USF Libraries
USF Digital Collections

Prototype and testing of a MEMS microcooler based on magnetocaloric effect

MISSING IMAGE

Material Information

Title:
Prototype and testing of a MEMS microcooler based on magnetocaloric effect
Physical Description:
Book
Language:
English
Creator:
Ghirlanda, Simone L
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Magnetic refrigeration
Magnetic materials
Gadolinium
Permanent magnet
Adiabatic demagnetization
Dissertations, Academic -- Electrical Engineering -- Masters -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: This thesis documents the work and research effort on the design, fabrication and testing of a magnetocaloric MEMS microcooler, focusing on the testing of the microcooler at low magnetic fields. The phenomenon of magnetocaloric effect (MCE), or adiabatic temperature change, which is obtained by heating or cooling magnetic materials due to a varying magnetic field, can be exploited in the area of magnetic refrigeration as a reliable, energy-efficient cooling system. In particular, its applications are being explored primarily in cryogenic technologies as a viable process for the liquefaction of hydrogen. The challenge for magnetic refrigeration is that the necessary MCE is most easily achieved with high magnetic fields (5-6 Tesla) provided by superconducting magnets. However, a significant magnetocaloric effect can be exhibited at lower magnetic fields (1-2 Tesla) by carefully controlling initial temperature conditions as well as by selecting, preparing and synthesizing the optimal fabrication process of Silicon (Si) wafers. A microcooler was integrated based on previous works of others and tested. Finally, testing of the magnetocaloric effect was conducted and results analyzed. Experimental results in these domains demonstrate that magnetic refrigeration can be part of the best current cooling technology, without having to use volatile, environmentally hazardous fluids. The MEMS magnetocaloric refrigerator demonstrated a ~ -12°C change in the temperature of cooling fluid at a magnetic field of 1.2 T.
Thesis:
Thesis (M.A.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Simone L. Ghirlanda.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 46 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001790339
oclc - 144611075
usfldc doi - E14-SFE0001500
usfldc handle - e14.1500
System ID:
SFS0025819:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001790339
003 fts
005 20070618145116.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 070618s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001500
040
FHM
c FHM
035
(OCoLC)144611075
049
FHMM
090
TK145 (ONLINE)
1 100
Ghirlanda, Simone L.
0 245
Prototype and testing of a MEMS microcooler based on magnetocaloric effect
h [electronic resource] /
by Simone L. Ghirlanda.
260
[Tampa, Fla] :
b University of South Florida,
2006.
3 520
ABSTRACT: This thesis documents the work and research effort on the design, fabrication and testing of a magnetocaloric MEMS microcooler, focusing on the testing of the microcooler at low magnetic fields. The phenomenon of magnetocaloric effect (MCE), or adiabatic temperature change, which is obtained by heating or cooling magnetic materials due to a varying magnetic field, can be exploited in the area of magnetic refrigeration as a reliable, energy-efficient cooling system. In particular, its applications are being explored primarily in cryogenic technologies as a viable process for the liquefaction of hydrogen. The challenge for magnetic refrigeration is that the necessary MCE is most easily achieved with high magnetic fields (5-6 Tesla) provided by superconducting magnets. However, a significant magnetocaloric effect can be exhibited at lower magnetic fields (1-2 Tesla) by carefully controlling initial temperature conditions as well as by selecting, preparing and synthesizing the optimal fabrication process of Silicon (Si) wafers. A microcooler was integrated based on previous works of others and tested. Finally, testing of the magnetocaloric effect was conducted and results analyzed. Experimental results in these domains demonstrate that magnetic refrigeration can be part of the best current cooling technology, without having to use volatile, environmentally hazardous fluids. The MEMS magnetocaloric refrigerator demonstrated a ~ -12¨C change in the temperature of cooling fluid at a magnetic field of 1.2 T.
502
Thesis (M.A.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 46 pages.
590
Adviser: Shekhar Bhansali, Ph.D.
653
Magnetic refrigeration.
Magnetic materials.
Gadolinium.
Permanent magnet.
Adiabatic demagnetization.
690
Dissertations, Academic
z USF
x Electrical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1500



PAGE 1

Prototype and Testing of a MEMS Microcooler Based on Magnetocaloric Effect by Simone L. Ghirlanda A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering Department of Electrical Engineering College of Engineering University of South Florida Major Professor: Shekhar Bhansali, Ph.D. Muhammad Rahman, Ph.D. Sangchae Kim, Ph.D. Elias (Lee) Stefanakos, Ph.D. Date of Approval: March 24, 2006 Keywords: magnetic refrigeration, magnetic ma terials, gadolinium, permanent magnet, adiabatic demagnetization Copyright 2006, Simone L. Ghirlanda

PAGE 2

To my family

PAGE 3

ACKNOWLEDGMENTS My advisor, Shekhar Bhansali, has been the source of inspiration and guidance throughout my graduate career. None of my success would have been possible without his efforts to teach me how to te st, present, and articulate ideas. I am also most grateful to my family in Italy: my parents who spent many years of my childhood inspiring me with life goals and teaching me a bout the value of education, who have shown their true love by allowing me to pursue my university studies away from home, and who have always been so clos e to me even from th e other side of the ocean; my sister Ilaria who I have always been able to count on whenever I needed; both of my grandmothers for thei r affection and phone calls; my uncle Alberto who has often talked to me and given me suggestions about the engineering professional life; and both of my grandfathers and great aunts who I wish could have s till been here today to share this moment. My sincere gratitude goes to my beau tiful girlfriend Amy who has always been there for me with love, support and motivation for these past four years across two states. I would also like to acknowledge my second family in Texas, Alan and Kaye Ostermann along with Lisa, Nathan and Seth who gave me the opportunity to come live in the US for a year during high school and c ontinue to be an important part of my life today. I would also like to thank everyone directly involved with the MEMS Microcooler Project at University of South Florida. In particul ar I am grateful to Senthil

PAGE 4

Sambandam, Luis Rosario, Bharath Bethal a, and Shantanu Shevade, for having the courage to initiate such a project. Sang Chae Kim deserves great credit for acting as lead researcher and keeping the project afloat duri ng this past year and a half. The projects only undergraduate student volunteer, Carl Adam s, also deserves my sincere appreciation for helping to make my ideas a reality as well as giving precious suggestions. I am also thankful to those who ha ve provided friendship, support, and/or inspiration including Stephen Bates, Eddi e Bettega, Roberto Cuneo, Matteo Fossati, Francesco Martinelli, Federico Mautone, Paolo Migliavacca, Federico Pippo, Marco Pollastri, Tony Price, Tony Rosado, Riccardo Si gnorelli, Omar Souza, Rodrigo Trejos and Silvia Volpini. I am thankful as well for the suppor t and guidance of my committee, Muhammad Rahman, Sang Chae Kim, and Elias Stefanakos of University of South Florida. I also wish to acknowledge the inspiration for gradua te studies given to me at my Alma Mater, The University of Texas at Austin, in particular by Prof essors Brian Evans, Thomas Milner, Theodore Rappaport, and Baxter Womack of the ECE Department. Finally, I wish to thank all those in the BioMEMS and Microsystems Lab who have made this thesis more than just a single persons research project. I am very thankful for all the feedback I have received from th em. In particular, I wish to thank Shyam Aravamudhan, Subramanian Krishnan, Kevin Luongo, Joe Register, Praveen Sekhar, Altagrace Sine and Andrew Farmer of the Center for Ocean Technology. This research was supported by NASA under the grant Hydrogen Research at Florida Universities (HYRES). SIMONE L. GHIRLANDA University of South Florida, May 2006

