close

Вход

Забыли?

вход по аккаунту

?

A hybrid high critical transition temperature superconductor filter/low-noise amplifier microwave integrated circuit

код для вставкиСкачать
lu g
ly ^
|^ 6
M I C R O C O P Y R E S O L U T I O N TEST CHART
\ A H O \ A 4 HtiW| Alt(H STANDARDS
M AM' AHP ML H Af NCT MATERIAL 10UM
,AN1.' a'nl ISO I f ST CHART No ? ’
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1+1
National Library
of Canada
Bibliotheque nationale
du Canada
Acquisitions and
Bibliographic Services Branch
Direction d e s acquisition:: et
d e s services bibliographiques
3 95 W ellington Street
O ttaw a, O ntario
K 1A 0N 4
395. rue W ellington
O ttaw a (Ontario)
K 1A 0IM4
NOTICE
AVIS
The quality of this microform is
heavily dependent upon the
quality of the original thesis
submitted
for
microfilming.
Every effort has been made to
ensure the highest quality of
reproduction possible.
La qualite de cette microforme
depend grandement de la qualite
de la these soumise au
microfilmage. Nous avons tout
fait pour assurer une qualite
sup6rieure de reproduction.
If pages are missing, contact the
university which granted the
degree.
S’il manque des pages, veuillez
communiquer avec I’universite
qui a confere le grade.
Some pages may have indistinct
print especially if the original
pages were typed with a poor
typewriter ribbon or if the
university sent us an inferior
photocopy.
La qualite d’impression de
certaines pages peut laisser a
desirer, surtout si les pages
originates
ont
ete
dactylographiees a I’aide d’un
ruban use ou si I’universite nous
a fait parvenir une photocopie de
qualite inferieure.
Reproduction in full or in part of
this microform is governed by
the Canadian Copyright Act,
R.S.C. 1970, c. C-30, and
subsequent amendments.
La reproduction, meme partielle,
de cette microforme est soumise
a la Loi canadienne sur le droit
d ’auteur, SRC 1970, c. C-30, et
ses amendements subsequents.
Canada
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A Hybrid High-Tc Superconductor Filter/Low-Noise Amplifier
Microwave Integrated Circuit
D an iel J. D rolet
A thesis subm itted to the 1.
!ty of G rad u ate Studies an d Research in partial
fulfillm ent of the requirem ents for the d eg ree of
M aster o f Electrical E n g in eerin g
O ttaw a-C arleton In stitu te o f E lectrical E n g in eerin g
D e p artm en t of E lectronics
Faculty o f E n g in eerin g
C arleton U n iv ersity
O ttaw a, C an ad a
M arch 1994
co C o p y rig h t
D.J. D ro let 1994
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
National Library
of Canada
Bibliotheque nationals
du Canada
Acquisitions and
Bibliographic Services Branch
Direction d e s acquisitions et
d e s services bibliographiques
395 W ellington Street
Ottaw a, O ntario
K1A 0N4
395, rue W ellington
O ttaw a (Ontario)
K 1A 0N 4
I
ft** Nt'tic
*•
The author has granted an
irrevocable non-exclusive licence
allowing the National Library of
Canada to reproduce, loan,
distribute or sell copies of
his/her thesis by any means and
in any form or format, making
this thesis available to interested
persons.
L’auteur a accorde une licence
irrevocable et non exclusive
permettant a la Bibliotheque
nationale
du
Canada
de
reproduire, preter, distribuer ou
vendre des copies de sa these
de quelque maniere et sous
quelque forme que ce soit pour
mettre des exemplaires de cette
these a la disposition des
persorines interessees.
The author retains ownership of
the copyright in his/her thesis.
Neither the thesis nor substantial
extracts from it may be printed or
otherwise reproduced without
his/her permission.
L’auteur conserve la propriete du
droit d’auteur qui protege sa
these. Ni la these ni des extraits
substantiels de celle-ci ne
doivent etre
imprimes ou
autrement reproduits sans son
autorisation.
ISBN
0 -3 1 5 -9 2 9 9 7 -9
Canada
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Name
Dissertation Abstracts International is arranged by broad, general subject categories Please select the one subject which most
neairly describes the content of your dissertation Enter the corresponding four digit code in the spaces provided,
SUBJECT C O D t
SUBJECT TERM
3 UMT
Subject Categories
THI HUMANITIKS AND SOCIAL SCIENCES
COMMUNICATIONS AND THE ARTS
Arc foiHirture
A rt H istory
f tntifTKJ
O cjno;
f imt A rts
JnforriMjIion ,j f ir*nrp
Jourfifltivrn
li b r a r y V
M rjiS C o m m u m c u h o n s
M u ‘, ir
V x x x h ( om m unK cition
ffwyitpr
0729
0*377
0900
0 3 /8
0357
07?J
03V 1
0399
0708
04 I 3
0459
0465
EDUCATION
Or*fM*r(il
A dm irnstK ition
A dult o n d C o fitin m n g
Aa»t< ullurul
Aft
0515
051 4
0516
0.01 7
0773
B tliti< |u o l f i n d M u l t i c u l t u r a l
B u s in e s s
C o m m u n ity G Jk?q«*
( u r n r u lu m f in d I n s tiu H io n
L u r ly th .M h o o c i
F U j m p f it a r y
f m 'i n c r *
G u i c k m u * <J»ul C o u n s r l i n y
H e a lth
H ig h e r
H u t o r y <>1
H w te * t< OIKMTItE',
In d u s tria l
t a n u u f K j r u n d ttte*rtiture*
M c jm e m u ttc s
0787
M usk
P h ilo s o p h y o f
H ty s K d l
0688
0
0
0
0
0
0
275
/7 7
518
524
?7 7
5 IV
0680
0745
0520
02*8
0 5 ? 1
(5 / 7 9
0280
052?
0 9 9 ft
0573
Psychology
R eading
Religious
Sciences
Secondary
Social Sciences
Sociology of
Special
Teacher Teaming
Technology
Tests a n d M e a su re m e n ts
Vocational
0525
0535
0527
0714
0533
0534
0340
0529
0530
0710
0288
0747
LANGUAGE, LITERATURE AMD
LINGUISTICS
la n g u a g e
G e n eral
Arw ient
linguistics
M odern
literatu re
G e n eral
C lassical
C om parative
M edieval
M odern
African
A m erican
A stan
( cjtiadujn (English)
( ancidian (French)
English
G erm anic
Latin A m e rta n
M iddle . t stern
R o m an te
Slavic a n d East E u ropean
0679
0289
0290
0291
0401
0294
0295
0297
0298
0316
0591
0305
0352
0355
0593
031 1
0312
0315
0313
03 i 4
PHILOSOPHY, RELIGION AND
THEOLOGY
P M usophy
Religion
G enerol
Biblical Studies
C lergy
History of
Philosophy of
Theology
0422
0318
0321
0319
0320
0322
0469
S0G AL SCIENCES
A m erican Studies
A nthropology
A rchoeofogy
Cultural
Physical
Business A dm inistration
G en eral
A ccounting
Banking
M an ag em en t
M arketing
C a n a d ia n Studies
Economics
G e n eral
A gricultural
C om m erce Business
Finance
History
Labor
Theory
Folklore
G e o g ra p h y
G erontolcxjy
H isto'y
G e n eral
0323
0324
0326
0327
0310
0272
0770
0454
0338
0385
0501
0503
0505
0508
0509
0510
0511
0358
0366
0351
A ncient
M edieval
M odern
Slade
A frican
A sia, A ustralia a n d O c e a n ia
C a n a d ia n
E uropean
Latin A m erican
M iddle Eastern
U nited States
History of Science
Law
Political Science
G e n eral
International ta w a n d
Relations
Public A dm inistration
Recreation
Social W ork
Sociology
G e n eral
Crim inology a n d Penolog,
D em ography
Ethnic a n d R o ao t Studies
Individual c.nd Family
Studies
Industrial a n d Labor
Relations
Public a n d Social W elfa re
Social Structure a n d
D evelopm ent
Theory a n d M eth o d .
T ransportation
U rban a n d Regional Planning
W om en s Studies
0579
0581
0582
0328
03 3 1
0332
0334
0335
0336
0333
0337
0585
0398
0615
0616
0617
0814
0452
0626
0627
0938
0631
0628
0629
0630
0700
0344
0709
0999
0453
0578
THE SCIENCES AND ENGINEERING
BIOLOGICAL SCIENCES
A y iK u lfu te
al
A g ro n o m y
A m n x iJ C u lfu ie a n d
N u tr itio n
Ammo! Pathology
fcxxf Si lem e a n d
Tr\ hnolcxiv
Forestry a n d Wildlife
Plant ( ultuft*
H a n t Pathology
Plant Physiolcxjv
R ang e M anagem ent
W o o d 1 t\h n o io i|y
fholcxjv
C•enem l
Anatomy
B'OAtljfjstn S
Botany
Coll
f i ology
f a to iu a lc x tv
I'lH-tk
liUirnilvKJV
rlugv
Mi» fnbiukX
'biohxJv
uliii
N eoiosi t«'iH u
l,\tM niX|H«phv
Phv sio!tX|V
Radtofion
VrtCMOivv Si <i*i 'i i*
/o o lo g v
BlLTf'I'S lit S
cWmkmoI
M r'du ill
04/i
0785
(14/ S
0 4 /6
0359
0 4 7B
(>4/9
0480
081 /
0 746
HEALTH AND ENVIRONMENTAL
SCIENCES
0306
Environmental Sciences
Health Sciences
0 777
028/
0308
030V
l>379
0329
0153
03 6v
C’ v i
0410
030’
031 ’
0416
(143 1
o fl; i
( V "8
04";
0 \* 6
o *<y0
EARTH SCIENCES
flilHJKNX henintry
li.KXvhemiifrv
G eo d esy
G eology
Geophysic s
Hydrology
M ineralogy
Pcjlt*ol>at{|!)y
Poleoecoiogy
Pcnrx.»ntology
P o h w o o lo g y
Palynolixjy
Physu ol G eo g ra p h y
Pftysicul O c e a n o g ia o h y
04 2S
p9s?<s
0370
0372
0373
0388
0411
0345
042o
0418
0985
04?7
0368
0415
0768
G e n e t ul
0566
Audioioqy
C hem otherapy
Dentistry
Educatrcri>
0300
0992
0567
0350
0^69
0758
0982
0564
034/
0569
0570
0380
H c 's p tfo l M c m o q e i n e n t
Hum an Dev< G pm ent
Immunology
M e d 'o n e a n d Su ig ety
M ental H ealth
N ursing
N utation
Dbstefrics a n d G ynei olixjy
C k iu p u tio n u l H ealth nncf
T herapy
O phtfuilm oluqy
Pathology
F h a n ro c o lo g y
Phar macy
Physical T herapy
Pubht H ealth
Radiology
Recreation
0354
0381
0571
0419
0572
0382
0573
0 5 /4
0575
Speech Path >togy
Toxtcology
H om e Economics
0460
0383
0386
PHYSICAL SCIENCES
P u re S cie n c e *
Chemistry
G eneral
A gricultural
Analytical
Biochemistry
Inorganic
N uclear
O rganic
Pharm aceutical
Physical
Polymer
Radiation
M athem atics
Physics
G en eral
Acoustics
Astronom y a n d
A strophysics
Atmospheric Science
Atomic
Electronics a n d Electricity
Elementary Particles a n d
High Energy
Flu'd a n d Plasm a
M olecular
N uclea.
O ptics
R adiation
Solid State
Statistics
0485
0 /4 9
0486
0487
0488
0738
0490
049?
0494
0495
0754
0405
0605
0986
0606
0608
0748
0607
0798
0759
0609
0610
0752
0756
0611
0463
A p p lie d S c ie n c e s
A pplied M echanics
C om puter Science
0346
0984
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Engineering
G e n eral
A ero sp ace
Agricultural
Automotive
Biomedical
C hem ical
Civil
Electronics a n d Electrical
H eat a n d Therm odynam ics
H ydraulic
Industrial
M arine
M aterials Science
M echanical
M etallurgy
M ining
N uclear
P ackaging
Petroleum
S anitary a n d M unicipal
System Science
G eotechnology
O p e ra tio n s R esearch
Plastics Technology
Textile Technology
0537
0538
0539
0540
0541
0542
0543
0544
0348
0545
0546
0547
0794
0548
0743
05 5 1
0552
0549
0765
0554
0790
0428
0796
0795
0994
PSYCHOLOGY
G en eral
Behavioral
Clinical
D evelopm ental
Experim ental
Industrial
P erso n ality
Physiological
Psychobiology
Psychometr ics
Social
0621
0384
0622
0620
0623
0624
0625
0989
0349
0632
04 5 1
®
Carleton University
Ottawa, Canada
K1S 5J7
Thesis contains black & white
illustrations which whei. microfilmed
may lose their significanceThe
hardcopy of the thesis is available
upon
request
from
Carleton
University Library.
The University Library
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
,,wr
The* undersigned hereby recom m end to the Faculty of G rad u ate S tudies and
Research acceptance of the thesis,
A Hybrid High-Tc Superconductor Filter/Low-Noise Amplifier
Microwave Integrated Circuit
subm itted by
Daniel J. Drolet, B. Eng.
in partial fulfillm ent of the requirem ents for the degree of
M aster of Electrical Engineering
Thesis Supervisor
Thesis Co-Supervisor
y
i c
j ^
j h
C hair, D epartm ent of Electronics
C arleton U niversity
M arch 1994
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
The increasing d e m a n d s for higher perform ance electronic circuits
in b o th d ig ital and analog a p p lica tio n s is lea d in g to the d e v e lo p m e n t and
im p lem entation of alternative technologies. O ne area of interest is the operation
of circu its a n d su b sy stem s at cryogenic tem p era tu re s to im p ro v e th eir sp eed ,
re liab ility , and p o w e r h a n d lin g re q u ire m e n ts. S e m ic o n d u c to r d ev ices for
m icro w av e applicatio n s such as the high electron m obility tran sisto r (IlhM T )
e x h ib it s u p e rio r p e rfo rm a n c e a t lo w e r te m p e ra tu re s d u e to im p ro v e d
transm ission gain, n o is ; figure, and frequency of operation.
S uperconductivity is a prom ising altern ativ e for im p ro v in g circuit
p e rfo rm a n c e . T he rec e n t d isc o v ery of h ig h te m p e ra tu re su p e rc o n d u c to rs
(HTSCs) w hich operate above the tem p eratu re of liquid nitrogen (77K) has m ade
its ap plication in analog and digital electronics m uch m ore feasible. Previous low
te m p e ra tu re su p e rc o n d u c to rs , re q u irin g e x p e n siv e co o lin g sy ste m s, h av e
d e m o n stra te d su p e rio r p erform ance
conventional circuits in term s of speed,
se n sitiv ity , a n d sw itc h in g cap ab ility . T he ex trem ely low loss p ro p e rtie s of
H TSCs, alo n g w ith the sim p ler refrigeration sy stem s necessary for o p eratio n ,
allo w h ig h Q re so n a to rs to be u se d in m ic ro w a v e c o m p o n e n ts su ch as
n a rro w b an d filters and low phase-noise oscillators.
T his th esis p re se n ts the d e sig n an d p e rfo rm a n c e o f a h y b rid
s e m ic o n d u c to i/s u p e rc o n d u c to r K -band filte r/a m p lifie r m icro w av e integ rated
circuit. T he circuit com bines the high frequer. y, high perform ance a d v an tag es of
a cryogenically cooled G aA s HEM T w ith the low loss, high Q p ro p e rtie s of a
n a rro w b an d HTSC filter and m atching netw orks.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGEMENTS
I w ould like to express m y sincere g ra titu d e to Prof. Jim S. W ight,
m y thesis supervisor, for his continuous encouragem ent a n d su p p o rt th ro u g h o u t
m y g ra d u a te studies at C arleton University.
I w o u ld also like to th an k Dr. M alcolm G. S tu b b s a n d M r. Rene
UouvifJe of the C om m unications R esearch C en tre (CRC) for p ro v id in g m e th e
o p p o rtu n ity to carry o u t this research.
Special th a n k s to D r. Jeffrey W. S m u k , c u rre n tly a t H ittite
M ic ro w a v e C o rp o ra tio n
in
M a ss a c h u se tts, fo r h is v a lu a b le te c h n ic a l
co n trib u tio n s, p a rtic u la rly in the early stag es o f th is w o rk . I also g rate fu lly
ack n o w led g e the assistance of the staff a t CRC, especially Ms. Sara M eszaros,
n o w w ith Bell N o rth e rn R esearch, for h e r c o n sta n t g u id a n c e a n d s u p p o rt
th ro u g h o u t this w ork.
Finally, I w ish to th an k m y p aren ts for th eir tolerance a n d su p p o rt
th ro u g h o u t m y academ ic career.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
TITLE PAGE........................................................................................................................i
ACCEPTANCE SH EET....................................................................................................ii
ABSTRACT.........................................................................................................................iii
ACKNOW LEDGEM ENTS.............................................................................................. iv
TABLE OF C O N T E N T S...................................................................................................v
LIST OF TABLES............................................................................................................... \iii
LIST OF FIGURES................................................................................................... ........iv
CHAPTER 1: INTRODUCTION
1.1 O v e rv ie w ......................................................................................................... 1
1.2 Thesis Objectives and O rg an izatio n ..........................................................<>
CH A PTER
2:
S E M IC O N D U C T O R
AND
SU PE R C O N D U C T O R
TECHNOLOGIES: BACKGROUND
2.1 In tro d u c tio n ....................................................................................................B
2.2 M icrow ave Integrated Circuits (M IC s).....................................................9
2.2.1 MIC T echnology.............................................................................9
2.2.2 MICs at Cryogenic T em peratures............................................... 13
2.3 Theory of S u p erco n d u c to rs......................................................................... IB
2.3.1 Low T em perature S uperconductors...........................................19
2.3.2 H igh T em perature S uperconductors......................................... 27
2.4 Fabrication and Processing of H T SC s.......................................................30
2.5 Cryogenic M easurem ent T echnique.......................................................... 37
2.5.1 TRL C alibration P ro ced u re.......................................................... 37
2.5.2 C ryogenic A p p aratu s.....................................................................39
2.5.3 C ryogenic M easurem ent T echnique.......................................... 41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
vi
Cl IA ITER 3: CRYOGENIC AMPLIFIER DESIGN A N D EXPERIMENTAL
RESULTS
3.1 In tro d u c tio n .................................................................................................... 46
3.2 Device C haracterization at 77K ...................................................................47
3.3 A m plifier Design P rocedure........................................................................ 54
3.4 M easurem ents and R esu lts.......................................................................... 59
3.4.1 H TSC /PH EM T A m plifier A' R e su lts....................................... 61
3.4.2 H FSC/HEM T A m plifier B’ R e su lts.......................................... 63
3.4.3 2nd Iteration A m plifier ’C ............................................................6 8
3.4.4 CA D Softw are Analysis of HTSC C irc u its ...............................70
3.5 S u m m a ry ..............................................................................................
72
CHAPTER 4: INTEGRATED SUPERCONDUCTING FILTER/AM PLIFIER
PERFORMANCE
4.1 In tro d u c tio n .....................................................................................................74
4.2 HTSC Filter Design P ro c ed u re.................................................................... 75
4.3 Filter M easurem ents and R esults................................................................80
4.3.1 19 G H z HTSC F ilte r......................................................................80
4.3.2 20 G H z HTSC F ilte r....................................................................... 85
4.3.3 A nalysis of HTSC R esonator C oupling S ections.................... 87
4.4 H ybrid HTSC F ilter/A m plifier M IC ......................................................... 89
4.4.1 19 G H z HTSC F ilter/A m plifier M IC .......................................89
4.4.2 20 G H z HTSC F ilter/A m plifier M I C .......................................95
4.5 S u m m a ry ...............................................................................................
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
vii
CHAPTER 5: CONCLUSIONS
5.1 Sum m ary and D iscu ssio n ............................................................................ 100
5.2 Future W o rk .................................................................................................... 105
APPENDIX I H TSC /PH EM T AMPLIFIER A SCHEMATIC 1>1ACIRAM........ ION
APPENDIX II H TSC/H EM T AMPLIFIER 'C SCHEMATIC D1ACKAM ......... I(W
APPENDIX III TOUCHSTONE LISTING OF HTSC FILTER/LN A .................
110
REFERENCES..................................................................................................................... 112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
viii
LIST OF TABLES
I able 2.1 Properties of various su b strate m aterials for M IC s................................. 10
Table 2.2 Therm al conductivity im provem ent from 300K to 77K.......................... 18
Table 2.3 C om parison of properties for a LTSC and H T S C .................................... 30
Table 2.4 Properties of HTSC substrates at 77K and 10 G H z ................................. 35
Table 3.1 Element values for sm all-signal FHR10X HEM T m o d e l........................51
Table 3.2 C om parison of HEMT and PHEM T devices at 20 G H z .........................58
Table 3.3 C om parison of design characteristics for th ree am p lifiers.................... 59
Table 4.1 R esonator dim ensi' ms for the HTSC filte rs............................................... 78
Table 4.2 S-param eters at the centre frequency for the 19 G H z HTSC filte r.... 82
Table 4.3 S-param eters at the centre frequency for the 20 G H z HTSC filte r.... 85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 1ST OF FIGURES
Figure 2.1 V arious planar transm ission line configurations...........................
Figure 2.2 Resistivity of bulk gold versus tem p eratu re....................................
Figure 2.3 Resistivity of thin-film copper versus tem p era tu re .......................
Figure 2.4 Rs of copper vs. tem perature at various freq u en cies....................
Figure 2.3 Chronological history of superconductivitv ....................................
Figure 2.6 M eissner effect in su p e rco n d u c to rs.................................................
Figure 2.7 Relationship betw een Tc, Jc, and I Ic in su p erco n d u cto rs............
Figure 2.8 Two-Fluid m odel for su p e rc o n d u c to rs...........................................
Figure 2.9 C u r rent-voltage relationship for SIS and |osephson ju n ctio n s...
Figure 2.10 Equivalent circuit m odel for a Josephson ju n ctio n .....................
Figure 2.11 C rystal structure for YBCO superconductors...............................
Figure 2.12 HTSC thin-film deposition techniques..........................................
Figure 2.13 Perform ance of Rs for STI's TBCCO thin-film ..............................
Figure 2 14 TRL procedure and ideal S-param eters for each sta n d a rd .......
Figure 2.15 Cold head of the cryocooler a p p a ra tu s .........................................
Figure 2.16 Refrigeration system for cryocooler s e tu p ....................................
Figure 2.17 Photograph of 1 herm kon split-black test fixture........................
Figure 2.18 R em easurem ent of "through" standard after c alib ratio n ...........
Figure 2.19 M easurem ent of a 0.101" line to verify calibration ......................
Figure 3.1 Cross-section of the structure for a HEMT and PI IE M T .............
Figure 3.2 M obility vs tem perature for FET and HEMT channel slruc lures
Figure 3.3 Sm all-signal equivalent circuit m odel for a MESEET...................
Figure 3.4 S-param eters derived from a HEMT m odel at 300K and 77K ....
Figure 3.5 Effect of source inductance on HEMT p e rfo rm a n c e ....................
Figure 3.6 Photograph of assem bled H TSC /H EM T am p lifier......................
Figure 3.7 Photograph of H TSC /H EM T am plifier B' in test f ix tu r e ...........
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
X
I igure
1 .8
M easured vs. predicted response of am plifier 'A’ ..................................62
Figure 1.9 M easured vs. predicted response of am plifier B '................................... 64
Figure 1.10 Block diagram for noise figure m easurem ent se tu p ........................... 6 6
Figure 1.11 M easurem ent of "through" loss in K-cables and test fix tu re ............. 6 8
Figure
1 .1 2
M easured S-param cters for 2nd iteration am plifier
'C ...................... 69
Figure 1.11 Perform ance of input and o u tp u t m atching n e tw o rk s ..................... 70
f igure 1.14 Results of F1TSC am plifier m atching n e tw o rk s................................... 71
Figure 4.1 AutoCAD layout of 2 " STI w afer containing HTSC filte rs................. 79
Figure 4.2 Photograph of 20 G H z HTSC m icrostrip filter...................................... 79
Figure 4.1 M easured vs. sim ulated response for 19 G H z HTSC f ilte r ................ 81
Figure 4.4 Kinetic inductance effect of filter response over te m p e ra tu re ...........84
Figure 4.5 M easured vs. sim ulated response for 20 G H z HTSC f ille r ................ 8 6
Figure 4.6 C oupling of resonator sections for 19 G H z and 20 G H z filte rs........ 8 8
Figure 4.7 Photograph of the HTSC filter/am plifier 'A' M IC................................90
Figure 4.8 M easured vs. sim ulated response of HTSC filter/am p lifier A '....... 92
Figure 4.9 Noise figures of the HTSC filter/am plifier M IC s.................
94
Figure 4 .1 0 M easured vs. predicted response of HTSC filter/am p lifier *B'...... 96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
l
CHAPTER 1: INTRODUCTION
1.1 Overview
C ryogenic tem p eratu res have been used to im prove the speed,
reliability, and perform ance of both digital and analog electronic system s. Low
tem p eratu res im prove the speed of digital circuits bv enhancing the m obility of
solid state devices, and im proves reliability by reducing the effects of heating
caused by higher p o w er densities. In high-speed analog electronics, cryogenics
h a v e also been im p lem en ted to m inim ize the th erm al noise in low noise
m icrow ave receiver system s. Low tem peratures provide both an im provem ent in
conventional sem iconductor device operation, and perm it the im plem entation of
su p e rc o n d u c to r devices. M oreover, w ith recent advances in su p e rco n d u c to r
technology, in tegration of sem iconductors and su p erco n d u cto rs can p ro d u ce
hybrid circuits w ith higher perform ance than conventional circuits at 77K.
