[15] B. Price and L. Bergstroem, Optical properties of pure ice at the SouthPole: scattering Appl. Opt. 36 (1997) 4181;

40 Bibliography

[16] L. Bergstroem, B. Price et al., Optical properties of pure ice at the SouthPole: absorption Appl. Opt. 36, 4168 (1997);

[17] P. B. Price et al. Proc. Natl. Acad. Sci. USA, 99 (12):7844, 2002;

[18] S. Boeser et al., Feasibility of acoustic neutrino detection in ice: Designand performance of the South Pole Acoustic Test Setup arXiv:0807.4676

[19] S. Boeser, Acoustic detection of ultra-high energy cascades in ice, PhD The-sis, University of Berlin, October 2006

[20] B. Semburg, Untersuchung von akustischen Sensoren zum Neutrino-Nachweis, Diploma Thesis, University of Wuppertal, May 2007

[21] B. Price, Attenuation of acoustic waves in glacial ice and salt domes, J.Geophys. Res.111, B02201, 2006

[22] F. Descamps f. t. Icecube Collaboration, Measurement of sound speedversus depth in Antarctic ice with the South Pole Acoustic Test Setup,arXiv:0811.1797

[23] J. Vandenbroucke et al.,Measurement of acoustic properties of South Poleice for neutrino astronomy, arXiv:0811.1087

[24] M. Podgorski & M. Ribordy, Acoustic sensor development for ultra highenergy neutrino detection, Proceedings of the 31st ICRC, Lodz 2009

[25] J. Fischer, Acoustic transducers for the South Pole Acoustic Setup, DiplomaThesis, University of Berlin, July 2006

[26] J. G. Learned, Acoustic radiation by chargedatomic particles in liquids,Phys. Rev. D1, 19:3293 3307, 1979

[27] L. G. Dedenko, Phys. Part. Nucl., 29:266, 1998

[28] M. Ribordy, Status report on acoustic neutrino detection, CHIPP Neutrinoworkshop, November 2008

[29] FERROPERM Piezoceramics URL: www.ferroperm-piezo.com

[30] R. Frei, Private Communications, 2009

[31] M. Bissok, Private Communications, 2009

[32] S. Boeser, R. Heller, R. Nahnhauer, M. Pohl, K. Sulanke and A. Hallgren,Development Of Glaciophones And Acoustic Transmitters For Ice, Int. J.Mod. Phys. A 21S1 (2006) 107.

[33] K. Graf, Teststrahlmessungen zur akustischen Neutrinodetektion: Ther-moakustische Schallerzeugung in Wasser, Diploma Thesis, University ofErlangen-Nurnberg, July 2004

Bibliography 41

[34] K. Graf, Experimental Studies within ANTARES towards Acoustic Detec-tion of Ultra-High Energy Neutrinos in the Deep-Sea, PhD Thesis, Univer-sity of Erlangen-Nurnberg, May 2008

[35] T. Karg, Detection of ultra high energy neutrinos with an underwater verylarge volume array of acoustic sensors: A simulation study, PhD Thesis,University of Erlangen-Nurnberg, April 2008

[36] O. Veledar, Hydrophone based calibrator for seawater acoustic detection ofUHE neutrinos, Proceedings of the International Workshop ARENA 2008

[37] AUTOSIGNAL 1.7 30-day trial versionURL: http://www.statcon.de/URL: http://www.clecom.co.uk/science/autosignal/

[38] V. Wesson, TN200410A - Dynamic Range, October 2004URL: http://www.seis.com.au/TechNotes/TN200410A_SNR.html

[39] J. B. Anderson, Digital Transmission Engineering, Wiley Interscience,Lund 2005

[40] A. H. Nuttall, Some Windows with Very Good Sidelobe Behavior, IEEETrans. ASSP, v29-1, Feb. 1981.