PAGE 5

TABLE OF CONTENTS LIST OF TABLES......................................................................................................iii LIST OF FIGURES....................................................................................................iv ABSTRACT .................................................................................................................v CHAPTER 1 INTRODUCTION .................................................................................1 1.1 Motivation ..........................................................................................................2 1.2 Approach ............................................................................................................3 1.3 Contribution and Impact ....................................................................................4 1.4 Overview of the Thesis......................................................................................6 CHAPTER 2 LITERATURE REVIEW OF MAGNETIC REFRIGERATION .........7 2.1 History of Continuous Magnetic Refrigeration.................................................7 2.2 Near-Room Temperature Magnetic Refrigerator ..............................................8 2.3 Conclusion .......................................................................................................11 CHAPTER 3 SYSTEM SETUP AND SPECIFICATIONS ....................................13 3.1 Magnet Specifications ......................................................................................13 3.1.1 Magnetic Field Measurements..................................................................14 3.2 GdSiGe Block Specifications and Testing .......................................................16 3.2.1 GdSiGe Block Properties ..........................................................................16 3.2.2 Cooling Chamber ......................................................................................18 3.2.3 GdSiGe Block Testing ..............................................................................19 3.3 Design of Microcooler .....................................................................................22 3.3.1 Wafer-to-Wafer Bonding ..........................................................................24 3.3.1.1 Anodic Bonding .................................................................................24 3.3.1.2 Fusion Bonding ..................................................................................26 3.3.2 Syringe Pump ............................................................................................26 3.3.3 Temperature Sensors Integration ..............................................................26 CHAPTER 4 TESTING AND RESULTS .................................................................28 4.1 Plexiglass Dummy Wafer Testing and Results ............................................28 4.2 Silicon Wafer Testing ......................................................................................31 4.2.1 Results for Silicon-to-Glass ......................................................................32 4.2.1.1 Si-to-Glass Testing Changing Flow Rate and Interval Time.............38 4.2.2 Results for Silicon-to-Silicon ....................................................................40 4.3 Discussion and Future Work ............................................................................41 i

PAGE 6

CHAPTER 5 CONCLUSION ...................................................................................43 5.1 Contributions ...................................................................................................43 5.2 Conclusion .......................................................................................................43 REFERENCES..........................................................................................................44 BIBLIOGRAPHY......................................................................................................46 ii

PAGE 7

LIST OF TABLES Table 3.1 Effect of materials interfering with the magnetic field..............................15 iii

PAGE 8

LIST OF FIGURES Figure 2.1 A schematic sketch of the AMR refrigerator...........................................10 Figure 3.1 Experimental setup for refrigeration testing.............................................13 Figure 3.2 Magnetic field vs. dist ance between poles...............................................15 Figure 3.3 Magnetization with temperature for GdSiGe disc....................................17 Figure 3.4 Magnetic entropy changes for GdSiGe di sc at various magnetic fields...18 Figure 3.5 Cooling chamber around magnetic poles.................................................19 Figure 3.6 MCE at different initial temperatures.......................................................21 Figure 3.7 GdSiGe temperature change.....................................................................21 Figure 3.8 Design of the microcooler........................................................................23 Figure 3.9 Microconnectors.......................................................................................23 Figure 3.10 Prototype of microcooling channels.......................................................23 Figure 3.11 Eight micro channels fabr icated on a Si wafer.......................................24 Figure 3.12 Schematic diagram of the experimental setup for anodic bonding........25 Figure 3.13 Temperature sensors integration...........................................................27 Figure 4.1 Plexiglas wafer.........................................................................................29 Figure 4.2 Plexiglas wafer results at initial temp. of -10 C.......................................30 Figure 4.3 Si-to-Glass microcooler at initial temp. of +5 C......................................32 Figure 4.4 Si-to-Glass microcooler at initial temp. of +3 C......................................34 Figure 4.5 Si-to-Glass microcooler at initial temp. of +1 C......................................35 Figure 4.6 Si-to-Glass microcooler at initial temp. of -3 C.......................................36 Figure 4.7 Si-to-Glass microcooler at initial temp. of -4 C.......................................37 Figure 4.8 Si-to-Glass microcooler at initial temp. of -8 C.......................................37 Figure 4.9 Si-to-Glass microcooler at initial temp. of -11 C.....................................38 Figure 4.10 Si-to-Glass microcooler with weak effect .............................................39 Figure 4.11 Si-to-Glass microcooler when magnet has not died out.........................40 Figure 4.12 Si-to-Si Microcooler at initial temp. of -1 C..........................................41 Figure 4.13 Configuration of magnetic refrigerator..................................................42 iv

PAGE 9

PROTOTYPE AND TESTING OF A MEMS MICROCOOLER BASED ON MAGNETOCALORIC EFFECT Simone L. Ghirlanda ABSTRACT This thesis documents the work and research effort on the design, fabrication and testing of a magnetocaloric MEMS micr ocooler, focusing on the testing of the microcooler at low magne tic fields. The phenomenon of magnetocaloric effect (MCE), or adiabatic temperature change, which is obtained by heating or cooling magnetic materials due to a varying magnetic field, can be exploited in the area of magnetic refrigeration as a reliable, energy-e fficient cooling system. In particular, its applications are being explor ed primarily in cryogenic technologies as a viable process for the liquefaction of hydrogen. The challenge for magnetic refrigeration is that the necessary MCE is most easily achieved with high magnetic fields (5-6 Tesla) provided by superconducting magnets. However, a significant magnetocaloric effect can be exhibited at lower magnetic fields (1-2 Tesla) by carefully controlling initial temperature conditions as well as by selecti ng, preparing and synthe sizing the optimal fabrication process of Sili con (Si) wafers. A microcool er was integrated based on previous works of others and tested. Finall y, testing of the magnetocaloric effect was conducted and results analyzed. Experimental results in these domains demonstrate that magnetic refrigeration can be part of the best current cooling technology, without having to use volatile, environmentally ha zardous fluids. The MEMS magnetocaloric refrigerator demonstrated a ~ -12 C change in the temper ature of cooling fluid at a magnetic field of 1.2 T. v

PAGE 10

CHAPTER 1 INTRODUCTION The fundamental motivation of magnet ocaloric refrigeration is to study alternative solutions for more efficient liq uefaction of hydrogen. Ge nerally this type of refrigeration requires the use of superconducting magnets able to produce strong magnetic fields in the range of 4-6 Tesla (T) as well as the use of ferromagnetic compounds based on rare earth elements such as gadolinium (Gd). However, superconducting magnets are very expensive and require extensive servicing which ultimately defeats the purpose of finding an e fficient alternative type of refrigeration. The primary advantages of this cycle are its relatively high cost-effectiveness ratio and environment friendliness. MEMS Microcoolers are proposed as a solution to the challenge of reducing magnetic fields: with MEMS coolers using ju st permanent or electro-magnets in the magnetic field range of only ~1 T would be feasible, as (a) they provide a large surface area for a given volume; (b) they redu ce the pole to pole distance resulting in adequate field intensity. In addition, MEMS Microcoolers add flexibility not only in the liquefaction of hydrogen, but in other r eal-world scenarios, such as household refrigeration and air-conditioning systems. Such small systems can also be used for Zero Boil-Off (ZBO) control. Sophisticated non-hazardous refrigerati on systems are difficult to implement, in part because they are likely to be extremely complex, perhaps requiring the synthesis and optimization of hundreds of di fferent forms of magnetocaloric materials with optimal properties. This thesis describes a method for implementing a sophisticated and effective magnetocal oric refrigerator by preparation of 1