T he p e rfo rm a n c e d e m a n d s on c o n v e n tio n a l se m ic o n d u c to r
electronics are being p u sh ed to physical lim its that cannot be exceeded d u e to
fu n d a m e n ta l law s of n a tu re and the in trin sic p ro p e rtie s of m a te ria ls . ! 11
Sem iconductor technology offers a trade-off of high speed capability versus high
p o w e r d issip atio n in digital circuitry. The speed is lim ited by the sw itching
ab ility of th e tran sisto r, as well as th e len g th s of the in ternal w irin g and
in terconnections, w hile the p ow er dissip atio n lim its the size of digital logic
circuits.
Sem iconductor com ponents operated at low tem peratures exhibit a
n u m b er of im provem ents, such as h ig h er Q (i.e. quality factor) circuits d u e to
low er resistivity and reduced substrate loss. Increases in operatin g sp eed s by a
factor of tw o have been reported in CM OS circuits d u e to the enhanced carrier
m obility, at low er supply voltages.l2! The ETA-10 su percom puter is one exam ple
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of an ap p licatio n w hich uses liquid n itro g en to e x p lo it th e p e rfo rm an c e
a d v an tag es of CMOS devices a t 77K.!5! As c ir:u it topology reduces in size and
p o w er d ensities increase, cooling is required to m aintain circuit reliability by
elim inating heating problem s. These heating effects contribute to h ig h er failure
rates and sh o rte r lifetim es in solid sta te d ev ices. A c o n tro lle d o p e ra tin g
te m p e ra tu re e n v iro n m e n t w hich c ry o g e n ics p ro v id e s w ill re d u c e a n y
degradation in the perform ance of tem perature-sensitive com ponents.
In m icrow ave electronics, active sem ico n d u cto r devices such as
Hi Is and high electron m ob ility tra n sisto rs (H EM Ts) h a v e d e m o n s tra te d
significant perform ance im provem ents at cryogenic tem peratures. These devices
have yielded increased carrier m obilities and tran sco n d u ctan ce values a t 77K,
w hich pro v id es a higher frequency of operation w ell into the m illim eter-w ave
frequencies. The h igher transconductance and low er resistance resu ltin g from
low tem p era tu re operatio n also results in higher g ain an d lo w er noise figure.
E xtrem ely low noise levels are required for rad io a stro n o m y an d d e e p space
co m m u n ica tio n s a t m icrow ave and m illim eter-w av e frequencies, w h e re the
incom ing signal is m ixed w ith cosm ic and atm ospheric noise com ponents. The
therm al noise, which is directly d ep en d en t on tem perature, is inherently reduced
by o p eratin g in a cryogenic environm ent. Recently, cooled HEM T devices h av e
been capable of m eeting these noise requirem ents as a less expensive alternative
to traditional com ponents such as MASERS and param etric am plifiers.!4-5!
A nother significant a d v an tag e of u sin g cryogenic technolo g y in
electronics is the possible application of superconductors. Since the discovery of
su p e rc o n d u c tiv ity in 1911, c o n tin u o u s research h a s been u n d e rta k e n in to
d eveloping new superconductors w ith higher transition, or critical, tem p eratu res
(Tt ). The early superconductors w ere m etallic elem ents such as niobium an d lead
w hich req u ire d liquid helium (4K b oiling p o in t) as a cry o g en in o rd e r to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
su p erconduct. Because of their extrem ely low transition tem peratures, these
m aterials are referred to as low tem perature superconductors (ITSC's). Apart
from exhibiting extrem ely low losses, another significant contribution of I.TSC's
h as b een the Josephson junction, w hich p ro d u c e s high -sp eed sw itch in g
capabilities in digital electronics faster than any sem iconductor device. The high­
speed sw itching also occurs w ith very low pow er dissip atio n , hence circuit
density can be increased and chip size can be reduced. Superconducting w iring
and interconnections in digital system s provide an o th er significant ad van tag e
ovei conventional technology, in th at disp ersio n is negligible. Pulses are
therefore able to propagate along a transm ission line with m inimal distortion.
T he Josephson junction has also found a p p lic a tio n in high
perform ance analog electronics to produce voltage-controlled oscillators (VC’t'ts)
at m illim eter-w ave frequencies.H O th er a p p licatio n s inclu d e a Josephson
p a r a m e tr ic
a m p lifie rl7! w hich y ield s extrem ely low noise levels, and
Superconducting Q uantum Interference Devices (SQUIDs)l8! which act as highly
sen sitiv e m agnetom eters. S uperco n d u ctin g -in su latin g -su p erco n d u ctin g (SIS)
m ixers are replacing conventional technology in high perform ance m icrow ave
system s such as for radiom etry and radar applications.!4!
In 1987, h ig h te m p e ra tu re su p e rc o n d u c to rs (HTSCs) w ere
discovered w hich operate at liquid nitrogen (77K boiling point) tem peratures.
These m aterials are ceram ics that typically exhibit extrem ely poor conductivity
above th eir transition tem perature. Liquid nitrogen refrigeration system s are
m ore feasible than the bulky, hard-to-handle liquid helium coolers d u e to a 60:1
im p ro v em en t in the latent heat of vaporization. Also, a HTSC' device can be
cooled using liquid nitrogen at an annual cost of $35, as com pared to liquid
helium system s, w hich cost upw ards of $50,000 annually.I,(,l
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
O n e su itab le ap p licatio n for HTSCs is in low p o w e r p assiv e
m icro w av e dev ices such as high Q reso n ato rs a n d filters. T hese h ig h Q
com ponents have been im plem ented in plan ar transm ission line circuits to yield
com parable perform ance to conventional w aveguide and dielectric resonators, at
a significant reduction in sized11'12' 13! W hile the cost of the refrigeration system
m u st be considered, the im plem entation of p lan a r HTSC filters into m in iatu re
filter banks required for applications such as ra d a r an d electronic w arfare can
offset the cost and inconvenience of cooling the entire system .
The im p le m e n ta tio n of a cryogenic refrig era tio n sy stem m u st
justify the cost of cooling the com ponents. Refrigeration system s can eith er utilize
a sh o rt d uration, inexpensive open-cycle m ethod such as a liquid nitrogen dew ar,
or a long d u ra tio n closed-cycle m eth o d such a s th e S tirling cryocooler an d
C iffo rd -M cM ah o n te c h n iq u e s w h ich re q u ire ele ctrica l p o w e r.!14! S h a rin g
refrigeration system s am ong several analog an d dig ital system s can red u ce the
cost o f achieving low tem peratures. The use of sup erco n d u ctiv ity in d e e p space
applications is especially prom ising d u e to the inherently cooled e n v iro n m en t in
w hich only passive radiators are required.
It is preferable to use liquid nitro g en refrig eratio n system s o v er
liquid helium to cool sem iconductor circuits n o t only because of the en o rm o u s
cost difference, b u t also because little advan tag e is gained at tem p eratu res low er
than 77K. D opant freeze-out occurs in G aA s FETs a t tem p era tu re s ap p ro ach in g
the liquid helium boiling point, w hich severely d eg rad es the device perform ance.
A lso, resistiv itie s in c o n v en tio n a l m etals su c h a s c o p p e r a n d g o ld reach
anom alous lim its for thin-film transm ission lines a t tem p era tu re s slightly below
77K. An excep tio n is in ex trem ely low n o ise a p p lic a tio n s, su c h a s ra d io
astro nom y , w here tem p eratu res of 4K are necessary to considerably low’e r the
therm al noise in the receiver front end.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
S uperconductors exhibit certain lim itations w hich need to be
overcom e before they are able to achieve the w ide spread use of sem iconductor
devices. C ne of the m ain disadvantages of superconductor technology is the lack
of a suitable three-term inal device to produce gain. W hile the tw o-term inal
Josephson junction is a m ature device for LTSCs, HTSC devices are still in the
developm ental stages. Therefore, a prom ising alternative is the integration of
both sem iconductor and superconductor com ponents into high perform ance
hybrid m icrow ave circuits. These hybrid circuits can exploit the advantages ot
g ain -p ro d u cin g solid-state devices and high Q, low loss HTSC s w hich both
o p e ra te b etter a t liquid nitrogen tem peratures. Thin-film HTSC' processing
tech niques have not yet reached the m atu rity of sem iconductor processing
technology, an d th erefo re su p e rco n d u c to r circuits a re q u ite expensiv e to
fabricate.
Several exam ples of these hybrid sem ico n d u cto r/su p erco n d u cto r
m icrow ave integrated circuits (MICs) have been reported since the discovery of
HTSCs. A m illim eter-w ave receiver system has been proposed which utilizes
su p erconducting antennas and filters to achieve low losses and high Qs, and
sem iconductor devices for the am plifiers and analog signal processors.l,r,l An
analog signal processing system w as developed w hich uses superconducting
tap p ed delay lines to achieve an attenuation com petitive w ith surface acoustic
w ave (SAW) devices.!16! A hybrid se m ic o n d u c to r/su p e rc o n d u c to r K u-band
oscillator w as designed at the C om m unications Research C entre (CRC) using a
high-Q HTSC linear resonator and m onolithic MIC (MMIC) am plifiers.!I7J A
sim ila r o scillato r h y b rid circuit w as d ev elo p ed at D u p o n t for C -band
application.!18! A Ku-band low noise am plifier (LNA) w as designed for operation
at 77K u sin g HTSC m atching n etw o rk s and a G aA s HfiM T.I17! Also, an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
integrated HTSC filter/G aA s LNA has been developed for possible application
in high sensitivity m icrow ave receivers.t19!
W hile th ese e x am p le s illu s tra te th e a b ility to in te rc o n n e c t
su p e rco n d u c to r and sem iconductor devices into a circuit, recent p a p e rs h av e
re p o rte d th re e -te rm in a l d ev ices u sin g hig h -T c s u p e rc o n d u c tin g m ate ria l.
Exam ples of these hybrid devices include the su p erco n d u ctin g FET (SuFET)l10!,
and the Superconductor-B ase Sem iconductor-Isolated T ransistor (SUBSIT).f2°l
1.2 Thesis Objectives and Organization
T he objective of th is w o rk is to in v e stig a te th e p o te n tia l fo r
c o m b in in g se m ic o n d u c to r a n d su p e rc o n d u c to r tec h n o lo g ie s in to a h y b rid
m icrow ave integrated circuit. The benefits of in tegrating the tw o technologies to
p ro d u c e h ig h e r p e rfo rm a n c e c o m p o n e n ts is e m p h a siz e d . S pecifically , a
n arro w b an d HTSC filter is integrated a t th e in p u t o f a H T S C /G aA s am p lifier
cooled to 77K for o p eratio n at K -band. Such filte r/a m p lifie r com binatio n s a re
useful at the front en d of m icrow ave receivers to low er the noise figure w hile
rem o v in g an y in te rm o d u la tio n p ro d u c ts d u e to s p u rio u s sig n a ls o u tsid e th e
frequency band. This p articular circuit configuration w as chosen to d em o n strate
the im p ro v e d p e rfo rm an ce of a n active se m ic o n d u cto r d evice a t cry o g en ic
te m p era tu re s in a low -noise am plifier, an d also to illu strate th e a d v a n ta g e o f
HTSCs for low loss, n arro w b an d filters. V fu rth e r objective o f th is w o rk is to
in v estig ate th e reliability o f v ario u s com m ercial C A D so ftw a re p ack ag es in
analyzing HTSC passive structures.
C h a p te r 2 p ro v id es back g ro u n d inform ation on v a rio u s form s of
conventional MIC technology and the advantages of o p eratin g M ICs a t cryogenic
tem p eratu res. An overview of th e theory of su p erco n d u c tiv ity for LTSCs a n d
HTSC s is presented, including the Tw o-Fluid m odel, L ondon a n d BCS theories.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
HTSC thm -film processing techniques and circuit fabrication are exam ined.
Finally, the cryogenic m easurem ent technique used in this w ork is described,
including an overview of the cryocooler apparatus and thec ilibration procedure
used to perform S-param eter and noise figure m easurem ents.
C hapter 3 presents the design of a K-band HTSC/C.aAs amplifier
MIC at 77K. Firstly, the characterization of a HEMT device at 77K is described.
Based on this characterization, the amplifier w as designed using 11TSC' m atching
netw orks. M easurem ents and results of this am plifier, as well as a m odified
version, are then presented. The final section of the ch ap ter discusses C A P
softw are used in sim ulating the HTSC matching networks.
The d e v elo p m en t of n arro w b an d HTSC m icrostrip filters is
presented in C hapter 4. A description of the filter design procedure is followed
by m easurem ents and results, and com parisons with CAD sim ulations. The filter
w as then integrated w ith the am plifiers described in C hapter 3 to produce the
h y b rid se m ic o n d u c to r/su p e rc o n d u c to r filte r/a m p lifie r MIC. S -param eter
m easurem ents and noise figures are presented for tw o variations of the hybrid
circuit.
T he final c h a p te r su m m arize s the p erfo rm an ce benefits of
com bining sem iconductor and high tem perature superconductor technology. A
review of the accom plishm ents of this w ork is presented, along w ith possible
m odifications and im provem ents to the circuits. The application of HTSCs in
future hybrid circuits is explored, as well as possible im provem ents in the design
and m easurem ent of these com ponents at cryogenic tem peratures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
CHAPTER 2: SEMICONDUCTOR AND SUPERCONDUCTOR
TECHNOLOGIES: BACKGROUND
2.1 Introduction
T he in te g ra tio n of se m ic o n d u c to rs a n d s u p e rc o n d u c to rs in
electronics applications has perm itted circuit d esigners to exploit the benefits of
each technology. C om bining these tw o technologies m ay be carried o u t o n three
d ifferent levels. I1'’) The first level involves th e com bin atio n of sem ic o n d u cto r
ch ip s w ith su p erco n d u cto r chips o r interconnections. T he next level in volves
integrating sem iconductor an d su p erco n d u cto r devices on the sam e chip. A t the
highest level of integration, the tw o technologies are com bined w ith in the sam e
activ e device. This thesis focuses on th e first level, as this is th e sim p le st to
fabricate, in w hich sem iconductor active devices w ill be integ rated w ith HTSC
passive structures on separate chips.
In th is c h a p te r, an o v e rv ie w of s e m ic o n d u c to r m ic ro w a v e
in teg rated circuit (MIC) technology w ill b e p resen te d , in clu d in g th e a re a s of
p erform ance im provem ent by o p e ra tin g circuitry a t cryogenic tem p eratu res. A
com parison o f the a d v an tag es and d isad v an tag es of M ICs v e rsu s o th e r circuit
fabrication technologies is p resented. Follow ing th is discussion, a n ov erv iew of
the theory of su p erco n d u cto rs for LTSCs a n d HTSCs is described. The v ario u s
p ro cessin g techniques for HTSC thin-film s w ill be d isc u sse d , a s w ell as the
m icrow ave characterizatio n of these film s. The final section w ill p re s e n t the
cryogenic a p p aratu s used for S-param eter an d noise figure m easurem ents, along
w ith the calibration m ethods used.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2 Microwave Integrated Circuits (MICs)
2.2.1 MIC Technology
M icrow ave in te g rate d circuits have increasingly replaced
conventional microwave circuits using coaxial and w aveguide com ponents due
to the m ore stringent dem ands on size and w eight for applications such as
satellite comm unications, electronic warfare, and radar systems. MICs can also
yield higher reliability, better perform ance, and low er cost along w ith the
significant reduction in size as com pared to these conventional technologies. A
basic MIC consists of planar transm ission lines on a single low loss dielectric
su b strate w hich form distributed m atching elem ents and interconnect the
discrete devices. These discrete devices include both passive com ponents such as
resistors, capacitors, and inductors, as well as active sem iconductor devices such
as diodes and transistors. These com ponents are usually attached to the substrate
in chip form and are connected into the circuit using w ire bonds. This type of
circuit is referred to as a hybrid MIC.
An extension of the hybrid MIC is the m iniature hybrid MIC
(MHMIC), wherein all passive com ponents are fabricated onto the MIC substrate
using thin-film processing. Only the active devices are added separately to the
circuit. M HM ICs p rovide a sm aller circuit topology allow ing for broader
ban d w idths, as well as lower assem bly costs and im proved circuit reliability
com pared to conventional circuits. The disadvantages include m ore complex
processing and a longer turnaround tim e for circuit fabrication.
Hybrid MIC technology offers the availability of a wide selection of
devices, including high-precision laser-trimmed resistors and stacked capacitors
w hich allow for a w ide range of capacitances while m aintaining a sm all base
area. Active com ponents include G unn and varactor diodes, as well as various
transistors such as FETs, HEMTs and HBTs. A significant problem with these
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
d iscrete active devices is the w id e variety of electrical characteristics b etw een
batches caused by slight variations in processing conditions. A lso, since these
devices are usually w ire bon d ed into the circuit, the variation in the placem ent
and length of the bond w ires can affect the circuit perform ance an d reliability.
A n o th er a d v an tag e of h y b rid M ICs is th e v a rie ty o f su b stra te s
which are available. The su b strate m u st have a uniform dielectric constant, w ith
a relatively high value to reduce the circuit size. It also m u st pro v id e m echanical
sta b ility from shock a n d v ib ra tio n , a n d h ig h th e rm a l c o n d u c tiv ity a n d
te m p e ra tu re sta b ility fo r p ro d u c in g th e v a rio u s c irc u it e le m e n ts a n d
stru c tu re s!211 A lum ina (AI2O 3) is one of the m ost suitable su b strate m aterials in
thin-film technology on the basis of the above criteria. Its su b strate p roperties are
listed in Table 2.1, along w ith som e oth er com m on substrate m aterials.
Alumina (99.5%
pure)
Sapphire
Quartz glass
GaAs
(hi ghl y
resistive)
Silicon
(p =1CP
Q-cm)
PTFE
Relative
Permittivity
£r/
Loss Tangent
(at 10 GHz),
tanS
9.8
0.0001
Specific
Thermal
Conductivity,
W/cm-K
0.37
9.4
3.8
12.9
0.0001
0.0001
0.002
0.42
0.017
0.46
11.9
0.015
1.45
2.1
0.0003
0.002
Table 2.1 Properties o f various substrate m aterials for MICs!21!
The relatively large d istrib u te d m atch in g n e tw o rk s e m p lo y ed in
M ICs are useful for narrow b an d w id th applications such as low noise am plifiers.
Also, the availability of high Q com ponents such as dielectric p u c k s a n d YIG
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sphere resonators in MICs can produce stable oscillators and narrow band filters
w ith Qs greater than one thousand.
A second type of MIC is the monolithic MIC (MMIC), in which all
active an d passive circuit elem ents and interconnections are fabricated on a
sem iconductor substrate in a com plete technological process. MMIC technology
has gained prom inence due to several factors, including: the production of large
diam eter sem i-insulating Si and GaAs wafers; the developm ent of the MBSIT.T;
an d the increased dem and for large quantities of MICs. The use of highly
resistive GaAs substrates allows for a higher frequency of operation by reducing
the leakage effects. The small size of MMJCs makes it possible to reproduce large
n u m b ers of identical circuits on a single w afer, and also allow s for broad
b andw idth applications. A nother advantage offered with MMICs is the potential
for u sin g custom ized circuit elem ents to achieve the desired perform ance. This
includes fabricating passive com ponents with specific values, as well as diodes
and FETs w ith various dim ensions to suit the required application. Also, as all
elem ents are interconnected by patterning thin-film metal layers rith e r than
using non-repeatable bond wires, the reliability of the circuit is enhanced.
D ue to the com m on su b strate that MMIC devices share, the
selection of active devices is lim ited to those which com ply w ith the doping
profile, hence certain devices such as PIN diodes are ^available in m any MMIC
applications. A nother disadvantage of MMICs is the lim ited pow er handling
capability of the GaAs substrates d u e to their poor therm al conductivity. Also,
the red u ced size of MMICs results in sm aller spacings betw een the circuit
elem ents, th u s creating circuit parasitics which becom e significant at higher
frequencies.
The m ost com m on forms of planar transm ission lines in MICs are
m icrostrip, stripline, and coplanar w aveguide, which a rt displayed in Figure 2.1.
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
The microstrip line is the most commonly used for its ease of fabrication,
compact size, low ground plane and radiation losses. The non-uniform dielectric,
consisting of air and the substrate, causes a quasi-TEM mode of propagation in
the microstrip. Values for the effective permittivity and the characteristic
impedance of microstrip lines are generally computed using CAD software due
to the complexity of the expressions. Another transmission line is coplanar
waveguide, which has the advantage that all metallization is on one side of the
substrate, allowing for low inductance grounding which is beneficial at
millimeter-wave frequencies. However, circuit topologies should be kept
relatively simple due to space limitations and difficulty in realizing complex
circuit geometries.
a) microstrip
Ground plane
c) stripline
b) coplanar waveguide
Ground plane
Ground plane
Substrate
7 Z 7 7 ttS (r r 7 7 7 7 s
Figure 2.1 Various planar transmission line configurations121'221
A stripline configuration reduces the radiation losses by confining
the electromagnetic fields between two ground planes. Stripline also supports
TEM propagation due to the uniform dielectric surrounding the conductor. These
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ad v antages are offset by the difficulties in fabricating and tuning stripline
circuits, since the elem ents are sandwiched between the ground planes.
The transm ission lines are formed by depositing the metal laver,
usually gold o r copper, onto the substrate using thin- or thick-film technology.
Thin-films, which will be used extensively in this w ork, are deposited onto the
substrate by various m ethods such a.' electron-beam evaporation and sputtering.
The film s are then patterned using etching or lift-off processing techniques.
H ybrid MICs usu£.liy require a single layer of m etallization, and are less
exp ensive as com pared to MMICs. A typical MIC can be fabricated and
assem bled w ithin one week. A MMIC usually requires at least ten masks and
complex processing to pattern all layers onto the substrate. The cost of fabricating
a few 3" wafers at an MMIC foundry using the same mask set can be upw ards ot
$100K, w ith a typical turnaround time of sixteen weeks.
The decision on w hether to use hybrid MICs or MMICs depends
upon the application, perform ance requirem ents, cost and tim e considerations.
MMICs are m ore suitable for applications requiring large quantities of small,
lightw eight, and broadband circuits. Conversely, MICs are m ore applicable to
sm all nu m b ers of low complexity circuits requiring low noise figure, narrow
b an d w idth or high power, as well as a short turnaround time. For these reasons,
and the fact that superconductor-based MMICs do not presently exist, all circuits
designed for this w ork were fabricated using hybrid MIC technology.
2.2.2 MICs at Cryogenic Temperatures
O perating MICs at cryogenic tem peratures offers a num ber of
benefits to the perform ance of the circuit. A cooled environm ent will increase the
Q factor of a circuit by reducing conduction and su b strate losses. Active
sem iconductor devices exhibit improved gain and higher frequency of operation,
as well as extremely low noise figures under low tem perature conditions. O ther
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
factors such as tem perature stability, circuit reliability, and pow er density benefit
from a cooled operating environment.
Transm ission line structures have three m ain loss m echanism s:
resistive conductor loss, dielectric loss, and radiation loss w hich are related to Q c,
Q a, and Qr respectively. Each m echanism degrades the overall Q of a circuit
according to equation 2.1. Recall that the quality factor describes the perform ance
of a resonant circuit, and is defined as the ratio of the tim e average stored energy
to the energy lost per second.
1
1
1
1
Q ' <5 + <5
(21)
T he resistive loss of n o rm al c o n d u c tin g m eta ls d e crea se s
significantly when cooled to liquid nitrogen tem peratures (77K). The resistivity
of bulk gold, for instance, drops to one-sixth of Jts room tem perature value, w hile
sim ilar decreases to one-eighth and one-tenth the room tem perature resistivity
occur for copper and alum inum , respectively.!23! The g rap h in Figure 2.2 displays
the decrease in the DC resistivity of bulk gold over tem perature.
In thin-films, the reduction in resistivity over tem perature is not as
significant due to a "size effect" which occurs w hen the conductor dim ension
approaches the mean free path of the carriers.!24! As the thin-film is cooled, the
electron mean free path increases, causing an increase in surface scattering. This
results in a higher resistivity as com pared to bulk m aterial. Figure 2.3 displays
the effect of tem perature on the resistivity of thin-film copper relative to its room
tem perature value. At 77K, the difference between the bulk resistivity and that of
a 1pm thick film is less significant as com pared to low er tem peratures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
10
ELECTRICAL
RESISTIVITT.
10E-6
O ba-c«
12
ChtmBcf Data,
200
1000
800
400
1200
Figure 2.2 Resistivity of bulk gold versus temperature123!
FILM THICKNESS
iOO'
i
10
*3
BULK
u
x
tpkj
Figure 2.3 Resistivity of thin-film copper versus temperature1241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MOO
16
Resistive losses become more severe at higher frequencies due to
the "skin effect" in which the fields and associated currents are confined to a very
thin layer of the conductor surface. This effect is measured by the penetration
depth, 5, as defined in equation 2.2 for good conductors. Based on this parameter,
the surface resistance, R«, is derived in equation 2.2 displaying a square root
dependence on frequency.