[41] National Instruments, Characteristics of Different Smoothing WindowsURL: http://zone.ni.com/reference/en-XX/help/371361B-01/lvanlsconcepts/char_smoothing_windows/

[42] K. Salomon, Simulation und Messung verschiedener Hydrophonenkom-ponenten zur akustischen Teilchendetektion, PhD Thesis, University ofErlangen-Nurnberg, Januar 2007

[43] D. Besson et al., Simulation of a hybrid optical/ radio/ acoustic extensionto IceCube for EeV neutrino detection, Int. J. Mod. Phys. A21S1, 2006

[44] R. Nahnhauer & S. Boeser, Acoustic and Radio EeV Neutrino DetectionActivities, Proceedings of the International Workshop ARENA 2005

[45] R. Brune Root, on abject-oriented data analysis framework. 2008URL: http://root.cern.ch

[46] Ch. Vogt, Ch. Wiebusch, K. Laihem, Status of the Aachen Acoustic LabHeapnet meeting, Amsterdam 2007

[47] S. Kuch, Aufbau eines Hydrophone-Teststands und vorbereitende Studienzur akustischen Detektion ultrahoch energetischer Neutrinos, Diploma The-sis, University of Erlangen-Nurnberg, October 2003

[48] P. Kollmannsberger, Studien zur akustischen Detektion ultrahochenergetis-cher Neutrinos mit Hilfe des piezoelektrischen Effekts, Diploma Thesis, Uni-versity of Erlangen-Nurnberg, Februar 2004

42 Bibliography

[49] Ch. Naumann, Development of Sensors for the Acoustic Detection of Ul-tra High Energy Neutrinos in the Deep Sea, PhD Thesis, University ofErlangen-Nurnberg, September 2007

A Acknowledgement

I would like to thank Prof. Christopher Wiebusch who gave me the chance to writemy Bachelor thesis abroad. It was an unforgettable experience to work at EPFLin an international team.

In addition, I would also like to thank Prof. Matthieu Ribordy who gave me thesubject for my project and was always available for my questions during my stay.

The help of Celine Terranova, Shirit Cohen and Levent Demirors, who werealways there for my theoretical and programming questions, was unforgettable. Ihave enjoyed the time with you and I would like to thank you as well.Furthermore, I really appreciate the assistance of Raymond Frei and Laurent Braunwho helped me in my error research and improvement.

I would like to thank all of my friends, but especially Amisha Patel and NoraTandberg working with me in the Cubotron and supplying me with cookies and teaduring breaks. I thank Patryk Sawicki who encouraged and supported me duringmy project.

Finally, I would like to thank my family for encouragement and financial support.

44 Acknowledgement

B Statement of independence

I declare that I have developed and written the enclosed Bachelor’s thesis entitled”Characterization of an acoustic sensor device” entirely by myself and have notused sources or means without declaration in the text. Any thoughts or quotationswhich were referred from these sources are clearly marked as such.

Lausanne, 9th June 2009

46 Statement of independence

47

C Appendix

C.1 The proceedings of the 31st ICRC, Lodz 2009

PROCEEDINGS OF THE 31st ICRC, ŁODZ 2009 1

Acoustic sensor development for ultra high energy neutrinodetection

Matt Podgorski† and Mathieu Ribordy∗

∗ High Energy Physics Laboratory, EPFL, CH - 1015 Switzerland†RWTH Aachen, visiting EPFL

Abstract. The GZK neutrino flux characterizationwould give insights into cosmological source evo-lution, source spectra and composition at injectionfrom the partial recovery of the degraded informa-tion carried by the ultra high energy cosmic rays. Theflux is expected to be at levels necessitating a muchlarger instrumented volume (>100 km3) than thosecurrently operating. First suggested by Askaryan,both radio and acoustic detection techniques couldrender this quest possible thanks to longer waveattenuation lengths (predicted to exceed a kilometer)allowing for a much sparser instrumentation com-pared to optical detection technique.We present the current acoustic R&D activities atour lab developing adapted devices, report on theobtained sensitivies and triangulation capabilities weobtained, and define some of the requirements forthe construction of a full scale detector.

Keywords: Ultra high energy neutrinos. Acousticdetection techniques. Acoustic sensor studies.