PAGE 11

magnetocaloric material GdSiGe, design and development of micro-fabrication processes for prototype microcoolers and validation of th e model with experimental results. This chapter begins by presenting th e motivation for magnetocaloric microrefrigeration, then briefly describes the a pproach and concludes with an overview of the results and contributions of the thesis. 1.1 Motivation Magnetic refrigeration has shown great promise as a viable process for liquefaction of hydrogen. It is a potentially compact, relia ble and efficient technology whose qualities are highlighted by the ab sence of hazardous or environmentally damaging chemicals (such as chlorofluorocarbons). Additionally, it is up to 60% efficient (based on the temperature of ope ration). On the other hand todays best commercial devices (vapor-compression refrig erators), which basically extract heat from a vapor using a compressor, achieve a maximum efficiency of about 40% [1]. Magnetic refrigeration can also be useful for heat dissipation in Zero Boil-Off (ZBO) cryogenic storage vessels. The miniatur ization of this effective refrigeration system could be a key technology for future Pi co-satellites. Also, because of its small size and lightweight it could be used to re-liquefy hydroge n in cryogenic storage tanks used for transportation and storage of hydrogen for space missions. In magnetocaloric refrigeration, the effec tiveness is measured in terms of the (a) cooling-capacity / magnetic-field ratio a nd (b) the volume of liquid extracting heat from (or dissipating it to) the ferromagne tic material in the least amount of time possible. In this regard, a miniaturized ma gnetocaloric cooling sy stem with a Silicon (Si) microstructure is needed when not using superconducting magnets. The use of MEMS technology enhances the surface area to volume ratio, improving the heat 2

PAGE 12

transfer between the fluid and the ferroma gnetic material, and enables the magnetic poles to be close to each other, resulting in greater field intensity. The improved heat transfer, with reduced distance between poles, enables operating at low magnetic field. This approach presents a pathway to evaluate inexpensive and environmentfriendly solutions to difficult vapor compre ssion refrigeration technology. It can even lead to common household refrigeration syst ems that can obtain competitive results utilizing permanent magnets. Such a system is developed and evalua ted in this thesis. 1.2 Approach Ferromagnetic materials heat up or cool down when inserted into a varying magnetic field. The second law of therm odynamics states that the entropy (or disorder) of a closed sy stem must increase with time. This is known as magnetocaloric effect (MCE) and it was origin ally discovered in iron by Warburg [2]. The MCE is intrinsic to all magnetic mate rials and is due to the coupling of the magnetic sub-lattice with the magnetic field, which reduces the spin entropy in the ferromagnetic material. This results in the a lignment of the electron spins in the atoms of the material with the direction of the magnetic field. To compensate for the effect and keep the process adiabatic (i.e. the to tal entropy of the system remains constant during the magnetic field change) the cons ervation of total en tropy leads to the enhancement of lattice entropy (i.e. the motion of the atoms becomes more random), thus raising the temperature of the material In the case of a magnetic refrigerator, the heat would be absorbed by liquid (or air) fl owing in the ferromagnetic material. Once the magnetic field is removed, the spin entropy is enhanced again and, the temperature of the material falls below that of its environment. This allows it to draw more unwanted heat, resulting in the cool down of liquid (or air) flowing in the 3

PAGE 13

refrigerant material. In other words, just as the compression of a gas, the isothermal magnetizing of a ferromagnetic material reduces its entropy, while vice-versa, demagnetizing (comparable to the expansion of a gas) restores the zero-field entropy of a system [2]. Designing and developing a valid alternat ive form of magnetic refrigeration presents several technical challenges in orde r to match current technology: the heating and cooling efficiency provided by existe nt MCE refrigerator s is usually highly dependent on the size of the applied magnetic field as well as the magnetic moments which tend to be largest in rare-earth elem ents. This would generally require the use of superconducting magnets (expensive and requiring extensive maintenance). The goal for the microcooler is to eliminate th e use of such magnets with no loss in cooling capability and, at a moderate co st without the use of environmentally hazardous chemicals. The MEMS microcooler meets thes e challenges through four technical optimizations: (1) preparation of magnetocalor ic material GdSiGe and its synthesis in different forms with optimal propertie s; (2) design and development of microfabrication processes for prototype microcoolers based on Silicon microstructure; (3) exploiti ng the peak of entropy of th e compound at significant low temperature; and (4) building an experimental setup to validate the concept. These conditions work together in optimizing soluti ons as part of the refrigeration process. This approach results in high cooling capab ility that can be adopted for the uses described above as a valid alternative to current existing technology, and also makes entirely new applications possible. 4

PAGE 14

1.3 Contributions and Impact The main contribution of this thesis is developing a MCE system to validate magneto-caloric refrigeration, under low ma gnetic fields, as a viable process for liquefaction of hydrogen. Several experiments, conducted in the research, demonstrate the feasibility of an effective MEMS micr ocooler based on magnetocaloric effect at a low magnetic field, and others suggest how th e approach could be used for real-world applications such as a household refrigerator. First, the properties of magnetocaloric material GdSiGe a nd its synthesis in different forms were studied by S.N. Sambandam et al [3]. In particular, the peaks of entropy of this compound are analyzed to come to the conclusion that significant MCE results can be obtained even at low ma gnetic field if working at significant low temperatures. Second, the design, development and test ing of micro-fabrication process for prototype micro-coolers was unde rtaken and is described elsewhere [4]. By analysis of previous research project in the same area, it is clear that one challenge is to be able to design and realize a system at the small s cale level. This thesis presents integration and testing of the microcooler. Several experiments were undertaken to validate individual concepts that ultimately led to th e assembly of a micro-cooler significantly smaller in size compared to what is current ly available. By exploiting the dimensions of the device, it is possibl e to obtain a remarkable MCE at a significantly low magnetic field (~ 1.2 T). The microcooler pres ented a challenge to its system analysis because in-situ temperature sensors are needed to be integrated in the system in order to validate and model the experimental results. This had b een done in related research [5]. 5

PAGE 15

The key driver for these experiments is high cooling capacity. Interestingly, the MEMS microcooler succeeds in obtaining a discrete MCE utilizing an electromagnet capable of producing a magnetic field of less than 1.5 T. This is a first step towards a better alternative to conven tional vapor compression refrigeration technology as well as household refriger ation, and opens up exciting avenues for future research. 1.4 Overview of the Thesis The thesis is divided into four parts: Literature Review (Chapter 2), System Set-Up and Specifications (Chapter 3), Testing and Results (Chapter 4), and Conclusion (Chapter 5). Chapter 2 reviews the history of application of the magnetocaloric effect and prior work in magneto caloric refrigeration. Chapter 3 presents a detailed description of the experiment settings and the system integration including electro-magne t, GdSiGe block, Silicon wafer, syringe pump, cooling chamber, bonding proce dure, and temperature sensors. Chapter 4 focuses on the testing procedures undertaken to characterize the performance of the MEMS microcooler as we ll as discussion of the results obtained. Chapter 5 discusses and reviews the majo r contributions of the MEMS microcooler and ideas for future research. 6

PAGE 16

CHAPTER 2 LITERATURE REVIEW OF MAGNETIC REFRIGERATION This chapter will review the history of continuous magnetic refrigeration and previous application work re sulting in an active magnetic regenerator cycle. After briefly reviewing specifications and setti ngs of the experiment, the results and discussion of it are presented. Finally, the chap ter concludes with a review of issues in competitive magnetic refrigeration that the MEMS microcooler attempts to address. 2.1 History of Continuous Magnetic Refrigeration The history of continuous magnetic refr igeration goes back to the work of Collins and Zimmerman [6], and Heer [7]. Th ey were the first ones to build and test magnetic refrigerators operating be tween 0.2 K 0.73~1 K by continuously magnetizing and demagnetizing iron ammonium alum. Their latest system (in 1953) was able to extract 12.3 J/s from the co ld reservoir at 0.2 K operating at 1/120 Hz frequency. However, few years had to pass before Brown [8] was able to realize a near room temperature continuously opera ting magnetic refrigerator. To accomplish this result, studies were conducted to e xplore magnetic refriger ation properties at significantly higher temperatures. This allowe d scientists to conclude that a much larger temperature span was possible. Brow n was able to attain a 47 Celsius no-load temperature difference between the hot inlet (46 C) and cold ou tlet (-1 C) flowing a combination of water and alcohol into a magnetocaloric bed made in Gd with a magnetic field change of 0 to 7 T. Following the early work of Brown, two other scientists, Steyert [9] and Barclay [10, 11] continued with design and development of the concept of active magnetic regenerator (AMR) refrigeration which was 7