(2.2)
At low temperatures, the surface resistance reaches anomalous limits as shown in
Figure 2.4 for bulk copper. This phenomenon is similar to the size effect, in which
the higher mean free path at cold temperatures and reduction in 8 at higher
frequencies causes surface scattering, thus limiting Rs to a minimum value. The
AC resistances of both copper and gold are approximately reduced by a factor of
four at 77K.
lOOi
A&2MAU2U2 .
1*10® Hz
10 -
AN0 MAL0 US
f»K >* Hz
ANOMALOUS
ff ** 10
in'2^ Hz
u*
01
Rm Tempt
10
100
1000
TCK)
Figure 2.4 Rs of copper vs. temperature at various frequencies1241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
A second consideration affecting the Q -factor is the con trib u tio n
d u e to the dielectric losses of the substrate. The loss tangent, tanS, is inversely
p ro p o rtio n a l to Qd, an d is defined as the ratio of the displacem ent current to the
c o n d u ctio n current. T he loss tan g en t of v a rio u s su b stra tes has been sh o w n to
d ecrease w ith tem p era tu re , th u s yielding a h ig h er Q j.l2*! This tren d w as also
o b serv ed in m ore recent experim ents of su b strates such as L aA lO i, w hich are
su itable for su p p o rtin g HTSC structures.!25!
T he final loss m echanism , radiation loss, becom es a factor at higher
m icrow ave frequencies. Also, since radiation loss is unaffected by tem perature, it
b e c o m e s sig n ific a n t a t low te m p e ra tu re s . A sid e from u sin g a s trip lin e
co n fig u ratio n , ra d ia tio n losses can be red u ced by rem o v in g transm issio n line
d isc o n tin u itie s, u sin g th in n e r su b stra tes, p ro v id in g sh ield ed en clo su res, and
m in im izing the circuit area.
P ow er d en sity o f MICs an d M M ICs im proves at low tem p eratu res
d u e to th e e n h an c ed th erm al co n d u ctiv ity of p u re m aterials, as d isp lay ed in
T able 2.2. This is n o t tru e for com posites an d m aterials w ith im p u rities such as
silver epoxy, how ever. C ircuit reliability also im proves at low te m p era tu re s by
ex te n d in g the lifetim e of G aA s devices, as found by R oeschJ26! FETs h av e been
th e m ain failu re m echanism of M ICs in w hich in terd iffu sio n of the junction
m e ta ls in to th e G aA s s u b s tra te s h a s been c au se d b y h e a tin g . C ry o g en ic
te m p e ra tu re s im plicitly p ro v id e in sen sitiv ity to ch an g es in ex tern al am b ien t
tem p eratu re, th u s m inim izing heating problem s.
A n o th e r im p o rta n t a d v a n ta g e of cooling M ICs is th e d ram a tic
im p ro v em en t in perform ance of active devices such as FETs and HEMTs. At low
te m p e ra tu re s , th e tra n s c o n d u c ta n c e a n d freq u e n cy of o p e ra tio n in crease
significantly, along w ith a c o rre sp o n d in g red u ctio n of the device noise figure.
These effects w ill be exam ined in m ore detail in C h ap ter 3.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
GaAs
lOx
Gold
8x
Alum ina
4x
Ag epoxy
0.5x
Table 2.2 Therm al conductivity im provem ent from 300K to 77Kl27l
2.3 Theory of Superconductors
Superconductivity w as discovered in 1911 by K am m erlingh O nnes,
w ho found that a sam ple of m ercury cooled to 4K usin g liquid helium yielded
/.ero resistance. For m any years following, discoveries of various niobium -based
alloy su p e rco n d u c to rs w ith h igher critical tem p era tu re s (Tc) w ere m ade. A
n u m b er of theories have been developed for superconductivity such as the TwoFluid m odel and L ondon theory!78! based on the M eissner effect, discovered in
the 1930's. The Bardeen-Cooper-Schrieffer (BCS) theoryl29! in 1957 introduced the
co n cept of C ooper pairs, w hile Josephson's theory!30! in 1961 ex p lain ed the
Josephson junction. An overview of the evolution of superco n d u cto rs and their
developm ent is displayed in Figure 2.5.
The second m ajor break th ro u g h occurred in 1986 w hen Bednorz
and
M u ller d isc o v e re d a c e ra m ic-b a se d s u p e rc o n d u c to r La 2-xBaxC u 0 4
(LaBaCuO) w hich attained a Tc above 30K.131! W ithin o n e year, C h u at the
University of H ouston broke the liquid nitrogen (77K) b arrier w ith the discovery
of YBaiCu^O? (YBCO) w ith a critical tem p eratu re o f 93K.l32l The discovery of
th ese high te m p e ra tu re su p e rc o n d u c to rs (HTSCs) h as sp a rk e d e n o rm o u s
research into finding new HTSCs such as Bi2SrCa 2C u 2 0 s (BSCCO) w ith TC=115K,
and ThBa 2C a 2C u 3 0 io (TBCCO) w ith a Tc of 125K. These superconductors could
find application in num erous areas including high Q resonators for filters an d
oscillators, RF an ten n as to exploit low loss, and h ig h speed interconnects in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
analog signal processors to remove loss and dispersion effects. This section
describes the basic theory of low temperature superconductors, and how these
concepts have been extended to explain high temperature superconductors.
140
T1 Ba Ca Cu O
B iSrC aC uO
TISrCuO
Y Ba Cu O
120
fi
s
100
2
Space Temperature
eo
Liquid Nitrogen
La SrCu O
Bi Ba KO
NbGe
N bN
Liquid Helium ~
Year
1900
1920
Meissner Effect
Discovery
J9K0
1960
1940
BCS
Theory
Josephson Effect
High Field/Current
Superconductivity
Figure 2.5 Chronological history of superconductivity1331
2.3.1 Low Temperature Superconductors
A material enters the superconducting state when the temperature
falls below the material's critical (or transition) temperature, Tc. A
superconductor is characterized by zero resistance, (i.e. the resistivity drops to
immeasurably small quantities within a very small temperature range), and
perfect diamagnetism. Diamagnetism, referred to as the Meissner effect in
superconductors, prevents m agnetic fields from penetrating inside a
superconductor as illustrated in Figure 2.6, and is demonstrated by magnets
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
levitating above superconducting disks. The Meissner effect distinguishes perfect
conductors from superconductors since, although both are able to prevent a
magnetic field from penetrating inside their interior, only superconductors can
expel a magnetic held already existing within the material.
(i> r>r
<ii) r<r.
Figure 2.6 Meissner effect in superconductors134)
A superconductor must not only operate below its critical
temperature, TC/ but it is also limited by a critical magnetic field intensity, He, and
a critical current density, Jc. The three parameters are interdependent according
to the three-dimensional surface shown in Figure 2.7. For instance, the farther
below its critical temperature a superconductor is operated, the greater the
current density and magnetic field it can withstand before reverting to a normal
resistivity state.
Superconducting materials can be classified as either Type I or
Type II superconductors. Type I superconductors have only one critical magnetic
field, below which surface currents cancel any magnetic held inside the material.
Type II superconductors, such as all ceramic HTSCs, have a second, higher
critical magnetic field due to the flux quantization within the material. Flux
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
q u a n tiz a tio n refe rs to the p e n e tra tio n of th e m a g n e tic field in to th e
superconductor in discrete dom ains.
Tcm ptvavjfe. keivn;
Figure 2.7 R elationship betw een TC/ JC/ and Hc in su p erconductors!VlI
Below Tc, the resistance declines to im m easurably sm all values d u e
to the form ation of C ooper pairs, in w hich tw o electrons interact w ith a phonon
(lattice vibration) to low er the effective energy of the pair. A ssum ing an infinite
m o m entum relaxation tim e (i.e. no electron scattering), classical m echanics and
electrom agnetics can be used to define the London equations. The relationships
for the force, F, acting on a charge in the presence of an electric field, E, and the
c u rren t density, J, of a n u m b er of superconducting electrons, n, are defined in
eq u ation 2.3 as
F = m ^ = qE
J = nqv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2.3)
22
w here m, q, and v represent the m ass, charge, and velocity of the electrons,
respectively. C om bining the tw o equations above yields
The M axw ell e q u atio n s p resen te d below show th e connection
betw een the electric field, current density, and m agnetic field.
„ _
3H
VxE = -Po-gjr
_ „
VxH -
.
J+
3D
(2.5)
A pplying the "curl" operator to both sides of equation 2.4 an d su b stitu tin g the
first M axwell's equation of 2.5 into the right side produces,
i ( V x j + -^ H ) = 0
X,
X,2 -----Unnq
(2.6)
X|.r referred to as the London penetration d ep th , defines the extent to w hich a
m agnetic field is able to penetrate the superconductor.
D ue to the M eissner effect, th e m agnetic field is n o t only tim einvariant, but zero w ithin the superconductor. L ondon's first equation is derived
from e q u a tio n 2.6 u sin g the M eissner effect to d escrib e th e field -cu rre n t
in teraction inside a su p erco n d u cto r. The second L ondon e q u a tio n follow s
directly from equation 2.4 by substituting the London penetratio n d ep th . The
tw o U>ndon equations are displayed in equations 2.7.
r, .
1
'" x 2
..
d)
d t %
1
„
^ E
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
<27)
L ondon's first equation can be rew ritten in term s o f the m agnetic field only.
T aking the curl o f the second M axw ell's equatio n in 2.5 an d neglecting the
displacem ent c u rren t at relatively low frequencies, the current density term of
L ondon's first equation can be replaced. N oting that the divergence of a m agnetic
field is alw ays zero, the H-field can be solved as a Laplace equation, expressed
below.
( 2 .8 )
W hile su p e rco n d u c to rs exhibit negligible losses at IXJ, losses
increase in proportion to the square of frequency. This phenom enon is explained
by the Tw o-Fluid m odel, as show n in r Igure 2.8, w hich assum es that the total
n u m b er of electrons, n, is divided into superconducting electrons, ns, and norm al
electrons, np.I34! The ratio n s/ n n varies w ith tem p eratu re below T, according to
equation 2.9. The superconducting electrons flow through the inductor, w hile the
norm al electrons flow in the resistor and are responsible for the AC losses.
R
nn
n
^
Hs
_
T
f
Y
Y
\
L
Figure 2.8 Two-Fluid m odel for superconductors
(2.9)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
The conductivity can be described in term s of a com plex quantity,
a=Oi-j<T2 , in w hich the resistive p art arises from the norm al electron conduction,
an d
02
represents the inertia of the superconducting electrons an d d e p e n d s on
th e L on d o n p e n e tra tio n d e p th . T he c o n d itio n a 2>XJi is v a lid
fo r all
su p e rc o n d u c to rs below Tc, hence the surface im p e d an c e Z s=Rs+jXs can be
represented as,
Rs =
(2-10)
D ue to th e te m p e ra tu re d e p e n d e n c e o f th e su p e rc o n d u c tin g
electrons, the L ondon p en etratio n d e p th is also d e p e n d e n t o n tem p era tu re .
E x p erim ents have p re d ic te d th a t Xl. v a rie s w ith te m p e ra tu re a cc o rd in g to
eq u ation 2.11, and becom es infinitely large as the tem p e ra tu re a p p ro a ch e s Tc.
This reflects the kinetic inductance effect, w hich becom es a factor w h e n operating
su p erco n d u cto rs close to Iheir Te. This will be dem o n strated in the HTSC filters
of C h apter 4.
( 2 . 11)
S u p erco n d u ctin g thin-film tran sm issio n lines e x h ib it v e ry low
disp ersion u p to and beyond m icrow ave frequencies, d u e to th e extrem ely low
surface resistance. N eglecting Rs, the p h ase velocity in su p erco n d u cto rs is only
d e p e n d e n t u p o n X[. an d film thickness for te m p e ra tu re s w ell b e lo w Tc (i.e.
n n/ n « l ) . Since X] is independent of frequency, the d ispersion can be considered
neg lig ible. T his p h e n o m e n o n m akes su p e rc o n d u c to rs a ttra c tiv e fo r u se as
interconnections in digital system s, delay lines, and filtering applications.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
The London theory is a sim ple phenom enological theory which
gives an accurate qualitative description of superconductors, but is less reliable
in p ro v id in g precise qu an titativ e results. The BCS theory provides a m ore
detailed, quantum m echanical explanation for su perconducting phenom ena.
Electrons norm ally experience C oulom b repulsion forces w hen they approach
each other. In a superconductor, BCS theory assum es the existence of additional
attractive forces w hich com pensate for the repulsive forces. These attractive
forces arise w hen tw o electrons exchange a phonon, w hich is a q uantu m of
th erm al energy in the lattice vibrations, to form a C ooper pair. If all the
conduction electrons form these C ooper pairs, in w hich the electrons have equal
and opposite m om entum , the energy of the system can be lowered. The average
distance at w hich the attractive and repulsive forces balance is know n as the
coherence length,
%.
The coherence length can span across m any other electron
pairs to form C ooper pairs w ith these other electrons, since they all have the
sam e q uantum mechanical state.
As predicted w ith the Tw o-Fluid m odel, norm ally conducting
electrons exist along with the superconducting electrons but at different energy
levels. The energy gap param eter, A, corresponds directly to the binding energy
of the C ooper pairs. It refers to the am ount of energy that m ust be added to each
su p erco n d u ctin g electron to break u p a pair and create in d iv id u al norm al
electrons. The d ensity of electron states in superconductors is zero inside the
en erg y gap, peaks at the ed g e of the gap, and decays to the norm al carrier
d en sity as the energy increases.!34! The energy gap decreases as the operating
tem perature rises, and becomes zero at Tc.
The peak in the density of states gives rise to the tunneling effect
w hich takes place in a superconductor-insulator-superconductor (SIS) junction.
N o c u rren t flow s in the junction until a voltage exceeding the energy gap is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
applied. A large surge in current is produced by the normal electrons, as
opposed to the Cooper-paired superconducting electrons, as shown in the
current-voltage relationship of Figure 2.9a). SIS junctions are suitable for
detectors and mixers in microwave and millimeter-wave applications, such as
radio astronomy in which high sensitivities are required.
CURRENT
Figure 2.9 Current-voltage relationship for a) SIS and b) Josephson junctions111
A special type of SIS junction is the Josephson junction, in w hidi
the Cooper pair superconducting electrons are the tunneling mechanism. These
junctions consist of a thinner insulating barrier to enable the Cooper pairs to flow
with no applied voltage, resulting in the I-V characteristic of Figure 2.9b). The
flow of current at V*0 is referred to as the DC Josephson effect. When a finite
voltage is applied, the junction behaves as a regular SIS junction. However, a
second phenomenon known as the AC Josephson effect occurs at very small
voltages, whereby an alternating current is produced by the electrons to radiate
this excess energy. Voltages on the order of a few microvolts w ill produce AC
currents in the microwave frequencies, where f = 2qV/fi. The Josephson effect is
possible due to the preservation o f long-range order between the two
superconductors, which occurs when each superconductor maintains a fixed
phase relationship between their phase functions, known as the phase coherence,
5q. The current flowing through a Josephson junction can be expressed as.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2qV
J = J0sin(60- - ^ - t )
(2.12)
The Josephson junction equivalent circuit model is displayed in
Figure 2.10, in which the total current comprises three components. The JosinSo
component represents the current due to the Cooper pairs at zero voltage, the
resistance is related to the flow of the normal electrons as in the SIS junction, and
the capacitance is determined by the insulator. Josephson junctions are used in
high-speed digital switching applications, and also provide the basis for
Superconducting Quantum Interference Devices (SQUIDS) which are extremely
sensitive magnetometers.
o
J
■»
JosinSo X
C
o
Figure 2.10 Equivalent circuit model for a Josephson junction!34)
2.3.2 High Temperature Superconductors
There is a continuing debate as to whether the same theories which
apply to LTSCs are also applicable to HTSCs. HTSCs exhibit the same properties
as traditional superconductors such as zero resistance, the Meissner effect, flux
quantization, and the Josephson effect. The BCS theory still applies in some cases,
such as the formation of Cooper pairs, but cannot explain whether it is due to the
electron-phonon exchange process or some other phenomenon.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Type II superconductors, including the ceram ic-oxides, consist of
three phases that depend on the m agnetic field intensity. At sufficiently sm all
fields, the magnetic flux is com pletely expelled from the m aterial according to
th e M eissner effect. A bove a critical field, Hci, the m ate ria l rem a in s
superconducting but the field partially penetrates the m aterial in q uantized flux
regions. A bove a second critical field, H c2 , the su p e rc o n d u c tin g sta te is
com pletely destroyed. Type II superconductors have been described u sin g the
C inzburg-L andau theory!36! w hich uses q u a n tu m m echanics to ex p lain h o w
m agnetic fields pen etrate into b u lk and film su p erco n d u cto rs, a n d how the
m agnetic structure is developed. This theory characterizes the superconducto r by
a w ave function to describe the superconducting electrons, and a m agnetic vector
potential. The focus of this theory is to determ ine solutions for these functions
which m inim ize the total free energy of a sam ple. The London penetration d ep th ,
Xi (T), and the coherence length, £(T), are show n to be tem p eratu re d ep en d en t,
varying proportionately to (T^-T)1/ 2. The G inzburg-L andau param eter, defined
as k--A,|.(T)/4(T), is in d ep en d en t of tem p eratu re an d d eterm in es the ex ten t to
w h ich
th e lo n g -ra n g e o r d e r o f the s u p e rc o n d u c to r e x ists. T y p e
su p erco n d u cto rs are defined for tc>
\h jl.
II
C eram ic su p e rco n d u c to rs su ch as
YBCO display values of ic greater than 200 at T=11K, th u s d em o n stratin g their
Type II tendencies.!14!
The key to superconductivity in the copper oxides lies in the b o nds
b etw een the oxygen and c o p p e r atom s. The oxygen form s a p y ra m id a l
p o lyhedron around the copper atom s, and it is the C u-O p lan es w hich allow
electrons to m ove freely through the ceram ic.!10! The YBCO h as a pero v sk ite
crystal s tru c 'ire , as does all HTSCs, in w hich consecutive planes of b ariu m are
separated by alternating planes of copper oxide and held together by planes of
yttrium , as show n in Figure 2.11. YBCO displays anisotropic properties in w hich
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the coherence length is considerably larger in the a,b plane than in the e-direction
of the crystal
O
C «vg«n
0
C ooear
C h ain
Cham
Chain
Figure 2.11 Crystal structure for YBCO superconductors1351
A popular contemporary theory for explaining the superconducting
phenomena in HTSCs is known as the "resonating valence bond" theory.1371 This
theory utilizes the fact that the conduction mechanism lies in the copper-oxide
planes. Q 1O2 is an insulator by itself, but when mixed with the other elements (Y,
Ba), the valence state of copper changes from +2 to an average of +2.33. The
copper and oxygen atoms exchange electrons by means of a sea of electrons, or
conduction band. When in the superconducting state, the conduction band is
filled with loose electrons, leaving positively charged holes in the lattice. The
holes are the primary charge carriers in ceramic superconductors.
Several other recent phenomenological theories to explain the
behaviour of HTSCs have been proposed. One of these, the bipolaron model,
provides the basis for software models of HTSC microstrip and coplanar lines.1381
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
This m odel pred icts the surface im pedance of the HTSC film b ased on the
m aterial’s London penetration d epth, w hich is tem p eratu re d ep en d en t. It also
relates how the effective dielectric constant of a m icrostrip o r coplanar line is
affected by kinetic inductance. A nother HTSC theory, the enhanced three-fluid
m odel, can m ore accurately describe the characteristics at tem peratures far below
T ;JW1This theory incorporates a tem perature d ep en d en t m om entum relaxation
time, t, to predict a phenom enological ’’coherence peak" observed in the real p art
of the com plex conductivity for YBCO thin-films and bulk m aterial.
A part from a h ig h er Tc, HTSCs experience a h ig h er en erg y gap
p aram eter than LTSCs, w hich should allow for higher o p eratin g frequencies in
su p erconductive electronics. O ne disadvantage of HTSCs, how ever, is a larger
London penetration depth, resulting in an increased kinetic inductance. This is
e v id en t from equation
2 .1 1 ,
w here A,l becom es large as the tem p era tu re rises
to w ard s Tc. O ther disadvantages of HTSCs as com pared to LTSCs include low er
c o h e re n c e le n g th s a n d e n e rg y sta te d e n sitie s . A c o m p a ris o n o f th e
superconducting properties of niobium and YBCO is p resented in Table 2.3.
Tc
XL(0 ), pm
A (meV)
U nm )
n ( / m 3)
Nb
9.2K
0.07
1.5
38
14.7e27
YBCO
90K
0 .2 2
-15
~2
2.6e2 7
Table 2.3 Com parison of properties for a LTSC an d HTSCl14-40!
2.4 Fabrication and Processing of HTSCs
Thin-films m ust exhibit several properties to ensure their suitability
for practical application of m icrow ave devices, including: availability on b o th
sides of large area substrates; low surface resistance, Rs; high p o w e r h a n d lin g
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
capability; and a surface topology com patible w ith stan d ard photolithographic
techniques.!411 Thin-film s are deposifed using various techniques, as show n in
Figure 2.12, such as electron beam co-evaporation, RF sputtering, pulsed laser
ablation, and m olecular beam epitaxy (MBE). Films m ay be applied using an insitu process in w hich the material is deposited onto a substrate m aintained above
the m aterial's crystallization tem perature. For an ex-situ process, the m aterial is
d e p o site d o n to th e su b stra te at a tem p e ra tu re below the cry stallizatio n
tem perature. The film is crystallized in a subsequent annealing procedure.
Presently, the TBCCO thin-film s have the highest reported Tt of
any practical superconductors, and hence w ere the m aterial of choice for circuit
fabrication in this w ork. Thallium-based superconductors exist in several phases,
such as Tl2 Ba2CaCu 2 0 g (2212), Tl2 Ba2Ca 2 Cu-iOio, and TlBa2C a 2C u 3O 4 . Thin-films
for this w ork w ere fabricated at Superconductor Technologies Inc. (STI) using the
Tl2 Ba2 CaCu 2C>8 com pound, w hich perm its for easier preparation at the expense
o f a slightly low er Tc. These STI thin-film s are fabricated in tw o stages, involving
th e d e p o sitio n of a p rec u rso r using laser ablation, and sintering u n d e r an
a p p ro p ria te th alliu m -o x y g en e n v iro n m e n t. T he p re c u rso r film s co n tain
stoichiom etric am o u n ts of calcium , barium , and copper, w h.ch are deposited
o n to the substrate. The a m o rp h o u s film is then an nealed at a tem p eratu re
betw een 830°C and 900°C in the presence of a controlled TI2O vapor to form the
TBCCO film w ith a 2212 phase. Special deposition facilities are required since
thallium is a highly toxic elem ent, although the TBCCO com pound is believed to
be m ore benign than its constituent elements. The sintering can be applied using
a lidded or sealed tube anneal, o r a flow -through furnace annealing process.!42!
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
Electron Beam Evaporation
Off-Axis Sputtering
wWyvyyw
-At*
Laser
-vWW\A/^“
T-c«U
Laser Ablation
Molecular Beam Epitaxy
Figure 2.12 HTSC thin-film deposition techniques1431
Patterning of the thin-film HTSCs requires a process that does not
degrade the film properties, is environmentally benign, and can produce the
desired feature size. The most common forms of patterning include the
application of stencil masks, wet etching or ion milling. STI currently uses an ion
miIImg process which allows for line widths of 25u:n and tolerances of ±0.5}un to
be achieved. Film thickness is specified to be O.^un ±0.1 pm which is sufficiently
greater than the London penetration depth of 0.3^m at 77KJ441
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
A key property in determ ining film quality is the surface resistance.
Rs is p ro p o rtio n a l to the sq u a re root of frequency for co nventional m etals
(eq u ation 2.2), b u t it increases in p ro p o rtio n to the sq u are of frequency for
su p erco n d u cto rs (equation 2.10). Even tho u g h Rs increases m ore rapidly w ith
freq u en cy in su p e rc o n d u c to rs, it can o u tp e rfo rm norm al c o n d u cto rs o v er
frequencies extending to h u n d red s of G igaH ertz. The surface resistance of STl's
TBCCO thin-film s decreases rapidly at
a
transition tem perature of typically 105K,
as sh o w n in Figure 2.13a). A t 77K and a frequency of 10 G H z, the Rs of these
TBCCO film s is approxim ately 0.25m£2, w hich is an im provem ent of alm ost forty
tim es as com pared to oxygen free h ig a conductivity (OFHC) copper u n d e r the
sam e conditions (Figure 2.13b)). The best surface resistances reported to date are
130g£2 and 181pQ a t 77K and 10 G H z for Tl2Ba2CaCu208 and TloBaiCn^CuiOio,
respectively.I41'45! F a tte m ed film s generally result in low er Tt and h ig h er Rs
c o m p ared to u n p a tte rn e d films, since the film s are slightly d e g ra d ed in the
etc h in g process. T his is especially tru e at the line edges, w h ere the c u rre n t
d en sity is highest.