I. I NTRODUCTION

The IceCube detector [1] may well soon identify thefirst ultra high energy neutrino of cosmogenic origin,following interactions of ultra high energy cosmic rayswith the cosmic microwave background [2]. Predic-tions for the cosmogenic neutrino flux, i.e. neutrinosfrom photo-disintegration, is at levels of the order ofEdN/dE ∼ 10−17 s−1 cm−3sr−1 at E = 1018 eV,resulting in 0.01 - 1 event / year / km3 in ice [3]. Thesepredictions strongly depend on the primary cosmic raycomposition [4]. Currently, the situation is uncertain:While the observed correlation of UHE CR sources withthe AGN distribution by AUGER [5] hints toward a lightcomposition (and in this case we lie close to the upperflux predictions), dedicated AUGER composition studiesfavor a composition turning heavier at UHE [6]. GZKneutrinos are astronomical messengers keeping trackof the original CR direction, GZK interactions mostlyoccur close to the source. In case of the existence of afew UHE cosmic accelerators located close-by (Gpc),the detection of a substantial flux of GZK neutrinosfrom these directions using a multi-messenger approachwould allow the possibility of pinpointing the nature ofthese CR accelerators.

The characterization of the GZK neutrino flux spec-trum, and thus the recovery of the degraded informa-tion information carried by UHE CR, would allow the

delineation of cosmological source evolution scenariosfrom source spectrum characteristics. To fulfill this goal,the event detection rate should be vastly increased.Therefore a much larger volume should be instrumentedwith an adequate technology for the detection of ultrahigh energy neutrino interactions. Two novel detectionmethods have been proposed, following signatures firstdiscussed by Askaryan [7], [8]. An interacting neutrinoemits a coherent Cherenkov pulse in the range of 0.1-1 GHz [9] close to the shower axis and a thin ther-moacoustic pancake normal to the shower axis. Bothdetection techniques are currently exploited by severaldetectors. In ice, both radio and acoustic emissions haverather large theoretical attenuation lengths [10]. Whilethis has been convincingly demonstrated for the radioemission, it is still a work in progress for the acousticemission and is one of the main goals for the SouthPole Acoustic Test Setup (SPATS) [11]. With the datacollected by the SPATS array, the sound speeds w.r.t. thedepth have been determined with great accuracy, meetingtheoretical expectations [12] and S-waves have beenfound as well. Unknown, however, remains the exactnature of the local source of noise and the exact valueof the attenuation length. Newest experimental resultshint toward a reduced pressure wave attenuation lengthon one hand and demonstrate favorable noise level below10 mPa on the other hand [13].

Contrary to salt and water, ice is unique. It allowsthe detection of three distinct signatures accompanyinga neutrino event: Optical Cherenkov light, coherent radioCherenkov and thermoacoustic emissions, thus firmlyestablishing the event origin by a strong background re-duction. A possible layout for the hybrid instrumentationof a large volume of order of 100 km3 at the SouthPole would consist of strings deployed one kilometerapart down to a depth of 2 km (radio and acousticattenuation lengths strongly vary with temperature andare decreasing with depth). Given the topologies forthe radio & acoustic emissions, a string should bedensely equipped with radio and acoustic devices withan option of supplementing it with PMT devices foroptical detection.10−103 interacting GZK neutrinos in100 km3 instrumented volume can be expected after 10years. The cosmogenic spectrum could be characterized(and consequently insights into the underlying physics),provided a high detection efficiency, a deep knowledgeof the local source of noise and good energy resolution.Thermoacoustic models and Monte Carlo simulations

2 MATHIEU RIBORDY et al. ACOUSTIC SENSOR R&D

predict that a signal from a neutrino with an energyE = 1018 eV will typically have an amplitude of 10mPa at a distance of one kilometer [14] (to be rescaledfor finite attenuation length). To keep a good S/N ratio,the sensitivity of the devices should be at the sub mPalevel. Also, a good pointing resolution may serve thepurpose of UHE point source search. Given the giantarray layout introduced above, an acoustic signal willbe recorded by a small number of acoustic devices. Itis therefore desirable to design acoustic sensor deviceswith pointing capabilities of their own.

In the next section, we present R&D activities whichare taking place at our lab in regard of sensor designand construction and discuss its performances.

II. R&D ACTIVITIES

The design and construction of a multi-channel sensorwas conducted at our lab, which use piezo transduc-ers (PZT) as sensitive elements. A noise level levelS/N < 5 mPa (S/N ≡ SRMS/NRMS) and a goodangular resolution were demonstrated, suggesting thepossibility for excellent vertex localization combiningthe responses from all sensor hits. The design, whichmust still be improved to meet our design goals, couldeventually allow for diffuse acoustic noise reductionthrough spectral shape analysis and accurate energyestimate of physical events.