PAGE 17

successively brought to life in several expe riments at different temperatures. One of these experiments performed by Zimm et al. [12] is described below. 2.2 Near-Room Temperature Magnetic Refrigerator Magnetic refrigeration is based on the magnetocaloric effect (MCE). In the case of ferromagnetic materials, the MCE corresponds to a warming as the magnetic moments of the atoms are aligned by the a pplication of the magnetic field and viceversa upon removal of the magnetic field. This paper reviews the descripti on and performance of a near-room temperature magnetic refrigerator develope d at the Astronautics Technology Center in Madison, WI based on previous similar work done at Ames Laboratory of Iowa State University. The unique aspect of the resear ch is that it uses the active magnetic regeneration concept of recent cryogenic de vices, but in contrast to the cryogenic case, the heat capacity of the fluid in the por es of the regenerator bed is compared to that of the solid matrix. The magnetic warming and cooling generated by the magnetocaloric effect can present almost no losses in soft ferro magnets such as gadolinium (Gd) based compounds, allowing for the possibility of an extremely efficient refrigeration process. There are two major challenges in the design of a refrigerator based on magnetocaloric effect. First, unless a strong magnetic field is applied (5 T or more) the magnetocaloric effect (MCE) is fairly small. Even with an optimal material such as gadolinium, a maximum adiabatic temper ature change of 11 K can be produced. Second, as the ferromagnetic material is a solid, it is not easy to pump it through heat exchangers, as in the case for gas and vapor cycle refrigerants. This second problem however can easily be solved by utilizi ng a heat transfer fluid, such as water, and regeneration (full cycle) can be obtained by flowing it 8

PAGE 18

through a porous bed of magnetocaloric mate rial which is alternatively magnetized and demagnetized by respectively insert ing it and removing it from the applied magnetic field. In this case an Active Magnetic Re generator Refrigerator (AMRR) was chosen, in which the magnetocaloric bed ge nerates refrigeration that appears to regenerate the bed itself using the thermal li nkage of the fluid. This particular version requires that the heat capacity of the fluid is much less than that of the bed so that the temperature gradient remains in the magnetocal oric bed. At first, the use of water as the heat transfer fluid seemed problematic, as water trapped in the pores of a bed of gadolinium spherical particles has a capacity eq ual to that of the refrigerant material. However, the volume of water flown through th e bed was held to less than that of the pore volume. This overcame the problem and a successful device could be designed on this basis. The AMRR tested (schematically shown in Figure 2.1) is made of two 1.5 kg beds composed of Gd spheres. The magnetic field is provided by conventional liquid helium immersed NbTi solenoid mounted in a warm bore dewar. The two magnetocaloric beds are alternatively inserted in the magnetic field of 1-5 T using an air cylinder drive. The magnetic refrigeran t material used in the experiment was gadolinium metal and it was only 93% pure. 9

PAGE 19

Figure 2.1 A schematic sketch of the AMR refrigerator [9] The flow starts in a primary loop with fluid being cooled as it passes through the demagnetized bed. Next, the water pick s up a thermal load by passing through the cold heat exchanger (CHEX). Heat then is extracted while the water is flown through the hot heat exchanger (HHEX) and returns to the pump inle t. After a standard time interval, the magnetized bed is remove d from the magnetic field to let the demagnetized ones insertion. At this point the flow direction is reversed and the 10

PAGE 20

thermal load supplied to th e CHEX by a secondary flow loop. The heat is removed from the HHEX by a secondary flow path using cooling water. Temperatures were measured at the inlet and outlet of each heat exchanger as well as inside each magnetocaloric beds in five locations. As expected, the cooling power is proportional to the magnetic field increase throughout the cycle and initially it is al so proportional to the fluid flow rate. However, a decrease in cooling power is also expected with increasing temperature because the adiabatic temperature change of gadolinium decreases with temperature below its Curie point [12]. In conclusion, as expected, a consid erable temperature change and cooling effect of 12 K was obtaine d only at a high magnetic fi eld (5 T). Up to 600 W of cooling power was generated in a 5 T ma gnetic field while only 200 W of cooling power was produced at 1.5 T. The efficiency of the system approaches 60% at 5 T (while only 30% at 1.5 T) of Carnot. However, this didnt take into account the power used to load the superconducting magnet. 2.3 Conclusion The AMR cycle described above has several positive features useful for practical application in the ar ea of magnetic refrigeration [2]. First, since the MCE of each individual particle of the magnetocal oric bed changes the entire temperature profile across the bed itself, the temperatur e span of a single stage can easily exceed that of the whole magnetic refrigerant. Second, because the bed is composed by particles which ultimately together act as thei r own generator, heat doesnt need to be transferred between two separa te solid assemblies, but rather only between each solid particle within the same bed via the ac tion of a fluid. Third, since the individual particles in the bed do not encounter the en tire temperature span of the stage, the 11

PAGE 21

magnetocaloric bed can be prepared into la yers of different magnetic material which ultimately could be optimized depending on the particular temperature range. However, the results obtained above, particularly in terms of temperature span, are substantial only at a high magnetic field (5 T), which requires the use of superconducting magnets. Therefore, although the work done gives a good insight on magnetic refrigeration, the current system ma kes the idea an unpractic al alternative to current near-room refrigeration systems, su ch as household refrigerators, and the air conditioning market. This combined with the characteristics of the refrigerant material used, Gd, and the potential for very high efficiency inspires further research in the area. 12

PAGE 22

CHAPTER 3 SYSTEM SET-UP AND SPECIFICATIONS This chapter focuses on all the work done prior to the experiment. In particular it gives details about the electro-magnet use d, the refrigerant material synthesis, the fabrication of the Silicon wafer, the syringe pump as well as the details of the experiment set-up such as cooling chambe r and temperature sensors integration. The overall system set-up is presented in Figure 3.1. Figure 3.1 Experimental setup for refrigeration testing 3.1 Magnet Specifications As stated earlier, one major difference in this project from past works in magnetic refrigeration, where superconduc ting magnets are usually the popular choice, is that an electro-magnet was select ed to be used. This choice was made for 13

PAGE 23

two main reasons. First, because as it was seen before, the use of superconducting magnets is not feasible as a choice for current household refrigeration or air conditioning systems because of its large size, high cost and extensive servicing requirements. Second, because an electro-magne t was available at no cost as part of the equipment of the Department of Elect rical Engineering and given its magnetic field capacity of up to ~ 1.7 T, it should be possible to simulate the effect of a permanent magnet which would be the ultimat e solution to create a technology able to compete with current existent devices. Low energy requirements would be crucial to create a winning alternative technology, therefore the choice to produce a system capable of operating with permanent magnets. The magnet utilized in the experime nt is a Varian electro-magnet with magnetic field capability of up to ~ 1.7 T. The main transformer enables approximately 42 A of current to charge two spiral coils, one in each magnetic disc and the magnetic field is created. The contro ller uses a feedback loop which relies on a Hall probe to be able to maintain the magnetic field at cert ain strength without unnecessary oscillations which coul d be caused by external agents. 3.1.1 Magnetic Fi eld Measurements Several measurements were done on the magnetic field alone to get more precise data and correlate the distance of the poles to the magnetic field produced. The maximum magnetic field is about 1.2 T at a gap of approximately 3 cm. This was about the minimum space needed in order to perform the experiment. It was established that the magnetic field drastically decreases with increasing pole to pole distance. Figure 3.2 shows the magnetic fiel d measurements as a function of distance between poles. 14