A n o th er factor w hich d e term in es th e q u a lity and usefulness of
su p erconductor film s is the critical current, Jc. The TBCCO film s from STI possess
a typical Jc of 5 x l0 5 A /c m 2 at 77K. The best j cs reported to d a te for TBCCO thin
film s a t 77K is 2 x l0 6 A /c m 2J42l Y ttrium based YBCO film s have reported Jts of
5X106 A /c m 2 a n d higher,I46) hence YBCO is m o re su itab le for h ig h -p o w er
perform ance than thallium com pounds, at the expense of a low er Tc and higher
Rs- H ig h e r p o w e r levels cause c u rre n t levels and m agnetic fields to increase
to w a rd s th eir critical lim its, th u s increasing the surface resistance and low ering
th e Q. STI specifies typical Q values of betw een 12000 and 15000 for each wafer,
w hich a re tested using a 5.6 G H z m icrostrip resonator. A 14 G H z TBCCO thinfilm lin ear reso n ato r m easured at CRC had an u n lo ad ed Q of 10500 at 77K,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
com pared to a sim ilar gold resonator providing a Q of 470.1471 A 20 GHz TBCCO
linear resonator produced an unloaded Q of 2800, which could be im proved by
adjusting the cover height to further reduce radiation loss.
100
o
E
Q)
O
c
10
2
M
oc
Q>
O
1
Limit on
Sensitivity
5
.1
0
50
100
150
Temperature, K
200
100
Temp = 77 K
E
(0
3
*
E
o
OFHCCu
c.
Patterned Resonator
Data
100
Frequency (GHz)
Figure 2.13 Performance of R* for STI's TBCCO thin-film144!
Single crystal films must be grown on substrates which share the
same perovskite structure and thermal expansion coefficient as the HTSCs. The
TBCCO thin-films at STI are deposited onto a lanthanum aluminate (LaAlOs)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
substrate. The substrates are presently available as 2" o r 3" diam eter w afers, with
thicknesses of 0.010" or 0.020”. The dielectric constant of LaAIO.i is typically 24,
w hile the loss tangent, tanS, is specified to be 3x10 s.l44l Both these substrate
param eters have been show n to decrease slightly at low er tem peratures.!-'! The
dielectric constant and loss tangent of various substrates com patible w ith 1ITSCs
are liste d in Table 2.4. D ue to the low loss tan g en t w hich is inversely
p ro p ortional to the Q-factor, Q values of a HTSC thin-film m icrostrip resonator
can a p p ro a c h 30000, a n d becom e lim ited m ain ly by ra d ia tio n loss. A
disadvantage of the LaAlC>3 substrate is the "twinning" problem , w hereby the a,b
crystal orien tatio n is random ly interchanged. This problem occurs w hen the
processing tem perature exceeds the tw inning tem perature of 450°C, thus causing
changes in the dielectric constant and surface roughness.
M ateiial
Relative
Dielectric loss
perm ittivity, r r
factor, tan8
24.0
3x10 ^
23 1
10 4
m agnesium oxide (MgO)
9.6
4x10^
R-plane sapphire (a-
10
1.5x10«
16.8
4x10 ^
lanthanum alum inate
(L aA l03)
lanthanum gallate
(LaGaOs)
a i 2o
3)
strontium lanthanum
alum inate (SrLaAlQ*)
Table 2.4 Properties of HTSC substrates at 77K and 10 C l 1/.I2S-48I
O th er substrates suitable for HTSC thin-film s include lanthanum
g allate (LaGaOs) w hich possesses sim ilar p ro p e rtie s to LaAlOs, b u t also
experiences tw inning. M agnesium oxide (MgO) offers a low loss tangent and a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dielectric constant which is close to that of conventional MIC substrates, how ever
dielectric losses are susceptible to increases due to w ater absorption. Sapphire
su b strates, w hich are used in MIC technology, w ould be ideal for HTSC
applications with their extrem ely low loss tangent. Therm al m ism atch w ith the
HTSC thin-films which causes cracking, and the fact that the alum inum low ers
the Tc d u rin g the annealing process lim it its present usefulness. H ow ever, the
addition of a buffer layer betw een the film and the sapphire substrate can reduce
these effects.
During circuit fabrication at STI, ohm ic contacts are deposited onto
the thallium films using e-beam gold deposition, creating a bondable contact
with a resistivity of 10*7 Q cm2, and an insertion loss of 0.01 dB at 10 GHz.I44! A
3pm polyim ide passivation layer is applied to protect the HTSC film and to allow
gold crossovers. A gold film is deposited onto the g ro u n d plane to allow for
substrate attachm ent to gold-plated kovar carriers using conductive silver epoxy,
or a low m elting point 150°C in d iu m /silv er solder. The backside gold is applied
in a "w indow frame" around the perim eter of the substrate in o rder no t to reduce
the Rs and hence degrade the Q. Dicing of the circuits requires caution d u e to the
brittleness of the LaAlC>3 substrates. The m ask process th u s requires various
layers for circuit fabrication: an ohm ic contact layer, a su p erconductor layer,
polyim ide passivation layer, gold on polyim ide, and a substrate lc /e r to define
the dicing.
C om puter files containing A utoCAD layouts are sen t to STI via
m odem . The m ask production, fabrication of films an d dicing is perform ed at
STI, w ith a norm al turnaround tim e of six weeks. The cost of a double-sided 2"
diced w afer w ith gold contacts is approxim ately $15K(US). U pon receipt, the
circuits are assem bled into the kovar carriers w hich are attached onto a centre
section of the test fixture.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.5 Cryogenic Measurement Technique
W ith the im provem ent of low tem perature refrigeration system s, it
h as becom e m ore feasible to cool sem iconductor devices in o rd er to achieve
b etter perform ance. The potential application of HTSCs in low loss m icrow ave
circuits has provided further incentive to utilize cryogenic technology. I lowever,
a significant problem has been the accurate m easurem ent of circuits inside
cooling system s. This requires proper calibration of the netw ork analyzer at low
tem peratures in order to rem ove the effects of the device test fixture and external
cables. T his section describes the T h ro u g h /R e fle c t/L in e (TRL) calibration
technique, as well as its application for low tem p eratu re m easurem ents. The
cryogenic m easurem ent apparatus used in this w ork is a 1st) described in detail.
2.5.1 TRL Calibration Procedure!49!
C alibration involves the cancellation of system atic erro rs in a
m easurem ent system in order to obtain accurate m easurem ents. These systematic
effects such as leakage, test port m ism atch, and frequency response at m icrowave
frequencies will affect m easured data. The effects are repeatable, and therefore
can be d eterm in ed for a n etw ork analyzer by m easuring a series of know n
stan d ard s. The system atic effects are contained in a tw elve-term e rro r model
w h ich p e rfo rm s an "error-correction" to su b se q u e n t m easu rem en ts. In a
c o n v e n tio n a l tw o -p o rt c alib ra tio n , th ree k n o w n
im p e d a n c e s an d o n e
transm ission stan d ard are required, im perfections in the calibration stan d ard s
cause residual effects which are the portion of the uncorrected system atic error
that rem ains after calibration is perform ed.
TRL is a tw o-port calibration technique th at relies on transm ission
lines rather than a set of im pedance standards. Transm ission lines offer a num ber
of ad v an tag es in th at they are the sim plest elem ents to realize in non-coaxial
m edia such as m icrostrip. Also, the im pedance of transm ission lines can be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
accurately determ ined from the physical dim ensions and m aterials. The basic
TRL c alib ratio n pro cess is d isp lay e d in F igure 2.14. A to tal o f six teen
m easurem ents are required to determ ine the tw elve u nknow ns. In a perfectly
balanced test system which ignores the different term inating im pedances as the
RF sw itch changes betw een ports 1 and 2, only eight e rro r term s are required.
9
T hrough
[5| = [J
Reflect
Sji —S22 —r
Oj
©
-V -m
*.
Line
Figure 2.14 TRL procedure and ideal S-param eters for each standard!49!
The TRL procedure initially assum es a balanced system in w hich a
th ro u g h connection is m ad e betw een p o rts 1 an d 2 w ith a sh o rt len g th of
transm ission line. The transm ission frequency resp o n se an d p o rt m atch are
m easured in both directions to pro d u ce four m easurem ents. N ext, an o p e n or
sh o rt reflection sta n d a rd is connected to each p o rt a n d th e tw o reflection
coefficient m easurem ents are perform ed. Lastly, a know n length of transm ission
line relative to the through is connected betw een p o rt 1 an d 2 such th at the four
m easurem ents of frequency response and p o rt m atch in each direction can be
m ade. These ten m easurem ents result in ten e q u atio n s, b u t since only eig h t
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
u n k now ns exist in the balanced system , the T of the reflect standard and the
p ropagation constant, y, of the line standard are determ ined as constants. To
correct for the different term inating im pedances at the RF sw itch, the ratio of
in c id e n t sig n a ls a re co m p u ted d u rin g the th ro u g h and line s ta n d a rd
m easurem ents. All twelve term s in the error m odel have thus been accounted for.
2.5.2 Cryogenic Apparatus
Split-Block
Test Fixture
Silicon Diode
Thermocouple
2nd Stage Cold
Station
Coaxial Cable
To Vacuum Pump/
Dry Nitrogen^
Coax Feedthrough
toN /A
Vacuum Port
1st Stage Cold
Station
zzzzzzzz
Coax Feedthrough
to N /A
Phosphor-Bronze
Bias Lines
Figure 2.15 Cold head of the cryocooler apparatus!47!
The cryogenic apparatus consists of a custom ized cold head on an
RMC C ryosystem s LTS-22-1R helium refrigerator. The refrigerator, o r cryocooler,
is capable of cooling a 0.25 W att therm al load to tem peratures below 10K. The
cryocooler is integrated w ith the HP8510C o r W iltron 360 netw ork analyzer to
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
perform accurate cryogenic m easu rem en ts u p to 30 G H z. S em i-rigid 0.118"
coaxial K-cables connect a split-block test fixture to coaxial feedthroughs w hich
a re connected to the test po rts of the n etw o rk analyzer. The cryocooler can
achieve a m inim um tem perature of 45K w ithin one h o u r w hen the tw o halves of
the test fixture are directly connected together. A d iag ram of the cryocooler
based ap p aratu s is illustrated in Figure 2.15.
W ater-Cix)led H elium C om pressor
Cold H ead
2nd Stage Cold
Station (-10K )
W ire W ound H eater
T em perature
C ontroller
Piston
—y ^ X -.T o V acuum P u m p /
D ry N itrogen
1st Stage Cold
Station (~77K)
Figure 2.16 Refrigeration system for cryocooler setup!47!
The cooling system o p erates by circulating high p ressu re helium
gas into and o u t of the cold head according to the G ifford-M cM ahon cycle, as
show n in the schem atic diagram of Figure 2.16. As the gas e n ters th ro u g h valve
A., it e x p an d s and absorbs heat. The resulting low p ressu re gas re tu rn s th ro u g h
valve B to the w ater cooled com pressor w here it gives u p its heat. In o rd e r to
achieve extrem ely low tem peratures, the cold head is evacuated using a v acu u m
p u m p to red u ce co n d u ctio n a n d convection h e a t loss. Wh»'n p e rfo rm in g a
m easurem ent, the tem p eratu re is stabilized usin g a silicon d io d e th erm o m eter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
an d a resistance h eater m ounted on the second stage cold station. W hen
retu rn in g to room tem perature, dry nitrogen is supplied into the cold head to
increase the heat transfer and to prevent condensation from form ing on the test
fixture substrates.
2.5.3 Cryogenic Measurement Technique
The TRL calibration technique allows the effects ot the test fixture's
m icrostrip lines, microstrip-to-coaxial launchers, and interconnecting cables to be
de-em bedded at cryogenic tem peratures.!*’! O nly the effects of the bond w ires
connecting the device u n d er test are not m athem atically rem oved. I'he gold
m icrostrip lines, having a characteristic im pedance of 5012, are patterned onto
0.010" thick alum ina substrates to form the sta n d ard s as well as the input and
o u tp u t lines of the m easurem ent test fixtures. The sta n d ard s include a zero
length through, open circuits on the input and o u tp u t ports, and a
0 .0 5 0 "
long
line sta n d a rd . The line stan d ard provides an l i p s delay and is designed lor
calibration of frequencies u p to 40.5 GHz. The alum ina substrates and centre pins
of the K-connector launchers are attached using an in d iu m /le ad solder to ensure
a solid connection at 77K. Test fixtures are fabricated using Therm kon 70M
(K 5 "„
M o, 15% C u) w hich possesses approxim ately the sam e therm al expansion
coefficient as alum ina.IS1I This alloy is easier to m achine than kovar and has
b etter therm al conductivity, b u t requires a longer gold plating procedure. A
p h o to graph of the split-block test fixture is displayed in Figure 2.17.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 2.17 Photograph of Therm kon split-block test fixture
To perform a TRL calibration on the netw ork analyzer, the th ro u g h
stan d ard is form ed by connecting the tw o halves o f the test fixture d irectly
together. Next, the line sta n d ard is m ounte 1 o n to a shim an d sa n d w ich e d
betw een the halves. The thro u g h and line sta n d a rd s are connected u sin g gold
ribbon bonds to en su re solid connections at 77K. Finally, the open reflection
sta n d ard s are attained by sep aratin g the tw o h alves of the test fixture. The
cooling cycle requires approxim ately fifty m inutes for the test fixture to reach
77K, and thirty m inutes to reheat back to room tem perature. To com plete a full
tw o-port calibration, roughly five h ours is needed for the th ree cooling cycles
plus w ire bonding time. Thereafter, each circuit requires u p to tw o h o u rs for the
assem bly, cooling cycle, and S-param eter m easurem ents.
Results of a
c a lib ra tio n
over a 1 G H z to 26 G H z frequency range are
displayed in Figure 2.18, in w hich the through sta n d ard w as rem easured after
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
the TRL calibration. The transm ission response show s a d ev iatio n of w ithin
±0.05dB a n d one degree in the angle, w hile the retu rn losses yield b etter than
40dB o ver the entire range. Ideal conditions for the th rough should p rodu ce a
tra n sm issio n re sp o n se of S2i = 1^0° a n d in p u t a n d o u tp u t reflectio n s of
Sn=S22=0. D eviations from the ideal conditions are m ostly d u e io variations in
the b o n d w ire connections as well as the drift of the netw ork analyzer d u rin g the
long calibration and m easurem ent cycles.
►Sgi
REF 0.0 rB
0 . 1 c B ''
0.0148 cB
0 .0
•
A
i-0
3 -354.01 m •
a
*
'
M
e ra
p ik
f
2 0 .0
♦
d B ^ d iv
I,
F - ?
0 .0 cfi
3 9 .0 dB^
—40.904 dB
i
i-
------
I ___ ____________
START
STOP
1.000000000 0Hz
96.000000000 GH2
Figure 2.18 Remeasurement of ’ through” standard after calibration
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
Measuring the same standard used to calibrate does not reflect the
true accuracy of this method since they have been defined by the calibration to be
ideal. Therefore, a 0.101" long microstrip line which is twice the length of the line
standard was measured. The results in Figure 2.19 show that the insertion loss is
better than 0.2dB up to 20 GHz, while the return losses are below 26dB for the
input and output. Sources of error are due to bond wires and network analyzer
drift, as mentioned above, along with the longei transmission line.
J.
V
0.0 dB
0.5 t®''
-0.1218 dB
0.0
A
^
*
90.0
-1S1.3 •
'2
srnrrr
STOP
H ^ 1 0.0 dB
A
10.0 dB"'
T -30.992 dB
1 .0 0 0 0 0 0 0 0 0 » t :
26.000000000 GUr
0 .0 dB
*
1 0 . 0 d B ''
-32.012 cJ3
Figure 2.19 Measurement of a 0.101” line to verify calibration
This TRL calibration technique provides accurate, broadband, de­
em bedded S-parameters of m icrowave devices operated at cryogenic
temperatures.1521 To reduce the length of calibration time, as w ell as to eliminate
the need tor bond wire connections, a cryogenic wafer prober is currently being
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
developed in-house at the Com m unications Research Centre. The probing station
will allow on-w afer m easurem ent of HTSC and sem iconductor circuits u p to the
m illim eter-w ave frequencies, at liquid nitrogen tem peratures (77K).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
CHAPTER 3: CRYOGENIC AMPLIFIER DESIGN AND
EXPERIMENTAL RESULTS
3.1 Introduction
Cryogenic cooling is used in m icrowave receiver system s to achieve
better perform ance in low noise am plifiers and m ixers. In radio astronom y, for
instance, cryogenic tem peratures are required to m inim ize system noise to levels
com parable w ith the natural lim its of galactic, cosm ic, a n d atm ospheric noise.
Param etric am plifiers and MASERS have been used to achieve su p erio r noise
perform ance to traditional MESFETs. W ith the a d v e n t of the H igh E lectron
M obility T ransistor (HEMT), perform ance co m p arab le to p a ra m p s can b e
achieved at a reduction in cost. The HEMT, as well as the Pseudom orphic HEMT
(PHEM T), p ro v id es m uch b e tte r noise p erfo rm an ce a n d h ig h er o p e ra tin g
frequency than the MESFET. These advantages are even m ore p ro nounced at
cryogenic tem peratures w here the m odulation doping of HEMTs can be further
exploited.
In
th is
c h a p te r,
th e
d e v e lo p m e n t o f a h y b rid
h ig h -T c
s u p e rc o n d u c to r/s e m ic o n d u c to r K -band a m p lifie r is d e sc rib e d . A b rie f
description of the principle of operation of HEMTs an d PHEM Ts is presented,
along w ith a sm all-signal device m odel for a HEMT at room tem p eratu re and
77K. Based on this m odel, an am plifier usin g HTSC passiv e stru c tu re s for
m atching has been designed. The final section presents m easured results of the
am plifier and com pares these to sim ulations. A n analysis o f the ability of various
CAD softw are packages to predict the perform ance of HTSC passive structures is
ilso presented.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
3.2 Device Characterization at 77K
The HEMT device was invented in 1980 by researchers at Thomson
and Fujitsu sim ultaneously153! The structure is physically identical to the
MESFET except for the active layer doping profile. The active layer is modulation
doped in the vertical dimension to create a "super lattice” to give the HEMT its
superior performance. The active layer contains a combination of doped AlGaAs,
a thin layer of undoped AlGaAs, and pure GaAs. A heterojunction forms at the
AlGaAs/GaAs interface as shown in Figure 3.1a). Electrons in the doped, higher
bandgap AlGaAs migrate through the thin layer of pure AlGaAs into the pure
GaAs. These carriers form a two-dimensional electron gas free of coulomb
scattering effects which are not obstructed by collisions with donor ions and
hence produce a higher mobility. The pseudomorphic HEMT or PHEMT is an
extension of the HEMT, wherein the undoped GaAs layer is replaced with the
lower bandgap material InGaAs as shown in Figure 3.1b). This results in a
further increase of the electron density and mobility of the two-dimensional
electron gas.
1.
So m e*
VS/S/WA .
Gate
______
Oram
______
E rl ,
^ /Z ////A
3 E
H"
^
I
/-JS 31
_ S
«•*AJrtaA*
AlGaAs
Undoped AlGaAs
s
| , Hetaroiunction
T
^
AlGsAs Sasesr
hi^ivmobrirty
Undoped GaAs
G‘V
X L *— J r
n* AlGaAs
motAs
ch an n el
GaAs SuHsr
Semnnsulatmg substrate
a)
b)
Suostrsw
Figure 3.1 Cross-section of the structure for a) HEMT*22! and b) PHEMT54!
Cooling FETs and HEMTs improves performance due to higher
carrier mobility, larger resistance in the semi-insulating substrate, and higher
conductivity in the active region. With higher mobility, the electron velocity
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
48
increases, resulting in a higher transconductance, gm, for the device. At the cutoff
frequency, the input current is equal to the output current, and is expressed in
terms of the device model elements as,
f
(3.1)
HEMT
ji
>
£
I
MESFET
100
300
Figure 3.2 Mobility vs temperature for MESFET and HEMT channel structures1271
From equation 3.1, it is evident that a higher fr w ill result from a higher
transconductance. The electron mobility improves at cold temperatures due to
the reduction of lattice vibrations. The mobility versus temperature is displayed
in Figure 3.2 for both a modulation doped GaAs HEMT and uniform doped
GaAs FET. This illustrates die benefit of the undoped GaAs layer in HEMTs for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
electron mobility a t low er tem peratures. A nother advantage of cooling the device
is a higher conductivity, which decreases the drain-to-source resistance, thereby
increasing the frequency of operation and reducing the noise. At tem peratures
below ~50K, how ever, extensive d o p a n t freeze-out occurs w hich severely
degrades device perform ance. The higher resistance of the sem i-insulating CiaAs
at low tem peratures helps reduce circuit parasitics such as substrate leakage
currents.
Perhaps the m ost im portant advantage of cooling devices is seen in
the noise perform ance. Noise figure, NF, is defined as the input signal-to-noise
ratio divided by the o u tp u t signal-to-noise ratio,
The noise characteristics for a FET o r HEMT depend on four noise param eters:
the m inim um noise figure, NFmin, optim um noise adm ittance, (g0+jb0), and the
eq u iv alen t noise resistance, r n. The noise figure can be expressed in term s of
these param eters as
(3.3)
The m inim um noise figure occurs w hen the source adm ittance (gs+jbs) equals the
optim um noise adm ittance, and is show n as,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
where Kf is a fitting factor to represent the quality of the chanr -' 'aterial. From
this equation, one can see that the higher transconductanr
iu lower gate
resistance of HEMTs at lower temperatures results in a lower NFmin.
The basic MESFET small-signal equivalent circuit is displayed in
Figure 3.3. The main intrinsic elements of this model are the gate-to-source
capacitance, Cgs, the drain-to-source and channel resistances, Rds and Rj, and the
transconductance, gm = dlds/dVg*. The extrinsic (parasitic) elements include the
source resistance, R*, the drain resistance, Rd, the gate resistance, Rg, and the
drain-to-source capacitance, Cds-
Gate
Source
Source
Figure 3.3 Small-signal equivalent circuit model for a MESFET^551
A small-signal model for the 0.25pm x 100pm Fujitsu FHR10X
HEMT has been derived by Smuk et al at both room temperature (300X) and
77K.I52! This model uses FETlink software which incorporates the standard FET
model, as shown in Figure 3.3, with additional parasitics. The model is based on
previous S-parameter measurements from 1 GHz to 20 GHz at bias conditions of
2V drain-to-source and 5mA drain current, in a source ground configuration. The
device model was found to provide an accurate agreement to measured data. I52)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
De-embedding was performed to the edge of the test fixture but did not include
wire bonds.
gm (mS)
Cgs (pf)
Ggs (mS)
Rds(ft)
Ri(Q)
x(ps)
Cgd (pO
Cds (pf)
Lg(nH)
Rg(Q)
Ld(nH)
Rd(Q)
Ls(nH)
Rs (ft)
fr (GHz)
297 K
26.9
0.116
0
334
2,2
1.9
0.015
0.058
0.135
1.0
0.089
2.8
0.021
0.8
32.7
77 K
42.5
0.113
0.077
232
0
0.8
0.021
0.063
0.161
0.5
0.148
5.6
0.024
1.0
50.5
Table 3.1 Element values for small-signal FHR10X HEMT model*52!
12.M
•■•00 4 .
ffco-GMZ
2ST99
Figure 3 4 S-parameters derived from a HEMT model at 300K and 77K
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Elem ent values are displayed in Table 3.1 for both 77K an d room
tem peratures. Note that by cooling the device, the transconductance, g m, show ed
a 58% increase, w hile the drain-to-source resistance, Rds- d ro p p ed by 30%. The
cutoff frequency for the device increased m ore than 5C% to 50 G H z. A plot of the
S-param eters is show n in Figure 3.4. At 20 GH z, the transm ission g ain increased
by 2.5dB after cooling, while the o u tp u t reflection coefficient d ro p p e d by 1.8dB
relative to room tem perature.
In this w ork, the a d d itio n of source ind u ctan ce to im p ro v e the
d ev ice stab ility w as investigated. T his so u rce in d u ctan c e p ro v id e s series
feedback to reduce the in p u t and o u tp u t VSWR. By a d d in g O .lnH source
ind u ctance to the FHR10X HEM T m odel, the stability factor, K, o f the device
increased from 0.54 to 1.1 at 20 G H z, w hile the gain d ro p p e d by 0.6dB. (The
stability factor is defined in term s of the scattering param eters, a n d m u st be
g reater than one for a transistor to be unconditionally stable).
Tw o sets of m easurem ents at 77K w ere also p e rfo rm ed on the
device
to confirm the effects of d ifferent bond w ire source inductance. The
increase in inductance w as achieved by lengthening, an d reducing the nu m b er of
bond w ires from the source to ground. As a rule o f thum b, the 0.0007" diam eter
g o ld b o n d w ires have an a p p ro x im a te in d u c ta n c e o f 0.8 n H /m m . T he
m easurem ents w ere taken from 1 G H z to 26 G H z, e n d th e S -param eters an d
stab ility factor are disp lay ed in Figure 3.5, along w ith th o se o f th e m odel
(FHRM ODL2 and FHRMEAS2 contain O .lnH a d d e d source inductance). The
d iscrep an cy betw een the m easu rem en ts and m odel o ccur d u e to th e fact a
d ifferen t device w as used in each case. Slight v a ria tio n s in th e p rocessin g
coi d it ions for different batches of devices can alter th eir characteristics slightly.