The setup for conducting the R&D activities consistsof a bath, topped with a support structure for oneabsolutely calibrated hydrophone (Sensortech SQ03),one homemade sensor and one transmitter. A datatakingLabView program interfaced to a National Instrumentboard is used for analog response digitization (12 bits× 12 channels, total 1.25 MHz), transmitter pulse gener-ation and4π relative orientation between the transmitterand the sensors for automatized sensor profiling.

The experimental setup shown in Fig. 1 consists of thehomemade hydrophone, a transmitter and the referencehydrophone. In the following, two different electric sig-nal shapes have been considered: a damped sin pulse anda gaussian pulse, resulting in a tripolar pressure pulse(the neutrino-induced thermoacoustic pulse is bipolar).

SQ03 hydrophone

homemade sensor

large water tank

transmitter

Fig. 1. Experimental setup.

A. Acoustic sensor design

The sensor consists of an aluminium pressure vesselhousing 4 channels to provide triangulation capabilities.The noise level at the amplifier input to 130 nV, reached

frequency [Hz]10000 20000 30000 40000 50000 60000 70000 80000 90000

pres

sure

[mP

a; S

/N=1

]

1

10

210

piezo 1piezo 2piezo 3piezo 4RMS piezo noise: 3.2mV

V/mPaµsensitivity SQ03: 7.94

Fig. 2. Pressure sensitivity as a function of frequency of damped sintransmitted pulses normalized toS/N = 1 ratio.

at the expense of some bandwidth reduction, peaking at22 kHz with∼90 dB amplification and sharply decreas-ing below 10 kHz and above 40 kHz. Whether that isoptimal has to be studied further. It is manufacturableat relatively low cost. While aluminium is an adequatemedium for use in a liquid water bath, it will be replacedby steel for application in ice (more adequate given bothimpedance and resistance to pressure).

B. Sensitivity calibration

Signals with peak frequencies in the range 10 - 90 kHzwere recorded with a sampling rate of 330 kHz. A strongfrequency correlation between the transmitted pulse andthe sensor response was observed. Due to the finite sizeof the bath tub and given the sound speed velocity inwater, only the first 150µs following the pulse arrivaltime were analysed in what follows to avoid reflexionartefacts.

With the collected data from the 4-channel sensor andfrom the commercial hydrophone, the absolute pressuresensitivity was calculated in the time domain using RMSvalues for signal and noise. Fig. 2 shows the absolutepressure sensitivity (defined asS/N = 1) w.r.t. thedominant frequency of the sent signals.

The measurements demonstrate the importance of thestate of surface coupling the PZT to the housing: Thepolished surfaces for piezos 1 and 3 show a response∼2 times stronger than piezos 2 and 4 coupled to thehousing through porous surfaces.

C. Triangulation

Time resolution is essential for triangulation andtherefore a digitization frequency of 100/200 kHzis required in order to reach cm resolution in alu-minium/steel. This suggests that a sensor design shouldinclude digital electronics with at least 200 kHz sam-pling rate per channel1, in order to reach 0.5-5 ms

1100 kHz (and therefore 200 kHz sampling rate) is by coincidencethe value above which the ice attenuation length drops quickly androughly the extension of the neutrino-induced thermoacoustic pulsespectrum.

PROCEEDINGS OF THE 31st ICRC, ŁODZ 2009 3

pulse start time resolution (depending on amplitude)which roughly corresponds to 2-20 (4π/102 - 4π/104)angular resolution with the current multi-channel sensor.The design of a new digital (0.2 MHz/channel) 4-channel amplifier board has been started, with long rangecommunication protocols. It does not yet include triggerlogic. Digitization is necessary for a viable acousticdetector design in order to avoid losses in km long cables(of order of 3 dB/100 m in high quality cables) and thuskeep both good sensitivity and time resolution. Once inoperation, this will allow to define the requirements forfuture efficient trigger concept at the sensor level.

degrees−200 −150 −100 −50 0 50 100 150 200

time

[ms]

0.62

0.63

0.64

0.65

0.66

0.67

−310×piezo 1piezo 2piezo 3piezo 4

Graph

Fig. 3. Time of first signal maximum as a function of the polarorientation of the sensor.