PAGE 24

Figure 3.2 Magnetic field vs. distance between poles At a distance of about 2 cm the poles are pulled together by the magnetic field strength. Also, any material in terfering with the field seems to decrease the strength of it. Different materials such as plastic, s ponge, and cardboard were tested in order to optimize the preparation of a cooling cham ber which would interf ere as little as possible with the magnetic field. Table 3.1 presents the results. Table 3.1 Effect of materials interfering with the magnetic field Surrounding Type Magnetic field without box (Tesla) Magnetic field in the box (Tesla) Plastic box 0.925 0.865 Plastic Box + Sponge 0.925 0.77 Card board/ Styrofoam 0.925 0.905 15

PAGE 25

As it can be seen from the table both plastic and sponge drastically decrease the magnetic field, while only cardboard and St yrofoam have a slight effect on it. This insight lead the idea of building a cooling chamber made in Styrofoam. 3.2 GdSiGe Block Specifications and Testing As described by S.N. Sambandam et al. [3] GdSiGe alloys were synthesized at the Ames Laboratory of Iowa State University by preparing mixtures of Gd (99.8% purity), Si and Ge (both 99.99% purity). The mixture was arc melted in an argon atmosphere on a water cooled copper hearth. Each mixture was re-melted at least six times. The resulting alloy button was turned over after each melti ng to achieve alloy homogeneity. The mass of the mixtures was chosen not to exceed 25 g to achieve highest possible solidification and cooling rates, and the la st melting step was always terminated by shutting down the power to the arc. All alloys are in itially examined for any impurity phases, and those that possess monoclinic crystal structure were heat treated at 1570 K for one hour. The heat trea tment was carried out in vacuum using an induction furnace. The alloy buttons were wr apped in Ta foils to avoid any kind of oxidation in the furnace. After the heat treat ment, the alloys were rapidly cooled by shutting down the power of the induction coil. These alloy buttons were then grounded and powdered. The powder was then compacted using room temperature iso-static pressing to yield th e 2 inch diameter discs [13]. 3.2.1 GdSiGe Block Properties Initially, studies and research were undertaken to better understand the GdSiGe block magnetocaloric properties at different initial temperature values. The synthesized powders of GdSiGe were char acterized for magnetic and magnetocaloric properties. The dependence of magnetization of GdSiGe on temperature at a magnetic 16

PAGE 26

field strength of 5 Gauss was measured and is plotted in Figure 3.3. It shows a paramagnetic to ferromagnetic transition be tween 260 K and 300 K. This indicates that the material is magnetic and has two composite phases that undergo transition. Figure 3.3 Magnetization with temperature for GdSiGe disc The magnetocaloric effect has been studied and is shown in Figure 3.3 by the plot of change in magnetic entropy with temperature, indicating maximum magnetic entropy change at the transition temperatures. This has been accomplished by indirectly measuring entropy change from heat capacity measurements. Heat capacity measurements have been made on the sa mples from 240 K to 300 K at different applied magnetic fields from 1 T to 5 T (10 kOe to 50 kOe). The magnetic entropy can be deduced using th e following Eq. (1) [13], dT T HTCHTC HTScT 0 1 2),(),( ),( 17

PAGE 27

where C(T, B) and C(T, 0) are the values of heat capacity in field B and in zero field, T being the temperature. It has earlier been demonstrated that this method yields comparable results derived from magnetization measurements [14]. Figure 3.4 Magnetic entropy changes for GdSiGe disc at various magnetic fields Figure 3.4 shows the entropy change calculated from the heat capacity measurements over a temperature range of 240 K to 340 K fo r varying applied magnetic fields from 1 to 5 Tesla. The peak of entropy change is observed near 262 K and 280 K. 3.2.2 Cooling Chamber A cooling chamber built in Styrofoam between the magnet poles was prepared utilizing metal plates on each side and on the top which were frozen by pouring liquid nitrogen. However, holes were cut th rough the chamber around the poles to completely expose them so that the ma gnetic field wouldnt get affected by the Styrofoam. This allowed controlling the temp erature inside the chamber and the initial 18

PAGE 28

temperature of the GdSiGe block, keeping it stable for an extended period of time. Furthermore, tests were performed with the magnet being switched on and off to make sure this alone didnt affect the temperature in the chamber when no GdSiGe block is exposed to the magnetic field. The cooling chamber set-up is shown in Figure 3.5. Figure 3.5 Cooling chamber around magnetic poles 3.2.3 GdSiGe Block Testing Changes in temperature were recorded in the GdSiGe block at an applied magnetic field of approximately 1.2 Tesla. More precisely, the peaks of entropy change described in Figure 3.4 were taken into account when pe rforming the testing procedure. Beginning at low temperatures (-16 C), a st able magnetic field of 1.2 T was applied directly to the GdSiGe bl ock for approximately 90 seconds. It took 19

PAGE 29

approximately 30 seconds for the electro-ma gnet to reach a maximum output of ~42 A. The temperature change was measured using a thermocouple, which has a 0.1 C resolution, through which data was logged ev ery second. As a result in Figure 3.6, the GdSiGe block initial temperature of -16 C ro se to 9 C. This is because the entropy of the block reduces, as the fields align th emselves, in presence of magnetic field thus dissipating heat. Successively, when the magnet was switched off and the magnetic field removed, demagnetization occurred and th e block temperature dropped to -12 C within approximately a minute and stabili zed at that temperature. This resulted because the entropy of the block increases by absorbing heat. The effect can be used for cooling of fluids in the microchannels It was noticed that the increase and decrease in temperature was not equal due to the heat, which was generated, from the increasing stage. Similar testing was performe d at -10 C, -6.5 C, 1.5 C, 4.5 C, and 7 C. However, as observed from Figure 3.6, changes of temperature progressively attenuated for higher initial temperatures. Th ese results can be correlated with Figure 3.4, which shows the entropy change. Entropy ch ange peak as shown in Fig. 3.4 exists at 262 K (-11 C), and the maximum temperature change of the block is observed around that peak in Fig. 3.7. Though an entropy change peak exists at 280 K (7 C), no appreciable temperature change was obs erved. This was due to the applied 1.2 T magnetic field, whose peak of entropy is observed only near 262 K. At higher magnetic fields, there would be appreciab le temperature change around 280 K. 20

PAGE 30

GdSiGe Temperature Change-20 -15 -10 -5 0 5 10 0 50 100 150 200 250 Time (s)Temperature (C) T_-15C T_-9.2C T_-4.4C T_1.6C T_4.9C T_7.2C On Off Figure 3.6 MCE at different initial temperatures GdSiGe Temperature Change-4 -2 0 2 4 6 8 -20 -15 -10 -5 0 5 10 Initial Temperature (C)Temperature Change (C) On Off Figure 3.7 GdSiGe Temperature Change 21