I low ever, the general trend of increasing stability, an d larger reduction in gain
over frequency is apparent in both cases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
S?2
* fHHWCt.2
Sll
’ fUM GKl
tMMEASl
su
‘ :>«K(W2
FMMFAS?
ri: 1
B: 21.MM
11.w
Figure 3.5 Effect of source inductance on HEMT performance
Certain cooled HEMT devices are known to be sensitive to light.
These HEMTs at cryogenic temperatures exhibit superior performance only
under illuminated conditions in which they are time-invariant, memoryless
devices.15*! In darkness, the threshold voltage shifts and distortion of the I-V
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
characteristics occurs. Illum inating, or heating the device u p to approxim ately
130K restores the initial device characteristics. For a Fujitsu FHR01X HEMT, it
w as reported that illum ination affects the I-V characteristics a t relatively low
drain voltages, w hile for V[>s>2V the effect is m inimal.Is7! The collapse of the I-V
characteristic in darkness is attrib u ted to the existence of DX centres in the
AlGaAs layer giving rise to trapping effects. A DX centre is a deep d o n o r level in
which electrons freeze out, or are "trapped”.
M easurem ents of a Fujitsu FHR10X HEM T w ere taken in both a
partially illum inated (i.e. cryocooler partially shields device from fluorescent and
natural light) and dark environm ent. W hen the device w as cooled to 77K u n d er
partial illum ination, a gate voltage of Vt,s=-0.564V w as required to achieve 5mA
d rain current at V'd s =2V . W hen the illum ination w as rem oved, the d rain current
increased to 7mA at the sam e Vgs- A second cooling cycle w as p erform ed in
which the HEMT w as in com plete darkness. To achieve Ips-S m A a t V d s = 2 V , a
slightly low er gate voltage of -0.584V w as required. Subsequently exposing the
device to illum ination caused no im m ediate appreciable changes in th e bias
c o n d itio n s. For th e am p lifie r d esig n s, S -p a ram e te r m e a su re m e n ts w ere
p erfo rm ed u n d e r p artial illum ination to rem ove the p ossibility o f electron
trapping, and to m aintain a consistent operating environm ent.
3.3 Amplifier Design Procedure
Low noise am plifiers are an integral co m p o n en t of m icro w av e
receiver system s in m inim izing the system noise figure. In very low noise
receivers, cryogenic am plifiers are im plem ented to achieve further noise figure
reduction. Not only does cooling the LNA im prove the noise tem perature, it also
increases the transm ission gain and provides a controlled tem perature operating
e n v iro n m e n t. C ry o g e n ic T em peratures a re a lso id e a l
fo r c o m b in in g
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sem iconductor devices with superconducting passive structures. In an amplifier,
for instance, the input and o u tp u t m atching circuits could be fabricated using
superconductors. This w ould reduce the losses of the m atching netw orks, and
therefore lower the overall amplifier noise figure.
A single-stage K-band am plifier w as designed for m axim um gain
using m icrostrip HTSC m atching networks. Although the design does not utilize
noise p aram eters to achieve optim um noise perform ance, the am plifier is
inherently low noise d u e to operation at cryogenic tem peratures. The procedure
for obtaining noise param eters at low tem peratures requires a tuning mechanism
w hich w ould be difficult and lengthy to im nlem ent in the present cryogenic
system . The overall circuit complexity was kept to i m inim um.
A narrow band design configuration w as used in which (In* input
and o u tp u t m atching netw orks consisted of single open-circuited shunt stubs.
The m icro strip line w id th s w ere d eterm in ed u sin g l,in eC alcIM, and are
d ep en d en t on the characteristic im pedance, the effective dielectric constant, and
the substrate thickness. To achieve optim um gain at the centre frequency, the
stubs w ere designed to conjugate m atch the device using S-param eter data from
the FHR10X HEMT m odel at 77K. Standard im pedance m atching techniques lor
single-stubs were used as follows:!1’8!
1) The complex conjugate of the input reflection coefficient, S t \ 4, at 20( A 1/
w as located on the Smith Chart.
2) A rotation along the constant VSWR circle w as m ade in the direction of
the load until it intersected th e C = l circle.
3) The d istan ce traversed in w av elen g th s aro u n d the Sm ith C hart
determ ined the placem ent of the stub aw ay from the device.
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
56
4) From the B=0 p oint, a rotation w as com pleted a ro u n d th e o u te r
perim eter of the Smith C hart until the adm ittance m atched th at from step
2). This rotation in w avelengths corresponded to the length of the stub.
The sam e procedure w as repeated for the o u tp u t reflection coefficient, S22*/ in the
ou tp u t m atching netw ork.
The initial transm ission lines an d stu b len g th s w ere o p tim ized
using T ouchstone[M to achieve m inim um input and o u tp u t reflections at 20 GHz.
The final optim ized values of the in p u t m atching netw ork d eviated by only 2% of
the initial predictions. The o u tp u t m atching stub v a rie d o v e r 50% from its
original value, indicating a high degree of insensitivity to changes in length.
Figure 3.6 Photograph o f assem bled H T SC /H E M T am plifier
The HTSC. m atching netw orks w ere fabricated at S upercon d u cto r
Technologies Incorporated (STI) using TBCCO thin film s grow n on both sides of
a 254pm thick lanthanum alum inate substrate, as described in C h a p te r 2. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TBCCO film becom es su p e rc o n d u c tin g a t a ro u n d 105K, and m ea su re m e n ts
sh ould be carried o u t well below this transition tem p eratu re (77K in this case) in
o rd e r to av o id kinetic in d u ctan c e effects. C ircu it assem bly m u st th erefo re
p ro v id e a very good therm al contact and therm al expansion m atch betw een the
circu it a n d carrier. C arriers w ere m ad e of go ld -p lated kovar, w ith the HTSC
circuits attach ed using e ith er a low tem p eratu re silver epoxy o r indium solder.
The silv er epoxy c o m p o u n d cures in five m inutes at 150°C, w hich sh o u ld not
d e g ra d e the TBCCO thin film s. The in p u t a n d o u tp u t m atch in g circuits w ere
epoxied o n to the carrier, on either side of a centre pedestal p ro tru d in g from the
carrier. T he sem ico n d u cto r device w as then m o u n ted o n to this pedestal using
silver epoxy. The gate and d rain of the device w ere connected to the gold contact
p a d s on the m atch in g n e tw o rk s usin g 0.0007" d iam etei gold w ire bonds. The
so u rc e p a d s of th e HEM T w ere g ro u n d e d to the c a rrie r via w ire b o n d s. A
p h o to g ra p h o f the assem bled am plifier is show n in Figure 3.6.
T he b ias c irc u itry consists of h ig h im p e d a n c e /lo w im p ed an ce
q u a rte r-w a v e tran sm issio n line n e tw o rk s on the g a te a n d d ra in side. T hese
n e tw o rk s p resen t an effective open circuit at the design frequency to prevent the
sig n al from p ro p a g a tin g into the bias circuit. T he high im p ed an ce line h as a
w id th of 25|im , w hich is the m inim um allow able feature size according to STI's
specifications.I44! To sim plify the analysis and assist in m in im izin g reflections,
the b ias feed line wras placed o p p o site the sh u n t stub. In the m id d le of each of
th ese netw o rk s, a 50ft resistor and a lOOpf capacitor w ere placed in series and
g ro u n d ed to the carrier to help stabilize the am plifier at the low er frequencies. A
2.2kft resistor w as a d d e d in the bias circuit on the gate side to protect against
excessive gate cu rren ts. The laser trim m ed chip resisto rs and m u ltilay er chip
capacitors w ere attached to the substrate using the low tem p eratu re silver epoxy.
The resistor and capacitor in the stabilizing configuration w ere connected w ith
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
wire bonds, and the resistor w as grounded to the carrier using epoxy. DC biasing
w as applied to the gate and drain via ultrasonic b o nded 0.0005" x 0.003" gold
ribbons. These ribbons connect to tw o filtercons attached to the centre section of
the test fixture, w hich in tum connect to the pow er supplies through phosp h o rbronze bias lines in the cryogenic test setup. D ue to the em phasis on design
sim plicity for these ..nplifiers, DC blocking capacitors on the in p u t and o u tp u t
5012 transm ission lines were om itted. Hence, it w as also possible to apply the bias
directly through the netw ork analyzer.
T w o slig h tly
d iffe re n t v a ria tio n s
of th e a m p lifie r w e re
im plem ented, designated as am plifiers 'A' and 'B'. A m plifier 'A' w as designed
and fabricated according to the preceding procedure. A schem atic diagram of
this am plifier is presented in A ppendix 1. H ow ever, d u e to a shortage of Fujitsu
FHR10X HEMTs, an d w ith the availability of PHEM T devices fabricated at
C om sat, it w as decided to substitute a PHEM T into this am plifier. W hile the
design w as based on a sm all-signal m odel for a HEMT, the PHEM T used h ad
identical gate dim ensions and also offered the possibility of b etter perform ance.
At 20 G H z, the gain and stability of a m easured PHEMT device, as com pared to
the HEMT m odel, is displayed in Table 3.2. Am plifier 'B' contained alterations to
the m atch in g netw o rk s, based on em pirical m o d ificatio n s of a p re v io u s
HTSC/H EM T LNA.I17! This involved adjusting the lengths of the open-circuited
stubs as well as the transm ission line betw een the stu b s and the transistor.
Results of both am plifiers are presented in the following section of this chapter.
1S2i 1 dB
K
PHEMT
6.5
1.13
HEMT
6.0
1.08
Table 3.2 Com parison of HEMT and PHEMT devices at 20 G H z
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A second iteration am plifier design (am plifier C) undertaken at a
later d ate attem pted to im prove on the stability. This new design incorporated a
resistor on the drain side to stabilize the device at the expense of transm ission
gain. The m atching netw orks were designed in the sam e m anner as the previous
am plifiers, with the exception that S-param eter data from the m easured I Il'MT in
Figure 3.5 w ith higher source inductance w as used. A nother feature incorporated
into this design w as the addition of DC blocking capacitors. The results of this
am plifier are presented in the next section. A brief description of each of the three
am plifiers is provided in Table 3.3.
Amplifier
Device
Stabilizing
Resistor
A
PHEMT
No
B
HEMT
No
C
HEMT
Yes
Table 3.3 Com parison of design characteristics for three amplifiers
3.4 M easu rem en ts and R esu lts
This section presents S-param eter m easurem ents at 77K of the
am p lifiers described in the prev io u s section and com pares the resu lts to
sim ulations using Touchstone™ . The original set of am plifiers (A and B) were
designed specifically for m axim um gain at 20 GHz. Noise figure results an* also
p resen ted for am plifier 'B'. The later version am plifier C
w as designed to
ad d ress the problem of stability in greater depth. An analysis of the individual
m atching n e tw ... ks using various CAD softw are packages for am plifier 'C , as
well as a previous am plifier,l,7J were conducted.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
S-param eter m easurem ents w ere perform ed using the cryogenic
technique described in section 2.5. The am plifier was inserted betw een the tw o
halves of the test fixture and connected with w ire bonds, as show n in Figure 3.7.
A significant step in line w idth occurs betw een the 5012 transm ission lines on the
alum ina substrates and the input and o u tp u t lines on the h ig h dielectric,
superconductor substrates. To minim ize any reflections d u e to the step junction,
tapered contact p ad s were used at the am plifier in p u t and o utput. A T hroughReflect-Line (TRL) calibration w as perform ed as described in section 2.5 over a 1
CiHz to 26 GH z frequency range. M easurem ents w ere perform ed using either the
11P8510C or the W iltron 360 Autom atic N etw ork Analyzer.
Figure 3.7 Photograph of HTSC/HEM T am plifier 'B' in test fixture
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.4.1 HTSC/PHEM T A m p lifier ’A' R esu lts
The am plifier based on a 0.25pm x 100pm Com sat PHHMT device
w as m easured at a bias condition of 2V drain-to-source voltage and fan A drain
current. These are the specified room tem perature bias values for optim um noise
perform ance. The m easured amplifier gain 1S2 1 1 in dB is displayed in the graphs
of Figure 3.8, along with predicted results. The gain reaches a peak value of 13dB
at 19 G H z with corresponding input and o utput reflections of -14dB and -7dB,
respectively. The discrepancy in centre frequency betw een the original design
an d this m easurem ent is attributed to the use of a different type of device
(PHEMT vs. HEMT). The Touchstone sim ulation predicts a slightly lower gain
p eak w hich is shifted d o w n in frequency by 500 M Hz. This sim ulation
incorporates device m easurem ents of a quarter micron gate length I’MFMT .it
77K along w ith sim ulated HTSC matching and bias networks.
The am plifier response w as found to be extremely sensitive to tire
value of source inductance. By varying the source inductance slightly in the
sim ulation to approxim ate the inductance caused by the w ire bonds, a closer fit
b etw een sim ulation and m easurem ent m ay be found. Lowering the source
ind uctance w ill cause a dow nw ard shift in frequency w ith a corresponding
increase in gain and decrease in stability factor, K. The discrepancy betw een
sim ulation and m easurem ent can be further attributed to the use of a different
dev ice in each case.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
—~©— Measured
-
-10
-
0 - - Touchstone
—
0
5
10
15
20
frequency (GHz)
25
30
0
5
— O — Measured
- E - - Touchstone
-10
-1 5
5
10
15
20
frequency (GHz)
25
30
Figure 3.8 a) and b) M easured vs. predicted m atching for am plifier A'
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
10
5—
-a
-10
0
5
10
15
20
frequency (Gl I/.)
25
Figure 3.8c) M easured vs. predicted gain of am plifier 'A'
A second gain peak occurs at a ro u n d 8 G H z in both th e
m e a su re m e n t an d sim ulation. This seco n d ary peak in dicates a poten tially
u n stable device, as confirm ed by the in p u t reflection ctH*fficient, IS n I w hich
sh o w s a tendency to w ard s oscillations. This w as likely caused by extraneous
effects in the bias circuit. An attem pt w as m ade in the second iteration am plifier
'C design to elim inate this instability problem .
3.4.2 HTSC/HEMT Amplifier 'B* Results
Am plifier ’B' w as designed and fabricated using the original Fujitsu
HEM T device w ith slight variations in the open-circuited stub lengths. Those
stubs w ere lengthened and placed closer to the transistor to shift the gain peak
u p w a rd s in frequency. R esults of this am plifier o v er a 1 G H z to 23 G H z
frequency range are displayed in Figure 3.9. The transistor w as biased at its room
tem p erature optim al noise bias condition of 2V drain-to-source voltage and 5mA
d ra in c u rre n t. T he g ain reach es a peak valu e of 7.8dB at 20.5 G H z.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
C o rrespo nding return losses are -7.5dB on the in p u t and -9dB o n the o u tp u t.
Touchstone predicted a m axim um gain of 9.7dB at 20 G H z w ith input and o u tp u t
m atching below -15dB. The low er gs.in peak and h ig h er frequency of the
m easured am plifier is attributed to a h igher source inductance for increased
stability. Touchstone predicted that an increase of 0.05nH source inductance for
the sim u la ted am p lifier w ould m ore closely m atch th e m easu rem en ts. A
secondary peak in gain em erges at the lower frequencies (10 GHz) w hich is again
attributed to effects in the bias circuit.
10
CO
T3
ri
<S>
-5
-©— Measured
El - - Touchstone
-10
10
15
frequency (GHz)
20
25
Figure 3.9a) M easured vs. predicted gain of am plifier 'B'
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t*5
l
'Vi
y
-5
OQ
TJ
-1 0
I
I
t
15 “
O--- Measured
- 1]
____ I_____
i
20
0
5
Touchstone
- ~ l - t
10
15
frequency (CIHa)
20
25
0 —
-10
-15
0
5
15
10
frequency ((iHs.)
20
Figure 3.9b) and c) M easured vs. predicted m atching of am plifier B'
N oise figure w as m easured for this am plifier using the setu p
sh o w n in Figure 3.10. An HP346C n o 5se source w as connected to the input
v acu u m feed th ro u g h p o rt of the cryogenic cham ber. The o u tp u t vacuum
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
feedthrough w as connected to the HP8970 noise figure m eter via an isolator an d
external mixer. An LO signal is generated using the 83570A RF plug-in connected
to a sw eep oscillator. The losses of the external m ixer and cables are rem oved by
directly connecting the noise source to the iso la to r/m ix e r on the o u tp u t an d
p erform ing a through calibration. The noise d io d e w as then connected to the
device u n d e r test (DUT) so the noise figure, N Fmt.as, could be m easu red . This
value includes the noise contribution from the K-cables and test fixture inside the
cham ber as well as the DUT.
NFm eas
1
N oise D iode
1
L
GDUT/ N Fpyy
M ix er/
Isolator
L
N F Meter
I
Figure 3.10 Block diagram for noise figure m easurem ent setu p
U sing the sta n d a rd expression for cascad ed noise fig u res, th e
m easured noise figure can be w ritten as,
N Fm, a, = NF, + (NFIXJT-1 )LS+
(NF -l)L j
1
^DUT
(3.5)
w here NF,=L, corresponds to the noise figure of the in p u t losses a n d NF0 is th e
noise figure of the o u tp u t losses. The DUT has a gain, G du T/ an d noise figure,
N I' d u i . The losses d u e to the in p u t and o u tp u t K -cables an d split-block test
fixture w ere characterized by connecting the tw o halv es of th e test fix tu re
together to form a through, and m easuring the transm ission response u sin g the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
o
netw ork analyzer. As the K-cables are of equal length, half of this loss, 1 1 , 1
is attributed to the input and thus directly contributes to the overall noise figure.
T he noise contribution of the o u tp u t side is reduced by the HUT gain. The
corrected DUT noise figure is obtained by rearranging equation 3 5 to give,
ME
NFmt,,s
n f d u i = — «—
(L-l)
n —
V,DUI
*
1 (>)
T he noise figure of the IIT SC /IIH M T am plifier IV w as ftniml
indirectly by determ ining the noise figure of the hybrid filter/am plifier circuit,
d escribed in C h a p te r 4. The noise figure of the stan d -alo n e am plifier w as
d eterm ined by subtracting the noise figure of the filter (i.e. insertion loss in dll)
from the total filter/am p lifier noise figure (in dB). At B>.8 C l I/,, and biasing of
5mA d ra in cu rren t at 2V, the am plifier noise figure w as found to be I.OdIV By
lo w erin g the d ra in cu rre n t to 3m A, a m inim um noise figure of 0.8dB w as
extracted for this am plifier. N ote that these values w ere m easured at frequencies
slig h tly offset from the am plifier peak, and therefore may not be the true
m inim um values.
An attem pt w as m ade at directly m easuring the noise ligure of the
am plifier at the gain peak of 20.5 GHz. This w as rather inaccurate, how ever, due
to relatively large variations above 20 G H z in the transm ission response, 15?t I ,
of the through loss in the K-cables and test fixture. These m easurem ents betw een
18 G H z and 21 G H z are show n in Figure 3.11. Also, the necessity for using bias
T's as a DC block further decreased the m easurem ent accuracy. I or this reason,
r.o noise figure m easurem ents were perform ed for the HTSCVl’HFMT am plifier
'A'.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
LOG nRG.
► REF--4.000dB
l.O O O dB 'D IU
S21 FORUARD TRANSMISSION
1 8 .0 0 0 0
21.0000
Figure 3.11 Measurement of "through" loss in K-cables and test fixture
GHz
3.43 2nd Iteration Amplifier 'C
A second iteration amplifier design was implemented to achieve an
improvement in the stability of the device and remove the secondary gain peak
present at lower frequencies. This involved adding a 300Q shunt resistor on the
drain side of the device to suppress the gain at frequencies outside the passband,
without degrading the noise figure. The resistor was connected to the 50Q output
line via a 1.9mm transmission line, and grounded to the carrier using conductive
epoxy. A schematic diagram of this amplifier is illustrated in Appendix H
The measured S-parameters displayed in Figure 3.12 show a peak
gain of 7dB at 15 GHz. This does not agree with the predicted peak response of
9dB at 20 GHz. While the input reflection at 20 GHz matched the predicted
response, the measured output reflection yielded a minimum of -18dB at 15 GHz.
The poor matching at the output caused the significant shift in the peak. This is
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
attributed to inaccuracies in modelling the resistive stabilizing network. For
instance, the laser-trimmed chip resistor (0.9mm x 0.5mm) was found to display
an appreciable reactive component at these frequencies. Re-simulating the
amplifier in Touchstone with this added reactance provided very close
agreement with the measurements. A 5% increase in the DC resistance at 77K as
compared to the room temperature value had a minimal effect on the amplifier
response.
lO.oooaeH
lO .O O O J B ^ ►O.OOOdB
LOGM
S ll
1 .0 0 0 0
GHZ
LOGM
2 6 .0 0 0 0 } 1 . 0 0 0 0
GHZ
LOGM
2 6 .0 0 0 0 1
GHz
2 6 .0 0 0 0
1 0 .OOOdEM ►O.OOOdB
1 .0 0 0 0
GHz
2 6 .0 0 0 0 ! 1 .0 0 0 0
Figure 3.12 Measured S-parameters for 2nd iteration amplifier ‘C*
To verify the performance of the matching networks, separate test
jigs were created to allow for measurement of the individual matching circuits.
Measurements and the corresponding Touchstone simulations are illustrated in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
the Smith Chart of Figure 3.13. While a phase shift is evident for the input
matching network, the output matching circuit consisting of the stabilizing
network and matching stub displays some resonance effect. This resonance is
caused by the reactance of the stabilizing chip resistor, as explained above. These
matching circuits contained slight deviations in their design as compared to
those of the measured amplifier.
1
Sll
IPMATCH
S22
c
o p iw t c h
v
f t : 1 8 .0 4 0 0
(?: 22.0000
2
S
1
2
Figure 3.13 Performance of input and output matching networks
3.4.4 CAD Software Analysis of HTSC Circuits
A comparison of various CAD software packages was undertaken
to simulate single-stub HTSC matching networks for a 14 GHz narrowband
HTSC/HEMT amplifier designed at the Communications Research Centre
(CRC).fl7! S-parameter measurements of separate input and output matching
networks were compared to simulations using Touchstone, EMSim™ by EEsof,
and Sonnet's EM™. Boh. EMSim and EM employ a full-wave analysis technique,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
w hile T ou ch sto n e utilizes a q u asi-static approach. The full-wTave analysis
considers the effects of coupling and surface waves, as well as radiation loss. The
results from 12 G H z to 16 G H z are displayed on the Sm ith C hart in Figure 3.14.
For the in p u t circuit, EMSim provides the closest m atch to the m easurem ents.
T ouchstone offered a better prediction of the o u tp u t m atching netw ork. It should
be n o ted th at the external bias circuitry w as om itted for the EMSim and EM
sim ulations for sim plicity of analysis. W hile EMSim required only 24 patching
seg m en ts to sim u late the in p u t m atching, EM u sed 142 segm ents an d thus
req u ired a m uch longer co m putation time. Touchstone, w ith its w ide range of
m odelling elem ents and su p erio r speed, proved a d eq u ate for analysis of these
sim ple HTSC passive structures.
= 5«1
L ib ra
C*
1
E M Sim
o —1i
Measurement
x 5.1
EM
rl: 12.0000
r 2.
?
15.C 000
Figure 3.14a) Results of in p u t HTSC m atching netw ork
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
L ib ra
4*
- - -
EM Sim
O
Measurement
X
522
EM
fl:
12.0000
f2
16. 0 0 0 0
.2
5
1
2
^
Figure 3.14b) Results of output HTSC matching network
3.5 Summary
This chapter presented the design, fabrication and testing of three
K-band HTSC/HEMT amplifiers. The amplifiers were based on a small-signal
model of a FHR10X device, which pred.cted a 2.5H” increase in gain at 20 GHz
and 77K, as compared to room temperature operation. The device stability was
highly sensitive to the source inductance caused by source-to-ground bond
wires. Increasing the source inductance by O.lnH in the device model to
approximate the longer bond wires improved the stability factor, K, from 0.54 to
1.08 at 20 GHz and 77K.
The amplifiers were designed for maximum gain by implementing
high temperature superconductor matching networks at the device input and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
o u tp u t. A m plifier 'A'
th iev ed a passband gain of 13dB at l l> G H z, with
corresponding m atching of -14dB and -7dB on the input and output, respectively.
A m plifier B' produced 7.8dB gain at 20.5 GHz, and noise figures of l.OdB with
5mA d rain c u rren t and 0.8dB w ith 3mA drain current at 19.8 G H z. The noise
figures for the am plifier w ere extracted from m easu rem en ts of the I1TSG
f ilte r /a m p lif ie r in te g ra te d
u n it. T hese v a lu e s ag ree w ith a K u-band
H T SC /H E M T am plifier developed at CRC which reported a m inim um noise
figure of 0.4dB at 13.2 GHz.I17! The results exceed the perform ance of a two-stage,
packaged FHR10X HEMT LNA developed at CRC which produced 1.9dB noise
figure a t 20 G H z and 297K.IWI
A second iteration H T SC /H E M T am plifier 'C w as designed lor
im proved stability by including a sh u n t resistor at the transistor o u tp u t. The
m easured response displayed a significant shift in the passband from 20 Gl lz to
15 G H z. These e rro n e o u s resu lts are m ainly a ttrib u te d to inaccuracies in
m odelling the chip resistor at high frequencies and 77K.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
CHAPTER 4: INTEGRATED SUPERCONDUCTING
FILTER/AMPLIFIER PERFORMANCE
4.1 Introduction
H igh T em perature Superconductors (HTSCs) have p ro d u ced very
high Q reso n ato rs d u e to th eir extrem ely low loss p ro p erties.!1M7,60i These
ach ievem ents in the quality factor m ake HTSC m aterials d esirab le for very
n arrow b andw idth filters. Thin-film plan ar circuits u sing conventional m etals are
considered too lossy to achieve the dem anding specifications on b an d w id th , such
as those required in satellite com m unications. N ow , filters fabricated using HTSC
p lan a r technology such as m icrostrip and co p lan ar w av eg u id e can achieve
com parable perform ance to w aveguide filters at a significant reduction in size.