Figure 3 demonstrates triangulation capabilities. Cou-pling between channels was found to provide in anysampled sensor positions 2 channels with signal withina factor 2 of the channel with the highest response. Theresolution will nevertheless depend on the individualsignal-to-noise ratios, but it shows potential for vertexreconstruction with a single sensor module.

D. Outlook

First positive results in the time domain were ob-tained. Absolute sensitivities are currently analyzed inthe frequency domain. A second 4-channel sensor ofsimilar design in a steel housing will be soon equippedwith digital electronics. Further sensor tests are foreseento happen next. Low temperature behavior test will beconducted in the laboratory and at large distances anddepths in the lake Geneva (∼ 400 m depth) to avoidreflexions, assess the acoustic noise characteristics andprobe the sensor design.

III. C ONCLUSIONS

Acoustic neutrino detection techniques should befurther developed, pushing the sensitivity at sub mPalevels, together with the characterization of noise sourceswhich may impede the applicability of the technique.Noise measurement in an open media (water / ice) arerequired to characterize the noise rate and its spectralshape in order to investigate improved trigger schemesrelying on signal processing within the sensors. Thesedevelopments have been started with digitization boardin the sensor, a necessary step for a viable acoustic arraydesign, where signal attenuation along km transmissioncable is excluded.

While it seems clear that sub mPa sensors shouldbe designed, the uncertain detection conditions at theSouth Pole make predictions concerning the detectionefficiency difficult. The potential can be dangerouslyspoiled in the case the local source of acoustic noisemimick a neutrino event. Further dedicated studies areon-going to ensure that it could be possible to distinguishthe event origin with high efficiency. The deployementof an acoustico-radio-optical hybrid detector would con-stitute a welcome option, allowing to reduce furtherpossible background noises. Also, complementing thehybrid radio-acoustic strings with a few optical DOMssuch as in IceCube would allow to unambiguouslytag neutrino events (at these energies, given a>100mabsorption length in the ice at these depths [15], photonsmay likely accompany a radio-acoustic signal in the caseof a neutrino event).

REFERENCES

[1] R. Abbasi et al., Nucl. Instr. Meth.A601:294-316,2009.[2] V. Berezinsky and G. Zatsepin, Phys. Lett. B 28 (1969) 423.[3] R. Engel, D. Seckel & T. Stanev, Phys. Rev. D64 (2001) 093010.[4] L. A. Anchordoqui, H. Goldberg, D. Hooper, S. Sarkar &

A. M. Taylor, Phys. Rev. D76 (2007) 123008.[5] J. Abraham et al. [Pierre Auger Collaboration], Astropart. Phys.

29 (2008) 188. [Erratum-ibid.30 (2008) 45][6] K. H. Kampert [Pierre Auger Collaboration], AIP Conf. Proc.

1085 (2009) 30.[7] G. Askaryan, JETP 14 (1962) 441; G. Askaryan, JETP 21 (1965)

658.[8] G. Askaryan, Sov. J. Atom. Energy 3 (1957) 921.[9] E. Zas, F. Halzen & T. Stanev, Phys. Rev. D45 (1992) 362.

[10] B. Price and L. Bergstrm, Appl. Opt. 36 (1997) 4181; L.Bergstrm, B. Price et al., Appl. Opt. 36, 4168 (1997); B. Price,J. Geophys. Res.111, B02201 (2006).

[11] S. Boeser et al., arXiv:0807.4676 [astro-ph]; S. Boeser et al., Int.J. Mod. Phys. A21S1 (2006) 107.

[12] J. Vandenbroucke et al., arXiv:0811.1087 [astro-ph].[13] F. Descamps et al. [IceCube coll.], in these proceedings.[14] D. Besson et al., Int. J. Mod. Phys. A21S1 (2006) 259.[15] M. Ackermann et al., J. Geophys. Res. 111, D13203 (2006).

Attached data CD 51

C.2 Attached data CD

Documents

Abstract - LPHE · 2009-06-18 · Abstract The characterization of ultra high energy neutrinos could provide insights into cosmological source evolution and cosmological accelerators