PAGE 31

3.3 Design of Microcooler As described in S.C. Kim et al. [4], the design of the microcooler is schematically shown in Figure 3.8. Eight fluidic micro-channe ls of trapezoidal shape, each 300 m wide, 150 m deep and 25 mm long were fabricated in a 1 inch 1 inch area on a 2 inch silicon (100) wafer. The mi cro-channels were designed to ensure an adequate mass of GdSiGe dedicated for c ooling. The channels had a minimal spacing of 2700 m between them. Micro Channels Liquid Figure 3.8 Design of the microcooler Microconnectors were attached to the inlet and outlet ports of the wafer with microchannels as shown in Figure 3.9. The wa fer was then attached to a 2 inch, 5 mm thick GdSiGe block using crystalbond TM 509. The adhesive was dissolved in acetone and sprayed on the GdSiGe surface, which was then allowed to evaporate resulting in a 4~5 m thin uniform layer of adhesive. The thickness of the resultant adhesive layer was dependent on the initial mixture of the adhesive and acetone [4]. 22

PAGE 32

Figure 3.9 Microconnectors The fabricated microchannels are shown in Figure 3.10 and 3.11 [4]. Figure 3.10 Prototype of microcooling channels 23

PAGE 33

Figure 3.11 Eight micro channels fabricated on a Si wafer 3.3.1 Wafer-to-Wafer Bonding Wafer-to-wafer bonding was a necessary last step needed to seal the face of created micro channels and be able to collect data when running fluid through the microchannels. Several ways to perform Si-to-Si wafer bonding were explored to ensure that the bond would w ithstand the low testing temp erature. They consisted primarily in the use of (a) super-glue, (b) potassium silicate, and (c) crystal bond materials. However, none of them were strong enough to prevent leaking from occurring when testing the flowing of fluids such as water or anti-freeze through the microchannels. The high pressure caused by the syringe pump, even at the slowest possible rate of flow, seemed to be t oo much for the different type of bonding materials to handle. Finally, anodic bonding and fusion bonding were explored. 3.3.1.1 Anodic Bonding Anodic bonding is a process which permits the sealing of a silicon wafer to a glass one. The two wafers are assembled and heated on a hot plate in a room atmosphere to 24

PAGE 34

temperature between 400-500 deg. A high vol tage and low current DC power supply is connected to the assembly such that the silicon is positive with respect the glass. Then by applying ~1.5 kV across the assembly for about 20 minutes, the glass seals to the metal. Figure 3.12 shows a drawing of the experimental setup. Figure 3.12 Schematic diagram of the experimental setup for anodic bonding During the process, the positive Na ions in the glass are thermally excited resulting in increased mobility and in their attracti on to the negative electrode on the glass surface. This causes their removal. The str onger bound negative ions in the glass form a space charge layer adjacent to the silicon surface. Initially the potential is uniformly distributed across the glass, but after the positive sodium i ons have drifted toward the surface a large potential drop occurs at the glass/anode interface. The resulting electric field between the surfaces pulls them into contact. This causes the bonding to begin below the negative electrode and to sp read across the surface. At the end of the process, the wafers are held together by a chemical bond which is irreversible [15]. 25

PAGE 35

3.3.1.2 Fusion Bonding In fusion bonding, the substrates are fi rst forced into intimate contact by applying a high contact force. Once in contac t, the substrates hold together due to atomic attraction forces (Van der Waal), which are strong enough to allow the bonded substrates to be handled. The substrates ar e then placed in a furnace and annealed at high temperature, after which a solid bond is formed between the substrates. 3.3.2 Syringe Pump In order to perform the fluid flowing part of the experiment a syringe pump was used. The CAVRO XP 3000 uses a ste pper-motor driven syringe and valve design to aspirate and dispense measured qua ntities of liquid. Both the syringe and the valve are replaceable. The syringe plunger is moved within the syringe barrel by a rack and pinion drive that incorporates a 1.8 degrees stepper motor and qua drature encoder to detect lost steps. The syringe drive has a 30 mm travel length and resolution of 3000 steps (3000 or 24000 steps for microstep-enabled firmware). The top of the syringe barrel attaches to the pump valve by a 1/4-28" fitt ing which is the size of the micro-tubes used for the flowing of liquid. 3.3.3 Temperature Sensors Integration In order to be able to monitor temp erature on the wafers inlet and outlet, digital temperature sensors were used. Howe ver, because of their size they could not be integrated directly into the wafer. Ther efore, a small rectangular piece of Plexiglas was used in order to create a temperatu re recording center. Two small channels were drilled through the Plexiglas while temp erature sensors were also inserted into the same channels by drilling holes from the other face of the rectangle. Micro26

PAGE 36

connectors were attached on both sides so th at liquid could flow directly from the wafer into the Plexiglas and temperature could be recorded without disturbing the cycle. Figure 3.13 Temperature Sensors Integration 27

PAGE 37

CHAPTER 4 TESTING AND RESULTS This chapter will review the testing procedure and results obtained as well as the future work inspired by the project. Firs t a Plexiglas wafer was used in order to analyze only the effect of the refrigerant material itself. Then Silicon wafer results are presented both for the anodic bonding case which uses Si-to-Glass as well as for the fusion bonding case which allows the use of Si-to-Si. Finally, discussion and comparison of the results take place and ideas for future work are briefly presented. 4.1 Plexiglas Dummy Wafer Testing and Results During the first liquid testing procedur e, a dummy wafer made in Plexiglas (an insulator) was chosen to be used. This choice allowed testing of only the cooling properties of the GdSiGe block without it be ing influenced by the conducting effect of Silicon (while wafer-to-wafer bonding solutions were explored). A small channel was created on the bottom surface of a small r ectangular Plexiglas block and this was mounted on top of the GdSiGe sample so th at the liquid could be flowing in contact with the sample itself. A mixture of anti-fr eeze and water was used as fluid in this experiment since initial system temperat ure was below 0 C. A syringe-pump with a 5.0 ml capacity was used in order to flow the solution through the channel. The coolant was pumped at a rate of 5 ml/min. That corresponds to a flow rate of about 83.3 L/s. The Plexiglas wafer is shown in Figure 4.1. 28

PAGE 38

Figure 4.1 Plexiglas Wafer The graph in Figure 4.2 describes the sensors response of changing temperature versus time at constant ma gnetic field. The initial temperature was stabilized around the peak of entropy change previously analyzed and found at -11 C for GdSiGe. The outlet liquid temperature was normalized with the inlet liquid temperature at magnet off in order to study th e results of the magnetic effect itself and not the results of the GdSiGe block be ing at an initial cold temperature. 29

PAGE 39

Plexiglass Wafer Results with GdSiGe Block at Initial Temp = -10 -5 -4 -3 -2 -1 0 1 2 135791113151719212325272931333537394143454749515355575961636567 Time (s)Temperature change (C) Cooling Effect Figure 4.2 Plexiglas Wafer Results at Initial Temp. of -10 C During the first few seconds of the process the magnet was off. However, inlet liquid temperature was below zero as a result of the residual liquid left into the microtube and the preparation of the cooling chamber through the use of liquid nitrogen which brought the liquid temperature below 0 C. This time however was used to make the adjustment needed on the outlet temperat ure to avoid consideration of the cooling process caused by the GdSiGe block itself ke pt at -10 C. In the plot, the first few seconds present a steep drop in temperature (of over 4 C) which is the result of demagnetization (since the magnet had been turned on shortly before). The magnet was turned on after approximately 10 s econds. At that point the inlet liquid temperature as expected began increasing. After approximately another 10 seconds the effect of magnetic heating began beco ming evident. It takes roughly 20 seconds for the magnet to reach its full magnetic fi eld capacity of ~1.2 T. As observed from the same plot, in this case, at the ideal starting temperature of -10 C, the heating 30