The cost of cryogenically cooling m icrow ave c o m p o n en ts can be offset by
im plem enting m any HTSC com ponents in a filter bank, for instance. Filter banks
are frequently used in w ide-band rad a r and com m unication system s to select a
m uch n arro w er ban d w id th signal. The current use of w av eg u id e technology in
filter banks requires a relatively large volum e.
In m icrow ave receiver system s, the low noise am plifier is th e key
co m p o n en t in reducing the noise figure an d im p ro v in g the sensitivity of the
s stem. At the receiver input, the am plifier reduces the noise contributions of all
fo llo w in g receiver co m p o n e n ts su ch as m ixers, filters a n d IF am p lifie rs.
1lo w ever, to avoid interm o d u latio n p ro d u c ts by signals o u tsid e th e d esired
frequency band, a b andpass filter at the receiver in p u t is desired. The losses d u e
to this filter will directly affect the overall noise figure. By im plem enting a HTSC
filter, how ever, the insertion loss can be reduced, th u s low ering the noise figure.
The first twfo sectio n s of th is c h a p te r p re s e n t th e d e sig n ,
im plem entation, and testing of high-Tc su p erconducting m icrostrip filters. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
filter responses at 77K are presented and com pared to predicted results using
T ouchstone and EMSim. The effect of tem perature variation on filter response is
investigated. Also, an analysis on the individual resonator sections of the filters is
p erfo rm ed to determ ine the coupling, as well as the accuracy of these CAD
softw are packages. M easurem ents and sim ulations are presented later in the
chapter.
The final section of the chapter reports the integration of the 11 I SC'
filters w ith their respective am plifiers described in the previous chapter. The S
p aram eter and noise figure m easurem ents are presented along with predicted
resu lts for both hybrid circuits. The problem of stability d u e to m ism atch
betw een filter and am plifier is addressed.
4.2 HTSC Filter Design Procedure
Tw o 3-pole C hebyshev filters w ere d esig n ed to p ro d u c e a
n a rro w b an d (2.3%) response at centre frequencies of W C H z and 20 ( i l l / to
coincide w ith am plifiers A’ and ’E’, respectively, described in C hapter 3. Both
d e sig n s im p lem en t q u arler-w av e p arallel-coupled m icrostrip resonato rs to
ach iev e a p a ssb a n d rip p le of O.OldB. A th ree-p o le configuration w ith a
m o d e ra te ly n a rro w b a n d w id th w as chosen for sim plicity of design and
im p lem en tatio n , as this w as a first iteration design using HTSC thin-lilm s.
S tan d ard design pro ced u re using parallel-coupled m icrostrip resonators w as
followed. I61i
The first step in the filter design procedure w as to determine- tinlow pass prototype elem ent values go to g 4 . For a Chebyshev response with an
od d n u m b er of sections, the first and last elem ents w ere chosen to be go -g 4 =l.
The rem aining elem ent values are determ ined using the Tables in [61J based on
the desired passband ripple. To convert the low pass prototype into a b jn d p ass
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
filter, a transform ation of the low pass frequency, o)’, to the bandpass frequency,
0 ),
is given as.
0)
2
( d '~ w
(4.1)
“b
w here o>i' is the cutoff frequency of the low pass filter,
w
is the fractional
b a n d w id th , and too is the centre frequency. The roll-off is determ ined from
(D*
grap hs of the attenuation, La, versus the frequency ratio, ——J6,l A three-section
to ,
filter w ith an attenuation of 15dB corresponds to a frequency ratio
——
(!)
= ±3.2 .
S ubstituting these tw o values into equation 4.1 yields the u p p e r an d low er
frequency points which correspond to an attenuation of 15dB.
Once the low pass prototype elem ents have been determ ined, they
can be used to calculate the im pedance inverters. These inverters are used to
convert the bandpass filter netw ork into one using only series tuned circuits. For
a C hebyshev filter w ith N sections, N +l im pedance inverters are required. D ue
to the symmetrical nature of Chebyshev filters with an odd num ber of poles, only
the first (N + l)/2 inverters need to be com puted. The in v erter sections are
expressed as.
(4.2)
After the im pedance inverters have been com puted, the even and o d d m ode
characteristic impedances, Zot. and Zo0, can be determ ined using the form ulas,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
From these im pedances, the resonator lengths, w id th s and gap
spacings w ere com puted using LineCalc for each of the coupled line sections. A
s h u n t capacitive loading at the e n d s of the open-circuited coupled lines
effectively lengthens each line. To offset this fringing capacita.ae, the physical
lengths of these resonators w ere reduced by an am ount proportional to the*
capacitive loading a t each end. The length reduction, Al, can be expressed as
w here eeff(f) is the adjusted effective dielectric constant at the centre frequency
d u e to dispersion effects.I62l N ote that this value is sim ply an initial estim ate,
since dispersion in superconducting lines is practically negligible.
The resonator dim ensions w ere im plem ented in Touchstone lor
fu rther optim ization using the m icrostrip coupled-line filter elem ent, MCTII,. A
sum m ary of the coupled-line section dim ensions is described in Table 4.1. All
optim ized dim ensions deviated from the initial design calculations by less than
1.5%, except for the gap spacings of the 20 G H z filter which d iffered by
approxim ately 10%. These filter elem ents w ere analyzed using a quasi-static
m ethod which is m ainly valid for substrate relative dielectric constants less than
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
18JMI For this reason, a full-wave analysis package. EMSim, w as used to verify
the Touchstone sim ulations. EMSim incorporates an integral equation technique
using G reen’s Function to account for surface w aves and radiation loss from the
resonator coupling sections.
m iddle sections
first and last sections
(pm)
(pm)
length
w idth
spacing length
w idth
spacing
237
19 GHz
973
943
75
968
83
20 GHz
79
917
237
914
84
953
Table 4.1 Resonator dim ensions for the HTSC filters
The filters w ere fabricated at Superconductor Technologies Inc.
using TBCCO thin-films on a 2" diam eter lanthanum alum inate (er=24) substrate
wafer. A layout of the tw o inch w afer, which includes the HTSC filters and
am plifier m atching netw orks, is show n in Figure 4.1. A polyim ide passivation
layer w as applied over the substrate to protect the HTSC thin films. The 7.5mm x
,13mm filters were assembled at CRC by m ounting the circuits onto gold-plated
kovar carriers using silver epoxy. Gold contact p ad s w ere included on the 50i2
HTSC input and o u tp u t lines to allow for wire bonding to the gold transm ission
lines on the test fixture. A photograph of the 20 G H z centre frequency HTSC
m icrostrip filter is show n in Figure 4.2. Experim ental and predicted results are
covered in the next section.
To exam ine in more detail the coupling in HTSC m icrostrip lines,
individual coupled-line sections were fabricated on the STI wafer. Both w ide gap
and n arro w gap sections w ere analyzed using T ouchstone and EMSim to
com pare with experimental results.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.1 AutoCAD layout of 2" ST1 wafer containing HTSC filters
i
*
/
Figure 4.2 Photograph of 20 GHz HTSC microstrip filter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
4.3 Filter Measurements and Results
The 3-pole HTSC m icrostrip filters were m easured at 77K using the
cryogenic m easurem ent technique as described in section 2.5. S-param eter
m easurem ents w ere obtained using the W iltron 360 A utom atic N etw ork
Analyzer over the frequency range of 16 GHz to 24 GHz. The pow er levels at the
o u tp ut ports of the netw ork analyzer were set to -30dBm to reduce the sensitivity
of the unloaded Q s of the resonators. At power levels above OdBm, the resonator
Q -factor is d eg rad ed d u e to the c u rren t densities a t the co n d u cto r edges
exceeding the critical current density, Jc, of the thin-films.l44! W hile the thin-film
TBCCO becom es superconducting below 105K, the filter response does not
stabilize until the tem perature d ro p s below 90K, d u e to the kinetic inductance
effect. The tem perature dependence of the filter response will be exam ined later
in the section.
4.3.1 19 GHz HTSC Filter
The insertion and return losses of the 19 G H z filter are displayed in
Figure 4.3, along with the predicted results using T ouchstone and EMSim. The
m easurem ents show a 0.78dB insertion loss at a centre frequency of 18.65 GHz,
with return losses of 17dB and 18dB at the filter input and output, respectively.
The T ouchstone response predicted an insertion loss of 0.56dB a t a centre
frequency of 18.8 GHz. EMSim predicted a significantly higher centre frequency
of 19 G H z with a corresponding 0.66dB insertion loss. The S-param eters of
m easured and sim ulated results at th eir respective centre frequencies are
displayed in Table 4.2. A further analysis of the filter w as perform ed using a
finite difference (FD) m ethod softw are package d ev elo p ed at M cM aster
University.I*'4! While the com putation time was extrem ely long (m ore than three
hours per frequency point using a SUN W orkstation), this softw are predicted the
centre frequency of the filter m ost accurately. Results w ere obtained over the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SI
frequency range 18 G H z to 19 G H z, w ith the centre frequency S-parum otors
displayed in Table 4.2. The finite difference m ethod assum ed a perfect conductor,
w hich explains the extrem ely low predicted insertion loss.
CO
"O
-15
-20
-2 5 -
-30
-40
17
17.5
18
21
18.5
Frequency [< 0 1 / j
0
-5 ■
t
t
-10
r .
..
o
measured
Touchstone
HMSim
-15
CQ
^ -20
i
'J -25
-30
-35
-40
17.5
18
1 8 .5
Frequency [<41/.J
Figure 4.3 M easured vs. sim ulated response for 19 G1 Iz I I IS( filter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
fo(GHz)
IS11I
IS211
IS22I
M easured
18.65
-17
-0.78
-18
Touchstone
18.8
-35
-1.56
-35
EMSim
19.0
-28
-0.66
-28
FD
18.6
-25
-0.1
-25
Table 4.2 S-param eters (in dB) at the centre frequency for the 19 G H z HTSC filter
To com pare the perform ance of the HTSC filters to co nven tio n al
m etals, a gold m icrostrip filter on LaAlOs w as sim u la ted in T ouchsto n e. A n
iden.ical configuration w as used for the gold filter, to p ro d u c e a n a rro w 2.4%
p assband at a centre frequency of 18.8 G H z. U pon o p tim iza tio n in T ouchstone,
the filter had a m inim um predicted insertion loss of 2.8dB at 297K and 1.7dB at
77K.
W hile b o th the T ouchstone a n d EMSim re sp o n se s e stim a te d a
h ig h er centre frequency, the quasi-static an aly sis w a s w ith in 150 M H z o f the
experim ental value, representing an erro r o f less than 1%. T he full-w ave analysis
u sin g EMSim show ed a 2% e rro r in the cen tre frequency. T his can be p a rtly
attrib u te d to a discrepancy in the relative dielectric constant, e r. W hile EMSim
did not predict the centre frequency of the filter response as w ell as T ouchstone,
th e full-w av e a n aly sis d id p re d ic t th e a sy m m e tric sh a p e of th e freq u en cy
resp o nse m ore accurately. R e-sim ulating in EMSim u sin g t'r=24.8 p ro v id e d an
excellent m atch to the actual response. The v ariatio n in er m ay be d u e to th e
tw in n ing effect of the lanth an u m alum inate su b strate, an d the passivation layer
(c,~3) p ro tectin g the TBCCO thin-film s. STI's m o st recen t sp ecificatio n s on
su b stra te p a ram eters reveals an em pirical value for
eT
o f 24.7. A n o th e r factor
w hich w as not accounted for in the C A D softw are is the negligible d isp ersio n in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
su p erconductors. H ow ever, this w ould result in a slightly low er effective
dielectric constant,
and hence an upw ard shift in frequency.
A significant cause of frequency ..nifts in filter responses is due to
the kinetic inductance effect of superconductors. This effect can result in a
d ram a tic change in the centre frequency d u e to the increased m agnetic
penetration d epth w hen the superconductor is operated close to its transition
tem perature, Tc. As the HTSC is cooled below Tt , the norm al electrons are
transform ed into superconducting electrons, w hose kinetic inductance results m
the rapid frequency shift.l111This dependence of the response on tem perature is
d isp lay e d in the g rap h of Figure 4.4. These results w ere d eterm in ed by
m easuring the 19 G H z filter response over a range of tem peratures from 'Mtk to
Tc. The advantage of using thallium -based thin films over the yttrium com pound
(YBCO) lies in the higher T0 value of TBCCO (I05K as com pared to ‘>0k tor
YBCO). W hen operating at 77K, a tem perature ratio of T /T , 0.74 achieved tor
TBCCO is superior to the T /T t =0.86 for YBCO. The kinetic in d iu ta iu e oltecl
should be m inimal in this case, as is dem onstrated in Figure 4.4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of/de
M t C R ^ r c >nr R E S O L 1r ' l O N TEST C HA R T
fniHlM 't.n
. 1 A \ ; ’ A h M ’ f-if f
A \'.'
I
Ni'b
i s , ' Tf s i
S1ANPAR[?S
M A U H 1A I
CHART
No
HV Uld
<3 i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
St
IS .8
—
. 18.ft
N
XWl
c
1 8 .4
h
z>
3
o
IS.2
c5j
u
\
18
!
1 7 .8
40
50
00
70
SO
‘10
!00
110
Ti-mpi-raliiri- (K)
F igure 4.4 Kinetic inductance effect of filter response o v er tem p eratu re
T he in sertio n loss of th e filter w as slig h tly d e g ra d e d from tinp re d ic te d v alu e d u e to an o v erestim atio n of the su rface resistance, Rs, lor the
HTSC thin films. The sim u lato rs assum e that Rs is p ro p o rtio n al to tin- square root
of freq u en cy w h ile su p e rc o n d u c to rs exhibit a f2 d e p e n d e n c e . Ify c h o o sin g a
m o d est value of resistivity (p=0.01pROtd), the Rs can be reasonably a p p ro x im ated
at h ig h e r frequencies. A n o th er factor responsible for th e d e g ra d atio n in insertion
loss acro ss th e p assb a n d is the inclusion of bond w ires at the filter in p u t an d
o u tp u t.
To im p ro v e th e filter in sertio n loss, th e re tu rn loss sh o u ld be
red u ced below 15dBJfiCil Insertion loss is now lim ited by m ism atches a! the input
a n d o u tp u t p o rts {i.e. bond w ires) as co n d u cto r losses are now sm all e n o u g h to
b e of the sam e o rd e r as m atch in g loss. The high retu rn losses have also been
a ttrib u te d to th e random tw in n in g of the LaAJO* su b strate.l^ ’l
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
The 3dB b a n d w id th for the filter w as d e te rm in e d to b e 2.5%, w hich
is slig h 'ly b ro ad e r than the 2.3% pred icted by T ouchstone. This can be a ttrib u te d
to ra d ia tio n losses w hich are not acco u n ted for in T o u c h sto n e b u t w ill h a v e a
significant effect in su p e rc o n d u c tin g filters w h e re c o n d u c to r losses a re m in im al.
EMSim, w hich co n sid ers rad iatio n loss, accurately p re d ic ts the 3dB b a n d w id th . A
p ro p e r c o v e r p laced o v e r the filter w o u ld re d u c e ra d ia tio n lo sse s, h e n c e
increasing the overall Q a n d p ro v id in g a n a rro w e r b a n d w id th resp o n se.
4.3.2 20 GHz IITSC Filter
A second filter w a s d e sig n e d for a 2.4% b a n d w id th a t a c e n tre
fre q u e n c y o f 20 G H z. T his filter also in c o rp o ra te s a th re e -p o le C h e b y sh e v
resp o n se u sin g q u a rte r-w a v e parallel-coup.
m icro strip reso n ato rs. T he d e sig n
p ro c e d u re w as identical to the 19 G H z filter, w ith the filter d im e n sio n s o p tim ize d
in T o u ch sto n e an d verified u sin g EM Sim . T he re s o n a to r d im e n sio n s a n d g a p
sp a c in g s are d isp la y e d in T able 4.1. A gain, th e p re d ic te d re su lts o v e re stim a te d
th e a ctu al cen tre freq u en cy d u e to a h ig h e r actual £r, a s sh o w n in th e g ra p h s of
E iguri 4.5. T he cen tre freq u en cy , in se rtio n a n d re tu rn lo sses a re d isp la y e d in
T able 4.3. D iscrepancies b e tw ee n m e a su re d a n d p re d ic te d resu lts a re a ttrib u te d
to the sa m e factors d iscu ssed in section 4.3.1. T his filter w ill b e in te g ra te d w ith
th e H T S C /H E M T a m p lifie r B' w h ich p ro d u c e d a p e a k g a in a t 20.5 G H z, as
d escrib ed in C h a p te r 3.
fo(GHz)
IS11 1
IS 2 1 1
IS22I
M easured
19.75
-12
-0.83
-13
T ouchstone
19.8
-33
-0.53
-33
EMSim
19.95
-16
-1.09
-16
fab le 4.3 S -p aram eters (in dB) at the centre frequency for th e 20 G H z H TSC filter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
So
t
-10
1
A
-15
Xle.iMireil
IiUk'hstoiU'
FM Sim
CD
T3
-20
-30
-35
-40
18
10.5
20
20.5
21
Frequency ( t i l l / )
-10
-15
-20
-25
-30
-35
-40
18
18.5
19
19.5
20
20.5
21
22
Frequency K i ll /)
Figure 4.5 M easured vs. sim ulated response for 20 (.1
\/.
11ISC' filter
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.3 A n a ly s is of HTSC Resonator Coupling Sections
To fu rth er ev alu ate the perfo rm an ce of th e HTSC thin film s, as well
as the1 ability of the C A D so ftw are to an aly ze them , single c o u p le d -lin e sections
identical to those used in the filters w ere m ea su re d a n d m o d elled . T he w id e g a p
sp acin g (i.e. ratio of g a p /lin e w id th > 1 0 ) of th e 19 G H z filter w as m e a su re d o v e r
thi' frequency ran g e of lb G H z to 24 G H z. T he resu lts are sh o w n in F ig u re 4.6a).
W hile T ou ch sto n e m atches the experim ental resu lts m ore closely a ro u n d 16 G H z,
.t u n d e re stim a te s the co u p lin g by ap p ro x im ately 3dB at 22 G H z. T his d e v ia tio n is
likely a ttrib u te d to the red u ced accuracy of T o u ch sto n e's c o u p le d -lin e m o d el for
d ielectric c o n sta n ts ab o v e 18. EMSim p ro d u c e d ill-co n d itio n ed re su lts a t lo w e r
fre q u e n cie s a ro u n d 12.5 G H z, b u t y ie ld e d a n e x ce llen t m atch a t fre q u e n cie s
ab o v e 18 G H z. T hese a cc u ra te re s u lts co u ld b e a ttrib u te d to th e in c lu sio n of
ra d ia tio n lo sses a n d su rfa c e w a v e s in th e fu ll-w a v e a n a ly sis w h ic h g a in
significance at h ig h er frequencies.
Eor the 20 G H z filter, in an aly sis of th e sm all g a p sp a cin g c o u p le d
line sectio n s (i.e. g a p /lin e w id th = 3) w as p e rfo rm e d from 16 G H z to 22 G H z at
77K. Both T ou ch sto n e and EMSim p red ict b e tte r c o u p lin g b elo w 18 G H z a n d less
c o u p lin g a b o v e 20 G H z, a s sh o w n in F ig u re 4.6b), w ith d e v ia tio n s fro m th e
m e a su re m e n ts w ithin IdB.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1i
li
A
lb
17
18
19
Ireq u i-i
EMSim
1 ouclistone
M easured
i
i
!
21
77
7^
• v ( ( ’. H /)
Figure 4.6a) C oupling for w id e gap resonator sec tions ol 10 Cii I/, lilter
0 r
i
i
o
-
a
-
M easured
Touchstone
EMSim
CO
a*
73
M
in
-10
-1 5
J
-20
16
17
18
19
20
frequency (C ill/,)
21
22
F igure 4.6b) C o u p lin g of n arro w gap resonator sections for 20 ( il 1/ filter
W hile th ese 3-pole, 2.3% b a n d w id th i ITSC filters h a d m o d est
sp ecificatio n s, m ore strin g e n t filter d e sig n s in the fu tu re will e m p h a siz e the
p o ten tial su p e rio r perform ance of HTSC technology as com pared to conventional
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
m etals. T hese designs, incorporating a larger n u m b er of poles to achieve very
n a rro w b an d sh a rp skirt resp o n ses (<1%), w ill allow for m ore efficient u se of
m icrow ave com m unication bands w ith m ore closely spaced channels.
4.4 H yb rid H T SC F ilter /A m p lifie r M IC
This section presents the design and p erform ance c f the integ rated
IITSC filte r/a m p lifie r MIC, based on the resu lts of th e a m p lifiers a n d filters
d escribed in the p rev io u s sections. T he 18.65 G H z cen tre frequency filter w as
in teg rated w ith a slightly m odified version of the H T S C /P H E M T am p lifier A'.
R esults of th is h y b rid c irc u it h a v e b een p u b lish e d in th e 23rd E u ro p e a n
M icrow ave Conference.!67! T he 19.75 G H z filter w as p a ire d w ith th e H T S C /
HEMT am plifier B’.
4.4.1 19 G H z H T SC F ilter /A m p lifie r M IC
A m plifier A' described in C h ap ter 3 h a d 13dB gain at 19 G H z, b u t
w as fo u n d to be p o ten tially u n sta b le a t certain freq u en cies. T h erefo re, th e
a m p lifie r w as m odified to achieve b e tte r stab ility by a d d in g a la rg e r so u rce
in d u c ta n c e . O p e n -c irc u ite d s tu b s w e re re -o p tim iz e d to ach iev e m in im u m
reflections at 19 G H z w hen m atching to m ea su re d S -p aram eters of a 0.25pm x
lOOgm C om sat PHEM T device at 77K. As a result of the increased stability, the
am p lifier had a lo w er p red icted gain of 9dB. A lso, a lOpf chip c ap acito r w as
a d d e d to the o u tp u t netw o rk to p ro v id e DC blocking, a n d th e b ias circuit w as
a lte re d slig h tly in o rd e r to rem o v e an y p o te n tia l c o u p lin g w ith th e filter.
S im u latio n s show ed th at a high ou t-o f-b an d reflection coefficient o f th e filter
w o u ld also create a p o ten tial stab ility p ro b lem at c e rta in freq u e n cie s w h e n
connected to the am plifier, an d th a t a section of tran sm issio n line b e tw ee n the
filter o u tp u t and am plifier in p u t w o u ld elim inate this problem . A s the filter an d
in p u t m atching netw o rk w ere com bined on the sam e su b strate, the tran sm issio n
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lin e w a s re a d ily in c lu d e d in th e final lay o u t. W hile th e a m p lifie i c o n ta in e d
v a rio u s d e sig n m o d ificatio n s from the initial version, a filter identical to th e I1*
G H z filter d escrib ed in th e p re v io u s section w a s used. A T o u ch sto n e listin g o!
th e filte r/a m p lifie r integ rated u n it is inclu d ed in A p p en d ix III.
<»
k
*
'
>
&
*
■4-
F igure 4.7 P h o to g ra p h of th e HTSC filte r/a m p lifie r A ‘ MIC
A p h o to g ra p h o f th e co m b in atio n HTSC filte r/a m p lifie r M IC is
s h o w n in F ig u re 4.7. T he o v erall c ircu it d im e n sio n is 10.5m m x 4.2m m . The
la n th a n u m a lu m in a te su b s tra te s w e re a ttac h ed to a g o id -p la te d k n v a r c arrier,
sim ila r to th o se u se d for th e a m p lifie rs, u sin g c o n d u c tiv e silv e r e p o x y . T he
P H E M T d ev ic e w a s e p o x ie d o n to the ele v ated p e d e sta l a n d co n n ec te d to th e
in p u t a n d o u tp u t m atch in g n e tw o rk s u sin g w ire b o n d s. T he c h ip re sisto rs an d
c a p a c ito rs w e re m o u n te d o n to g o ld c o n ta ct p a d s o n th e s u b s tra te as before.
B iasing w a s a p p lie d th ro u g h filterco n s a tta c h e d to the circuit test jig via th e
p h o sp h o r-b ro n z e bias lines in th e cryogenic test se tu p . To
ip p re ss any ex tern al
RF sig n a ls w hich m ig h t p ro p a g a te a lo n g the b ias lines from th e p o w e r su p p lie s
c a u s in g o sc illa tio n s, O .lgf c e ra m ic c a p a c ito rs w e re a d d e d in s h u n t to th e
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
filtercons on both the gate and drain Ferrite b ead s w ere also inchided o n th e b ias
feed lines to assist in rem o v in g ex tran eo u s RF signals.
The filte r/a m p lifie r MIC w a s c o n n ec te d in to th e c ry o g e n ic test
se tu p an d m easu red u sin g the W iltron 360 A u to m atic N e tw o rk A n aly zer a t 77K.