PAGE 40

process led to a temperature change of about 5 C (from -4C to +1C). As well see in the next section, this was improved by the us e of a Silicon wafer exploiting its heat conducting properties. 4.2 Silicon Wafer Testing The second part of the experiment was conducted on a real Silicon wafer which is an excellent thermal conductor. Both the anodic bonding version (which presented a Glass wafer to Silicon wafer combination mounted on top of the GdSiGe block) and the fusion bonding one (Silicon wa fer to Silicon wafer mounted on top of GdSiGe block) were tested. These two so lutions were expected to produce better results than the ones obtained by the Plex iglas wafer because Silicon is a good heat conductor. In order to prevent excessive pres sure which ultimately could cause leaks, a lower flow rate of 33.3 L/s was chosen. This would allow a testing procedure of about 2 minutes and 30 seconds to take place. First, liquid nitrogen was used to obtain the desired initial temperature conditions Then, once the system temperature had stabilized, the mixture of water and antifreeze would flow through the system for about 10 seconds in order to eliminate a ll residual liquid left into the tubes and microchannels whose temperature would be affected by the overall chamber initial conditions. This would ensure that any new change in temperature in the system should be attributed only a nd exclusively to the varyi ng magnetic field and not to other causes. At approximately 10 seconds, the magnet was turned on to reach its maximum output ( ~1.2 T ) in about 20 s econds, while liquid still flew through the system at a constant rate. Once reached its maximum output capability (indicated by the controller reaching 42 A), the magnet wa s left on for 10 seconds for magnetization and then off /demagnetization for the rest of the time since the cooling effect was the ultimate interest in the expe riment. Another important aspect to take into account is 31

PAGE 41

that the GdSiGe block is sealed and placed far from where the liquid nitrogen was poured, therefore any change in temperature following the initial temperature system stability must be only due to the changing magnetic field. 4.2.1 Results for Silicon-to-Glass The testing of Si-to-Glass combinat ion was analyzed in several sessions. Beginning at a temperature of about 5 C, the first testi ng session was performed and the results are presented in Figure 4.3. The plot of temperature vs. time shows the response of the sensors on liquid flowing in to a Si-to-Glass wafer combination. Si-to-Glass Microcooler Results at Initial Temp. = +5-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 17131925313743495561677379859197103109115121127133139145151 Time (s)Temperature Change (C) MCE ( C ) Figure 4.3 Si-to-Glass Microcooler at Initial Temp. of +5 C As it is observed from the figure, a sl ight cooling effect is noticeable. In particular, as the magnet was turned on and maximum magnetic field capacity reached after approximately 20 seconds, a heating burst is evident on the excel 32

PAGE 42

graph (over 4 C). This is because GdSiGe is not a continuous heat source but an instantaneous one. However, because far fr om the ideal initial condition temperature of -11 C (change of entropy peak) as seen before, no actual cooling effect is provided once the magnet was turned off (demagnetization) but only an attenuation of the heating effect. This effect is strictly due to demagnetizati on as the process was started once the initial temperature ha d stabilized to approximately 5 C (and no more liquid nitrogen was provided during the process), th erefore, if anything, the overall chamber temperature condition could slightly increas e with time but definitely not decrease except because of the MCE caused by the magnetic field applied on the GdSiGe block. As the experiment was repeated, with initial temperature conditions lowered to about 3 C and the same exact procedure t ook place, a further decrease in heating effect during demagnetization was recorder as presented in Figure 4.4. However, this change in fluid temperature does not corres pond to a true cooling effect yet either but more to a strong weakening of the h eating moment. That again can only be attributed to the demagnetization process since no more liquid nitrogen was added during the process. 33

PAGE 43

Si-to-Glass Microcooler Results at Initial Temp. = +3-0.5 0 0.5 1 1.5 2 2.5 3 17131925313743495561677379859197103109115121127133139145151 Time (s)Temperature Change (C) Series2 Figure 4.4 Si-to-Glass Microcooler at Initial Temp. of +3 C Continuing in the experiment procedur e, initial system temperature condition was lowered to about 1 C. After a usual heating peak appearing at around 20 seconds into the experiment (magnet at full capacity), the magnet was turned off as in the previous case. However, this time a slight true cooling effect took place with almost a 1 change (at outlet) below the fluid temperature meas ured at the microcooler inlet. Figure 4.5 presents the results. 34

PAGE 44

Si-to-Glass Microcooler Results at Initial Temp. = +1-1.5 -1 -0.5 0 0.5 1 1.5 2 17131925313743495561677379859197103109115121127133139145151 Time (s)Temperature Change (C) Series2 Figure 4.5 Si-to-Glass Microcooler at Initial Temp. of +1 C Proceeding in the experiment at ini tial temperature conditions below 0 C (273 K), the MCE provided by the GdSiGe block combined with the heat conduction properties of the Silicon wafer became more evident and result in a larger temperature difference span. More precisely, an increas ing cooling effect leading to a maximum temperature difference span of almost 2.5 C was obtained at initial temperature conditions of -3 C as presented in Figure 4.6. 35

PAGE 45

Si-to-Glass Microcooler at Initial Temp.= -3-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 17131925313743495561677379859197103109115121127133139145151 Time (s)Temperature Change (C) Series2 Figure 4.6 Si-to-Glass Microcooler at Initial Temp. of -3 C Lower initial temperatures (closer to the change of en tropy peak value) lead to even better results. Specifica lly, the microcooler led to coo ling effects of over 5 C at initial conditions of approximate ly -4 C, almost 11 C at initial conditions of about 8 C and 12 C at initial conditions of -11 C. Figures 4.6, 4.7 and 4.8 (below) document the results. 36

PAGE 46

Si-to-Glass Microcooler at Initial Temp. = -4-7 -6 -5 -4 -3 -2 -1 0 1 2 3 17131925313743495561677379859197103109115121127133139145151157 Time (s)Temperature Change (C) Series2 Figure 4.7 Si-to-Glass Microcooler at Initial Temp. of -4 C Si-to-Glass Microcooler at Initial Temp. = -8-12 -10 -8 -6 -4 -2 0 2 4 6 17131925313743495561677379859197103109115121127133139145 Time (s)Temperature Change (C) Series2 Figure 4.8 Si-to-Glass Microcooler at Initial Temp. of -8 C 37

PAGE 47

Si-to-Glass Microcooler at Initial Temp. = -11-14 -12 -10 -8 -6 -4 -2 0 2 4 6 17131925313743495561677379859197103109115121127133139145 Time (s)Temperature Change (C) Series2 Figure 4.9 Si-to-Glass Microcooler at Initial Temp. of -11 C 4.2.1.1 Si-to-Glass Testing Changing Flow Rate and Interval Time Further testing was conducted in the Si -to-Glass combination to analyze the effect of flow rate as well as analyzing the role played by the time interval between each cycle. Just like the electro-magnet doesnt instantaneously charge to full magnetic field capacity in the moment it is switched on, once turned off it takes over a minute for it to completely die out. If this standard time is not completely elapsed then the MCE response both in heating and cooling effect is not as efficient as in the previous scenario. Figure 4.10 shows the temp erature sensors response versus time in this scenario at an initial system temperature of ~ -11 C and at the same flow rate used before. 38

PAGE 48

Si-to-Glass Microcooler at Initial Temp = -11-3 -2 -1 0 1 2 3 4 16111621263136414651566166717681869196101106111116121126 Time (s)Temperature Change (C) Adj MCE ( C ) Figure 4.10 Si-to-Glass Microcooler with weak effect Figure 4.11 shows the effect more clos ely. The magnet was initially turned on for magnetization and then off for demagneti zation. It was then again turned on for magnetization without the necessary standard time elapsed in between and as a result the heating effect was just about half the initial one. 39