W hen the tran sisto r w a s biased to its p rev io u s o p e ra tin g c o n d itio n s o f 2V drain*
to -so u rce v o lta g e a n d 6m A d ra in c u rre n t, the c irc u it o sc illa te d a t 20.2 G H z,
w hich w as noticeable in th e o u tp u t reflection coefficient. T he o scillatio n s w e re
m ainly a ttrib u te d to v a ria tio n s betw een th e actual PH EM T a n d th e d ev ic e u sed
in th e sim u la tio n s, w h ereb y the o u t-o f-b an d filter o u tp u t im p e d a n c e re s u lte d in
d ev ic e in stab ility . T he o scillatio n s s u b s id e d w h e n th e d ra in -to -so u rc e v o lta g e
w as re d u c e d to 1.5V. M e asu red re su lts a re d is p la y e d in F ig u re 4.8 o v e r th e
freq u en cy ra n g e o f 17 to 21 G H z. The gain reaches a p e a k v a lu e o f 6.4dB a t 18.9
C l Iz, b u t sh o w s a significant rip p le across th e p a ssb a n d . A t th e cen tre freq u en cy
of 18.7 G H z, th e gain is red u ced below 6dB w ith c o rre sp o n d in g in p u t a n d o u tp u t
re tu rn kisses o f less th a n 5dB. T his rip p le is m a in ly c a u se d b y a m is m a tc h
b etw een the filter p a ssb a n d a n d the p e ak of the am plifier, w h ic h w ere offset from
each o th e r by o v e r 300 M H z.
The sim u la ted version of th e filte r/a m p lifie r c ircu it h a d a p re d ic te d
p a ssb a n d g ain of 8.4dB at a cen tre freq u en cy o f 18.8 G H z, w ith c o rre sp o n d in g
in p u t a n d o u tp u t m a tc h in g o f -22dB a n d -23dB, resp ec tiv e ly . T he sim u la tio n
in c o rp o ra te d the T o u ch sto n e filter re sp o n se as o p p o se d to th e EM Sim v e rsio n ,
d u e to its g re a te r flexibility in sim u la tin g a s w ell as its closer a p p ro x im a tio n to
th e m e a su re d resp o n se. E ven th o u g h th e p re d ic te d filter's p a s s b a n d d id n o t
directly coincide w ith the p e a k of the am p lifier resp o n se , th e excellent re tu rn loss
o f th e filter p ro v id e d a good m atch in the overall circuit.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
O'*
5
0
-5
Y >
i'
CQ
TJ
v .
I"
Measured
-10
Touchstone
-15
r—
-20
-25
-50
.1 .
-55
17
17.?
18
18.5
M
10.5
frequency (Cl 1/)
20
20.c
21
r
-5
-10
sa
3,
rt
r-j
«1
-15 (~
<>
C/j
Measured
Touch:.! cm-
-20
-25
J
-50
17
17.5
18
18.5
10
10.5
frequency ((>11/.)
20
20.5
Figure 4.8 M easured vs. sim ulated a)Sl1 and b)S22 of 1 FI'SC filter/a m p lifie r 'A'
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
10
f ' A
h
\
o — Measured
- o - - Touchstone
I:
CQ
n
<1,
-5
i '
*10
-IS
A A 1__ 1__
-20
17
17.5
18
18.5
19
19.5
frequency (GHz)
20
20.5
21
Figure 4.8c) M easured vs. sim ulated gain o f HTSC filte r/a m p lifie r 'A'
W hen d esig n in g the filters for in teg ratio n w ith the am p lifier, the
c en tre freq u en cies w e re d e lib erately set slig h tly lo w e r th an th e p e a k o f th e
a m p lifie r resp o n se , since the gain roll-off w a s m o re g ra d u a l on th e lo w er
freq u en cy edge. For instance, w ith a 19 G H z p e a k a m p lifie r re sp o n se , cen tre
freq u e n cie s o f 18.8 G H z an d 19 G H z w ere ach iev ed w ith T o u c h sto n e a n d
F.MSim, respectively. This w ould help com pensate for any variatio n in dielectric
constant, noting that
£r
has been found to decrease at lo w er tem p eratu res,l25l th u s
c au sin g an u p w a rd shift in cen tre frequency. The s h a rp e r g ain roll-off at th e
u p p e r frequency ed g e of the am plifier p assb an d w o u ld considerably d isto rt the
filte r/a m p lifie r passband response.
The discrepancy betw een the p redicted an d m easu red p erform an ce
of th e integrated u n it is clearly reflected in the co m p ariso n of th e m easu red and
sim u la ted filter responses in Figure 4.3, as th eir cen tre frequencies d iffer b y 150
M H z. As the filter w as fabricated o n the sam e w afer as th e filte r/a m p lifie r MIC,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
n o o p p o rtu n ity existed to m odify the filter to m o re acc u ra tely m atch tin*
am p lifier response. Therefore, a second iteration filter design accounting tor the
frequency shift w ould be required to solve the m ism atching problem .
to
I i l t e r / A m p l i f i e r 'A'
Fi l te r / A m p l i f i e r '15' SmA
Fi l te r / A m p l i f i e r It- 2 mA
I
CO
3
U.
z
IK
is.5
i4.s
20
20 S
frequency ( ( it I/)
Figure 4.9 N oise figures of the HTSC' filte r/a m p lifie r Ml( s
N oise figure for th e HTSC filte r/a m p lifie r m icro w av e integ rated
c irc u it w a s m e a su re d at 77K u sin g the tech n iq u e d escrib e d in C h a p te r 3. by
c o n n e c tin g th e n o ise so u rc e to the in p u t v acu u m fe e d th ro u g h port an d the
o u tp u t p o rt con n ected to the noise figure m eter, as in Figure 3.10, the overall
n o ise figure of the system can be m easured. The noise figure of tin* DU I can be
d e te rm in e d once th e K-cable an d test fix tu re losses h ave been at c o u n te d lor.
N o ise fig u re for th e filte r/a m p lifie r in te g ra te d u n it w as m e a su re d o v e r a
freq u en cy ran g e e x te n d in g from 18.4 to 19.2 G H z in o rd e r to coincide w ith the
n a rro w p assb an d . The o p tim u m noise bias c o n d itio n s at room te m p era tu re <4 2V
a n d 6m A w ere app lied to the d rain , as no oscillations occurred in this frequency
range. T he resu lts are d isp lay ed in Figure 4.9, sh o w in g a m inim um noise figure
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of 1 SdB at 18.9 c;i Iz. V arying the drain cu rren t above and below 6m A p ro d u ce d
no im p rovem ent in noise figure, confirm ing th at this w as also the o p tim u m noise
bias condition at 77K.
4.4.2 20 G H z H T SC F ilte r /A m p lifie r M IC
A second filte r/a m p lifie r in teg rated circu it w as d ev elo p ed u sin g
the inS C Y M F M T am p lifier B' described in C h a p te r 3 an d the 20 G H z H i .
filler p re se n te d in section 4.3.2. T he filter a n d a m p lifie r w ere m o u n te d on
sep arate carriers for individual m easurem ent, and then interconnected w ith w ire
b o n d s for the overall circuit. The com bined circuit w as m easured at 77K o v e r the
frequency range of 16 G H z to 24 G H z. The tran sisto r w as biased w ith 5m A an d
2V a p p lie d to the d ra in , and d isp lay e d no o scillation p ro b le m s a t th ese b ias
values.
M easured and T ouchstone resu lts are sh o w n in F igure 4.10. T he
m ea su re d gain at the 19.8 G H z p a ssb a n d cen tre freq u e n cy w a s 7.2dB w ith
c o rre sp o n d in g m atch in g of -17dB at th e in p u t a n d -8dB on th e o u tp u t. T he
T ouchstone sim ulation predicted 8.8dB of gain at an identical centre frequency of
19.8 G i l / . O nce again, the d iscrep an cy b etw een th e m ea su re d a n d p re d ic te d
resu lts is a ttrib u te d to m ism atching betw een th e filter a n d am p lifier respo n ses.
T he lo w e r fre q u e n c y e d g e of th e tra n s m is s io n re s p o n se ,
IS 2 1 I, c le a rly
d e m o n stra te s this problem The filter p assb an d is 700 M H z low er th an th e p eak
am p lifier response at 20.5 G H z w hich c o rre sp o n d s to an e rro r of a lm o st 4%. For
th e sim u la ted h y b rid circuit, the a m p lifie r w as to p ro v id e a p eak at 20 G H z
w hich w o u ld pro p erly m atch the filter w ith a p red icted centre frequency o f 19.8
G 11/ in T ouchstone and 20 G H z in EMSim.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sa
\U \IM IU \i
-HI
I oiii lislorn-
S.
-15
-20
-25
18
1 8.5
10
1 0 .5
20
20 5
I r i ' q u e n i v K .11/)
i
,
i f
ca
rl
rl
S i
M iM su rrd
-10
!<>(<( h s t o n c
-15
-20
18
1 8 .5
10
1 0.5
20
205
21
215
22
f r i - q u e m v (< H i / )
F ig u re4.10 M easured vs. predicted a)S ll an d b)S22 for H'fSC filter/a m p iific
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
M easured
Touchstone
18
,8 3
19
193
20
203
frequency (GHz)
21
213
22
fig u re 4.10c) M easured vs. pred icted gain of HTSC filte r/a m p lifie r 'B'
To co rrect th e m ism atch , th e am p lifie r re sp o n se c o u ld b e sh ifted
d o w n w a rd in freq u e n c y by a p p ro p ria te ly re d u c in g th e a m o u n t o f s o u rc e
in d u ctance. T his m ay be achieved by sh o rte n in g th e b o n d w ires w h ich co n n ect
the source of the tran sisto r to the go ld -p lated carrier (i.e. g ro u n d ). A lo w e r so u rce
u d u c ta n ce w o u ld also increase th e p e a k tran sm issio n g ain , hen ce re d u c in g th e
a m p lifie r sta b ility . To o ffse t a n y s ta b ility p ro b le m , a sp e c ific le n g th of
tra n sm issio n line could b e in c lu d e d . T h is w o u ld b e m o re e asily re a liz e d b y
in te g ratin g the filter and in p u t m atch in g circuit on a sin g le su b stra te , as w a s th e
case for th e p rev io u s filte r/a m p lifie r MIC. A n a ltern a tiv e so lu tio n to e lim in a tin g
the m ism atch b etw een th e filter an d am p lifier is to red e sig n the filter to ach iev e
the req u ired h ig h er centre frequency, rath e r th an alter th e ex istin g am plifier.
T he noise fig u re o f th is HTSC filte r/H E M T a m p lifie r in te g ra te d
unit w as m easu red at v ario u s bias cond itio n s, w ith resu lts for Ios=3m A a n d 5m A
d isp lay e d in Figure 4.9. M in im u m n o ise fig u res a t 19.8 G H z of 1.7dB a n d 1.5dB
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c o rre sp o n d to d ra in c u rre n ts of 5m A an d 3m A, respectively. M easu rem en ts
p erfo rm ed at 10mA yielded significantly d e g ra d ed noise figures. These results
are c o n siste n t w ith noise figures m ea su re d for a K u-band HF.MT low noise
a m p lifier at 77K w hich p ro d u ced o p tim u m noise figures at a d ra in c u rre n t of
3.5mA.I47! C o n sid erin g th at the device w a s m atched for m axim um gain rath er
th an o p tim u m noise perform ance, these initial results a re very en couragin g . An
im p ro v e m e n t in th e m atc h in g b e tw ee n filter an d a m p lifie r sh o u ld fu rth e r
im p ro v e noise figure.
4.5 Summary
T his c h a p te r p resen ted the d e v elo p m e n t an d resu lts of tw o high
te m p e ra tu re su p e rco n d u c tin g filter/lo w -n o ise am p lifier hybrid circuits. Two 3p o le n a rro w b a n d (2.5%) C h eb y sh ev m icro strip filters w ere fabricated u sin g
TBCCO HTSC th in -film s on a LaAlO^ su b stra te. T he filters p ro v id e d O.SdB
in sertio n loss a n d retu rn losses of 18dB and 13dB at centre frequencies of 18.05
G H z an d 19.75 G H z, respectively. CA D sim ulations using a quasi-static app ro ach
(T o u chstone™ ) a n d a full-w ave analysis (E M Sim ™ ) p red icted a lo w er centre
freq u e n cy for b o th filters. T his w as a ttrib u te d to a h ig h e r actual d ielectric
c o n stan t d u e to the tw in n in g effect of the LaA lO i su b strate. As a co m p ariso n to
the p erfo rm an c e of the HTSC filters, a sim ilar gold filter d esig n ed on LaAIOj
p red icted 2.8dB insertion loss at room tem p eratu re and 1.7dB at 77K.
T he lo w e r frequency filter w as in te g rate d w ith a m p lifie r A' to
p ro d u c e th e first k n o w n HTSC filte r/lo w -n o ise am p lifier h y b rid circuit at Kb a n d . T he tran sm issio n resp o n se achieved a peak valu e of 6.4dB at 18.9 (ill/.,
w ith c o rre sp o n d in g noise figure of 1.8dB. The 19.75 Gf Iz HTSC filte r/a m p lifie r
B' in te g ra te d u n it p ro v id e d 7.2dB gain and 1.7dB noise figure at the nom inal
d ra in b ia s c o n d itio n of 2V a n d 5m A , and 1.5dB n o ise fig u re at 3m A d ra in
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
c u rre n t. These are excellent resu lts for a first ite ra tio n d e sig n , a n d c o m p a re
favourably to a C -band HTSC filter and G aA s LNA h y b rid ci cuit w hich rep o rte d
a noise figure of l.SdB.I1^
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5: CONCLUSIONS
5.1 Summary and Discussion
T he p o ten tial for com bining se m ico n d u cto r and su p e rc o n d u c to r
co m p o n en ts to form hybrid m icrow ave integrated circuits has been investigated.
A t p re se n t, th e g rea test benefits of these h y b rid circu its can be o b tain ed bv
u tilizin g active sem ico n d u cto r devices and passive su p erco n d u c tin g stru ctu res.
T he processing of three-term inal active devices u sing sem iconductor technology
is a m a tu re a re a, w h e re a s sim ila r su p e rc o n d u c tin g d e v ices are still at the
d e v e lo p m e n t stages. M oreover, the p e rfo rm an c e of th ese d ev ices im p ro v e s
c o n sid erab ly w h e n o p e ra tin g at cryogenic tem p era tu re s, m aking them su itab le
fo r in te g ra tio n w ith s u p e rc o n d u c to rs . HTSC p a ssiv e c o m p o n e n ts su ch as
reso n ato rs an d filters have d em o n strated m uch low er losses and h igher Q s than
sim ilar circuits fabricated usin g conventional technology. T heir p erfo rm an ce is
c o m p a rab le to dielectric reso n ato rs and w a v eg u id e filters trad itio n ally used in
n a rro w b a n d a p p licatio n s, su ch as for satellite c o m m u n icatio n s HTSC p lan a r
circu its, h o w ev er, are cap ab le of m eeting the n a rro w b a n d w id th req u ire m e n ts
w ith a significant reduction in size.
T his w o rk has presen ted the design and d e v elo p m en t of a hybrid
H T S C /G a A s K -band a m p lifie r a n d filter a t 77K. T he filter an d m a tc h in g
n e tw o rk s w ere fabricated usin g HTSC thin-film s, w hile th e am p lificatio n w as
accom plished w ith a HEM T device. This filter/a m p lifie r integrated unit pro v id es
a basic co m p o n en t for the front end of a m icrow ave receiver by m in im izin g the
sy stem noise figure w hile rem oving any sp u rio u s signals and noise o u tsid e the
d e sire d freq u en cy b a n d . T his K -band s e m ic o n d u c to r/s u p e rc o n d u c to r hy b rid
circu it w a s the first k n o w n a tte m p t at in teg ratin g the tw o technologies in this
circuit c onfiguration at such a high frequency.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p te r 2 p resen te d an o verview of m icro w av e in te g rate d circuit
(M IC )
te c h n o lo g y , a n d
th e
p e rfo rm a n c e
im p r o v e m e n ts
a t c ry o g e n ic
te m p e ra tu re s. C ooled M ICs exhibit an increase in re so n a to r Q -factors d u e to
low er resistivities and su b strate losses. Low te m p e ra tu re o p e ra tio n also red u c es
th e o n -ch ip p o w e r d en sity by e n h an cin g the th erm al c o n d u c tiv ity in m aterials,
in creases the circu it reliability by e lim in a tin g d ev ic e failu res d u e to h e a tin g
effects, a n d p ro v id e s a sta b le te m p e ra tu re e n v iro n m e n t. P e rh a p s th e m o st
significant a d v an tag es of cooling M ICs is the im p ro v ed perfo rm an ce of the active
d e v ic es su ch as FETs a n d H EM Ts. T hese d e v ic e s e x h ib it h ig h e r g a in a n d
o p e ra tin g freq u en cy d u e to th e e n h an c ed c a rrie r m o b ility , a n d lo w e r n o ise
figures attrib u te d to the intrinsically low er therm al noise.
T he fo llo w in g s e c tio n e x a m in e d
th e d e v e lo p m e n t o f lo w
te m p e ra tu re su p erco n d u c to rs, in clu d in g the T w o-Fluid m o d el, a n d th e L o n d o n
and
BCS th e o rie s . T h ese th e o rie s w e re e x te n d e d to h ig h te m p e r a tu r e
su p erco n d u cto rs, as well as so m e m ore recent ex p lan atio n s into th e b e h a v io u r of
HTSCs. Follow ing, the thin-film fabrication a n d processing tech n iq u es for H TSCs
w e re d isc u sse d . T he c h a p te r c o n clu d ed w ith a d e sc rip tio n of th e c ry o co o ler
a p p a ra tu s an d TRL calibration tech n iq u e d e v elo p ed a t CRC.I50! T his test se tu p
p ro v id e s accu rate S -p aram eter m ea su re m e n ts at cryogenic te m p e ra tu re s b y d e ­
e m b e d d in g the test fixture and K-cables.
C h a p te r 3 re p o rte d the d e sig n a n d d e v e lo p m e n t o f tw o K -b an d
H T S C /G aA s cryogenic am plifiers. HEM T devices o p e ra te d at 77K h a v e y ield ed
3dB im p ro v em en ts in gain, along w ith significantly lo w e r o u tp u t im p ed an ce. Sp a ra m e te r m easu re m e n ts of the am p lifier sh o w ed rea so n a b le a g re em e n t to th e
T o u ch sto n e sim u latio n s, w hich revealed a h ig h sen sitiv ity to b o n d w ire so u rce
in d u ctan ce. F ew er b o n d w ires co n n ectin g th e so u rc e of th e d ev ic e to g ro u n d
resu lte d in a h ig h er inductance, hence red u c in g th e in p u t a n d o u tp u t reflection
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
coefficients an d p ro v id in g stability for the am plifier. A seco n d ary peak at the
lo w e r fre q u e n c ie s w h ic h affected th e a m p lifie r sta b ility w a s c a u se d by
e x tra n eo u s effects of the bias circuitry T he elim ination of this potential stability
p ro b lem w a s the objective of a second iteration am plifier design. H ow ever, d u e
to m o d e llin g inaccuracies of the resistive sta b ilizin g n e tw o rk , the am p lifier
y ield ed po o r results.
N o ise figures of less th an 1 dB for the 20 G H z a m p lifie r w ere
ach iev ed . T his is a p ro m isin g re su lt c o n sid erin g th e d evice w as m atch ed for
o p tim u m gain ra th e r th an m in im u m noise figure. M atching the device noise
p a ra m e te rs w o u ld im p ro v e the noise figure, h o w e v er the tu n in g m echan ism s
re q u ire d to o b tain these noise p a ra m eters a t low te m p e ra tu re s a re extrem ely
d ifficu lt to im plem ent. N oise m o d els for v ario u s devices at low te m p e ra tu re s
h a v e b e e n d e v e lo p e d for a p p lic a tio n in e x tre m e ly low n o ise c ry o g e n ic
am plifiers.!57-68'69!
W h ile th e se a m p lifie rs h a v e d e m o n s tra te d th e p o te n tia l for
in teg ratin g su p erco n d u cto rs and sem iconductors, there is no conclusive evidence
th a t th is h y b rid c ircu it can p ro v id e b e tte r p e rfo rm an c e th an a co n v en tio n al
c o o led G aA s am p lifier. T he im p ro v e m e n ts in losses of th e HTSC m atch in g
n e tw o rk s com pared to conventional m etals at 77K is m inim al for these relatively
sh o rt tran sm issio n lines. A lso, th e high dielectric co n stan t of 11TSC su b stra te s
c o n sid e ra b ly re d u c e s th e line w id th s, c re a tin g a larg e d isc o n tin u ity w h en
co n n ectin g the circuit into the test fixture. This effect w as partially accounted for
b y im p le m e n tin g ta p e re d tran sitio n s at th e e n d s of the 50£2 in p u t and o u tp u t
HTSC lines. The m ain a d v an tag es of HTSCs becom e e v id en t in high Q reso n ato rs
a n d n a rro w b a n d w id th filters w ith g reater n u m b ers of coupling sections.
T he HTSC m icrostrip filters desig n ed in th is w ork had m easu red
in s e rtio n lo sse s of a p p ro x im a te ly 0.8dB at K -band fre q u e n c ie s, for 3dB
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
bn n d w id th s less than
2 .5 % .
Both Touchstone and EMSim sim ulations p redicted a
higher centre frequency, w hich m ay be attributed to a low er estim ated dielectric
constant. W hile T ouchstone pro v id ed a closer estim ate to the cen tre frequency,
EMSim predicted the asym m etric shape of the response m ore accurately. Kinetic
in d u ctance, w hich causes frequency shifts w h en o p e ra tin g n e a r the tran sitio n
tem p eratu re, w as m inim ized by u sing TBCCO thin-film s w ith a Tc g rea ter th an
100K.
The ability of these com m ercial CA D softw are packages to analyze
11TSC s tru c tu re s w as fu rth e r ex am in ed by m e a su rin g in d iv id u a l re so n a to r
co u p lin g sections o f the filters. EMSim show ed excellent ag reem en t for a w id e
gap section above 18 G H z, b u t produced erroneous resu lts a t a frequency of 12.5
GI Iz. T ouchstone predicted the w ide gap coupling accurately below 18 G H z, b u t
u n d erestim ated the coupling a t h ig h er frequencies. T he im p ro v ed m o d ellin g of
EMSim a t h igher frequencies is likely d u e to the con sid eratio n of rad iatio n loss
an d surface w aves w hich this full-w ave analysis softw are is capable of hand lin g .
Eor th e n arro w gap coupling section, both T ouchstone a n d EMSim resu lts w ere
w ithin ld B of the m easurem ents over the 16 G H z to 24 G H z frequency range.
These softw are packages, along w ith S onnet's EM, w ere also u sed
to sim u late the frequency response of HTSC m atching n e tw o rk s for am plifiers.
EMSim and T ouchstone pro v id ed the closest m atch to m easu rem en ts of an in p u t
an d o u tp u t m atching circuit, respectively. W hile EM :s m o re u se r friendly th an
EMSim, it requires a longer c o m putation tim e w ith n o a p p a re n t im p ro v em en t of
results. A lthough EM and EMSim did not consider th e effects of th e bias circuits
in their analyses in o rd er to shorten the com putation tim e, T ouchstone a p p ea rs to
be su fficient for m o d ellin g th ese sim p le HTSC p a ssiv e s tru c tu re s d u e to its
sim plicity of m odelling and superior co m putation speed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
104
T he HTSC filte r/a m p lifie r h y b rid M ICs p ro v id e d excellent resu lts
fo r a first ite ra tio n design. T he in te g ra te d u n it c o n ta in in g a PHEM 1 d e v ic e
c o n sisted of the HTSC filter a n d in p u t m atch in g n e tw o rk on a single su b stra te,
illu stra tin g a n a d v a n ta g e o f u tilizin g HTSC m atch in g circuits in the am plifier. A
m e a s u re d tra n s m is s io n re s p o n s e of 6dB at th e c e n tre freq u e n cy d e v ia te d
s ig n ific a n tly fro m th e 8dB p re d ic te d by T o u c h sto n e , m ain ly c a u se d In a
d o w n w a rd shift in frequency of the filter response. T his m ism atch w as e v id en t in
th e p a s s b a n d rip p le , w h ic h co u ld be su p p re ss e d by im p ro v in g th e m a tc h in g
b e tw e e n th e filter o u tp u t a n d a m p lifie r in p u t. D ue to th e h ig h o u t-o t-b a n d
reflectio n coefficient of th e filter, the a m p lifie r m u st be u n c o n d itio n a lly stab le
o v e r th e e n tire freq u en cy ra n g e to p re v e n t oscillations. A sin u la r h y b rid circuit
u tiliz in g a H E M T dev ice p ro v id e d a h ig h er gain of 7.2dB at the cen tre frequency,
w ith less rip p le in th e p a ssb a n d . This resu lt co u ld be im p ro v ed by red u c in g th e
700 M H z d isc re p a n c y b e tw e e n th e filter a n d a m p lifie r c e n tre freq u en cies. By
lo w e rin g th e so u rc e in d u c ta n c e of th e dev ice to in crease the g ain a n d shift the
a m p lifie r re s p o n s e d o w n in fre q u e n c y , re s u lts c lo se r to th e VdB p re d ic te d
p a s s b a n d g ain co u ld b e a ch iev e d . Im p ro v ed m a tc h in g b e tw ee n th is filter an d
a m p lifie r c o u ld b e m o re re a d ily rea liz ed b y c o m b in in g th e filter a n d in p u t
m atch in g n e tw o rk o n a single su b strate.