PAGE 49

Si-to-Glass Microcooler at Initial Temp. = -11-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 1357911131517192123252729313335373941434547495153555759616365 Time (s)Temperature Change (C) Series2 Figure 4.11 Si-to-Glass Microcooler when magnet has not died out Flow rate was both doubled and halved for previous experiment settings without significant chan ges in the heating or cooling effect. This is most likely due to working at such small scale. As previously described the MEMS system enhances the surface area to volume ratio, improving the he at transfer between the fluid and the ferromagnetic material and this corresponds to non detectable temperature change within a change in flow rate. A normal size system like the one seen in Chapter 2 would account for changes in liquid flow ra te given its much lower surface area to volume ratio. 4.2.2 Results for Silicon-to-Silicon A final experiment was conducted us ing the Si-to-Si wafer combination obtained through fusion bonding. A similar pr ocedure was followed. However, this 40

PAGE 50

time the bonding didnt hold for long and a pproximately after 25 seconds into the testing session there was a leaking between the wafers. Figure 4.10 illustrates the result. Real Wafer Testing with GdSiGe block at Initial Temp = -1-15 -10 -5 0 5 10 135791113151719212325272931333537394143454749515355575961636567 Time (s)Temperature Change (C) Cooling Effect Figure 4.12 Si-to-Si Microcooler at Initial Temp. of -1 C Before the leaking took place, the he ating process seemed to be greatly improved compared to the Glass-to-Si wafer combination even if only at an initial temperature of -1 C (pretty far from the p eak of entropy change). This is due to the fact that Glass is not as good of a heating conductor as Silicon. This would predict even better results in the cooling effect and this result opens up exciting avenues for future research. 4.3 Discussion and Future Work Three different testing procedures were undertaken to an alyze the cooling properties of the designed microcooler. The first one involved a Plexiglas wafer from 41

PAGE 51

which only the magnetocaloric properties of GdSiGe were exploited. In the following testing sessions magnetocaloric changes in the fluid were recorded for different initial system temperatures where a Glass-to-Si combination was used. Exploiting the peak of entropy of GdSiGe a remarkable cooling effect of about 12 C was obtained. Even better results are suggested by the third experiment where Si-to-Si combination was adopted, although the process couldnt be fu lly completed because of leaking in the bonding between etched wafer and bare wafer. The results are encouraging and sugge st that a refrigerator could be constructed using two such microcoolers (magnetocaloric beds) with two heat exchangers, a pump and several valves as shown in Fig. 4.11 [ 16]. The microcoolers are analogous to the expansion stage and compression stage of a gas refrigeration system. Figure 4.13 Configuration of magnetic refrigerator 42

PAGE 52

CHAPTER 5 CONCLUSION In this thesis, a MEMS microcooler based on magnetocaloric effect was presented and evaluated. This chapter summari zes the contributions of the thesis and places MEMS microcoolers in the gene ral context of innovation for magnetic refrigeration. 5.1 Contributions MEMS microcoolers are a significant advance in applications of the magnetocaloric effect (MCE) and magnetic refrigeration. The MEMS microcooler is the first method to effectively applying the MCE of Gd based compounds at relatively low magnetic fields. Testing and results in Chapter 4 showed that this approach is indeed powerful, outperforming traditional ma gnetic refrigeration that is based on the use of superconducting magnets. 5.2 Conclusion The exploitation of the magnetocaloric effect, the relative small size of the device, and initial temperature conditions of the system combine to produce a novel methodology for the heating and cooling of fluids without the use of volatile, environmentally hazardous fluids. Not only is this method environmentally friendly, but it supports continual innova tion. Thus, this thesis is a first step in optimizing the study of low-energy requiremen t solutions to the need for liquefaction of hydrogen and household refrigeration. 43

PAGE 53

REFERENCES [1] E. Cartlidge, Attractive advance towards magnetic refrigerator, Physics World Jan. 9, 2002 [2] V.K. Pecharsky, and K.A. Gschneidner, Jr., Magnetocaloric effect and magnetic refrigeration, Journal of Magnetism and Magnetic Materials 200 (1999) 44-56 [3] S.N. Sambandam, S. Bhansali, and V.R. Bhethanabotla, Study on magnetocaloric GdSiGe thin films for microcooling applic ations, TMS Annual Meeting, Charlotte, NC, March 14-18, 2004 [4] S. C. Kim, B. Bethala, S. Ghirlanda, S. Sambandam, and S. Bhansali, Design and Fabrication of a Magnetocaloric Microcooler, ASME International Mechanical Engineering Congress and Exposition IMECE 2005-82720, Orlando, FL, 2005 [5] B. Bethala, Bulk Silicon Based Temperature Sensor, Masters Thesis, University of South Florida, Department of Electrical Engineering, October 31, 2005 [6] S.C. Collins, F.J. Zimmermann, Phys. Rev. 90 (1953) 991 [7] C.V. Heer, C.B. Barnes, J.C. Daunt, Rev. Sci. Instr. 25 (1954) 1088 [8] G.V. Brown, J. Appl. Phys. 47 (1975) 3673 [9] W.A. Steyert, J. Appl. Phys. 49 (1978) 1216 [10] J.A. Barclay, W.A. Steyert, U.S. Patent No. 4,332,135, June 1, 1982 [11] J.A. Barclay, U.S. Pa tent No. 4,408,463, October 11, 1983 [12] C. Zimm, A. Jastrab, A. Sternberg, V. Pecharsky, K. Gschneidner, Jr., M. Osborne, and I. Anderson, D escription and performance of a near-room temperature magnetic refrigerator, Advances in Cryogenic Engineering Vol. 43 [13] Pecharsky, V.K., and Gschneidner Jr K.A., 1997, Giant Magnetocaloric Effect in Gd5(Si2Ge2), Physical Review Letters 78, No. 23, pp. 4494-4497 [14] Tegus, O., Bruck, E.H., Zhang, L., Dagul a, W., Buschow, K.H.J., and Boer, F.R. de, 2002, Magnetic-phase Transitions and Magnetocaloric Effects, Physics B (319) 174-192 [15] http://www2.polito.it/ricer ca/thin-film/Activity/MEMS/anodic_bonding.htm 44

PAGE 54

[16] Rahman, M.M., and Rosario, L., 2004, Thermodynamic Analysis of Magnetic Refrigerators, Proceedings 2004 ASME International Mechanical Engineering Congress and Exposition Vol. 3, Anaheim, California 45

PAGE 55

BIBLIOGRAPHY X. Bohigas, E. Molins, A. Roig, J. Te jada, and X.X. Zhang, Room-Temperature Magnetic Refrigerator Using Permanent Magnets, IEEE Transactions on Magnetics, Vol. 36, No. 3, May 2000 P. Clot, D. Viallet, F. Allab, and A. Kedous-Lebouc, Member, IEEE J.M. Fournier, and J.P. Yonnet, A Magnet-Based Device for Active Magnetic Regenerative Refrigeration, IEEE Transactions on Magnetics, Vol. 39, No. 5, September 2003 A.J. DeGregoria, L.J. Feuling, J.F. Laatach, J.R. Rowe, J.R. Trueblood, and A.A. Wang, Test Results of an Active Magnetic Regenerative Refrigerator, Advances in Cryogenic Engineerin, Vol. 37, Part B S.J. Lee, J.M. Kenkel, V.K. Pecharsky, and D.C. Jiles, Permanent Magnet Array for the Magnetic Refrigerator, Journal of Applied Physics, Vol. 91, No. 10, 15 May 2002 P. Shirron, E. Canavan, M. DiPirro, M. J ackson, T. King, J. Panek, and J. Tuttle, A Compact, High-Performance Continuous Magne tic Refrigerator for Space Missions, Cryogenics 41 (2002) 789-795 46