A n im p re ss iv e m in im u m n o ise fig u re of 1.5dB ach iev e d for th e
H T S C /H E M T a m p lifie r 'B’ a n d filte r illu s tra te s th e p o te n tia l fo r h y b rid
se m ic o n d u c to r/s u p e rc o n d u c to r circuits in low noise ap plications. Im p ro v em en ts
in filte r in s e rtio n lo ss a n d m a tc h in g b e tw e e n th e filter a n d a m p lifie r c o u ld
p o te n tia lly lo w e r th is v a lu e close to ld B . T his type* o f p e rfo rm an ce w o u ld m ak e
th e s e c irc u its s u ita b le a s lo w n o ise , n a rro w b a n d w id th c o m p o n e n ts for
ch an n e lize d receivers in satellite co m m unications.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
105
5.2 Future W ork
Im p ro v em en ts for the HTSC filte r/a m p lifie r MIC m u st focus on
achieving b etter stability of the device at low tem p e ra tu re s w ith o u t significant
d e g ra d a tio n in the gain an d n o ise figure. V ary in g th e b o n d w ire so u rc e
ind u ctance to accom plish this task is highly sensitive to slight ch an g es in b o n d
w ire len g th a n d n u m b e rs. A lte rn a tiv e fo rm s of sta b ility su c h a s re sistiv e
stabilizing n etw o rk s have been investigated, ho w ev er these m eth o d s w ill req u ire
a m o re d e ta ile d a n a ly sis in to freq u e n cy a n d te m p e ra tu re effects o f c h ip
com ponents.
A nother area of im provem ent lies in th e d esig n o f HTSC filters. A
second iteration filter design w ould allow the d esig n er to p red ict th e frequency
response w ith greater confidence. W hile the filters desig n ed in th is w o rk yielded
mi>derately n arro w b an d w id th s, the high-Q capabilities of HTSCs can be fu rth er
ex p lo ited in ex trem ely n a rro w b an d ap p lica tio n s (<1%), w h ich req u ire h ig h ero rd e r filters to achieve a steep er roll-off. To attain th e h ig h est p o ssib le Q s for
HTSC reso n ato rs an d filters, a p ro p erly d esig n ed cover sh o u ld be im p lem en ted
to red u ce radiation loss. R ather th an m atching the filter's o u tp u t to a 50Q system ,
the filter could be inherently m atched to the in p u t im pedance of the am plifier.!70!
T his w o u ld elim in a te the need for a n in p u t m atch in g n e tw o rk a n d th ere fo re
reduce the area of th e overall hybrid circuit.
D ue to restrictions on su b stra te dielectric co n stan t in co nventio n al
so ftw are packages, an d th e u n iq u e p ro p e rtie s of su p e rco n d u c to rs, cu sto m ized
HTSC so ftw a re w ill be req u ire d to an aly ze m o re com plex stru c tu re s. K inetic
in d u c ta n c e , w h ic h re p re se n ts th e te m p e ra tu re d e p e n d e n c e o f th e L o n d o n
p en etratio n d e p th , m ust be considered w h e n o p e ra tin g a t te m p e ra tu re s close to
Tt . The b eh av io u r of the surface resistance for thin-film su p e rco n d u c to rs sh o u ld
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b e a c c u ra tely m o d elled , as well as the lim itatio n s o n c u rre n t d e n sitie s a n d the
d e g ra d a tio n ot Q -factors at high p o w e r levels.
T he p ro ce ssin g of H TSCs m ust c o n tin u e to im p ro v e to m eet the
d e m a n d for h ig h e r critical c u rre n t d e n sitie s a n d lo w er su rface resistance. Also,
su b stra te q u a lity m u st b e im p ro v ed to e n su re th e h ig h e r p erfo rm an ce criteria for
H TSC s a re m et. LaAlO^ p ro v id es good therm al m atch in g p ro p erties to th e 1ITSC
th in -film s , h o w e v e r it p o se s p ro b le m s d u e to a h ig h d ie le c tric c o n s ta n t,
"tw in n ing" p lan e s, a n d b rittlen e ss w h ich m akes via holes difficult to im p lem en t.
S u b stra te s su ch as R -plane sa p p h ire p ro v id e extrem ely low su b stra te loss a n d a
lo w e r dielectric c o n stan t, b u t d o n o t p ro v id e a good th erm al m atch to th e I ITSC
thin-film s.
T h e c u rr e n t c ry o g e n ic m e a s u re m e n t facility at CKC is b e in g
a u g m e n te d b y a c ry o g e n ic w a fe r p ro b e r, w h ic h w ill a llo w for o n -w a fe r Sp a ra m e te r m e a su re m e n ts at 77K. T his p ro b in g statio n w ill sig n ifican tly red u ce
th e c alib ratio n tim e, a n d im p ro v e m ea su re m e n t accuracy by e lim in a tin g th e nt*ed
fo r w ire b o n d s. T he accu racy o f n o ise fig u re m e a su re m e n ts u sin g th e c u rre n t
cry o g e n ic s e tu p sh o u ld be im p ro v e d by u tiliz in g a h ig h e r excess n o ise ratio
(EN R ) n o ise d io d e . H ig h e r EN R s w ill c o m p e n sa te for th e sig n ific an t losses
in c u rre d in th e K -cables, test fixture, an d m ixer, esp ecially at high frequencies.
F or a c c u ra te m e a s u re m e n ts of e x tre m e ly low n o ise d ev ices, a h o t-co ld n o ise
so u rc e u s in g liq u id n itro g e n a s th e "cold" refe ren c e w o u ld p ro v id e th e best
resu lts.
F u tu re a p p lic a tio n of s e m ic o n d u c to r /s u p e r c o n d u c to r h y b rid
c ircu its w ill e x te n d to M M ICs, w h erein su p e rc o n d u c tin g layers can be d e p o site d
o n to a G a A s s u b s tra te . F u rth e rm o re , in te g ra tio n o f s e m ic o n d u c to rs a n d
s u p e rc o n d u c to rs in to sin g le h y b rid d ev ices su ch a s th e S u p e rc o n d u c tin g IT T
(SuF E T ), a n d R e so n a n t T u n n e lin g T ra n s is to r (RTT), a re c u rr e n tly b e in g
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
d ev elo p ed .I,0J The SuFET will be capable of p ro v id in g large-signal gain in digital
ap p lica tio n s w ith very low p o w e r d issip atio n , w hile the RTT w ill be u se fu l for
very high frequency (THz) applications w ith carrier tra n sit tim es o n th e o rd e r of
a few picoseconds.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
108
APPENDIX I HTSC/PHEMT AMPLIFIER fA' SCHEMATIC
DIAGRAM
Hi
Hi
I
$
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
109
APPENDIX II HTSC/HEMT AMPLIFIER 'C SCHEMATIC
DIAGRAM
£ £
§ > ,,1
h lli-^ T H i'
1 - j"
IH 1
5
*
i
i
*
£
£
sat
.2 1
to eu
3or
J5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
110
APPENDIX III TOUCHSTONE LISTING OF HTSC FILTER/LNA
!FXL LNA.CKT
!SIMULATION OF HTSC FILTER/ PHEMT-AMPLIFIER MIC
!MAY '93
DIM
LNG UM
COMO /KOH
VAR
LI-549
Dl\602
L2-893
D2\727
WA-75
SA-237
LA-973
WB-83
SB-943
LB-966
1in*-1200
CRT
MSOB ER-24 H-254 T-0.8 RHO-O.Ol RGH-0
TAND TAND-0.00003
! input matching network Cor amplifier
MLIN 1 2 M—86 L-100
MCROS 2 3 4 5 Hl-86 W2-86 W3-86 W4-25
MLEF 3 W-86 L*D1
MLIN 4 6 W—86 L'Ll
MLIN 5 7 W-25 L-1100
mte* 7 8 9 wl-25 w2-500 w3-25
MLEF 8 W— 500 L-787
CAP 8 IS C-10
RES 15 0 R-50
mlin 9 10 w-25 1-600
mbend 10 11 w-25 ang-45 m-0
mlin 11 12 w-25 1-677
rea 12 13 r-2000
mleC 13 w—350 1-350
0EF2P 1 6 IPMATCH
! output matching network Cor amplifier
MLIN 10 11 W—86 L*L2
MCROS 11 12 13 14 Ml—86 M2-86 W3-86 M4-2S
MLEF 12 M—86 L~D2
MLIN 13 15 W— 86 L-1040
cap 15 26 c-10
MLIN 14 16 W-25 L-1100
mte* 16 17 18 wl-25 w2-500 w3-25
MLEF 17 M—500 L-787
CAP 17 25 C-10
RES 25 0 R-50
mlin 18 19 w-25 1-500
mbend 19 20 w-25 ang-60 m-0
mlin 20 21 w-25 1-600
matap 21 22 wl-25 w2-350
mleC 22 w-350 1-350
DEF2P 10 26 OPMATCH
! simulated cryogenic HTSC/PHEMT amplifier 'A'
ipmatch 40 41
S2Pb 41 42 45 ../fhrl0x/LNF100 2.S2P
ind 45 0 1-0.05
OPMATCH 42 43
DEF2P 40 43 AMP
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ill
! 3 -p o le Chebyshev b an d p ass n u c r o s t n p
MLIN X 2 W-86 L-500
MCFIL 2 3 W'WA S'SA
MCFIL 3 4 W'WB S'SB
MCFIL 45 W*WB S'SB
MCFIL 56 W'WA S*SA
MLIN 6 7 W-86 L ' l m e
DEF2P 1 7 BANDPASS
filter
L'LA
L'LB
L'LB
L'LA
! simulated HTSC filter/amplif.ter MIC
BANDPASS 1 2
MLIN 2 3 W-86 L'line
AMP 3 4
DEF2P 1 4 FIL_AHP
! measured filter/amplifier MIC
s2pd 1 2 0 ../fillna/fplna3.s2p
def2p 1 2 meas
FREQ
sweep 16 22 0.05
OUT
FIL_AMP DB[S11] GR1
FIL_AMP DB(S21] GR2
FIL AMP DB[S22] GR3
MEAl DB(Sll) GR1
MEAS DB(S21) GR2
MEAS DB[S22] GR3
GRID
FREQ 16
GR1 -30
GR2 -20
GR3 -30
22 2
S S
10 5
S 5
OPT
FREQ 18.8 19.1
AMP MAG[SI1]-0
AMP MAG[S22]-0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
REFERENCES
1
Sim on, R. a n d Sm ith, A., S u p erco n d u c to rs - C o n q u e rin g T oday s NowFrontier. N.Y., Plenum Press, 1988.
2
Sun, J. Y-C, e t al, "Subm icrom eter-C hannel CM OS for l.ow -T em poraturc
O peration", IEEE Trans, on Electron Devices, Vol. ed-34, No. 1, Jan. 1987.
3
Vacca, T. et al, "A C ryogenically C ooled CM O S VLSI S u p erco m p u ter",
VLSI System s D esign, June 1987, pp. 80-88.
4
W einreb, S., "L ow -N oise C ooled GASFET A m plifiers", IEEE. T rans, on
MTT, Vol. MTT-28, No. 10, Oct. 1980, pp. 1041-1053.
5
W einreb, S. e t al, "Cryogenic, HEMT, L ow -N oise Receivers for 1.3 to 43
G H z Range", IEEE MTT-S Digest, 1988, pp. 945-948.
6
R ichards, P.L. and H u, Q., "Superconducting C om ponents for Infrared and
M illim eter-W ave Receivers", Proc. of the IEEE, A ug. 1989, pp. 1233-1246.
7
O lss o n , H .K . a n d C la e so n , T., "Low n o ise Jo s e p h s o n p a ra m e tric
am plification an d oscillations at 9 GH z", J. A ppl. Phys., Vol. 64, 1988, pp.
5235-5243.
8
C larke, J. et al, "T unnel junction dc SQUID: Fabrication, o p e ra tio n , and
perform ance", J. Low Tem p. Phys., Vol. 25, Oct. 1976, pp. 99-144.
9
B lu n d ell, R. et al, "SIS-m ixer receiv er w ith sin g le a n d a rra y tu n n el
junctions for the 140 G H z to 170 G H z range, Proc. 14th EMC, pp. 581-586.
10
M ad d o ck s, H., "YBa2C u 3C>7 : T he S u p e rc o n d u c to r of the F uture", IEEE
Potentials, Dec. 1993, pp. 30-33.
11
T akem oto, J.H. e t al, "M icrostrip R esonators a n d Filters U sing High-Tc
S u p e rco n d u ctin g T hin Film s on LaAlOV, IEEE T rans, on M agnetics, Vol.
27, No. 2, M arch 1991, pp. 2549-2552.
12
Bonetti, R.R. an d W illiam s, A.E., "Prelim inary Design Steps For Thin-Film
S up erco n d u ctin g Filters", IEEE MTT-S D igest, 1990, pp. 273-276
13
M a tth a e i, G .L . a n d H e y - S h ip to n , G .L ., " H ig h T e m p e r a tu r e
S u p e rco n d u ctin g 8.45 G H z B andpass Filter for the D eep Space N etw ork",
IEEE MTT-S Digest, 1993, pp. 1273-1276.
14
H in k e n , J.H ., S u p e rc o n d u c to r E le ctro n ics, S p rin g e r-V e rla g , B erlin,
H eidelberg, 1989.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
K roger, H. et al, "S u p e rc o n d u c to r-S e m ic o n d u c to r H y b rid D evices,
C ircuits, and Systems", Proc. of the IEEE, A ug. 1989, pp. 1287-1301.
16
W ith ers, R.S. a n d R alston, R.W ., "S u p e rc o n d u c tiv e A n a lo g S ignal
Processing Devices", Proc. of the IEEE, vol. 77, A ug 1989, p p . 1247-1263.
17
S m uk, J.W., S tubbs, M.G., W ight, J.S., "H ybrid S e m ic o n d u c tiv e / H ig h
T em p eratu re S uperconductive Ku-Band O scillator an d A m plifier MICs",
IEEE MTT-S Digest, 1992, pp. 485-488.
18
Shen, Z-Y. et al, "H igh Tc Superconductor and 11I-V Solid State M icrow ave
H ybrid Circuits", IEEE Trans, on A pp. S.C., M ar. 1993, pp. 2832-2835.
19
B hasin, K.B. e t al, "P e rfo rm a n ce of a Y B aC uO S u p e rc o n d u c tin g
F ilter/G aA s Low N oise A m plifier H ybrid C ircuit", IEEE MTT-S D igest,
1992, pp. 163-165.
20
H uang , Q. et al, "Fabrication and C haracterization o f H T S /S em icondu cto r
T hree Term inal Device", IEEE MTT-S Digest, 1993, p p . 1957-1960.
21
H offm an, R.K., H a n d b o o k of M icrow ave In te g ra te d C ircu its. A rtech
H ouse, 1987.
22
C ollin, R.E., F o u n d a tio n s for M icro w av e E n g in e e rin g , 2 n d e d itio n ,
M cG raw -H ill, 1992.
23
King, I.A., M aterials H an d b o o k for H ybrid M icroelectronics. Boston MA,
A rtech H ouse, 1988.
24
Keyes, R.W., "The Role of Low T em p eratu res in the O p e ra tio n of Logic
C ircuitry", Proc. of the IEEE, Vol. 58, Dec. 1970, p p . 1914-1917.
25
K o n o p k a, J. a n d W olff, I., "D ielectric P ro p e rtie s o f S u b s tra te s fo r
D epostion of H igh-Tc T hin Film s U p to 40 G H z", IEEE T rans, on MTT,
Vol. 40, Dec. 1992, pp. 2418-2423.
26
Roesch, W.J., "Therm o-Reliability R elationships of G aA s ICs", 1988 G aA s
1C Sym p., pp. 61-64.
27
K irschm ann, R.K., "Cold Electronics: an O verview ", C ryogenics, M arch
1985, p p . 115-122.
28
L o ndon, F. a n d L o ndon, H ., "The E lectro m ag n etic E q u a tio n s o f th e
S uperconductor", Proc. Roy. Soc. London (A), 1935, p p . 71-88.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
B a rd e e n , J., C o o p e r , L .N . a n d S c h rie ffo r, J.R .,
S uperconductivity", Phys. Rev. 108,1957, p p . 1175-1204.
30
Jo sep h so n , B.D., "Possible N ew Effects in S u p e rco n d u c tiv e T unneling",
Phys. Lett., July 1962, pp. 251-253.
31
B ednorz, J.G. a n d M uller, K.A., "Possible H igh Tc S uperconductivity in the
Ba-La-Cu-O System ", Z eitschr. Phys. B 6 4 ,1986, pp. 189-193.
32
C h u , C.W . e t al, "S uperconductivity a t 93K in a N ew M ixed-P hase Y-BaC u -O C o m p o u n d System at A m bient Pressure", Phys. Rev. Lett. 58, 1987,
p p . 908-910.
33
N ise n o ff, M .,"Space A p p lica tio n s for S u p e rc o n d u c to rs", IEEE MTT-S
W orkshop, Ju n e 1990.
34
V an D uzer, T. a n d T urner, C.W ., P rinciples o f S u p e rco n d u c tiv e IV vices
a n d C ircuits. NY, Elsevier, 1981.
35
F itzg erald , K., "S uperconductivity: Fact vs Fancy”, IEEE S p ectru m , May
1988, p p . 30-41.
36
G in zb u rg , V.L. a n d L andau, L.D., Zh. Eksp. Tcor. Fiz. 20, 1950, p. 1044.
37
A n d e rs o n , P.W . e t al, "R eso n atin g -V a le n ce -B o n d T h e o ry o f P h a se
T ran sitio n s a n d S u p erco n d u ctiv ity in La 2CuC>4 -Based C om pounds",P h y s.
Rev. Lett., Ju n e 1987, p p . 2790-2793.
38
V e n d ik , O . a n d K ollberg, E., "Softw are M o d els HTSC M icro strip an d
C o p la n a r Lines", M icrow aves an d RF, July 1993, p p . 118-120.
39
K obayashi, Y. e t al, "Phenom enological D escription of 'C oherence Peak' of
H igh-T c S u p e rc o n d u c to rs b y Im p ro v ed T hree-F luid M odel", E u ro p ean
M icrow ave C onference, 1993, pp. 596-599.
40
D io n n e , G .F., "N ew T w o -F lu id S u p e rc o n d u c tio n M odel A p p lie d to
P en etratio n D ep th an d M icrow ave Surface Resistance", T rans, on A pplied
S u p erconductivity, M arch 1993, pp. 1465-1467.
41
H o lste in , W .L. e t al, "Epitaxial Tl2 Ba2C a 2 C u iO jo Film s W ith Very Low
M ic ro w av e S urface R esistance U p To 110K", IEEE T rans, on A p p lied
S up erco n d u ctiv ity , Vol. 3, No. 1, M arch 1993, p p . 1197-1200.
42
G inley, D.S. e t al, "Synthesis an d Processing o f Thin Film s in the Tl-Ca-BaC u-O System ", IEEE Trans, on A pp. S.C., M arch 1993, pp. 1201-1204.
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
"T he o ry
of
115
43
G io v an n ella, C., "HTSC S u p e rco n d u c tiv e T hin-F ilm s fo r C ryoelectro n ics:
A S urvey of Recent A chievem ents and N ew T rends", Ju n e 1991,
44
S u p e rco n d u c to r T echnologies Inc., HTSC U p d a te, A pril 1993.
45
H o lstein , W .L. e t al, "Tl2Ba2CaCu208 Film s W ith V ery L ow M ic ro w av e
S urface R esistance U p To 95K", A ppl. P hys. L ett., Vol. 60, N o. 16, A p ril
1992, p p . 2014-2016.
46
A n d re o n e , A. et al, "Surface R esistance o f S u p e rc o n d u c tin g F ilm s b y a
M ic ro s trip R in g R e s o n a to r T e c h n iq u e " , T ra n s a c tio n s o n A p p lie d
S u p e rco n d u ctiv ity , Vol. 3, No. 1, M arch 1993, p p . 1453-1456.
47
S m u k , J.W ., " H y b rid S e m ic o n d u c tiv e /S u p e r c o n d u c tiv e M ic ro w a v e
In teg rated C ircuits", C arleto n U niversity P h d . Eng. T hesis, 1991.
48
Talvaccio, J. a n d W agner, G.R., "H igh-Tc Film D e v elo p m en t for E lectronic
A pplications", SPIE Vol. 1292,1990.
49
P ro d u c t N o te 8510-8, "A p p ly in g th e H P8510B TRL C a lib ra tio n fo r N o n C oaxial M easurem ents", H ew lett-P ack ard , P alo A lto, 1987.
50
S m u k , J.W ., S tu b b s , M .G ., W ig h t, J.S., "A n E n h a n c e d M ic ro w a v e
C h a ra c te riz a tio n T e c h n iq u e for C ry o g e n ic T e m p e ra tu re s" , E le ctro n ics
Letters, Vol. 26,1990, p p . 2127-2129.
51
Pavio, I. an d H y d e, D., "Effects o f T h erm al E x p an sio n M ism atch o n S o ld er
A ttach ed G aA s M M ICs", IEEE MTT-S D igest, 1988, p p . 241-244.
52
S m uk, J.W ., S tubbs, M .G., W ight, J.S., "S -P aram eter C h a ra c te riz a tio n a n d
M o d e llin g o f T h re e -T e rm in a l S e m ic o n d u c to r D e v ic e s a t C ry o g e n ic
T e m p e ra tu re s", M icro w av e a n d G u id e d W av e L ett., Vol. 2, N o. 3, M ar.
1992, p p . 111-113.
53
K irschm ann, R.K., Low T em p era tu re E lectronics. IEEE P ress, NY, 1986.
54
L ester, L. e t al, "0.15gm G a te -L e n g th D o u b le R ecess P s e u d o m o rp h ic
H EM T w ith fm ax of 350 G H z", IEDM C onference, 1988, p p . 172-175.
55
W alk er, J.B., H ig h -P o w er G aA s FET A m p lifiers. A rte c h H o u s e , B oston,
1993.
56
P ospieszalski, M .W . e t al, "N oise P a ram e te rs a n d L ight S en sitiv ity o f L ow N o ise H igh-E lectron-M obility T ra n sisto rs a t 300 a n d 12.5K", IEEE T rans,
o n E lectron D evices, Vol. ed-33, N o. 2, Feb.1986, p p . 218-223.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
116
57
Pospieszalski, M.W. et al, "FETs and HEMTs a t C ryogenic T em peraturesT heir P roperties an d Use in Low -N oise Am plifiers", IEEE Trans, o n M TL,
Vol. 36, No. 3, M arch 1988, pp. 552-560.
58
G onzalez, G., M icrow ave T ransistor A m plifiers. Prentice-1 tail, N |, P>84.
59
C u h aci, M. a n d C larke, B., C o m m u n ic atio n s R esearch C en tre, p riv a te
com m unication.
60
S h en , Z.Y. e t al, "H igh Tc S u p e rc o n d u c to r-S a p p h ire R eso n ato r w ith
E xtrem ely H ig h Q -V alues u p to 90K”, IEEE MTT-S D igest, 19U2, pp. W3196.
61
M a tth a ei, G., Y oung, L., Jones, E.M.T., M icrow ave Filters. Im p ed an ce
M atching N etw orks, and C oupling S tructures. M cG raw-Hill, 1964.
62
E d w ard s, T.C., F oundations for M icrostrip C ircuit D esign, John W iley and
Sons, 1981.
63
EEsof E lem ent C atalog, EEsof Inc., 1989.
64
C ai, K., M cM aster U niversity, priv ate com m unication.
65
T alisa, S.H. e t al, "M icrow ave S u p erco n d u ctin g Filters", IEEE T rans, on
M agnetics, Vol. 27, No. 2, M arch 1991, pp. 2544-2547.
66
T alisa, S.H . e t a l, "P re se n t a n d P ro je cted P e rfo rm a n c e o f llig h T e m p e ra tu re S u p e rc o n d u c tin g Filters", IEEE MTT-S D igest, 1991, p p .
1325-1328.
67
D rolet, D., M eszaros, S., Stubbs, M .G., W ight, J.S., "A H ybrid H igh-Tc
S u p erco n d u cto r K-Band A m plifier w ith Filter", 23rd E uropean M icrow ave
C onference, 1993, pp. 781-783.
68
P o sp ieszalsk i, M .W ., "M odeling of N oise P a ram e te rs of MESEETs an d
M O D FETs a n d T heir F requency an d T e m p e ra tu re D ependence", IEEE
MTT-S D igest, 1989, pp. 385-388.
69
P o sp ie sz a lsk i, M .W ., "C ry o g en ically -C o o led , HFET A m p lifie rs a n d
Receivers: State-of-the-A rt and F uture T rends", IEEE MTT-S Digest, 1992,
p p . 1369-1372.
70
Schw ab, W. an d M enzel, W., "A Low -N oise A ctive B andpass Filter", IEEE
M icrow ave and G uided W ave Lett., Vol. 3, No. 1, Jan. 1993, pp. 1-2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
5 568 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа