close

Вход

Забыли?

вход по аккаунту

?

Optical control of microwave integrated circuits using high-speed photoconductive switches

код для вставкиСкачать
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may
be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand corner and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in
reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly
to order.
U n iv ersity M icrofilm s in te rn a tio n a l
A Beil & H ow ell in form ation C o m p a n y
3 0 0 N o rth Z e e b R o a d . A nn Arbor. Ml 4 8 1 0 6 -1 3 4 6 U SA
3 1 3 /7 6 1 -4 7 0 0
8 0 0 /5 2 1 -0 6 0 0
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
O rd e r N u m b er 9425126
O ptical control o f m icrow ave integrated circuits using h igh-sp eed
p h otocondu ctive sw itches
Saddow, Stephen Edward, Ph.D.
University of M aryland College Park, 1993
C o p y rig h t © 1993 by S ad d o w , S te p h e n E d w ard . A ll rig h ts reserv ed .
300 N. Zccb Rd.
Ann Arbor, MI 48106
R ep ro d u ced with perm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
T itle o f D issertation:
O ptical C ontrol o f M icrow ave Integrated C ircuits
U sing H igh-Speed P h o to co n d u ctiv e S w itches
S tephen E dw ard Saddow . D octor o f P h ilo so p h y , 1993
D issertation directed by:
Prof. Chi I I. Lee. P ro fesso r o f E lectrical E ngineering
D epartm ent o f E lectrical E ngineering
A n optoelectronic attenuator suitable fo r the optical control o f m icrow ave
integrated circu its is presented.
H igh-speed photoconductive sw itches are em bed d ed
in plan ar m icrow ave transm ission lines and sem ico n d u cto r laser diodes (L D s) are used
to control the m icrow ave signal level on these high-speed lines.
W ith a silicon
co p lan ar
to
w av cguide-photoconductive
m icrow ave
atten u atio n
has
been
sw itch
achieved
(S i:C P W -P C S ).
w ith
a
up
fibcr-pigtailcd
45
dB
laser
of
diode
having 144 m W o f optical pow er.
E dge-coupled F abry-P erot A lG aA s/G aA s sem ico n d u cto r LD s. as w ell as both
silicon
and
gallium -arsenide
(G aA s)
C P W -P C S s.
w ere
developed
for
the
o p toelectronic attenuator. W hen conventional gain -sw itch in g tech n iq u es are used. LD
peak o u tp u t pow ers greater than
optical
^ -sw itc h in g
schem e is used.
1 W have
6
been d em onstrated, and w hen an
W o f peak pow er has been achieved in tens o f
picoseconds.
T he L D /G aA s:C P W -P C S interaction w as carefully studied using both discrete
broad-area LD s and a tunable w avelength titan iu m -sap p h ire laser.
A lso studied w as
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
the d ep en d en ce o f the G aA s:C P W -P C S o n -state resistan ce (Ron) w ith respect to
applied static electric field and device tem p eratu re; for a fix ed LD w av elen g th . R on
could be d ecreased by a factor o f three for a tem p eratu re ch an g e o f 20 °C.
T his is
superior to w h at can be achieved w ith conventional LD w av elen g th tu n in g techniques.
A “ h y b rid " o p toelectronic attenuato r w as constructed from a S i:C P W -P C S and
a co m m ercially available fiber-pigtailed LD. T he S i:C P W -P C S ex h ib ited an optical
saturation effect; exploiting this effect w e achieved 45 dB o f atten u atio n at 1.7 G H z.
M easu rem en ts m ade using a vector netw ork an aly zer sh o w that the atten u ato r
p erform ance can be explained by a classical p lasm a ab so rp tio n argum ent, w hereby the
m icrow ave signal is attenuated by the optically induced so lid -state plasm a.
F inally, silicon carbide (SiC ) PC sw itch es m ade on 6 H -SiC su b strates w ere
used so that the P C and photovoltaic (PV ) resp o n se o f this m aterial co uld be
d eterm ined w hen both lateral and vertical PC sw itch g eo m etries w ere used.
D eep
level tran sien t spectro sco p y (D L T S ) m easu rem en ts w ere m ad e on both tw o -sid ed and
ab ru p t-ju n ctio n SiC p-n diodes.
T his w ork has confirm ed th e prev io u sly po stu lated
"D -ce n ter." a native defect observable o nly in the presence o f boron.
T h e D -center
trap d en sity , cross section, and activation energy w ere id en tified , w ith the trap residing
at 0.58 eV ab o v e the valence band.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Optical Control of M icrowave Integrated Circuits
Using High-Speed Photoconductive Switches
by
S tephen E dw ard S addow
D issertation subm itted to the Faculty o f the G raduate School
o f T he U niversity o f M aryland in partial fulfillm ent
o f the requirem ents for the degree o f
D octor o f P hilosophy
1993
A dvisory C om m ittee:
P rofessor
P rofessor
P rofessor
P rofessor
P rofessor
Chi H. Lee. T hesis A dvisor
Julius G oldhar
C hristopher C. D avis
K aw thar Zaki
W endell T. H ill
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
0 C opyright by
Stephen Edw ard S addow
1993
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Dedication
In M em ory o f E dw ard M . S addow . Sr.
To K aren for her patience and unw avering support.
ii
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Acknowledgm ents
I first w ould like to thank Dr. Chi H. Lee for his guidance, su p p o rt and
technical insight during this research endeavor. I am esp ecially grateful to Dr. L ee for
his k indness, patience and integrity, and for the excellent scien tific en v iro n m en t th at
he has established a t the university.
O ur technical and personal d isc u ssio n s w ere
indeed one o f the highlights o f this experience.
I w ould like to thank all o f the past and present m em b ers o f the U ltrafast
O ptoelectronics L aboratory.
In particular. I w ish to thank D r. Bruno T h ed rez w h o se
p rofound insight into the m any technical problem s we en co u n tered w as tru ly an
inspiration.
I am also grateful to Dr. Luke H uang for h is assista n ce in d ev ice
m odeling. Eric Funk for his technical support and reading o f the m a n u scrip t, and Pak
C ho for his help in perform ing the UV laser experim ents.
I am also grateful to Dr.
Julius G o ld h ar for his enthusiasm and technical support. T h e assistan ce o f Y uan L iu
for operatin g the N diglass laser and Steve Y ang for his insight into laser d iode p h y sic s
is also sincerely appreciated.
F inally. I en jo y ed co llab o ratin g w ith Y i-S e m Lai and
Dr. W ei C ao on th eir high-tem perature sup erco n d u cto r experim ents.
I am grateful for the financial su p p o rt provided by th e A rm y R esearch
L aboratory, and u 'ould like especially to thank Dr. Joseph P. S attler for su p p o rtin g m y
fellow ship at the university.
T he support o f Louis J. Jasp er. Jr.. h as indeed b een
invaluable, as well as the support o f W illiam L. V ault and D r. Jam es M . M cG arrity. I
thank D r's. C hristian Fazi. R oger K aul. and F. Barry M cL ean for th e ir tech n ical
insight.
I also thank Dr. R ichard P. Leavitt for providing G aA s M B E m aterial.
I
w ould like to thank M ary M orrone for keeping m e o u t o f o rg an izatio n al d ifficu lties
and. o f course, for her assistance.
A special thank you goes to the T echnical
P ublishing S taff for their ever tim ely and professional m an u scrip t ed itin g .
F in ally . I
iii
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
am indebted to T im o th y M erm agen for providing an en d less su p p ly o f sh a d o w m asks
and d ev ice fabrication ideas, and to C h risto p h er M erm agen for his help in d ev elo p in g
the L D M T .
I have had the good fortune to w ork w ith so m any o u tstan d in g p eo p le d u rin g
this project, w hom I w ould like to ackno w led g e now :
K evin B oulais (N av al S urface
W arfare C enter) fo r his technical d iscu ssio n s, insight, and softw are assistan ce. Dr.
P enny P o lak-D ingels (L aboratory for Physical Sciences) for use o f h er T i:S ap p h ire
laser. John C a cc io la (A n ritsu/W iltron) for the use o f the W iltron V ecto r N etw ork
A nalyzer. D r. M ichael M azzola (M S State U niversity) for innum erable technical
discussions and for assistance w ith the D LTS m easurem ents. Dr. Phil N eu d eck
(N A S A ) for fabrication o f SiC p-n diodes. Dr. John P alm o u r (C ree R esearch Inc.) for
fabricating the S iC PC sw itches. A rm and B alckdjian (A R L -F t. M o n m o u th ) for
assistin g in silico n device fabrication. Dr. D ave M azzoni (U n iv ersity o f M D ) for his C
pro g ram m in g assistance, and Dr. A ryc R osen (D av id S a rn o ff R esearch C en ter) for
pro v id in g the 1-cm A IG aA s laser dio d e array. 1 also thank Dr. Jam es M asi (W estern
N ew
E ngland C ollege). C harles S palding (E lectronic
C oils.
Inc.).
and
R obert
K inasew itz (U S A rm y A R D E C ) for helping to shape m y technical sk ills o v er the
years.
N o n e o f th is w ould have been possible w ith o u t the years o f su p p o rt and
encourag em en t offered by m y fam ily, and I w ould like to esp ecially th an k m y p aren ts
and grandparents for all they have done for m e and for teach in g m e the v irtu es o f
honesty and integrity.
Finally, to the one w ho has perhaps sacrificed m ore th an I
during the p ast n in e years, m y w ife K aren.
I f not for her en co u rag em en t, unselfish
dev o tio n , and endless supply o f h o t tea. none o f this w ould have been po ssib le.
iv
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Table O f Contents
S ection
Ea&st
L ist o f T a b le s ............................................................................................................................................. viii
L ist o f F ig u re s............................................................................................................................................... ix
C h a p te r 1. I n tr o d u c tio n ...........................................................................................................................1
§1.1 O v e rv ie w ................................................................................................................................. 1
§ 1.2 O bjectives o f the R e se a rc h ............................................................................................... 4
§ 1.3 S um m ary..................................................................................................................................7
C h a p t e r 2. M ic ro w a v e O p to e le c tro n ic A tte n u a to r S c h e m e ................................................. 11
§2.1 O verview o f M icrow ave A tte n u a to rs.......................................................................... 13
§2.2 R eflective O ptoelectronic A tten u ato r S c h e m e ......................................................... 16
§2.3 A bsorptive O ptoelectronic A ttenuator S c h e m e ....................................................... 23
§2.4 S um m ary............................................................................................................................... 33
C h a p te r 3. L a s e r D io d e D e v e lo p m e n t............................................................................................. 34
§3.1 M ulti-S ection Q uantum -W ell Laser D io d e s............................................................. 36
§3.2 L aser D iode D esign. F abrication, and P erfo rm a n ce...............................................45
§3.3 O ptically O -S w itched T w o-S ectio n L aser D iode E x p e rim e n t...........................51
§3.4 Laser D iode M icrow ave Im pedance M atch in g T ran sfo rm er...............................57
§3.5 S um m ary............................................................................................................................... 63
C h a p te r 4. P h o to c o n d u c tiv e S w itc h e s.............................................................................................64
§4.1 P hotoconductive Sw itch P h y sic s.................................................................................. 65
§4.2 Planar M icrow ave T ransm ission L in e s ......................................................................69
§4.3 C P W -P C S D esign. Fabrication, and E v a lu a tio n .....................................................73
§4.4 C P W -P C S E lectrical and O ptical C h a ra cteriza tio n ................................................ 77
§4.5 S um m ary............................................................................................................................... 84
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
C h a p te r 5. G a A s :C P W -P C S C h a r a c te r iz a tio n ...........................................................................86
§5.1 Laser D iode M e asu rem en ts............................................................................................87
§5.2 T i:S ap p h ire Laser M e asu rem en ts.................................................................................90
§5.3 G aA s:C P W -P C S C onductive-M ode P lasm a M o d e l...............................................96
§5.4 S u m m ary .............................................................................................................................101
C h a p te r 6 . O p to e le c tro n ic A tte n u a to r E x p e r im e n ts ..............................................................103
§6.1 N d:G lass L aser M easu rem en ts....................................................................................105
§6.2 B road-A rca Laser D iode M e a su re m e n ts................................................................. 108
§6.3 F iber-P igtailed Laser D iode E xperim ents - R.F V oltage M e a su re m e n ts.... 114
§6.4 F ib er-P igtailed Laser D iode E xperim ents - V N A M e a su re m e n ts..................119
§6.5 S u m m ary .............................................................................................................................126
C h a p te r 7. S ilicon C a r b id e P C S w itch D e v e lo p m e n t............................................................ 127
§7.1 M aterial Properties o f 6 H -S iC .....................................................................................129
§7.2 SiC P hotoconductive S w itch D esign an d F a b ric a tio n .........................................131
§7.3 P hotovoltaic E xperim ents U sing L ateral S w itc h e s ..............................................134
7.3.1 N ear B and-G ap M e a su re m e n ts.................................................................. 135
7.3.2 H igh-Speed P icosecond M ea su re m en ts....................................................139
7.3.3 P hotovoltaic E xperim ental D isc u ssio n ..................................................... 142
§7.4 P hotoconductive SiC Sw itch E x p erim e n ts..............................................................147
§7.5 SiC D eep-L evel T ransient S pectroscopy M e a s u re m e n ts ...................................152
§7.6 S u m m ary .............................................................................................................................162
vi
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Chapter 8. Conclusions and Future R esearch..................................................................... 163
§8.1 C o n c lu s io n s ....................................................................................................................... 163
§8.2 Future R e se a rc h ................................................................................................................ 164
A ppendix A: SiC D LTS C a lc u la tio n s............................................................................................... 172
R e fe re n c e s .................................................................................................................................................. 179
C irricu lu m V itae
vii
R ep ro d u ced with perm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
List o f Tables
N u m b er
P age
3.1 M B E 330 S Q W G R IN S C H Laser S tru c tu re...............................................................................40
7.1 C om p ariso n o f S iC P roperties to Si
and G a A s....................................................................... 131
7.2 C om parison o f D -C en te r Param eters as M easured H ere and R eported in the
L ite ra tu re .............................................................................................................................................. 161
A .l M inority T rap D L T S D a ta .............................................................................................................172
A .2 M ajority T rap D L T S D a ta .............................................................................................................176
viii
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
List o f Figures
N um ber
1.1
H ybrid O ptoelectro n ic A ttenuator S c h e m a tic ..........................................................................6
2.1
R eflective O ptoelectronic A ttenuator S c h e m a tic ................................................................. 20
2.2
L o ss-L ess T ransm ission L ine T w o-P ort S c h e m a tic ............................................................ 20
2.3
R eflectiv e A ttenuator T heoretical P re d ic tio n .........................................................................22
2.4
A bsorptive O p to electro n ic A ttenuator S ch em atic................................................................ 25
2.5
A bsorptive A tten u ato r T heoretical Prediction versus P lasm a D e n s ity .........................31
2.6
A bsorptive A tten u ato r T heoretical P rediction versus LD P o w e r....................................32
3.1
S Q W -G R IN S C H L aser D iode M aterial S tru ctu re................................................................ 39
3.2
L aser D iode P erform ance versus Q uan tu m W ell N u m b e r ................................................ 41
3.3 (a) B road-A rea L aser D iode C ontact G e o m e try ...................................................................... 43
3.3 (b) R idge-W aveguide L aser D iode C o n tact G e o m e try .........................................................43
3.4
T w o-S ection L aser D iode S c h e m a tic ....................................................................................... 44
3.5
T w 'o-Scction L aser D iode T heoretical U I C u r v e ................................................................. 44
3.6
P hotograph o f B road-A rea Laser D iode C leaved B a r.........................................................46
3.7
P hotograph o f R idge-W aveguide L aser D iode C leaved B a r ............................................ 46
3.8
B road-A rea L aser D iode E xperim ental L /l C u rv e................................................................47
3.9
B road-A rea L aser D iode O utput S p e c tru m ............................................................................ 47
3.10 T w o -S ectio n L aser D iode E xperim ental LI I C urve versus B ia s....................................... 49
3.11 F acet Im age o f 150 (am S tripe B road-A rea L aser D io d e.................................................... 50
3.12 P hotograph o f P ackaged Laser D io d e ...................................................................................... 50
3.13 L aser D iode O ptical O -S w itching E xperim ental S e t- U p ...................................................54
3.14 T y p e-4 L aser D iode O -S w itched O u tp u t P u lse..................................................................... 56
3.15 T y p e -5 L aser D iode O -S w itched O u tp u t P u lse..................................................................... 56
IX
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
3.16
L D M T E lectrical C ircuit S ch em atic........................................................................................ 58
3.17
L D M T M icrostrip C ircuit L a y o u t............................................................................................58
3.18
L D M T an d 50 Q M icrostrip C ircuit S'] ( C om parison (T h e o re tic a l)............................ 61
3.19
L D M T and 50 Q M icrostrip C ircuit S | | C om parison (E x p e rim e n ta l)........................ 61
3.20
U I C urve W ith L D M T and 50 Q M icrostrip C ircuit D riv ers..........................................62
4.1
M icrostrip and C o p lan ar W aveguide G e o m e trie s...............................................................72
4.2
G aA s:C P W -P C S C ro ss-S e c tio n ............................................................................................... 75
4.3
S i:C P W -P C S C ro ss-S e c tio n ...................................................................................................... 75
4.4
Photograph o f 50-pm G ap G aA s and Si:C PW -PC S D ev ic es.......................................... 76
4.5
M easured S i] o f G aA s:C PW -P C S (D ark V a lu e )................................................................ 80
4.6
M easured S t | o f S i:C P W -P C S (D ark V alu e )....................................................................... 80
4.7
E xperim ental Set-U p used to m easure C PW -PC S PC C arrier L if e tim e .....................81
4.8
M easured G aA s:C P W -P C S PC C arrier L ifetim e (a) IR. (b) G reen, (c) B o th ............. 82
4.9
M easured S i:C P W -P C S PC C arrier L ifetim e (a) 1.054 p m . (b) 797 n m ....................83
5.1
E xperim ental Set-U p used to m easure R on w ith 300 p m L aser D io d e.........................89
5.2
M easured R an w ith 300 pm Laser D iode versus T e m p e ra tu re........................................89
5.3
E xperim ental Set-U p used to m easure Ron w ith T irsappiiirc L aser............................... 92
5.4
M easured R on w ith T i:sapphirc versus P hoton E nergy and T e m p e ra tu re ...................92
5.5
M easured Ron w ith T i:sapphirc versus P hoton E nergy and E lectric F ield ..................94
5.6
M easured a ( c n r ') versus P hoton Energy o f S em i-In su latin g G aA s.............................94
5.7
M easured Ron w ith T i:sapphirc versus P hoton E nergy and L aser P o w e r ...................95
5.8
G aA s: C P W -P C S Schem atic V iew used in C o n d u ctiv e-M o d e P lasm a M o d e l
5.9
T heoretical and E xperim ental Ron versus Photon E nergy and T e m p e ra tu re ..............99
5.10 T heoretical and E xperim ental Ron versus Photon E nergy and E lectric F ie ld
99
100
6.1
E xperim ental Set-U p U sed to M easure 6-2 i w ith N d :g lass L a s e r................................. 10?
6.2
O scillogram o f A ttenuated RF W aveform U sing N d :g la ss L a s e r ................................107
x
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
6.3
O scillogram o f A ttenuated R F W aveform and LD P ulse U sin g 300 pm L D .....
110
6.4
RF A ttenuation versus LD Pow er U sing 300 p m L D .................................................
110
6.5
O scillogram o f A ttenuated R F W aveform U sin g 1-cm L inear LD A rra y ............
112
6.6
RF A ttenuation versus LD Pow er U sing 1-cm L inear LD A rra y ............................
112
6.7
RF A ttenuation versus R F P ow er U sing 1-cm L in ear LD A rr a y ............................
113
6.8
Experim ental S et-U p U sed to M easure 5 t | w ith F ib er-P ig tailed L D ....................
117
6.9
RF A ttenuation versus L D Pow er U sing Fibcr-P igtailed L D ....................................
117
6.10 T heoretical and E xperim ental 5->t versus R o n ................................................................
118
6.11 M easured S i | w ith VNA versus L aser Pow er (10 p m x 10 p m S p o t)...................
1n
6.12 M easured S \ | w ith VNA versus L aser Pow er (10 p m x 10 p m S p o t)...................
1a
6.13 M icrow ave A ttenuation versu s LD Power U sing F iber-P igtailed L D ...................
123
6.14 M easured S t ) w ith VN A versus L aser Pow er (10 p m x 2.5 m m S p o t) ................
124
6.15 M easured S i | w ith VN A versus L aser Pow er (10 p m x 3.5 m m S p o t) ................
124
6.16 M axim um S i | o f 45 dB for LD P ow er o f 144 m W ......................................................
125
7.1
Schem atic V iew o f Lateral 6 H-SiC PC S w itc h .............................................................
133
7.2
Schem atic V iew o f V ertical 6 H-SiC PC S w itc h ...........................................................
133
7.3
SiC PC S w itch PV R esponse E xperim ental S etu p .......................................................
136
7.4
M easured PV Response o f 10 pm G ap Lateral SiC PC S w itc h ...............................
136
7.5
SiC PC S w itch PV R esponse versus Beam Position E xperim ental S e tu p ...........
137
7.6
SiC PC S w itch PV R esponse versus Beam Position at 308 an d 43 0 n m .............
137
7.7
M easured PV Response o f 3 mm Lateral SiC PC S w itch versus W avelength....
140
7.8
M easured a f c n v 1) versus P hoton Energy o f P-T vpc S i C .........................................
140
7.9
P icosecond PV R esponse (a) 266 nm. (b) 450 nm. (c) 53 2 n m . (d) 705 n m .......
141
7.10 SEM Im age o f Lateral S iC PC Sw itch C ontact R e g io n .............................................
146
7.11 Proposed SiC PC S w itch E nergy Band D iagram fro PV R e s p o n s e .......................
146
7.12 SiC PC S w itch PC R esponse Experim ental S e tu p ......................................................
160
R ep ro d u ced with perm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
7.13
M easured L ateral SiC PC Sw itch PC R esponse versus A p p lied B ia s........................150
7.14
M easured V ertical SiC PC Sw itch PC R esponse versus A pplied B ia s ......................151
7.15
M easured V ertical SiC PC Sw itch PC C arrier D ecay L ifetim e at 431 n m
7.16
S am ple o f M easured SiC D LTS M inority C a rrie rT ra p S p e c tru m .............................. 155
7.17
S am ple o f M easured SiC D LTS M ajority C a rrie rT ra p S pectrum .............................155
7.19
A rrhenius Plot o f SiC 0.58 eV T r a p ...................................................................................... 158
8.1
M onolithic O ptoelectronic A ttenuator S ch em atic ............................................................. 171
151
xii
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Chapter 1
Introduction
§1.1 O verview
Since the discovery o f picosecon d p h o to co n d u ctiv ity by Ja y aram an an d Lee
[ 1]. the co m b in atio n o f m icrow ave and ultra-fast op tical tech n o lo g ies has advanced
sign ifican tly in recent years [2],[3],
O p toelectronic integrated circu its, o r O E IC s.
co ntinue to m ake im pressive strides [4],[5], w hile the d ev elo p m en t o f M onolithic
M icrow ave Integrated C ircuits (M M IC s) [6 ] has n o w p ro g ressed to th e p o in t w here
M M IC s arc being placed into both com m ercial an d m ilitary sy stem s [7 ],[8 ],
The
in tegration o f these tw o techn o lo g ies to perform hig h -sp eed m icro w av e system s
functions are seen by som e to represent the n ex t-g en cratio n o f electro n ic integrated
circu its [9],
In fact, a new in tegrated circuit d ev ice n o m en clatu re for such a
technology has been suggested by S im o n s [10];
she refers to these circu its w ith
com bined M M IC and O E IC com ponents as optical m o n o lith ic m icro w av e integrated
circ u its (O M M IC s).
T he advan tag es o f M M IC s
for m icrow ave sy stem s
use are
num erous,
esp ecially at m illim eter-w av e (M M W ) frequencies w here trad itio n al m icrow ave
techniques are im practical.
For exam ple, the large scale in teg ratio n o f m icrow ave
sub sy stem s u sing M M IC s allow s for the M M IC to be placed d irectly o n the an tenna
substrate, thus reducing the need for intervening co n n ectio n s and th ereb y reducing
system co m p lex ity and cost. For less am b itio u s system s, su ch as au to m o b ile co llisio n
av o id an ce sensors [11]. M M IC s can provide m uch h ig h er reliab ility at a greatly
reduced cost since all com p o n en ts are m o n o lith ically integrated onto a sin g le substrate
and the p roblem s associated w ith w irin g to g eth er discrete co m p o n en ts a re elim inated.
1
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
M M IC fabrication uses the sam e technology that is used to fab ricate high
speed
O E IC s.
For
exam ple,
alum in u m -g alliu m -arsen id c/g alliu m
arsenide
(A lG aA s/G aA s) h etcrostructures are em ployed to m ake activ e m icro w av e d evices,
such as I-IEMTs [12]—[14] and H B Ts [15]; in O E IC s. A lG aA s/G aA s q u an tu m w ells
are em ployed to form the active regions o f laser diodes, optical w av eg u id es, and
p hotodetectors [16].
T hus, O E IC and M M IC circuit to p o lo g ies can, in p rin cip le, be
m ade tog eth er on the sam e substrate to form very high speed circuits. E v en the basic
building block o f m o st M M IC designs, the M E S FE T . has been co n tro lled o p tically
[17], as w ell as the G unn diode [18]. T his technology has m ade som e m o d e st success
in recent years and all indications are that even higher levels o f system in teg ratio n will
be achieved in the n ea r future [19].
As further ev id en ce o f the m icro w av e potential o f opto electro n ic d ev ices, laser
diodes w ith relaxation frequencies up to 30 G H z [20]. p h o to d etecto rs w ith b an d w id th s
on the order o f 100 G H z [21], and electro-optic in ten sity m o d u lato rs w ith b an d w id th s
o f 6 G H z [22] and even 20 G H z [23], have been reported. T hus, not o n ly can O E IC s
be fabricated w ith M M IC devices to form O M M IC s. there are inherent ad v an tag es
(high-speed, large bandw idth, light w eight, high reliability, infinite iso latio n , etc.) to
such an integration that is attractive to the system designer.
H ow ever, im provem ents in M M IC control techniques, especially for sy stem s
co m p o sed o f m u ltip le M M IC s. such as phascd-array radar, arc needed if M M IC s are to
b eco m e co m m ercially viable.
T his is especially true for M M IC s o p era tin g in noisy
en v ironm ents w here use o f an optical control signal can p rovide m ore reliab le control
due to the inherent im m unity o f optical fibers to r f interference [24],
In ad dition.
M M IC s w hich o perate at rem ote sites from a cen tral statio n , or that are placed onto
surfaces w ith lim ited access (such as sen so rs placed on the skin o f aircraft [2 5 1), are
d ifficu lt to control w ith all-electrieal/m icrow avc sig n als due to the high sig n a l loss and
->
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
large size o f the control com ponents [26], O bviously this is ag ain a situ atio n w here
o p tical control o f m icrow ave system s has a m ajor ad v an tag e o v er traditional
techniques. N ow that w e have discussed the need to control m icrow ave circuits, and
w hy optical techniques are preferable over traditional m ethods for certain applications,
a rev iew o f the various control circuits frequently used is in order.
There are several w ays to control m icrow ave circuits. P hase sh ifters are often
ap p lied to shape antenna beam patterns [27]. perform intcrferom etry [28], and provide
isolation betw een various m icrow ave ports [29], All o f these arc im p o rtan t m icrow ave
control functions that are typically perform ed using all-electro n ic or m icrow ave
techniques [30].
H ow ever, there are oth er techniques that are frequently em ployed,
such a s m icrow ave sw itches [31] and attenuators [32].[33].
A lthough these traditional techniques arc w idely used d u e to th eir adequate
perform ance, and ease o f fabrication and /o r system insertion, they all su ffer from one
o f the aforem entioned draw backs.
In particular, w ith all o f these techniques there is
little o r no isolation betw een the control signal and the m icrow ave signal to be
co ntrolled.
T his can represent a serious problem
for sy stem s w here m ultiple
m icrow ave integrated circuits (M M IC s) are to be co n tro lled [34]. such as in large (or
even m odest sized) phased array antennas.
In such system s the n u m b er o f radiating
elem ents, and, hence, M M IC s. can appro ach tens o f thousands [10].
A n additional difficulty w ith all-electronic or m icrow ave control is the routing
o f the control signal to the M M IC m odule.
T his requires either bulky w ires, in the
case o f electrical control signals, o r w aveguiding co m ponents in the case w here
m icro w av e signals perform the control function.
A third draw back has to do w ith
tim in g errors, or w hat is m ore com m only referred to as tim ing jitte r. T h is is a problem
w h en a high degree o f m icrow ave control accuracy required.
A ll electrical system s
J
R ep ro d u ced with perm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
ca n
typically
provide
control
w ith
tim ing jitte r
in
the
n anosecond
regim e:
opto electro n ic techniques can provide sim ilar control w ith pico seco n d jitte r [3],
A s discussed earlier, M M IC s and O E IC s are u su ally fabricated w ith th e sam e
m aterial structure.
It therefore seem s reasonable that an atten u ato r for M M IC
applications utilizing O EIC techniques is feasible from a fabrication p o in t o f view .
H ow ever, this is only true provided that the atten u ato r req u irem en ts o f fast turn-on
speed an d laser d iode activation arc met.
R ecent w ork on high speed laser d io d es indicates that these d ev ices can be
m odulated w ell beyond 10 G H z [20] and could thus serve as the trigger m ech an ism for
a M M IC attenuator.
All that is then req u ired is to generate a so lid -state p lasm a to
atten u ate the m icrow ave signal by altering the r f im pedance o f the tran sm issio n line.
B ulk G aA s photoconductive sw itches can b e fabricated w ith d y n am ic resistan ces o f
less than 1 Q [35],
T hese d ev ices also have the p roperty th at their tu rn -o n tim e is
equal to the rise tim e o f the optical trigger, n am ely a laser p u lse o f suitable p o w er and
w avelength.
§1.2 Objectives o f the Research
The three principal advantages o f op tical control o f m icrow ave circ u its w ere
ju s t discussed: high degree o f isolation, n oise im m unity, an d h ig h -sp eed control.
T hese three m ain points form the basis o f this research, nam ely to d em o n strate an
o p toelectroinc attenuator suitable for co n tro llin g M M IC s.
T his atten u a to r sh o u ld be
controllable w ith sub-nanosecond precision and p ro v id e g reater th an 30 dB o f
m icrow ave attenuation.
T he attenuato r m u st be all so lid -state (i.e.. based on
sem iconductor com ponents only) to perm it eventual M M IC insertion, an d m ust
require reasonable optical pow er levels to ach iev e w ell co n tro lled atten u atio n levels.
4
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
In ad d itio n , the attenuation m ust be achieved to r tim e scales on the o rd e r o f a
nanosecond; therefore, atten u atio n levels that can o nly be su stain ed for p ico seco n d
tim e fram es are unacceptable (this design co n sid eratio n helps to red u ce the laser p o w er
requirem ent). T herefore, m aterials w ith carrier reco m b in atio n lifetim es on the o rd er o f
1 ns. or longer, are to be considered.
In add itio n to develo p m en t o f a su itab le atten u atio n elem en t, the o p tical driv er
(i.e., laser diode) m ust eith er be developed to d o th e jo b , o r ex istin g laser diode
technology assessed and a suitable candid ate d ev ice selected.
F inally, n ew m aterials
that m ight greatly im prove the perform an ce o f the o p to electro n ic atten u a to r m u st be
investigated; these new m aterials m ight also help to reduce the co st o f the a tte n u a to r
by perm ittin g device m iniaturization, im proving d ev ice han d lin g and fabrication, or
p erm itting the optoelectronic attenuator to p roperly function in harsh en v iro n m en tal
cond itio n s, such as w h at is encountered for rem ote m icrow ave sites.
A n y new
m aterials m u st be adequately characterized , both electronically as w ell as o p tically , so
that recom m endations for future d evelopm en t can b e in tellig en tly m ade.
Finally, the optoelectronic attenu ato r m ust work: it is n ice to perform d etailed
and thorough research, how ever, it is eventually the success o f th is project th a t w ill
interest M M IC designers to em ploy this technique on w orking M M IC system s.
A s a preview o f the next chapter. Figure 1.1 show s a pictorial v ie w o f the
proposed optoelectronic attenuator.
An im bedded PC sw itch resides w ith in a h igh­
speed m icrow ave transm ission line and is the atten u atin g elem en t.
A sem ic o n d u cto r
LD is used to both activate the atten u ato r and to co n tro l the atten u atio n level.
Either
free-spacc optics or fiber-optics (not show n in figure) m ay be used to d e liv e r the
optical po w er to the PC sw itch.
Before we discuss the o p to electro n ic a tten u a to r in
further d etail, a sum m ary o f the results attain ed d u rin g this research effo rt is helpful to
put our w ork into proper context.
5
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
V,
PC Sw itch
Frcc-Space O ptics
L aser D iode
Figure 1.1 D iagram o t'th e proposed hy b rid o p to electro n ic atten u ato r schem e.
Z0 = Z l = transm ission line ch aracteristic im pedance.
S em iconductor LD activates im bedded PC sw itch to attenuate r f signal.
6
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§1.3 Sum m ary
D uring the course o f this research project, m ost o f the o b jectiv es w ere m et.
A ttenuation levels exceeding 45 dB w ere achieved using less than 145 m W o f p o w er
from
a com m ercially
available
laser d io d e
[36];
hence, an
all-sem ico n d u cto r
opto electro n ic attenuator that requires m odest laser p o w er levels has been su ccessfu lly
dem onstrated.
In C hapter 2. tw o atten u atio n schem es are presented along w ith
estim ates o f their anticipated
perform ance.
H igh-speed
p h o lo co n d u ctiv e
(PC )
sw itch es, fabricated on both gallium arsen id e (G aA s) and silicon (Si), have been
d eveloped using a coplanar w aveguide (C P W ) contact g eom etry: suitable m icro w av e
perform ance has been achieved w ith these devices. T he dev elo p m en t o f these h ig h ­
speed coplanar w avcguide-P C sw itches (C P W -P C S s) is described in C h ap ter 4. along
w ith the relevant physics o f such high-sp eed devices.
If the optoelectronic atten u ato r is to be realized, a su itab le laser source m ust be
developed: this laser m ust be a sem ico n d u cto r d evice (i.e.. laser diode, o r LD for
short) and m ust be triggerable in less than a nanosecond w ith p ico seco n d tim ing jitte r.
T h is is a rather stringent design requ irem en t, esp ecially in light o f the fact that
consid erab le optical pow er is norm ally req u ired for PC sw itching ap p licatio n s [37],
C onseq u en tly , in-depth research w as perform ed to ensure that a suitable LD w as
available to m eet the optoelectronic atten u ato r design requirem ents:
d iscu sses the progress m ade in this area.
C hapter T h ree
In particular, w e have achieved the h ig h est
rep o rted broad-area LD output pow ers u sin g an optical O -sw itching schem e, w h ere the
L D output w as increased to g reater than 6 W from ap p ro x im ately IW . and the p ulse
w idth reduced from 100 ns to 60 ps [38]. W e have also used the op tical O -sw itch in g
tech n iq u e to better understand the LD d y n am ics [39] so th at further im p ro v em en ts can
b e m ade. U sing a tw o-section broad-area LD [4 0 ].[4 1] w e dem o n strated that su ch a
7
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
device can be turned on in less than 500 ps: thus, su b -nanosccond o p eration o f an
optoelectronic attenuator seem s feasible. Finally, the light output o f a broad-area LD
has been increased by greater than 37 percent using m icrow ave m atch in g techniques.
Since A lG aA s/G aA s laser diodes (LD s) arc to be used as the laser source, a
d etailed in vestigation o f the interaction betw een a G aA s:C P W -P C S and the LD was
perform ed [42]; a titanium sapphire (T h sap p h irc) laser w as used to m easure the PC
sw itch perform ance as a function o f laser photon energy.
optoelectronic
attenuator
perform ance,
the
In o rd er to im prove the
G aA s:C PW -P C S
perform ance
was
sim ultaneously m easured as a function o f sw itch tem perature and applied electric
field. O ur results show that a factor o f three im provem ent in the o n -state resistance
(and. conversely, conductance) can be achieved, for a fixed photon energy, by tuning
the C P W -P C S tem perature over a 20 °C range.
C om parable LD tem perature tuning
w ould cut this im provem ent factor in half. D etails o f this investigation are described
in C hapter 5.
Since the G aA s:C PW -P C S perform ance appeared to be poor due to the
m aterials short (5 ns) PC carrier lifetim e, the S i:C P W -P C S u'as used to set-up the
hybrid optoelectronic attenuator (the S i:C P W -P C S PC carrier lifetim e w as ~1 pis).
O ptoelectronic attenuator experim ents w ere perform ed in several w ays: w ith pulsed
optical illum ination (m easured on a
1 G H z bandw idth analog oscillo sco p e) to
attenuate a 500 M H z r f w aveform [43], and w ith C W illum ination to atten u ate C W
w aveform s from a vector netw ork analyzer (V N A ) [30]. T hese ex p erim en ts perm itted
us to finally assess the relevance o f the tw o attenuation schem es p resen ted in C h ap ter
2; thus, the exact nature o f the attenuation m echanism for such a device has been
determ ined. T hese results are presented in C h ap ter 6 w here w e see that the atten u ato r
not only provides greater than 45 dB o f m icrow ave attenuation, b ut that this can be
8
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
accom p lish ed w ith less than 145 m W o f optical p ow er by p ro p erly m atch in g the laser
beam profile to the S i:C P W -P C S .
A lthough silico n is not the usual m aterial o f ch o ice for M M IC s d u e to low er
ca rrie r m obility a n d lack o f sem i-insu latin g b ehavior, co n sid erab le research and
d ev e lo p m e n t has been conducted to develop silico n M M IC s [44], esp ecially at
frequencies below X -band ( - 1 0 G H z).
In ad d itio n , su p erlatticc stru ctu res can be
em p lo y ed in G aA s to increase the carrier lifetim e into the m icro seco n d reg im e [45].
T herefore, d em onstration o f a silicon-based o p to electro n ic atten u ato r is valid for
M M IC ap p lications; it can be directly used in silicon M M IC s. and su p erlattices can be
used to ach iev e sim ilar perform ance in an all G aA s-b ased o p to electro n ic atten u ato r
using a G aA s:C P W -P C S and A lG aA s/G aA s LD.
O ne o f the draw backs o f both G aA s and Si is th eir po o r therm al and electric
field breakdow n properties com pared w ith w id e-b an d -g ap sem ico n d u cto rs such as
diam o n d and silicon carbide (SiC ). S ince d ev ice-g rad e SiC is now av ailab le [46].[47].
the suitability o f S iC as a substrate m aterial for the o p to electro n ic atten u ato r w as
assessed.
D uring these investigations, w hich arc d escrib ed in C h ap ter 7. several
im portant revelations w ere m ade; a high-speed, efficien t p h o to v o ltaic (PV ) effect was
observed
and
is
P hotoconductivity
tim e [49]—[51 ].
being
studied
for
use
in
sen so r
m easurem ents w ere m ade o n SiC PC
ap p licatio n s
sw itches
[48], [49],
for the
first
A lthough w e w ere unable to take ad v an tag e o f the high-ficld
break d o w n potential o f bulk SiC (due to low m aterial resistivity), th eir hightem p eratu re capability w as observed; a SiC sw itch d issip ated m ore than 11 W o f static
p ow er and, although the device w as literally glow ing red -h o t d u rin g this tim e, only the
sw itch m ount failed.
R econnection o f the SiC PC sw itch verified th a t the device
survived such an ex trem e operating con d itio n .
N eith er G aA s nor Si d ev ices w ould
have dissip ated such static pow ers and survived to be used again. F inally, in an effort
9
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
to
im prove
the
intrinsic
resistivity
o f device-g rad e
S iC .
deep
level
transient
sp ectroscopy (D L T S ) m easurem ents were co n d u cted to ch aracterize so m e o f th e native
d efects in this m aterial. W e have identified a deep -lev el (0.58 eV ) trap th at w e believe
to be a hole trap [52]; how ever, the capture cro ss section an d activ atio n energy for an
e lectro n trap also correlates w ith this defect.
W hat is u n u su al is th at this trap lies
belo w the room -tem peralure Ferm i level in th e m aterial.
T his su p risin g resu lt has
been reported by the U niversity o f E rlangen (G erm any) a n d others [53],[54], and been
d esignated as the ” D -center." O ur results serv e as an in d ep en d en t co n firm atio n o f this
result and. as such, provide a valuable piece o f inform ation to the SiC co m m u n ity .
10
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Chapter 2
M icrowave O ptoelectronic Attenuator Schem e
S ince the m otivation for this research was d iscu ssed in C h a p te r 1. vve w ill now
d ev elo p the basis fo r the optoelectronic attenuator. T he adv an tag es o f our schem e arc
as stated previously: high-speed control, a h igh degree o f isolation b etw een the control
and m icro w av e signals, and virtually jitte r-fre e operation.
In the first section o f this
chapter, a general review o f m icrow ave attenuators w ill be p resented, w ith an
em phasis on M M IC atten u ato r circuits. T his will b e follow ed by a d iscu ssio n o f how
an o p tically co n tro lled atten u ato r w orks.
A ttenuators
are
usually
classified
attenuators and abso rp tiv e attenuators.
in on e o f tw o ca te g o rie s:
reflective
T his classificatio n is also relevant to the
optically con tro lled attenuator, and w e w ill see th at th ere are tw o related physical
interpretations o f h o w such a device m ig h t attenuate a m icro w av e signal. In fact, how
we classify the atten u ato r is dependen t on how w e in terp ret th e in teractio n o f the
optical and m icrow ave energy; we can postulate that the o p to electro n ic atten u ato r
m ight function in eith er a reflection or absorption m o d e.
o p eratio n
m ay
be
valid,
experim ents
are
n ecessary
to
S in ce either m o d e o f
d eterm in e
the
correct
interpretation: th ese experim ents are p resen ted in C h a p te r 6.
P h otoconductivity in a sem icond u cto r m edia is th e basis for th e o p to electro n ic
attenuator, as w ell a host o f o th e r useful devices, an d h a s been th o ro u g h ly studied by
Lee [55]. Sim ply stated, optical photons im pinging on a sem ico n d u cto r su b strate are
co n v erted to electro n -h o lc pairs, w hich in turn co n stitu te an clectro n -h o le (i.e.. solidstate) plasm a [55].
H ow this solid-state plasm a, w h ich for b o th o p to electro n ic
atten u a to r schem es is generated w ithin the sem ico n d u cto r substrate o f a h igh-speed
11
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
ph otoconductivc (P C ) sw itch, affects the p ropagation o f a m icro w av e signal is w hat
differentiates the tw o atten u atio n schem es, w hich w e w ill now describe.
First, if w e view the solid-state plasm a/m icro w av e signal in teractio n from a
circuit analy sis standpoint, then the p lasm a is seen as altering th e local co m p lex
im pedance o f the m icrow ave transm ission line at the spatial lo catio n w h ere it is
induced.
T his im pedance m ism atch w ill, in turn, alter the p ro p ag atio n o f the
m icrow ave signal by inducing a partial reflection o f the m icro w av e signal back
tow ards the signal source.
In addition to th is partial reflection, so m e en erg y will be
absorbed by the real co m ponent o f the im pedance m ism atch (i.e.. d ue to resistiv e
losses).
T hus, upon introduction o f an o ptically induced im p ed an ce m ism atch , the
transm ission line load experiences a decrease in signal stren g th , w h ich ap p ears to be
an atten u atio n to the load. For clarity, w e w ill refer to an atten u a to r that o p erates in
this fashion as a “reflective attenuator." even though a sm all p o rtio n o f the m icro w av e
pow er is also absorbed.
W e can also view the optically induced plasm a/m icro w av e in teractio n using a
m ore rigorous electrom agnetic approach, w hereby M axw ell's eq u atio n s are solved
directly. In this case, the p h ysics o f the piasm a is directly acco u n ted for as a ch an g e in
the m aterial pro p erties o f the m icrow ave m edia, nam ely as a ch an g e in the m a te ria l's
conductivity (and hence, perm ittivity).
In this case, the plasm a acts to increase the
m a te ria l's loss tangent, w hich, in turn, acts to absorb the m icro w av e signal.
A s w as
the case for the reflectiv e attenuator, there m ay also be an increase in signal reflectio n
from the lossy m edia due to any optically induced im pedance v ariatio n s: ho w ev er, the
fraction o f pow er th a t is reflected is no rm ally m uch less than that w hich is ab sorbed.
T herefore, w e w ill refer to the decrease in signal stren g th resu ltin g from th e direct
interaction o f the solid-state plasm a w ith the m icro w av e signal as an ab so rp tiv e
12
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
attenuation m echanism , and w ill refer to optoelectronic atten u ato rs th at operate o n this
principle as “ absorptive atten u ato rs."
In C hapter 6, optoelectronic atten u ato r experim ents will be d escrib ed an d the
actual attenuation m echanism responsible for the o p to electro n ic atten u ato r (w h ich
turns o u t to be absorptive) determ ined.
H ow ever, both interpretations had to be
explored not only to develop the op to electro n ic attenuator, but to pro p erly u n d erstan d
how it w orks. T herefore, the basis for both schem es w ill be d iscu ssed in d etail and
their respective perform ances estim ated.
To put this w o rk into pro p er persp ectiv e.
M M IC atten u ato r techniques will no w be briefly review ed.
§2.1 Overview o f M icrowave Attenuators
In the introduction to this chapter, we briefly m entioned tw o m icrow ave
attenuation schem es w here the attenuation m ay be caused by eith er an ab so rp tio n o f
the m icrow ave signal by a solid-state plasm a, or by a reflection o f the m icrow ave
signal from an optically induced im pedance m ism atch.
B efore d iscu ssin g h o w the
optoelectronic attenuator w ill be im plem ented, a b rie f overview
attenuators is in order.
o f m icrow ave
Since an excellen t review o f th e relev an t M M IC atten u atio n
techniques has been presented by Bahl and B hartia [56], w e will point out th e co m m o n
features o f all attenuators and contrast these w ith the key features o f the opto electro n ic
attenuator.
For a m icrow ave attenuator to be useful to the system d esig n er, it m ust
function w ith a m inim um o f reflected pow er being re-directed back to the m icrow ave
source. T his is im portant since high levels o f reflected po w er m ay resu lt in a change
o f the source characteristics, such as frequency “ load p u llin g " [57] and. in ex trem e
cases, large reflected pow er levels m ay cause dam age to the source activ e elem ent(s).
13
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T h u s, useful attenuators can provide ad eq u ate attenuation su fficien t to m eet th e system
ap p licatio n , w hile m in im izin g the am ount o f reflected pow er seen by the source. T his
w ould im ply that an ab so rp tiv e attenuator w ould be an "id eal" can d id ate for all M M IC
ap p licatio n s, especially for low -level signal operatio n . H ow ever, atten u atio n o f highlevel signals can pose difficu lties due to therm al heating in the ab so rp tiv e atten u a to r
w hich, in turn, alters its characteristic im pedance and thus increases the reflected
p ow er b ack tow ard the source.
In principle, any arbitrary im pedance m ay be m atched to an o th er i f an
im pedance transform er, w h ich is ideally m any w avelengths in length [58], is used.
T hus, as the atten u atio n in an absorptive atten u ato r increases, the im pedance m ay also
change a n d a m atching netw ork m ight be used to m inim ize reflectio n s.
H ow ever,
even if th e w avelength o f th e m icrow ave signal in the tran sm issio n line is in the
m illim eter-w ave range, this still im plies that rather long m atch in g n etw o rk s m ay be
necessary.
T herefore, absorptive attenuators, and esp ecially
those
rely in g
on
im pedance m atching netw orks to reduce p o w er reflections, arc less d esirab le for
M M IC integration due to th e lim ited am ount o f M M IC real estate.
In ad d itio n ,
ab so rp tiv e attenuators dissipate considerable m icrow ave pow er, and the n eed to
d issipate the resulting therm al energy aw ay from tem p eratu re-sen sitiv e activ e d evices
is obvious.
R eflective attenuators, such as the type d iscussed by W eb er [59], can be used
to circu m v en t these pro b lem s, w hereby the reflected m icrow ave p o w er is re-d irected
to a m atched load w ith the a id o f a non-reciprocal transm ission line elem ent, such as a
circu lato r [60] or transm it an d receive (T /R ) m odule [61]. T hus, the m atch ed load can
be eith er off-chip. o r at a convenient location on-chip that is far aw ay from any
tem perature-sensitive com ponents.
H ow ever, the principal d isad v an tag e o f this
approach is the need for a m icrow ave circu lato r (or T /R sw itch), w hich n o rm ally is
14
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
com p rised o f a three-port (or h ig h e r order) tran sm issio n line stru ctu re and a n o n ­
reciprocal elem ent, typically a ferrite resonator, th at perm its p o w er flow in o n ly o ne
d irec tio n (eith er from port 1 to port 2, or port 2 to port 3. etc.). A lthough co n sid erab le
p rogress has been m ade recently to synthesize circu lato rs using active ele m e n ts for
M M IC s [61], these devices do n o t have su fficien t isolation to p rev en t the le ak ag e o f
so m e o f the reflected pow er from reaching the m icrow ave source.
A gain, for low -
p o w er applicatio n s, this m ay be acceptab le, but for h ig h -p o w er ap p licatio n s th is no n ideal perform ance is cause for concern.
S im ply stated, for low -p o w er ap p licatio n s w here m icrow ave sig n als are to be
attenuated, use o f existing atten u a to r schem es (either ab so rp tiv e or reflec tiv e ) is
acceptable. H ow ever, w hen high -p o w er levels are involved, su ch as w ith h ig h -p o w e r
rad ars and senso rs [62], these trad itio n al app ro ach es have serious draw backs.
M ore
im portantly, they cannot be ac tiv ated rem otely w ith o u t sig n ifican t control tim in g jitte r:
thus, their su itab ility for sim u ltan eo u s control o f m u ltip le m icro w av e circu its is in
q uestio n . A lth o u g h other optical control sch em es have p ro v en their relative w o rth for
so m e applications [17],[18],[34], these are all fairly co m p licated schem es that e ith e r
require additional device fabrication steps or specialized device to p o lo g ies.
In
ad d itio n , these techniques m ay be slow , im p ly in g that h ig h -sp eed control m ay n o t be
p o ssib le [63].
S ince every m icrow ave system , and for that m atter every M M IC . co n tain s
tran sm issio n lines, controlling the relativ e atten u atio n by optical injection o f a so lid state p lasm a w ithin these pre-ex istin g transm ission lines ap p ears to be a co n v e n ie n t
technique.
T he optical pow er le v els that arc required to ach iev e useful a tten u a tio n
v alu es are belo w the threshold le v els for n o n -lin ear fiber o p tic effects [64]. (W e w ill
see in C h ap ter 6 that less than 150 m W o f laser pow er resu lts in 45 dB o f a tten u a tio n
a t 1.7 G H z.)
T herefore, a single laser source can be used to o p tically co n tro l the
15
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
a tten u a tio n
of
m ultiple
M M IC
transm ission
lines
usin g
standard
llbcr-optic
tech n iq u es. T herefore, the control signal "tim ing error" associated w ith this approach
is equal to the erro r in cleaving the individual fibers to the correct length, an d to
d ifferen ces in the therm al exp an sio n o f the individual fibers.
Since both o f these
errors in tim in g can be on th e order o f a few picoseconds (a 1-cm discrep an cy in fiber
length y ields a 33-ps tim ing error), tim ing errors o f less than a few pico seco n d s are
easy to achieve. N o w that the m otivation for developing an o p to electro n ic atten u ato r
has been presented, the tw o possible o ptoelectronic atten u ato r schem es m entioned
earlier w ill now b e discussed in further detail, starting w ith the reflective atten u ato r
schem e.
§2.2 Reflective O ptoelectronic Attenuator Scheme
T h e reflective attenuator schem e is based on the fundam ental electro m ag n etic
circu it principle o f partial w ave reflection from an im pedance discontinuity.
If a
m ic ro w a v e signal is propagating on a transm ission line o f characteristic im pedance Z0.
and en co u n ters a different characteristic im pedance Z. w here Z = Z0 ± AZ. and A Z is
the d ev iatio n in im pedance from Z0. the m icrow ave signal w ill be p artially reflected
by. and absorbed in. this transm ission line discontinuity.
The voltage reflection
co efficien t, f . m ay be used to represent this situation and is defined as
r =
Z - Z C)
--------- .
z +z0
(2.1)
T herefore, if Z = Z0. then f = 0 and there is no voltage (i.e.. signal) reflection.
W hen Z = 0 (i.e.. short circuited transm ission line case) then T = - 1 . im plying that the
16
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
voltage trav elin g w ave is reflected w ith a 180° phase sh ift (note that, for the cu rren t
traveling w ave, T = +1. and no phase shift is incurred). I f T is in betw een th ese tw o
extrem es, then a fraction o f the m icrow ave sig n al will also be absorbed du e to the real
(i.e..
resistive)
com ponent o f the
com plex
d iscontinuity
im pedance, Z. w here
Z = [{R + ja iL ) /(G + j ( i ) 0 ] /z . and R. L. G , and C are the tran sm issio n line d istrib u ted
resistance, in ductance, conductance, and capacitance, respectively.
T hus, from the
point o f view o f the transm ission line load, the m icrow ave signal am p litu d e has
decreased due to the partial reflection and. hence, an effective atten u atio n has been
realized.
F or clarity, we w ill refer to this attenuation schem e as a “ reflective
atten u a to r." even though a sm all fraction o f th e m icrow ave signal is also absorbed.
L ater w e w ill see that for the reflective atten u ato r schem e to w ork w ell, a m ajo rity o f
the signal is reflected, thus further ju stify in g o u r choice o f this term inology.
W e constru cted the reflective opto electro n ic atten u ato r by fabricating a
m icrow ave transm issio n line on a photoco n d u ctiv e (PC ) substrate.
PC m aterials
convert light (i.e.. photons) into electrical ch arg e (i.e.. clcctro n -h o le pairs).
W e refer
to this p h o to -in d u ced charge as an electro n -h o le. or solid-state, plasm a. T he induced
plasm a d ensity /Vc alters the m aterial’s co n d u ctiv ity c . since a is proportional to ,VC by
th e sim ple relatio n a - e\iNa. w here e is the electronic charge an d p is the sum o f the
electron and hole m obilities.
Thus, illu m in atio n o f the PC substrate w ith the
appropriate light source alters the substrate co n d u ctiv ity at the illum ination point,
w hich in turn alters the electrical resistance o f the p h o to co n d u cto r at this spatial
location.
S ince this is the substrate o f a planar m icrow ave transm ission line, this
resistance ch ange alters the local characteristic im pedance o f the tran sm issio n line at
th e point o f illum ination. T herefore a sh u n t resistance change is induced at the point
o f illum ination, and the laser beam pow er and profile can be used to accu rately control
th is localized im pedance change.
17
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Thus, all th a t is required to turn on th e reflective atten u a to r is to create the
p hoto-induced solid-state plasm a betw een the cen ter conductor and g round p lan e o f
the high-speed transm ission line. Tw o o f the m ost co m m o n ly used PC m aterials are
silicon (Si) and gallium -arscnidc (G aA s) (the d ev elo p m en t o f h ig h -sp eed PC sw itch es
using these m aterials w ill be outlined in C h ap ter 4).
B ulk PC sw itch es can be
fabricated w ith off-state (i.e., dark) resistances on the o rd er o f a few m egohm s, w ith
dynam ic on-state resistance values o f less than 1 Q [35], T hese d ev ices also have the
property that their turn-on tim e is equal to the rise tim e o f the optical trig g er - n am ely,
a laser pulse o f suitable pow er and w avelength.
R ecent w ork on h igh-speed laser
diodes indicates th a t these devices can be m o d u lated well beyond 10 G H z [65] and
could thus serve as the high-speed trigger m ech an ism for the reflective attenuator.
T herefore, it seem s plausible that a reflective o p to electro n ic atten u ato r can be su itab ly
constructed.
The basic concept for the reflective atten u a to r is illu strated in F igure 2.1. H ere
we have show n ideal transm ission line and lu m p ed circu it elem ents.
In the reflective
atten u a to r o ff-state, a m icrow ave signal pro p ag ates dow n the tran sm issio n line a n d is
absorbed by the m atched load. Z L. w here Z ^ = Z0. (Z0 is the tran sm issio n line
characteristic im pedance).
W hen the m icro w av e signal is to b e attenuated, a trig g er
signal is sent to the laser diode (L D ). T he LD . responding in tens o f p ico seco n d s to
the trig g er signal [39], activates the im bedded photoco n d u ctiv e (P C ) sw itch, w hich
then chan ges the shunt im pedance o f the tran sm issio n line at th e p oint o f LD
illum ination. S ince the value o f shunt on -state resistance. V?on. is p ro p o rtio n al to the
LD p eak output pow er and beam profile, the d egree o f m icrow ave atten u atio n can be
accurately controlled.
In the lim it that the sh u n t im pedance goes to zero, all the
m icrow ave pow er is reflected back to the source w ith no ab sorption ta k in g place in the
sh u n t im pedance. The sw itch conducts for a tim e equal to the laser pulse w idth, and
18
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
op en s in a tim e eq u al to the PC ca rrie r recom bination lifetim e. xr [55].
H ence, the
photo -in d u ced sh u n t im pedance w ill be m aintained fo r at least the d u ratio n o f the laser
pulse, and th erefore is fully controllable.
T he
reflective
attenuator
perform ance
can
be
predicted
using
standard
transm ission line th e o ry . The m ost straightforw ard ap p ro ach is to use tw o -p o rt theory,
and therefore A B C D m atrices, to m odel this circuit. A ssu m e th a t a v ariable resistance
is placed in shunt across an ideal loss-less tran sm issio n line, as show n in F igure 2.2.
W e can now calculate the resulting attenuation a t port 2 induced by this shunt
resistance from elem entary tran sm issio n line theory.
The ch aracteristic eq u atio n s for
this (and any general) tw o-port netw ork are
vi = --I'V: + B-U
(22)
i] = C v , + D-U
T herefore, once the A B C D m atrix elem en ts are k n o w n
for each o f the
com ponents in the tw o-port netw ork, the overall tran sfer function o f the n etw ork can
be determ in ed . T h e A B C D m atrix o f a shunt ad m ittan ce. Y. is easily calcu lated and
yields A = 1, B = 0 . C = Y. and D = 0. w here the sh u n t ad m ittan ce is the reciprocal o f
the sh u n t on-state resistance. Y = ! /R on. The co rresp o n d in g A B C D m atrix elem en ts o f
a loss-less tran sm issio n line o f length / arc A = cos0, B = j Z 0sinG. C = y T 0sinO.
and D = cosO. w here 0 = 2nI/X is the electrical length o f th e lo ss-less tran sm issio n line
and X is the w av elen g th o f the propagating m ode.
For th e reflectiv e optoelectronic attenuator m o d el w e assu m e a purely resistive
shunt resistance, denoted by Ron.
A ssum ing a 50-£2 tran sm issio n system , the
m icrow ave scattering param eters for the circu it sh o w n in F igure 2.2. w hose A B C D
m atrix elem ents hav e ju s t been described, can be co m p u ted as follow s [58].
19
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Frec-Spacc O ptics
Laser D iode
F igure 2.1 H ybrid reflective optoelectro n ic atten u ato r schem atic, d etailin g all
principal atten u ato r com ponents: Z 0 =
= 50 Q.
= 20-Z,og|o(vt/vj).
on
F igure 2.2 T w o -p o rt m odel o f a loss-less transm ission line w ith im b ed d ed shunt
resistance. Ron = 1/K. The tw o -p o rt boundary is denoted by the d ash ed box.
20
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T he voltage reflection coefficient, F, is defined as: T = vy'v,-. w here vr and v(are the reflected and incident voltages, respectively. S n is the scatterin g co efficien t
that describes the am o u n t o f voltage reflected from a given m icro w av e circu it elem en t,
and is defined as N| | = T.
T he insertion loss in a m icrow ave circuit is usually rep resen ted by the voltage
scattering coefficient elem ent A (, w hich is the ratio o f the vo ltag e at p o rt 2 to port 1
(hence the subscript "'21"). M ore precisely, Nt] = vt/vj, w here v t and Vj arc the tw oport through (i.e., o utput) and incident voltages, respectively. W e can represent .S’21 in
term s o f pow er, w ith units o f dB 's. and therefore defin e the atten u atio n in th e tw o-port
netw ork as
( 2 . 3)
N ow all that needs to be d o n e is to derive .S^i in term s o f the A B C D m atrix for
a shunt elem en t em bedded in a loss-less transm ission line; th en we can calcu late the
expected circu it attenuation for a g iven shunt resistance, Ron. T h is is a sim p le exercise,
and the result is .S^i = 2/(2+Z0 T) [58]: therefore. .SSi for the reflectiv e atten u ato r is
(2.4)
w here Y = l/R 0n. an d Ron is the PC sw itch on-state resistance.
21
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
60
CQ
40
■a.
20
10
Ron (H)'
Figure 2.3 R eflective optoelectronic attenuator theoretical p rediction for atten u atio n .
I .^21 I . versus shunt on-state resistance. Ron. due to reflection co efficien t f .
Z0 = 50 Q.
22
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
F or reference. IS^i I >s plotted versus Ron in Figure 2.3.
C learly. (?on =3 Q
should result in an atten u atio n o f =20 dB ; this "20-dB d esig n g o al” is indicated in the
figure by the dashed lines. W e therefore sec from this p red ictio n w hat is required for
the reflective atten u a to r to reach the 20-dB design goal: Ron ~ 3 Q . In C h ap ter 5 w e
will d iscuss a detailed series o f experim en ts using a high-speed G aA s PC sw itch to
d eterm ine if a LD can reduce the sh u n t resistance o f a G aA s high-speed PC sw itch
from =1 M Q to =3 Q . N ow that the b asis for the reflective o p to electro n ic atten u ato r
schem e has been adequately described, w e w ill discuss how the ab so rp tiv e atten u ato r
operates.
§2.3 Absorptive O ptoelectronic A ttenuator Scheme
A s m entioned in the introduction to this chapter, the atten u atio n m echanism for
the o p toelectronic atten u ato r m ay be du e to plasm a ab sorption in the m icrow ave PC
sw itch. T h e basic p rinciples behind this m ode o f attenuation have been stu d ied rather
extensively in the literature [55].[66].
In this section we w ill ex plore the physics
behind th is m ode o f o peration and w ill, as we did in F igure 2.3 for the reflective
attenuator, estim ate the perform ance on e can expect from the ab so rp tiv e o p to electro n ic
atten u ato r schem e.
If w e assum e that a TE M (tran sv erse electro m ag n etic) w ave is pro p ag atin g
through a dielectric m edium o f p erm ittivity, et,. w here b indicates that the p erm ittiv ity
is due to bound electrical charges (i.e.. electro n s and holes for a sem ico n d u cto r), the
introduction o f additional charge into the sem iconductor w ill alter the m a teria l's
dielectric properties: lie n e e , a correct expression for the effective perm ittivity, e. d ue
to the con trib u tio n o f b oth bound charg es, s^, and free ch arg es in the solid-state
23
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
plasm a. e p. m ust be derived.
A n ex cellen t treatm ent o f the classic p ro b lem o f
transverse electrom agnetic (T E M ) w ave pro p ag atio n thro u g h lossy m etals is p resented
by Rarno et cil [66] and. since the present situ atio n (i.e.. w ave p ro p ag atio n th ro u g h a
lossy dielectric) is analogous to this treatm en t, it will su ffice to c o v e r o n ly the
im portant features o f this theory in this chapter.
T h e present situation is show n picto rially in F igure 2.4.
W e see th at a T E M
w ave is in c id en t on a dielectric slab o f finite co n ductivity, a . w here the m o st general
case o f a com plex conductivity is considered: a = a' - jo " .
S ince the effectiv e
perm ittivity, e, is related to a by O h m 's L aw (cr = /cue), it too is co m p lex .
W e w ill
first d isc u ss TEM w ave p ropagation through a sim ple linear isotropic dielectric m ed ia
o f perm ittiv ity e.
A ssum ing that the T E M w ave is tim e-h arm o n ic, w e can im m ed iately w rite
dow n the follow ing relations for the TEM w ave:
E = E xe~j h x .
H = H re ' ,izy . and
(2.5)
j k = ( a + ;'P ) z.
w here
and /7V arc the electric and m agnetic field v ector am p litu d es, resp ectiv ely , k
is th e co m p lex propagation co n stan t (also a vector q u an tity , as d enoted by the bold
type face), and a and P are the propagation co nstant in radians per m eter and
atten u atio n coefficient in nepers p er m eter, respectively.
S u b stitu tio n o f k into the
harm onic field equations o f eq u atio n (2.5) y ield s, for ex am p le, E = E te x p { - a c —j P r j .
w here clearly the w ave experiences an atten u atio n due to the d ecaying ex p o n en tial
factor e x p { - a r } . and propagates harm on ically as exp { - / P r |.
24
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
F ree-S pacc O ptics
Lossy D ielectric
( e = e' -y e")
L aser D iode
----------------------
F igure 2.4 H ybrid "ab so rp tiv e" optoelectronic atten u ato r d ep ictin g T E M w ave loss in
a lossy d ielectric o f perm ittivity e. £ x and / / v are the com plex field am p litu d es, and
e = e’ - ye" is th e lossy d ie lec tric's effective p erm ittivity.
25
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
T his is the case for m ost sim ple lossy d ielectric m edia. W e w ill see in a lew
m o m en ts th a t for lossy dielectrics, su ch as for a so lid -state p lasm a in a typical
sem ico n d u cto r, som e interesting effects occur.
T o m odel the so lid -state plasm a, w e
start by d escrib in g the m otion o f an electro n in the sem ico n d u cto r m aterial. T he w ellknow n eq uation o f m otion for the electron is as follow s:
3v
i)i* — =
d t
-cE - ii f v v ,
(2.6)
w here m * is the effectiv e electron m ass in the sem ico n d u cto r, w hich is 1.08«/o fo r
silicon [67]. w here m Q is the electron rest m ass (9.1 x l t H 1 kg), v is the electron d rift
v elocity du e to applied electric field E . c is the electro n charge (1 .6 x 10~19 C). and v
is the electro n collision frequency in the plasm a (essen tially , th e electro n m otion is
harm onic w ith E except for the dam ping term m * vv).
U sing equations (2.5) and (2.6), plus M axw ell's E quations, w e w ill derive the
co n trib u tio n o f the solid -state plasm a to th e overall se m ico n d u cto r perm ittivity. From
eq u a tio n (2.6). the electron velocity v can be d eterm in ed if w e assu m e that the velocity
is tim e harm onic (i.e.. ex p f-y c)/] dependence):
m * { v + y o i)
S ince the cu rren t flow associated w ith this electro n m otion is prim arily due to
co n d u ctio n cu rren t, then J = - c N c\ . w here a new term . .Vc. h as been introduced to
describ e the solid-state plasm a electron density, w here the u n its o f ;VC are cm - 3 .
S u b stitu tio n o f eq uation (2.7) into J yields
26
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
w here w e sec that the conductivity, a. is indeed com plex.
S u b stitu tio n o f eq uation
(2.8) into M a x w e ll's equations gives
(2.9)
From equation (2.9). we can now derive the effective p erm ittivity o f a sem ico n d u cto r
con tain in g a solid -state plasm a.
For a tim e-h arm o n ic T E M w ave, m an ip u latio n o f
(2.9) directly results in the w ell-know n com plex perm ittiv ity o f a lossy dielectric
m edia:
co n
( 2 . 10)
w here
s = e1 - j s "
and
a
new
introduced; w p2 = Ncc 2/(m *z0).
term ,
cop,
perm eability o f free space, and c =
the
propagation
plasm a
frequency,
has
been
The general form o f the co m p lex p ropagation
con stan t is j k = a + /(5. w here k2 = w -p (e '
T herefore
the
constant,
- /e " ).
( p 0E0 )- '-
w hich
p ^ p „= 4tt
x
10- 7 H /m is the
is the speed o f light in vacuum .
d escrib es
the
p ropagation
of
an
electrom agnetic w ave through a solid-state plasm a, is
27
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T here are m any types o f plasm as [68]: th erefo re, the lim its o f eq u atio n (2.11)
are o f interest to us since this w ill sim plify the analysis co n sid erab ly . For a so lid -state
plasm a, w e can assu m e that the collision frequency v = v,|,//c. w here r,|, * I 0 7 cm /s is
the therm al carrier velocity and /c = 100 A is the th e m ia l m ean-free scatterin g length,
respectively, E(, = e 0. w hich is the perm ittivity o f free space. Since j k = a + _/[}. a and
(5 can now be w ritte n using eq u atio n (2.11)
(2 . 12)
w here a and [J are the electrom agnetic w ave atten u atio n and pro p ag atio n coefficien ts
w ithin the solid state plasm a, respectively,
a rep resen ts the total losses experien ced
by the p ro p ag atin g electrom agnetic w ave [66] and. for the case o f the so lid -state
plasm a, w e w ill assum e that the dom inant loss m ech an ism is plasm a abso rp tio n .
W ith the use o f eq uation (2.10) to d escrib e the so lid -state plasm a p erm ittiv ity
and equation (2.12) to describe the w ave propagation, the p ow er lo ss ex p erien ced by
the w ave w ithin the solid-state plasm a can be estim ated .
U sing the P oynting vector.
S = E x H to com pute the actual pow er loss th ro u g h the p lasm a,
w e find that
S = Ex2/2 r|re x p { - 2 a z } . w here q r = ( |V e r) '- is the ch aracteristic im p ed an ce o f the
m edium .
T hus, for a p ropagation length o f c = 1 cm . the w ave is attenuated by an
am o u n t equal to the exponential factor e x p { - 2 a ] . E xpressing the p ow er loss in term s
28
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
o f th e m icrow ave sc atterin g param eters. Sy, w here i and j refer to the m easu rem en t
port for a 2-port netw ork as described in the last section, the atten u a tio n from p o rt 1 to
port 2. p er cm . is g iven by
= 10-Z.og 10 exp to.
S2i ( d B / c m )
We can
no w
plot the anticip ated
(2.13)
attenuation
o f the m ic ro w a v e signal
propagating th ro u g h the absorptive o p to electro n ic atten u a to r as a function o f the
p lasm a density. N c (and hence as a function o f the p lasm a frequency. a)p): this result
is sh o w n in F igure 2.5 for the d esign frequency. / = 3 G H z (co = 2nf).
attenuation
values
of
20 dB
are
predicted
for
p lasm a
d en sitie s
N ote that
on
the
o rd er o f 9 x 1 0 16 cm - -.
P lasm a d en sities o f this m ag n itu d e arc readily ach iev ab le using optical
illum ination w ith laser diodes [55], and a sim ple an aly sis can be em p lo y ed to predict
the required laser peak p ow er to achiev e the attenuation values sh o w n in F ig u re 2.5.
T h is analysis is straightforw ard and is basically a m atter o f co u n tin g the n u m b e r o f
p hotons absorbed by a one centim eter-th ick sem ico n d u cto r slab an d estim a tin g the
p ercentage con v erted into elcctro n -h o le pairs:
p _
M.D -
1-T pr(
(2.14)
0 (1 - R ) - X r
w h ere P L[) and /l Sp0t are the required laser diode peak p o w er and sp o t size, R is the
reflectivity o f the laser beam at the sem ico n d u cto r surface, q is the internal qu an tu m
efficien cy o f the m aterial [69], w hich is an estim ate o f the co n v ersio n efficien cy o f
29
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
absorbed photons into electron-hole pairs, hv is the photon en erg y , and r r is the PC
carrier recom bination lifetim e o f the photo-induced carriers.
F or silicon. R is approxim ately 0.3.
A ssum ing /lspot = 10 p m x 100 pm
and r) = 0.1. w hich is reasonable for an indirect band -g ap m aterial [70]; the laser
pow er required to induce the attenuatio n values o f Figure 2.5 w as co m p u ted and is
show n in Figure 2.6.
T he analysis described in this section predicts atten u atio n v alues o f = 20 dB for
plasm a d ensities that are on the o rd er o f 9 x 1016 cm - 3 , and it has been sh o w n that the
predicted laser diode pow er requirem en t o f 750 mVV to ach iev e these v alues is
reasonable.
It should be noted that for a plasm a d en sity o f 1 0 15 c m -3 , the
corresponding plasm a frequency is 1.8 x [ 0 13 rads/s ( o r / p = 2.86 TH z):
T he design
frequency is 3 G H z; thus the necessary condition for plasm a ab so rp tio n (cap > m) is
met [66],
30
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
0
-40
S
(dB/cm)
-20
-60
-80
1E+15
1E+17
1E+16
1E+I8
A'e (cm '3 )
F igure 2.5 T heoretical prediction for attenuation. .S’t j . versus solid-state plasm a
density. ;VC. due to plasm a absorption. / = 3 GH z.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
-10
spot
-20
^
-30
23
T3
-40
E
= 10 fjni x 100 pm
-50
-60
-70
-80
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
F igure 2.6 T heoretical prediction for atten u atio n . 5 2 |. versus laser d iode pow er. P l q
for plasm a den sity show n in Fig. 2.5.
32
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§2.4 Sum m ary
In this chapter, w e have presented the basis for tw o hybrid o p to electro n ic
a tte n u a to r schem es. T w o optically induced ab so rp tio n m ech an ism s, a reflectio n from
an im pedance disco n tin u ity and an ab so rp tio n in a so lid -state plasm a, h av e been
p rese n ted , as w ell as the projected perfo rm an ce from each schem e.
It w as o bserved
that the reflective atten u a to r predicts 20 d B o f atten u atio n for an o p tically induced
sh u n t resistance in the m icrow ave tran sm issio n line o f 2.5 £X
T he ab so rp tiv e
a tte n u a to r p rediction is th at 20 dB /cm o f atten u atio n can be ach iev ed fo r plasm a
d en sitie s o f ap p ro x im ately 1 0 ' 7 cm - 3 , a n d that the an ticip ated iascr d io d e pow er
req u irem en t for such a plasm a density is ap p ro x im ately 750 m W .
W e have co n trasted optically con tro lled m icro w av e d ev ices w ith all-electrical
versio n s, and have pointed out the situ atio n s w here optical co n tro l is p referab le.
In
o rd e r for the opto electro n ic attenuator to be realized, laser diode so u rces m u st n ot only
be available to provide the necessary' o u tp u t pow er, b ut m ust be cap ab le o f hig h -sp eed
m o d u la tio n so that the optoelectronic atten u ato r can provide hig h -sp eed m icro w av e
control.
Since the laser diode is therefore a critical elem en t to both o p to electro n ic
a tte n u a to r schem es, considerable research w a s perform ed to en su re that a su itab le laser
w'as available for th is application: therefore, the laser diode research and d ev elo p m en t
co n d u c te d as part o f this w ork is presented in the next chapter.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Chapter 3
Laser Diode Development
Im p lem entation o f the optically con tro lled
atten u ato r requires
th a t two
principal sem ico n d u cto r d evices be developed. O ne is the h ig h -sp eed p h o to co n d u ctiv e
(P C ) sw itc h , w hich is the attenuating elem ent. T h e o th er is the se m ico n d u cto r laser
dio d e w h ich is the control device for the attenuator.
In the ideal situation, a single
laser d iode (L D ) w ould em it sufficient o ptical pow er to create an clectro n -h o le p lasm a
w ith in the high-speed PC sw itch to atten u ate pro p ag atin g m icro w av e sig n als.
In
ad d itio n , the laser o u tp u t w ould have sufficient o ptical p o w er to ac h iev e the
atten u a tio n values that m eet the m icrow ave sy stem desig n specificatio n s.
L astly, the
LD w ould be easy to turn o n and require a low level o f input electrical p o w er to
op erate.
M uch
w ork
has
been
done
to
d evelop
high
speed
LDs
for
the
te lec o m m u n icatio n s industry [71]. Som e o f these devices co n sist o f two LD co n tact
reg io n s, the first p roviding gain, w hile the second is reverse b iased and m odulated
w ith the inform ation signal [40],[72]. To d ate, g ig ah ertz data rates have been ach iev ed
w ith th is approach [41], indicating that sub -n an o seco n d LD turn-on tim es are feasible.
H ow ever, th ese are typically low -pow er devices (= 100 m W ). w hich are m ade by the
fo rm ation o f an optical ridge w aveguide on the LD surface: th is w av eg u id e d ecreases
the m o d u la to r section capacitance (by red u cin g th e co n tact area), th u s ach iev in g high­
speed p erform ance (the ridge m ainly serves as a lig h t-co n fin in g region to ensure
sin g le-m o d e laser operation).
C onversely, very high pow er LD s have been dev elo p ed (in our lab at the
U n iv ersity o f M aryland, and by others), that use a "broad-area" LD geo m etry ,
fo r
34
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
exam ple, w e have m easured peak optical pow ers in ex cess o f 1 W at X = 840 nm . In
addition, w e have ^ -sw itc h e d tw o-section b ro ad -area LD s w ith an estim ated peak
p ow er o f greater than 6 W w ith a pulse w idth o f <30 ps [38].
W e w ill discu ss in detail (C hapter 4) the relationship b etw een the PC sw itch
m aterial and the corresponding LD em issio n w avelength. The band-gap en erg y is 1.1
eV for Si an d 1.41 eV for G aA s, corresp o n d in g to ch aracteristic w av elen g th s o f 1.13
and 0.88 p m . respectively. T hus, G aA s/A lG aA s LD s w ith X as low as 780 nm should
allow intrinsic photon absorption to o ccu r at the surface o f a Si PC sw itch.
For the
G aA s PC sw itch, the LD w avelength m ust be tailored to m eet the intrinsic ab so rp tio n
condition; fortunately this can be acco m p lish ed w ith standard LD qu an tu m -w ell
desig n techniques.
In sum m ary, the critical LD param eters for p ro p er o p eratio n o f the m icrow ave
attenuator a re the follow ing: (1) X m ust be m atched for both Si and G aA s attenuators.
(2) the o p tical intensity m ust be sufficien t to achieve useful plasm a d en sities,
and (3) turn -o n tim es m ust be at sub-nanosecond speeds so that high m o d u latio n rates
can be achieved.
T hese requirem ents are quite stringent and m ay not be fully
realizable w ith traditional LD triggering schem es.
A s a co n sequence, several LD
schem es w ere studied and are presented in this chapter. Before d iscu ssin g the various
LD types, a b rie f description o f the physics o f both m u lti-sectio n and qu an tu m -w ell
lasers is presented.
T hen several LD im p lem en tatio n schem es will be discussed.
F inally, research that w as undertaken to im prove the overall efficiency o f h ig h -p o w er
broad-area LD s w ill be presented: a 37-p ercen t increase in laser diode efficien cy has
been dem onstrated.
35
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§3.1 M ulti-Section Q uantum -W ell Laser Diodes
D uring this research project, several types o f ed ge-coupled, F abry-P erot.
quantum -w ell laser d io d e s w ere developed for th e o ptoelectronic atten u a to r b ecause o f
their relatively high quantum efficiency [39] a n d ease o f fabrication.
Since excellen t
treatm ents o f the various types o f LD s are available in th e literature [74].[75], only the
basic structure o f quantum -w ell lasers will be d iscu ssed here.
T he basic LD structure co nsists o f a p -n ju n c tio n , a F abry Perot cavity th at is
form ed by cleaving o f the sem iconductor cry sta l, and ohm ic contacts w h ich arc
necessary for electrical connection.
LDs are current-controlled d ev ices that m ay be
operated in either a C W o r pulsed m ode, w h ere therm al d issipation d ictates the
possible m ode o f operation. Sandw iched betw een th e p -n ju n c tio n is the active layer,
w here the recom bination o f electrons and holes results in th e generation o f photons.
The ac tiv e region m ay be com posed o f eith er a single q u an tu m well (S Q W ) or
m ultiple quantum w ell (M Q W ) structure. The m aterial su rrounding the active region
serves as an optical w aveguide, and this w aveguide can co n sist o f cither a graded
index region o r an abrupt index region, know n respectively as G R IN S C H (gradedindcx separate confined heterostructure) or S T IN S C H (stcp -in d ex separate confined
heterostructurc) optical structures.
Figure 3.1 is a schem atic diagram of th e gain m edium o f a G R IN SC H S Q W
laser structure used to fabricate LDs during th is research.
T ab le 3.1 lists the actual
m aterial profile o f this G RIN SCH LD structure, and indicates for each lay er the
corresponding thickness, m olar fraction o f A1 present, and p u rpose.
In this case, the
active region consists o f a 100-A -w ide SQW . an d th e optical intensity is guided by the
A lG aA s cladding layers, as show n.
An M Q W laser ty p ically d iffers from the SQ W
laser sh o w n in Figure 3.1 in that the SQW activ e region is replaced with a M Q W .
36
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
O therw ise, a G R IN S C H M Q W laser is identical to the S Q W structure sh o w n in Figure
3.1.
T he basic physics o f both S Q W and M Q W structures is well trea ted in the
literature [74],[76], w ith an excellent com p ariso n b etw een the two ty p es g iv en in
F igure 3.2 [76].
B asically, the difference b etw een the tw o types o f q u an tu m -w ell
lasers can be sum m arized w ith the follow ing three ap p ro x im ate relatio n s [76]:
£m qw
=
(3.1)
N'Ss q w
(3.2)
Ah. M O W
=
'
'Ah. S Q W
•
(3.3)
w here gsQW anc^ £ m q w are ^ie S Q W and M Q W m odal gain, resp ectiv ely . f { is the
relaxation oscillatio n frequency, a is the differential gain s dg/dn (w h ere n is the
carrier density in the laser active region), P Q is the photon d en sity , and t p is th e photon
lifetim e. >Ah,MQW and -Ah.SQW ^
the threshold current d en sitie s for th e M Q W and
SQ W lasers, respectiv ely , and N is the nu m b er o f w ells.
N ote th a t eq u atio n (3.1) show s that M Q W lasers can operate w ith gain factors
that are appro x im ately N tim es that o f a sim ilar S Q W d ev ice , w hich is ev id en t in
Figure 3.2; for N = 5 and a current density J o f 500-A /cm 2, we see that th e m odal gain
o f this M Q W is approxim ately 68 percent greater than for th e SQ W ( N = 1) case.
Equation (3.2) indicates that j r is m axim ized w hen the d ifferen tial gain a is m ax im u m ;
this is im p o rtan t for applications requiring short pulse w idths an d is an in stan ce o f an
M Q W structure being preferable to an SQ W structure. The penalty one p ay s for using
an M Q W LD is show n in equation (3.3) and ag ain ev id en t in Figure 3.2; th e M Q W (N
- 5) thresh o ld cu rrent d ensity is five (5) tim es g reater than th at for a co m p arab le S Q W
37
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission .
( N = 1) device.
T hus, to realize th e 68-percent m odal g ain increase o f a five-w ell
M Q W LD, one m u st tolerate a fivefold increase in the th resh o ld cu rren t d en sity o v e r a
S Q W LD.
T hus, the basic difference betw een SQ W and M Q W L D s is sim p le; for h ig h pow er and h igh-speed applications, M Q W d ev ices are p referred , and the pen alty one
pays for such a choice is a higher thresh o ld current. D uring this research p ro ject w e
used both ty p es o f gain m edia to en su re that the optim um LD w o u ld be av ailab le for
the o p toelectronic atten u a to r experim ents (discu ssed in C h a p te r 6 ).
In ad d itio n to the choice o f gain m edia, there is a ch o ice o f LD co n tac t
geom etries for discrete devices: ridge w aveguide and b ro ad -area ty p es are co m m o n ly
used (sh o w n in F igure 3.3).
T he essential differen ce b etw een the tw o d ev ice
geom etries is as follow s. In the ridge w aveguide laser, a m e sa is form ed on the p -sid e
o f the w afer by etching a ridge, typically o f 0.7 p m depth and 5 to 10 p m w idth. T his
ridge provides an in dex-guiding structure in the lateral d irec tio n ; w ith p ro p er d esig n ,
only a single lateral m ode can be supported in su ch a structure. The rid g e also reduces
the
electrical
capacitance
of
the
contact
to
perm it
h ig h -sp eed
m odulation.
U nfortunately, since the LD gain volum e is also reduced, th is lim its the am o u n t o f
laser p ow er from these devices. O n the o th er hand, b ecause b ro ad -area lasers have a
w ide stripe geom etry (typically >50 pm ), they can em it h ig h e r pow ers b ecause o f th eir
relatively large gain volum es. H ow ever, m u ltiple lateral m o d e s resu lt from the broad
stripe con tact geom etry, so that these lasers are o f little in terest to th e co m m u n icatio n s
industry (besides the highly astigm atic o utput beam , b ro ad -area lasers lack m odal
purity for w avelen g th d iv isio n m ultiplex in g app licatio n s).
Thus, ridge w av eg u id e
lasers are m o st useful for applications such as fiber optic co m m u n ic atio n s, but i f high
optical p ow er levels are required, b road-area lasers are m ost useful.
38
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
P+ G aA s
P A1 G a , As
X
I -X
P Al G a , A s
x
I -x
AI G a
S Q W a c t i v e la y e r
X
As
1-X
N A1 G a, As
x
l-x
3,Ny'GaAs.buffer^layer..
N + G aA s substrate
Al G a
x
N -
As
G r a d e d r e g i o n ( p - t y p c a b o v e S Q W , n - tv p e b e l o w ) ,
l- x
Al G a
x
As
T r a n s i tio n r e g i o n ,
l- x
Figure 3.1 G R IN S C H LD device m aterial structure. A ctive area is S Q W o f
w idth 80 to 100
A.
Al concentration (x) and layer th ick n esses arc show n in T able 3.
39
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Tabic 3.1 MBE 330 SQ W GRINSCH Laser Structure
L ayer
T hickness
C om position
G raded
(A)
A lxG a |. xA s
xl - > \2
C ontact
2.000
0.0
—
Be
-5 0 0
G rade
1.500
0 -> 0.7
yes
Be
50
C lad-p
10.000
0.7
—
Be
10
G rin-p
1.500
0.7 - > 0 .3
yes
—
—
80
0.3
—
—
—
8 0 -1 0 0
0.0
—
—
—
80
0.3
—
—
—
G rin-n
1.500
0.3 -> 0.7
yes
—
—
C lad-n
8.000
0.7
—
Si
5
T ransition
1.500
0 -> 0.7
yes
Si
5
B uffer
2.000
0.0
—
Si
30
>100 pm
0.0
—
Si
30
Step
Q -W ell
S tep
S ubstrate
D o p in g 1
C oncentration
x 1017 cm -3
1 n-type d opant = Si. p-type dopant = Be.
40
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
100
N =5
£
u
a
0
100
200
300
400
500
J (A/cm2)
F igure 3.2 LD perform ance as a function o f q uantum w ell num ber. N. from A rak aw a
and Y ariv [76]: M odal gain. a . versus current d e n s ity ../. for a quantum w ell o f w idth
L y = 100
A.
41
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Since o u r application requires high p o w er levels and d o cs n o t dep en d o n m odal
p urity for its operation, w e chose to in v estig ate broad-area lasers.
S ingle discrete
b road-area L D s w ere fabricated in our lab w ith pu lsed output p o w ers in ex cess o f 1 W
(fo r drive currents o f 10 to 12 A ). The d ev ice co n tac t g eo m etry can be se g m en ted in
su ch a w ay as to o ffse t the higher cap acitan ce associated w ith the broad contact
geom etry. In this case, the broad area co n tac t is d ivided into sm a lle r sectio n s that arc
electrically isolated. For illustration, a tw o -sectio n device is sh o w n in F igure 3.4. In
th is case, the gain section is electrically d riven in th e sam e w ay as a sin g le-sectio n
device. H ow ever, if the second section is rev erse biased w ith a dc voltage, then this
section acts as a saturable absorber, w hich can be used to rap id ly m odulate th e laser
d io d e output. If a m odulation signal is ap p lied to the ab so rb er section, it b leach es this
section and leads to a m odulated output (w e w ill sh o w in S ectio n 3.3 that w e can use
this approach to O -sw itch these devices).
It is th eo retically possible to achieve
gig ah ertz m odulation rates w ith this tw o-sectio n LD schem e [40],[77],
U sing a num erical sim ulation co d e d ev elo p ed by T h e d re z and Y an g in our
laboratory [38],[39],[78], we can predict the theoretical L U cu rv e for this type o f
device, as show n in F igure 3.5. For this sim u latio n , th e stripe w id th w as 300 p m . w ith
m o d u lato r and overall device lengths o f 200 p m and 1 m m , resp ectiv ely .
The
m o d u lato r bias is 0, - 4 . and - 8 V . as show n. N otice that in creasin g the se co n d section
b ia s sim ply shifts the L U curve to hig h er cu rren t values (i.e.. the thresh o ld c u rre n t is
increased). T h is is caused by the increased in tracav ity loss in tro d u ced by th e reverse
bias on the second section.
In the next section w e w ill sh o w th a t this is an accurate
prediction o f the m easured perform ance o f a tw o -sectio n dev ice o f th e sam e
d im ensions.
42
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
C r/A u p+ c o n ta c t ( - 0 .2 5 um th ick )
P+ G a /\ s
--------
I’ A l G a , A s
P Al G a , As
I Al G a
S Q W a c tiv e la y er
As
??tN^OW.bufFtalayCTftlSBS
N + G aA s s u b strate
A u -G c /A u n+ co n tact ( - 0 .3 u m th ick )
F igure 3.3 (a) B road-area LD dev ice fabricated using the m aterial stru ctu re o f Fig.
3.1. B road-area strip e w idth. IV. is as indicated.
C r/A u p + c o n ta c t ( - 0 .2 5 um th ick )
P AI G a , As
S Q W A c tiv e L a y e r
I Al G a
-------
As
N + G aA s s u b strate
A u -G c /A u n + co n tact { - 0.3 u m th ick )
F igure 3.3 (b) R idge-w aveguide LD device structure usin g m aterial in Fig. 3.1.
Ridge w id th . IV. is as indicated (ty p ically 5 to 10 pm ).
N ote that the rid g e has been etch ed - 0 .7 p m into the clad d in g layer.
43
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
P +co n tact(C r-A u )
G ain section
(section I)
C la d d in g region
M o d u lator section
(section 2)
P h o to c o n d u ctiv e
sw itching gap
(10 p m )
Active region
C o n tac t (Au-Gc/Au)
F igure 3.4 Tvvo-scction LD geom etry.
-4 V
0.4
-8 V
0
i
3
4
5
6
C u rren t (A)
Figure 3.5 T heoretical tw o -sectio n LD LU curve.
44
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§3.2 Laser Diode Design, Fabrication, and Perform ance
In su p p o rt o f the optically controlled atten u ato r project, m any LD s were
fabricated an d tested at the U niversity o f M aryland.
T he m aterial w as g ro w n at the
L aboratory for P hysical Sciences (L P S ) u n d er the Jo in t P rogram for A dvanced
E lectronic M aterials.
B oth SQ W and M Q W lasers, o f the broad-area and ridge
w aveguide geom etry types, w ere fabricated. In addition, one-, tw o- and three-section
devices w ere investigated, w ith the tw o section broad-area LD used as the prim ary
research tool for the photoconductive sw itching ex p erim en ts d iscussed
in later
chapters. F igure 3.6 is a photograph show ing a cleaved b ro ad -area LD bar. T he stripe
is 150 (am w ide and the bar w as cleaved to achieve a cav ity length o f 450 (am. Figure
3.7 is a photograph o f a cleaved bar o f laser devices sh o w in g two ridge w aveguide
d evices: the leftm ost device is a three-section device, an d the other is a one-section
device.
A typical optical pow er versus drive current plot ("'U I C urve") is sh o w n in
F igure 3.8 for a one-section SQ W broad-area LD. T he dev ice w as cleaved to form a
1-m m -long F abry-P erot cavity w ith a contact w idth o f 500 jam. Figure 3 .9 show s the
spectral o u tp u t o f this device: its spectral peak is at 879.5 nm with an FW HM
linew idth o f less than 1 nm.
45
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
F igure 3.6 C leav ed bar show ing four 150 p m -strip e o n e-sectio n b ro ad -area LD s.
Figure 3.7 C leaved bar show in g tw o rid g e-w av eg u id e LDs:
a three-sectio n device (left) and a o ne-section d ev ice (right).
46
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
1.6
0.8
0.4
0
8
6
4
0
10
C u rren t (A)
F igure 3.8 O ne-section 500 -p m b ro ad -area pulsed L / I Curve.
LD op eratio n is in the p u lsed mode.
5
4
J
0
877.5
878
878.5
879
879.5
880
880.5
881
/, (nm)
F igure 3.9 C orresp o n d in g spectral o u tp u t o f F ig u re 3.8.
T he LD is biased ju st ab o v e threshold.
47
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited without perm ission.
To ach iev e h igh-speed atten u ato r trigger rates, the m o d u latio n c a p a b ilitie s o f a
tw o-section device w ere assessed.
F igure 3.10 sh o w s the U I cu rv e o f a tw o -sectio n
device as a function o f th e applied reverse bias on the ab s o rb e r section.
The stripe
w idth is 300 p m w ith a 2 0 0 -g m abso rb er section length. The overall d e v ic e length is
1 mm.
B ecause o f the quantum -confined Stark effect [79], an in crease o f the
m odulato r section voltage results in an increase in the device threshold cu rren t w hile
the slope o f the LU curve is unaffected [80]. T he am ount o f o p tic al pow er that can be
m odulated (assum ing a 100-percent m odulation d ep th ) is e v id e n t from th e figure: for a
current o f 3 A . the m odulated o p tical output fo r V2 = - 8 V is ap p ro x im ately 200 m\V.
To investigate the turn-on speed o f th e tw o-section L D . we p erfo rm ed a sim ple
tim ing experim ent. The current pulse used to bias the gain se ctio n w as sy nchronized
w ith a sh o rter pulse, w h ich can be used to forw ard-bias the o th erw ise reverse-biased
second section. W e then m easured the tim e d elay betw een th e pulse used to forw ardbias the second section and the LD o p tical o u tp u t pulse.
W ith th is sch em e w e
m easured a tim e delay o f less than 500 ps, w hich indicates th a t tw o-section broad-area
L D s can have sub-nanosecond turn-on tim es.
The near field im age o f a lSO -pm b ro ad -area laser o u tp u t is s h o w n in Figure
3.11.
N ote that there a re m any lateral spatial m odes e v id e n t in this im age, as is
expected
from a b road-area laser o f this size.
Figure 3 .1 2 is a p h o to g rap h o f a
3 0 0 -p m -strip c LD that w as fabricated and packaged in o u r laboratory. T h e two SM A
co nnectors are electrically connected to cith er the gain or m o d u la to r se ctio n s v ia goldcoated alu m in a contact pads and gold wire bonds.
The L D is m o u n ted by indium
so ld er to a A u-plated O F H C C u h eat sink for im proved th e rm al co n d u c tiv ity (fro n t o f
m ount show n in figure).
The laser m ount is then placed o n a th e rm o -electric (TE)
co o ler to provide tem perature control for both w avelength tu n in g and to extend the
device lifetim e.
48
R e p ro du ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
1.2
1
-4 V
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
C urrent (A )
F igure 3.10
U I curve o f tw o-section 3 0 0 -p m -strip e LD versus lA.
D evice length = 1 mm.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
Figure 3.11 Im age o f 150-(im -w ide bro ad -area LD output o n front facet sh o w in g
fairly uniform lateral optical em ission w ith som e lateral m ode structure.
Figure 3.12 Photograph o f m ounted 500-(im b ro ad -area LD.
50
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§3.3 O ptically Q-Switchcd Tw o-Section Laser Diode Experim ent
In S ection 3.2 it w as noted th a t a tw o-section device can p erm it rapid LD tu rn ­
on tim es provided that a fast electrical signal is used to m o d u late the ab so rb er section.
T he fundam ental lim it o f this schem e is basically related to the d ifficu lty in g enerating
a fast (i.e.. picosecond) electrical p ulse to m o d u late the ab so rb er sectio n [81 ]—[83].
U sing a high-pow er, m o de-locked N d:glass laser sy stem [84], w e assessed the
fundam ental d y n am ics
o f tw o-section
b ro ad -area
LDs
[38].[39],
U sing
this
experim en tal technique, we generated O -sw itched pulses w ith g reater than 6 W o f
p eak p o w er and pulse w idths less th an 30 ps from m o n o lith ic tw o -sectio n lasers. For
reference, w ith conventional gain-sw itching tech n iq u es these sam e d ev ices produced
less th an 1 W o f peak pow er in tens o f nanoseconds.
T w o broad-area LDs. devices A and B. w ere investigated.
W e used a lift-o ff
tech n iq u e to p attern the tw o electro d es and w et etched the cap lay er to achieve
electrical isolation (see F igure 3.3). The etched g ap w as 10 p m for b o th devices. The
stripe w id th for device A w as 100 p m . w ith a total length o f 775 p m . T h e length o f
the m odulator section was 175 p m . The active region w as co m p o sed o f a single
100-A G aA s quan tu m w ell, separated from a 1000-nm A l04Ga()6A s clad d in g layer by
a 150-nm graded region. T he stripe w idth for d ev ice B w as 300 p m . w ith a 1-mm total
length and a 2 0 0 -p m m odulator section. An M Q W active region w as used, con sistin g
o f five 60-A G aA s quantum w ells separated by 80-A A ln 2G a08A s barriers, w ith a
grad ed region o f 200 nm and a cladding layer o f 1000 nm .
T he second section o f the LD was rev erse-b iased and used as an optical
m o d u la to r [72], w hile the first section provided gain.
T he gain sectio n w as pulsed
w ith 100-ns-w iae electrical pulses at a repetition rate o f 500 Hz.
W e m odulated the
51
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
loss in the second (i.e., m odulator) sectio n by using the g ap b etw een the electro d es as
a p h o to co n d u ctiv e sw itch [38],[85],
F igure 3.13 show s a schem atic d iag ram o f th e o p tic ally ^ -s w itc h e d LD
experim ent.
Illum inating the gap w ith 527-nm lig h t creates p h o to carricrs that
electrically short the tw o electrodes to g eth er.
Since the gain sectio n h as a low
im pedance (typically a few ohm s) in co m p ariso n w ith the rev erse-b iased m o d u lato r
section, the creation o f this near short circu it results in a d rastic electrical ch an g e
o ccu rrin g m ainly in the m odulator sectio n , w hich ch an g es from a h ig h to low
a b so rp tio n state in a tim e com parable to th e illum inating p u lse w idth.
T he light source for the sw itchin g pulse w as a m o d e-lo ck ed N d :g lass laser
co u p led to a regenerative am plifier [84].
Pulses 10 p s w id e w ere generated at a
repetition rate o f 500 H z and could be co m p ressed to 1 p s w ith a p air o f gratin g s. The
1.0 5 4 -g m beam w as doubled into the g reen and focused onto the sw itch in g gap. An
average p o w er o f 32 jiW w as available fo r sw itch in g w ith th e 10-ps g reen p ulse, w ith
25 p W availab le from the 1 ps green pulse.
W e m easured the ^ -s w itc h e d p u lse w idth by m ix in g the d io d e an d N d:glass
beam s in a L iI 0 3 crystal. W hen the 1-ps sw itch in g pulse is used, the cro ss-co rrelatio n
o f th e tw o beam s allow s one to record th e actual shape o f the d iode laser p u lse d irectly
[8 6 ], w hile a d econvolution step m ust be perform ed w hen the 10-ps pulse is used.
R esults from the cross-correlatio n m easu rem en ts are sh o w n in F igure 3.14 (a)
for device A , and F igure 3.15 (a) for dev ice B. T he m o d u lato r sectio n s w ere reverse
biased to the poin t w here the LD o u tp u t w as q u enched.
F or each valu e o f ap p lied
voltage, the d iode gain section w as b iased to ju st below th resh o ld , since this
arran g em en t proved to yield the best results. A s can be seen from these figures, tw o
d ifferent (7-sw itching behaviors w ere ob serv ed . Pulses from the ty p e B laser sh o w a
train o f pulses separated by the cavity ro u n d -trip tim e. xn . A large asy m m etry (a fast-
52
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
risin g edge w ith a relatively slow decay tim e) w as o bserved in all cases.
Peak 0 -
sw itched po w er w ith the 10-ps sw itching pulse, w hich was ded u ced from av erag e
p o w er m easurem ents, w as as high as 6 W.
T he type A laser did not exhibit a distinct p u lse train. S in g le-lo b e pulses w ith
F W H M values as low as 21 ps w ere m easured. T h is is to our kn o w led g e the sh o rtest
p ulse w idth generated to date by a broad-area laser.
T aking into acco u n t the pulse
w idth o f the 1.054-pm pulse in the interpretation o f the cro ss-co rrelatio n traces, w e
found that no noticeable reduction in pulse w idth w as recorded d u rin g 0 -sw itch in g
w ith either the 10-ps o r 1-p s green pulse (this was also the case w ith the type B laser).
P eak pow ers deduced from average po w er m easu rem en ts w ere also as high as 6 W for
m easured pulse w idths betw een 20 and 30 ps (for a d ev ice bias o f 380 m A ). T o o u r
know ledge, our m ethod has achieved the largest p ulse energy m easured from a
m on o lith ic d ev ice in such a tim e scale.
A theoretical m odel that accurately pred icts the 0 -sw itc h e d LD p erfo rm an ce
has been dev elo p ed in our laboratory by T hedrez [38] and Y ang [39]. S in ce w e are
interested in know ing if o u r results check with theoretical m odels, w e present a
co m parison o f theory w ith experim ent, b ut d o not give the d etails o f the theory here.
F igure 3.14 (b) show s the theoretical 0 -sw itch cd o u tp u t pulse for th e type A laser, and
co m parison w ith Figure 3.14 (a) show s excellent ag reem en t. For the type B laser, the
theoretical resu lt is show n in Figure 3.15 (b). w here again ex cellen t theoretical
ag reem en t is dem onstrated. W hat is novel about this theoretical cu rv e is the fact that
d io d e features th a t are on the order o f r rt have b een accurately m odeled.
O ne can
accurately m o d el such features by using a set o f traveling w av e eq u atio n s for the
o p tical intensity and a rate equation for the gain, w here as in previous m odeling
techniques a distributed loss approxim ation was assu m ed that o u r results have proven
to b e invalid for our ex perim ent [38].
53
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
O scillo sco p e
Lock-in
am plifier
f s i
W?
Photodiode
\- y
plotter
I Interference filter
f
LiQj crystal
T w o-section
LD
Infrared filter
Green filter
M otorized
Translation
Stage
Fequency
Doubler
(K.DP crystal)
/w
V
- /
.06 um beam
—
►—
7*
Figure 3.13 O ptical ^ -sw itc h in g ex p erim en tal setup.
54
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
T he features evident w ith the T ype B laser can be q u alitatively ex p lain ed as
follow s. W hen the optically ^ -s w itc h e d pulse is ex cited in the LD cavity, it o scillates
in the cav ity w ith a period equal to t rt.
If, after o n e round trip , the cav ity gain is
depleted, th en the O -sw itchcd pulse structure is sim ilar to th a t d isp lay ed by th e T ype A
laser: a sim ple pulse shape w ithout any additional features.
H ow ever, if the 0 -
sw itched p u lse has insufficient intensity to co m p letely d ep lete the gain in the LD
cavity, then additional pulses will experience gain an d , upon being am p lified by this
gain, be observ ed at the LD output as features o n the O -sw itch ed p ulse.
A s an
illustration, the cavity length for the T ype B laser is 1 m m . w hich co rresp o n d s to a t rt
o f =12 ps.
A s Figure 3.15 show s, this value is in qu alitativ e ag reem en t w ith the
observ ed perio d icity o f the ^ -sw itc h e d pulse features.
W e have dem onstrated operation at high pow ers (u p to 6 \V) and sh o rt pulse
w idths (as sh o rt as 21 ps) from a single broad-arca LD . T hese results are im p o rtan t for
the optoelectro n ic attenuator since this is a potential m eans for activ atin g the
atten u ato r w ith high-pow er optical pulses from a laser diode: th e sh o rt p u lse w idth
indicates th a t this can be done at a high repetition rate (the repetition rate is lim ited by
the C P W -P C S PC carrier decay lim e and not the laser driver). R esearch is co n tin u in g
in our lab to dem onstrate sim ilar perform ance u sin g gain sw itch in g , w h ich differs
from the O -sw itching technique in that an electrical pulse is used to rap id ly bias the
gain section w ithout the need for the optical sw itch in g pulse.
55
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
1
0
a
.6
.0
4
.4
2
2
0
0
0
100
60
0
160
Time (ps)
60
100
160
Time (ps)
00
(b)
F igure 3.14 T y p e // broad-area laser experim ental (a) a n d theoretical (b) O-svvitched
p ulse output. B ias current is 400 m A , and m o d u la to r section voltage is - 5 V.
i
1 " /I
0
.6
.0
.6
.4
4
2
2
0
0
0
100
60
0
160
Time (ps)
100
ISO
Time (ps)
(a)
F igure 3.15
so
(b)
T ype B broad-area laser ex perim ental (a) an d th eo retical (b) ^ -s w itc h e d
pulse output. B ias current is 3000 m A , and m o d u lato r sectio n voltage is 0 V.
56
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§3.4 Laser Diode M icrow ave Im pedance M atching Transform er
In the last section w e dem o n strated that o p tically ^ -s w itc h in g a sin g le tw osection LD c a n be used to d rastically increase the laser o u tp u t pow er and reduce the
pulse w idth. O bv io u sly , this is a com plicated m eth o d s and i f it is p o ssib le to im prove
the o u tp u t po w er o f LD s by m o re traditio n al techniques, then th is should be tried. O ne
o f the b ig g e st factors lead in g to poor L D efficiency is the sim ple fact that 5 0 -Q
electrical driv ers are ty p ically used to activ ate the lasers, an d these lasers generally
have o n-state im pedances o f 1 to 5 f i. C learly , m o st o f the electrical d riv e p o w er is
reflected from the LD by th is large im pedance m ism atch an d therefore w asted. T he
p o w er tran sfe r to a lo w -im pedancc device can be im proved by the use o f m icro w av e
m atching te ch n iq u es that are w ell co v ered in the literatu re [87].
R ecen tly , sim ilar
te ch n iq u es h av e been em p lo y ed to m atch L D s to h ig h -freq u en cy drivers so th at highfrequency m o d u latio n can be efficiently ach iev ed [88].[89],
T hese are trad itio n ally
n arro w -b an d w id th techniques, w here the m atched b an d w id th is ty pically <10 p ercent
o f the m o d u latio n frequency (e.g., 1-GH z sig n als m atch ed o v er 100-M H z b andw idth).
F ast-risetim e curren t pulses on the o rd er o f 1 ns, w ith pulse w idths o f 50 to 100
ns. are used to tu rn on o u r L D s (the F ourier frequency co n ten t o f th ese p u lses is <1
G H z). In o rd er to im prove th e overall efficien cy o f h ig h -p o w er LDs. w e m u st m atch
the gain section im pedance o v er a broad b andw idth (=1 G H z), and th is b an d w id th
co v ers frequencies from dc to 1 G H z. T h erefore, w e u n d erto o k to m atch th e LD gain
se ctio n from 300 M H z to 1 G H z using co m m ercially av ailab le m icrow ave d esig n tools
(A C A D E M Y [90]). In o u r d esig n , w e assu m ed a resistiv e dio d e o n -state im p ed an ce o f
5 Q and perm itted an in-band ripple o f 0.5 dB .
T he im p ed an ce tran sfo rm er d esig n
selected w as a q u arter-w av e transform er, since th is typically y ield s b ro ad -b an d
m atch in g [87],
57
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
MSUO
P3
ER* 11.4
H =0.1
T = 0.003
RHO = 0.68
RGH = 0.0
CAP
CAP
ci i
C2 i
CA P
CAP
C3 i
C4 i
P2
MLIN
M STEP
T1
W = W1
T6
T2
T7
W1 = W1 w = W 2 W1 = W 2
T3
T8
W = W3 W1 = W 3
T4
T9
W = W4 W1 = W 4
W 2 = W2 L = L2
L = L3
L = L4
l = L1
M UN
M STEP
W 2=W 3
MLIN
M STEP
W2 = W4
MLIN
M STEP
W2 = W5
MLIN
T5
W = W5
L = L5
F igure 3.16 L D M T electrical schem atic g enerated by A C A D E M Y [90],
L um ped capacitors C , (0.6 pF). C t (12 pF). C 3 (12 pF. and C 4 (7.5 pF ) red u ce the
L D M T length from 5 to 3.7 cm . A 5 -Q real L D im pedance w as assum ed.
0 33 c
1 3 cm
3.7 cm
F igure 3.17 L D M T m icrostrip layout g enerated by A C A D E M Y [90].
L um ped capacitors. C , to C 4, w ere soldered to ground w here indicated. T h e laser
diode w as w ire-bonded to P2.
58
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
T he laser diode m icrow ave tran sfo rm er (L D M T ) electrical circu it sch em atic
generated by A C A D E M Y is show n in Figure 3.16. Since w e w anted to m atch low lrcquency com ponents, w e used discrete lum ped capacitors (C , to C4 in the figure) to
reduce the overall electrical length to 3.7 cm . The built-in A C A D E M Y o p tim izatio n
routine w as used to optim ize the design, an d the final m icrow ave layout w as g en erated
(F igure 3.17).
Figure 3.18 show s the calculated value o f
for the L D M T and a
reference 50-Q m icrostrip line. N ote that A C A D E M Y predicts that w e can im prove
the m atching o f a 5-Q real im pedance with the L D M T over a b an d w id th o f
approxim ately
1 G H z, w hich is quite im pressive co n sid erin g th at this includes
frequency com ponents as low as 200 M H z.
T he L D M T was realized w ith m icrostrip line and fabricated o n co p p er-clad
D U R O ID 6010
[91];
a reference 50-Q m icrostrip line w as fabricated as w ell.
A
100-gm -stripe broad-area LD w as then used in m easuring the perform ance o f both
circuits. T he laser was m ounted o n a suitable heat sink and th en w ire bonded to each
o f the circuits.
Figure 3.19 show s the value o f S | | m easured on a sc alar n etw ork
a n aly zer (SN A ). N ote that the SN A data sh o w that the L D M T im p ro v es the m atching
to the LD o v e r a bandw idth o f =500 M H z.
T he discrepancy betw een the m easured and calculated values is m o st likely due
to tw o factors; the LD w irc-bonds are highly inductive, and the real LD im p ed an ce is
highly reactive (i.e., capacitive).
To sim plify o u r design, a n d since no d ata w ere
available concerning the actual LD on-state im pedance, w e chose a 5-Q real diode
im pedance as a first try.
I f the LD m odel can be adjusted to acco u n t for these tw o
factors, an im provem ent o v er a 500-M H z bandw idth is expected.
U ltim ately,
to
determ ine
w heth er the
LD M T
can
im prove
the
diode
p erform ance, one m ust com pare the L i l curves for the sam e laser being d riv en by both
circuits. F igure 3.20 show s the m easured result from such an ex p erim en t. N o te th a t at
59
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
a d riv e cu rren t o f 175 m A , a 37-perccnt increase in the op tical intensity h a s been
ac h iev ed w ith the L D M T. T h is certainly proves that lo w -freq u en cy , b ro ad -b an d w id th
m atch in g o f L D s is a possible technique fo r im proving o verall LD efficiency.
O ne o f the research g oals o u tlin ed in the p rev io u s ch ap ter w as the need to
d ev e lo p an all-scm ico n d u cto r-b ased opto electro n ic atten u ato r th a t could be o perated
w ith m inim um electrical pow er. If the Z ,//cu rv e is im proved by 37 p ercent fo r a fixed
bias level, th en the current required to ach iev e a given o p tical pow er level should
th erefo re be reduced by this am ount. T hus, this p ortion o f the L D d ev e lo p m e n t effort
is o f extrem e im portance for the o p toelectro n ic attenuator; the L D M T p erm its the use
o f M Q W L D s since w e can take advantage o f the high gain (and th erefo re h ig h output
pow er) o f these LD s w hile sim ultaneously reducing the electrical driver req u irem en ts
to acceptable levels.
60
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
5
0
•5
-10
-15
-20
-25
LDMT
-30
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
Frequency (GHz)
F ig u re 3.18 A C A D E M Y [90] prediction o f the reflected pow er. .S^ |. o f th e L D M T and
5 0 -Q reference line, assum ing a 5 -Q real load im pedance. N ote that a 1-G H z m atch ed
bandw idth is p red icted .
5
0
5
-10
-15
LDMT
-20
-25
-30
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Frequency (GHz)
F ig u re 3.19 M easured S \ |, o f the L D M T and 5 0 -Q reference line term in ated w ith
b ro ad -area LD. N ote that the LD m atching h as been im proved o v er a 5 0 0 -M H z
bandw idth. The sharp peak o n the 50-Q d a ta is d ue to a pack ag in g reso n an ce.
61
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
300
200
c
LDMT
100
'•'*
0
50-Q line
200
100
300
C urrent (m A )
F igure 3.20 M easured LI I curve using the L D M T and 50-Q referen ce line to d rive the
sam e LD. N ote: an increase o f 37% in the o ptical pow er, at 175 m A . has been
ach ie v ed w ith the L D M T . 100-ns pulses at 1-kHz PRF w ere used to d rive the laser.
62
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
§3.5 Sum m ary
S everal LD types (i.e., ridge w aveguide, broad area, m u ltiple section. SQ W ,
M Q W . etc.) w ere fabricated and evaluated for use as the lig h t so u rce for the o ptically
controlled attenuator.
We have fabricated discrete b ro ad -area L D s with peak optical
pow ers on the o rd er o f 1 W , and D -sw itched these sam e d evices to increase the pow er
to 6 W. In ad d itio n , these ^ -sw itc h e d resu lts arc believ ed to be the highest energy and
shortest pulsc-vvidth results from discrete devices to d ate.
W e hav e d em onstrated a
m icrow ave im pedance m atching transform er, the L D M T . and sh o w n that the overall
LD efficiency can be im proved by as m uch as 37 percen t for a fixed bias level.
N ow
that
the
optoelectronic
atten u ato r
control
d evice,
nam ely
the
se m ico n d u cto r LD . has been developed, w e w ill now d escrib e the design, fabrication,
and evalu atio n o f the attenuation elem ent.
63
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
Chapter 4
Photoconductive Switches
P erhaps o n e o f the sim plest sem ico n d u cto r d evices is the bulk p h o to co n d u ctiv e
sw itch, w hich is m ade up o f two m etallic co n tacts placed on a photo co n d u ctiv e
substrate, such as a sem iconductor w afer.
L ight then illum inates the intervening
sem iconductor m aterial and is absorbed and co nverted into elcctro n -h o le pairs.
P hotoconductors can thus be used to convert optical en erg y into electrical signals.
T hese photoconductive devices are co m m o n ly referred to as photo co n d u ctiv e
(PC ) sw itches; ap plications for. and the d ev elo p m en t of. PC sw itches have been well
covered in the literature by Lee [3],[551- W hen th ese devices are fabricated in su ch a
w ay as to su p p o rt high-speed signal generation, m o d u latio n , and p ropagation, then
these types o f PC sw itches can be used in m icrow ave sy stem s.
M any ex am p les o f
such high-speed PC devices and system s ap p licatio n s m ay be found in the literature
[3].
In particular, the fabrication o f planar m icrow ave tran sm issio n -lin e structures
onto PC substrate m aterials yields high-speed PC sw itch es that arc em b ed d ed in the
m icrow ave structure [92],
Since a treatm ent o f the basic physics o f PC sw itch es is essential to the
o ptim ization o f the optoelectronic attenuator, the relevant physics o f PC sw itch es will
be given at the beginning o f this chapter.
N ex t, w e rev iew planar m icrow ave
transm ission-line structures and discuss their potential u sefu ln ess w ith regard to the
optoelectronic attenuator.
Finally, w e describe th e actual design, fabrication, and
characterization o f G aA s and silicon coplanar w av cg u id e-P C sw itches (C P W -P C S s)
that w ill be used for constructing the hybrid o p to electro n ic attenuator.
64
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
§4.1 Photoconductive Switch Physics
T he theory d escribing th e dynam ic PC sw itch resistan ce has been d ev elo p ed by
C. H. L ee [55]. T h e PC sw itch on-state resistance, Ron. o f a d ev ice w hose o p era tio n is
based on an o p tic ally induced conductivity ch an g e (i.e., “co n d u ctiv e m o d e ” o f
operation) is giv en by
* on =
(4 - o
w here /gap is the PC sw itch electrode spacing. N is th e total n u m b e r o f p h o to -in d u ced
electrons g enerated by the optical pulse. <j is the electro n ic charge, and p is the
electron m obility. The electro n m obility o f bulk G aA s at room tem p eratu re is = 8,800
cm -/V -s. w hile th e surface m obility can b e as low as 1000 c m 2/V 'S.
T h erefo re, the
path taken by the ph o to cu rren t is an im portant PC sw itch co n sid eratio n .
W ork by F unk and o th ers [35] has show n th a t PC sw itc h es can han d le large
currents and be triggered in picoseconds.
T he q u estio n is, can this be d o n e w ith a
conventional (or new ) laser diode design?
T he laser radiation h a s a w av elen g th , k .
corresp o n d in g to a photon en erg y (E = h v = h c/k) that satisfies o n e o f th e follow ing:
( 1) ph o to n energy, h v , is greater th an the sw itch m aterial b an d -g ap en e rg y , i.e.,
h v > £ g,
(2 ) p h o to n energy is less than the b and-gap en erg y , b ut g rea ter than th e energy
differen ce betw een a suitable im purity level an d the co n d u c tio n ban d . i.e..
Et < h v < £ g,
(3) p h o to n energy is less than the b and-gap en erg y and no su itab le im p u rity
levels ex ist, i.e..
£ c > hv.
65
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
The first abso rp tio n m ech an ism is referred to as in trin sic p h o to n ab so rp tio n
[94]; a single p h o to n has su fficie n t energy to p ro m o te a v alence electro n into the
co n d u ctio n band, thus generating a n electro n -h o lc p air in the PC sw itch . T h is is the
m o st efficient ab so rp tio n m echanism , and. as a result th e ab so rp tio n d ep th in the
se m ico n d u cto r
is
very
shallow
(the
photon
ab so rp tio n
co efficien t,
a.
varies
e x p o n en tially w ith th e distance from the surface). A PC sw itch u sin g this ab so rp tio n
m echanism is usu ally fabricated w ith both co n tacts on the sam e sem ico n d u cto r
surface, and is thus referred to as a su rface sw itch (sin ce the co n d u c tiv e path is parallel
to the sw itch surface).
D esigns based on the second type o f ab so rp tio n m ech an ism rely on the
presence o f im p u rities w ithin the se m ico n d u cto r cry stal; these im p u rities, an d /o r traps,
m ediate the ab so rp tio n o f the incid en t photons. Since the en erg y levels o f th ese traps
arc w ithin the se m ico n d u cto r band-gap, the ab so rp tio n is m ore u n ifo rm ly distrib u ted
w ith in the se m ico n d u cto r bulk th en w as the case for in trin sic ab so rp tio n .
T his
absorption m echanism can result in fairly efficien t p h o to n to elec tro n -h o le pair
generation, d ep en d in g on the ex trin sic im purity, ag ain sin ce th e ab so rp tio n is m ore
uniform ly d istrib u ted through the bulk crystal.
absorption
m echanism
have
co n tacts
T hus, PC sw itch es that rely on this
fabricated
on
o p p o sin g
sid es
of
the
sem ico n d u cto r cry stal; this type o f PC sw itch is referred to as a b u lk PC sw itch. T his
absorption m ech an ism is usually referred to as ex trin sic ab so rp tio n [94],
T he third ty p e o f ab so rp tio n is referred to as tw o -p h o to n ab so rp tio n ; the energy
o f a single p h oton is insufficient to prom ote electron tran sfer, and th erefo re two
photons m ust use th eir co m b in ed energy to g enerate an elec tro n -h o le pair.
The
absorption o f ph o to n s in this case is non linear (d ep en d en t on o p tical intensity
squared) and is thus som ew hat inefficient, especially for low in ten sity lasers such as
66
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
the sem ico n d u cto r LD . PC sw itches using this tech n iq u e are also called b u lk sw itches
since the photon absorption is also uniform ly distrib u ted in the se m ico n d u cto r bulk.
T herefore, the relevant abso rp tio n m echanism is o f fundam ental im portance,
since both the sw itch and laser so u rce chosen are d eterm in ed by this factor. I f intrinsic
photon absorption (ty p e 1) is used (/iv > £ g). then the d epth a t w hich the p h o to n s are
absorbed by the m aterial is very shallow .
For pulsed p o w er ap p licatio n s, surface
sw itches have dem onstrated a m inim um on-state resistance value o f > 2 0 0
[95].
H ow ever, these arc large d evices (1 -5 m m gap spacing); the lo w er resistan ce lim it for
surface sw itches w ith gap lengths on the order o f m icrons m ay be m uch less.
S ince the cry stal structure in the bulk is generally su p erio r to that o f the surface
(because o f surface d efects and recom bination centers, w hich resu lt in a low er surface
m obility), bulk
PC
sw itches
can. in principle, achieve
m uch
lo w er d y nam ic
resistances, w hich m ay be as low as 1 Q [35]. S ince tw o pho to n ab so rp tio n (type 3) is
inefficient, the m ost suitable bulk absorption m echanism
for the o p to electro n ic
atten u a to r is extrinsic absorption (type 2); i.e., £ t < h v < E z . T h is is esp ecially true if
the LD and PC sw itc h are m ade from th e sam e m aterial (h v = £ g). I f a qu an tu m well
laser is used, then the laser energy can be designed to m eet the laser en erg y co n d itio n s
sp ecified for both ty p es I and 2.
T herefore, trap-m ediated (extrinsic) ab so rp tio n is believed to be the best ch o ice
for an all-G aA s optoelectronic attenuator, provided that a su itab le trap in the bulk
m aterial is used. O ne likely candidate is the well k now n EL2 trap in G aA s [96], w ith
a co rresponding AE = 0.83 eV (AE = Ec - £ trap). PC sw itches have been su ccessfu lly
fabricated w ith this trap [97], and since this level is alw ay s p resen t in G aA s, no special
d o p in g o f the su b strate is required.
I f the o p toelectronic allenualor uses a silico n PC sw itch , then a surface PC
sw itch (type 1) can b e used, since the surface q u ality o f p assivated Si is m uch b etter
67
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
than that o f G aA s.
In addition, the low er band-gap o f Si (*1.1 eV ) su g g e sts that
efficien t surface sw itching is p ossible w ith an A lG aA s LD. sin ce hv > Eg. O ne other
reaso n to try usin g a Si PC sw itc h in the atten u ato r is w orth m en tio n in g now .
A PC sw itch characteristic o f im portance is the PC ca rrie r recom bination
lifetim e, x r. For G aA s, Tr is * 1 - 5 ns [98], w hile in silico n the lifetim e is * 1 -1 0 ps
[67],
T his is an im portant m aterial property because the life-tim e d eterm in es how
often the PC sw itch m ust rc-absorb photo n s for the laser to m aintain a co nstant density
o f photocarriers, yV, and it also d eterm ines the speed at w hich the sw itch can open. For
exam ple, a 100 ns optical pulse is incident on both a G aA s and Si PC sw itch.
Fvery
1-5 ns. the G aA s PC sw itch loses 1/e o f its photocarricrs to reco m b in atio n ; these
carriers m ust be replenished by the laser for a constant level o f p h o to co n d u ctiv ity to be
m aintained. O n the other hand, the Si sw itch does not lose carriers a s quickly, so that
additional carriers are continuously created, thus providing m ore p h o to carriers for
co nduction.
T his phenom enon is usually referred to as photo co n d u ctiv e gain [99],
ex p ressed as
o;n = t r / t , ,
(4.2)
w h ere t r is the PC carrier lifetim e and Tt is the carrier transit tim e; r t = /„ap/(pjT).
w here p is the carrier m obility. E is the electric field (E = P7/cap), and /uap is the PC
sw itc h gap. T hus, if the carrier tran sit tim e is shorter than the PC carrier life tim e (i.e.,
xt < xr), then G on can exceed one, and the PC sw itch can be said to have an effective
PC gain.
For the S i:C P W -P C S , if xr * 1 ps.
V = 1 V. p = 1400 c m -/V -s , and
/cap = 10 pm . then Gon * 1400.
T herefore, although G aA s is the preferred materia! for th e C P W -P C S because
o f its com patibility w ith standard m icrow ave integrated circu its, silicon m ay be the
68
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
best ch o ice for dem onstration o f the opto electro n ic atten u ato r concept. If this p ro v es
to be valid, then the lifetim e in G aA s can be increased by techniques th a t use
sup erlatticc structures [45],[100].
It w as for these reasons that both G aA s an d Si
C P W -P C S s w ere fabricated and characterized during this research p roject.
§4.2 Planar M icrowave Transm ission Lines
P hotoconductivity can be thought o f as the prim ary "intrinsic" ch aracteristic o f
the PC m edium , w here as long as the m edium is a p h o toconductor, o ne can ex p e ct the
m edium to behave as discussed: above £ g rad iatio n yields intrinsic ph o to n ab so rp tio n ,
w h ich in turn generates PC carriers in a sh allo w depth from the PC sw itch surface.
C onversely, below £ g radiation yields ab so rp tio n depths that m ay include the en tire
crystal thickness, an d this type o f ab so rp tio n m ay be d ue to various ex trin sic
m echanism s, but im purity-level-m cdiated ab sorption is the usual situ atio n .
T he next issue is how to take su ch a m edium and fabricate a practical PC
sw itch. T he m ost im portant elem ent in su ch a device is the electrical contact stru ctu re.
If tw o contacts are placed on the sam e sem ico n d u cto r surface and a p h o to cu rren t is to
How betw een them , the laser chosen w ould typically em it rad iatio n th at is ab o v e £,,.
since this w ould create PC carriers betw een the contacts. H ow ever, if th e co n tacts are
placed on opposite sides o f a sem icondu cto r crystal, then the laser ch o sen w o u ld em it
radiation below £ g so that bulk PC carriers w ould be injected b etw een the co n tacts.
O f co u rse, this is not alw ays the case, since effects such as surface reco m b in atio n m ay
d ictate th a t radiation below £ g should be used even for surface sw itches.
[n addition to these considerations, the contact geom etry d ictates w hat ty p e o f
ap p licatio n the PC sw itch can be used for.
T h e contact (o r sw itch) gap. L. is also
69
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
critical; i f L is too sm all, then a large vo ltag e cannot be “ held o f f ' by the P C sw itch,
and if L is too big, th en a large am o u n t o f o p tical en erg y is req u ired to clo se the PC
sw itch. T his is perh ap s the m ost im portant co n sid eratio n in PC sw itch d esig n ;
it is
q u ite o b v io u s w hat choice m ust be m ade fo r a p articu lar ap p licatio n ("h ig h -sp ee d "
im plies surface sw itch w ith sm all gap, "h ig h pow er" im plies b ulk sw itch w ith large
gap. etc.).
Since the o bjective o f this research is to o p tically control a m icro w av e signal
w ith a la ser light source, it is quite reason ab le to assum e th at the b est contact geo m etry
to use is one that is com patible w ith m icrow ave dev ices.
T h u s, any p la n ar (i.e..
parallel to the sem iconductor surface) tran sm issio n -lin e geo m etry that can su p p o rt
trav elin g electrom agnetic w aves in the m icrow ave p o rtio n o f the frequency spectru m
is acceptable.
An ex c ellen t sum m ary o f th e m ost tech n o lo g ically im p o rtan t planar
m icro w av e tran sm issio n lines is given by Rivera an d Itoh [101].
An ex cellen t
co m p ila tio n o f the relevant planar transm issio n -lin e an aly ses has been ed ited by Itoh
[ 102],
A s an illustration o f the im portance o f contact g eo m etry and laser w av elen g th
o n a m icrow ave PC sw itch design. Figure 4.1 com pares tw o m icro w av e tran sm issio n
line types that have been fabricated on a PC substrate, su c h as G aA s o r Si. T h e figure
show s th e im portant differences as they affect m icrow ave PC sw itch d esign:
T he
m icro strip PC sw itch requires radiation belo w £ g so that carriers arc g enerated
b etw een the center (i.e.. top) co n d u c to r and the integral ground plane. T h e co p lan ar
w av eg u id e PC sw itch (or C P W -P C S for brev ity ) req u ires radiation above £ g so that
carriers are created on the surface betw een th e coplanar strips.
70
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
S ince our goal is to dem onstrate a hybrid o p to electro n ic attenuator, m ean in g
th a t d isc re te devices arc to be config u red w ith frcc-spacc an d /o r o p tical
co m p o n en ts, then the C P W -P C S is naturally the best choice.
fiber
T h ere are o th e r p lan ar
stru ctu res w here both contacts reside on the surface, su ch as co p lan ar strip lin e and
slo t line, b u t these geom etries are eith er m ore d ifficu lt to realize or p ro v id e less
elec tro m ag n e tic interference protection [103].
In ad d itio n . C P W can be o p era ted at
h ig h e r frequencies than these other structures [104],
71
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
PC Substrate
e r = relative perm ittivity.
fxTl = m etallization.
E d en o tes electric field lines.
Figure 4.1 C om parison betw een m icro strip line (left) and C P W (right).
The electric field lines are as show n: w ave pro p ag atio n into page.
N ote that carriers travel through bulk in the m icro strip PC sw itch and
on the surface in the C P W -P C S .
72
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§4.3 CPW -PCS Design, Fabrication, and Evaluation
The coplanar transm ission-line geom etry facilitates the use o f b ro ad -area laser
diodes that can be ea sily focused onto th e gap(s) betw een th e co planar conductors.
B oth S i and G aA s can b e used as the substrate m aterial, w ith the prim ary differen ce
being the ca rrier recom bination lifetim e o f 5 ns fo r G aA s and 10 p s for Si.
We
d esig n ed a coplanar w aveguide transm ission line for h ig h -sp eed app licatio n s using
stan d ard m icrow ave design tools [105]. T he ch aracteristic im pedance w as d esig n ed to
be 50 Q , w ith th e spacing betw een the center co n tact and the tw o outer co n tacts being
5. 10. o r 20 pm.
Figure 4 .2 show s a schem atic representation o f the G aA s:C P W -P C S . A highly
doped p ++ (A , =
1 0 19 cm - 3 ) epitaxial lay er w as g ro w n by m o lecu lar beam epitaxy
(M B E ) so that o h m ic contact to the substrate co u ld be m ade.
C r-A u co n tacts w ere
then deposited an d patterned w ith a metal lift-o ff technique, a n d the epitaxial lay er w et
chem ically etched betw een the m etalizations for electrical isolation.
T he dark
resistance o f the 10- and 20-pm sw itches w as ty p ically >1 M Q .
In all cases the substrate thickness w as approxim ately 500 pm . and th e overall
dev ice length w as chosen to be 1.6 cm.
T his length p erm its sw itch activ atio n by
v ario u s laser diode sources, such as linear one-dim ensional array s or m u ltip le discrete
laser d io d e s. S ince the con tact deposition thickness was only -2 5 0 0 A, the C P W -P C S
c e n te r
conductor has
a rather substantial
dc resistance
at
room
tem perature,
typically «20 to 40 Q. A lthough this parasitic resistance is a problem for the C P W PCS ap p licatio n itself, a gold plating p ro ce ss can be used to greatly reduce this
unw anted resistance by m aking the cen ter con d u cto r cross section m uch larger.
It
sh o u ld be noted th a t this unw anted d c resistance can be properly accounted for w h en
73
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
the o p to electro n ic atten u ato r experim ents arc p erform ed.
A ground plane was
d ep o sited for an o th er application [106] and the C P W -P C S designed for this case.
Figure 4.3 show s a cross section o f the S i:C P W -P C S that w as used to fo rm the
hybrid atten u ato r circuit discussed in C h ap ter 6 . F or ohm ic co n tac t to be ach iev ed to
the h igh resistivity substrate (p si > 6 x 103 Q -cm ). th e co n tact region w as ion
im planted w ith boron, w hich creates a P++ region.
A lu m in u m contacts w ere then
deposited onto the ion-im planted regions to form th e co p lan ar w av eg u id e structure.
C urrcnt-voltagc m easurem ents indicate that ohm ic co n tac t to the su b strate w as indeed
achieved.
Figure 4.3 show s tw o m ounted G aA s and S i:C P W -P C S d ev ices. So that both
the G aA s and S i:C P W -P C S w ould be properly co n n ected into a hybrid atten u ato r
circuit, the cop lan ar structure w as flared to p erm it co n n ectio n via an SM A con n ecto r.
T he IVIL ratio w as kept con stan t so that a 5 0 -Q ch aracteristic im pedance w as
m aintained at the design frequency o f 3 G H z. C o n sequently, the ov erall dev ice length
w as 1.6 cm , resulting in a large parasitic cen ter co n d u cto r resistance (-4 0 and - 6 0 Q
for the G aA s and Si devices, respectively). F ortunately this p arasitic resistance w o u ld
not be present in an operational device, since the d evice length n eed only be eq u al to
the laser spot size.
74
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
L = IO |im
M
M
~i rri
;
Cr-Au Contacts
ssssssssssssssssss
, p++ Epi-Layer
( N A - 5 x i O ISc m - 3 )
500 pn>
Semi-Insulating G aAs Substrate
Ground Plane
F igure. 4.2. S chem atic view o f the G aA s:C P W -P C S . Total d ev ice length (n o t show n)
is 1.6 cm . p ++ epi-lay er w as w et ch em ically etched to R jurk ^ • M Q .
L = 10 pm
Silicon Dioxide
. AI Contacts
\
250 pm
Boron Implanted Region
\
n
a - lxlOl9c m -3 )
Silicon Substrate ( P - 6x10"* n - c m )
F igure 4.3. S chem atic v ie w o f the S i:C P W -P C S . Total d ev ice length (n o t sh o w n ) is
1.6 cm . C o n tact region ion im planted to achieve ohm ic su b strate contact, as show n.
R jark =13 k Q ach iev ed for I = 10 pm . G ap reg io n s passivated w ith 4000 A o f S iO ; .
75
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Figure 4.4 M ounted G aA s (left) and Si (right) C P W -P C S devices.
L = 50 p m , overall length = 1.6 cm.
76
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
§4.4 CPW -PCS Electrical and O ptical Characterization
T he first step in the characterizatio n o f the C P W -P C S d evices is to m easure
th e ir m icrow ave perform ance.
A Scalar N etw ork A n aly zer (SN A ) w as used to
m easu re the insertion loss (IL ) in the off-state (see Figure 4.2 for the G aA s and 4.3 for
the S i:C P W -P C S ) as a function o f frequency.
T he G aA s:C P W -P C S displayed
ap p ro x im ately 3 dB o f l L at the desig n frequency o f 3 G H z, w hile the S i:C P W -P C S
d isplayed a higher level o f IL (~6.5 dB at 500 M H z). T he C P W -P C S d esig n assum ed
a su bstrate thickness o f 500 p m an d a dielectric co nstant o f 13.1. w hich are both
values for the G aA s:C P W -P C S .
Since the high-resistivity Si substrate w as only
250 p m thick, and Si has a low er dielectric co nstant (11.7), the increased IL in the
S i:C P W -P C S
is not unexpected.
Fortunately, all the o p to electro n ic attenuator
ex p erim en ts are unaffected by these values o f IL. since it is th e relative increase in onstate attenuation that is im portant. In addition, w hen the high level o f dc resistance is
accounted for, the S i:C P W -P C S IL is - 4 dB, w hich is accep tab le for this investigation.
O ne o f the m ost im portant characterization m easu rem en ts for PC sw itches is
th a t o f the PC carrier lifetim e (w hich, o f course, indicates w h eth er the PC sw itch is
functioning properly).
T his m easurem ent can be done w ith the sim p le dc circuit
show n in F igure 4.7. In this circuit, the reference (or ch arg in g resistor, as it is referred
to in the pulsed pow er com m u n ity ) is used to set a cu rren t reference across the
o scillo sco p e load. R$, w h ich is 50 Q . A dc bias, typically a few volts, is applied to the
circuit, and one m easures the decrease in the voltage across th e load that occurs w hen
the PC sw itch is illum inated w ith laser energy. The recovery tim e o f the vo ltag e drop
is u sed to determ ine the PC carrier lifetim e.
F igure 4.8 show s the m easured PC carrier lifetim e for th e G aA s:C P W -P C S ; the
top trace is the PC response, and the bottom trace is the laser pulse shape m easured
77
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
w ith a p h o to d io d e. In part (a) o f the figure, the response to n ear-IR laser en erg y from
a m ode-locked N d.'glass laser [84] (the sam e laser described ea rlier in S ectio n 3 .3) is
show n ( k = 1.054 pm ).
N ote that the voltage recovers to its d ark value in
approxim ately 3 0 ns (1/e value).
Part (b) o f this figure sh o w s th e resp o n se to green
laser energy (k = 0.512 p m ) th at is from the sam e laser but is frequency d o u b le d by a
LiC>3 crystal. H ere the PC carrier lifetim e is show n to be 5 ns for th is m aterial.
T he difference in the tw o m easu red lifetim es is straig h tfo rw ard ; the IR
response is the bulk PC ca rrie r recom bin atio n lifetim e w h ile the green resp o n se is the
surface PC ca rrie r recom bination lifetim e. W e conclude this b ecau se o f the ab so rp tio n
depth, a (cm -1 ), o f the in c id en t p h otons w ith in the G aA s cry stal [107]: a ~ 1 0 4 cm -1
in the green and thus the ab so rp tio n depth ( 1 /a ) is about I p m . w hereas a - 1 0 c m -1 at
1.054 p m . w h ich yields a n absorption depth o f ap p ro x im ately 1 m m . Since the depth
o f PC carrier generation is equal to th e absorption depth, w e can see th at the carrier
dynam ics are n o w a function o f the laser w avelength: i.e.. sh o rt w av elen g th s yield
surface carrier effects, and longer w av elen g th s yield b ulk carrier effects.
T his
correspondence is o bviously useful for correlating m easured d evice p erfo rm an ce w ith
the actual m aterial topology.
In a sim ilar fashion, the PC carrier lifetim es are sh o w n in Figure 4.9 for the
S i:C P W -P C S. w h ere this tim e only th e PC response is show n.
P art (a) o f the figure
show s the m easured lifetim e in th e !R (k = 1.054 pm ) (a 5 0 0 -M H z r f w av efo rm w as
also p ro p ag atin g through the circu it - hence the broad line w id th ) and in the red ( k =
0.797
p m ).
In this case,
we can estim ate the bulk and surface PC carrier
recom bination lifetim es to be - 2 ps and 350 ns. resp ectiv ely .
T hese rath er long
lifetim e values are consisten t w ith the fact th a t the S i:C P W -P C sw itch is fabricated on
very pure silicon (p ~6 x 103 Q -cm ). H ow ever, because o f th e in d irect b an d -g ap o f Si.
78
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
the a for Si at these w avelen g th s is still an order o f m ag n itu d e less than th at fo r G aA s.
even though this radiation is considered to be "ab o v e b an d -g ap " [108].
79
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Maikcr - UJ
CQ
-o
Ficq
= 3 GHz
= -2 .8 7 dB
-10
/(G H z )
F igure 4.5 M easured S^ \ o f G aA s:C P W -P C S .
N ote that the insertion loss is - 3 dB at 3 G H z.
M arkerFicq
SI,
LLI
= 5 0 0 M Hz
= -6 .5 4 dB
-10
-14
0.0
2.5
0.5
3.0
/(G H z )
F igure 4.6 M easured 5 t | o fS i:C P W -P C sw itch.
N ote that the insertion loss is - 6 .5 dB at 500 M Hz.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
C P W -P C S
scope
10X
N d rg lass
X = 1.054 p m
L ase r
I
10 ps
- 2 0 m J/p u Ise
P R F = 5 00 H z
Figure 4.7 E xperim ental setup used to m easure C P W -P C S PC carrier lifetim e. t r.
N d:glass laser is show n as an exam ple; o th e r lasers w ere used to m easure
t r at various w avelengths.
81
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
(a)
PC
PD ■
(b)
PC •
PD-
(C)
PC
PD-
Figure 4.8 M easured PC carrier lifetim e, t r, o f the G aA s:C P W -P C S .
T op trace: PC response, PpC. B ottom trace: P hotodiode response. ^PD(a)
(b)
IR response (1.054 p m ), w h ere Tr<bUik ~ - 0 n s- ar>d
green response (0.512 pm ), w here "^surface ~5 ns, and (c) both.
82
R e p ro du ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
(b)
■ —
M L -
ih M M iM IIM
F igure 4.9 M easured PC carrier lifetim e, r r, o f the S i:C P W -P C S :
(a) IR response (1.054 p m ), w here
- 2 p s, and
(b) red response (0.797 p m ), w here t r surfacc - 3 5 0 ns.
83
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§4.5 Sum m ary
A s this ch ap ter show s, the high-speed m icro w av e PC sw itch es for the
optoelectronic atten u ato r should be fabricated using a coplanar w av eg u id e geom etry.
F o r surface sw itching, w hich is used in the hybrid op to electro n ic atten u ato r, this
geom etry m akes the m o st sense.
F or a m o nolithically integrated versio n o f the
atten u ato r, fabricating the high-speed PC sw itch w ith a m icro strip tran sm issio n -lin e
geom etry w ould be m ore useful.
In this w ork, the o p toelectronic atten u ato r w as synth esized w ith w hat has been
called the C P W -P C S , an d this PC sw itch w as fabricated o n both G aA s and Si. T h e
su b se q u en t PC ca rrie r lifetim e m easurem ents show w hy the C P W -P C S w as fabricated
o n these m aterials: G aA s has a fast (~ 5 —30 ns) response, w hile Si has a m uch slo w er
(-3 5 0 ns - 2 p s) response. G aA s is the preferred m aterial since it is m ost co m p atib le
w ith m icro w av e integrated circuits. H ow ever, Si is useful because it allo w s the laser
pow er requirem en ts to be low ered since th e PC carriers arc not lost to reco m b in atio n
a s fast, and thus a hig h er level o f PC co n d u ctiv ity can be m aintained.
Since G aA s
N IPI stru ctu res can easily be grow n by M B E techniques, the effectiv e PC carrier
reco m b in atio n lifetim e can be engineered to also be in the m icro seco n d regim e [109].
T hus the true rationale for the S i:C P W -P C S is clear: i f this lo n g -lifetim e m aterial can
prove the feasibility o f the optoelectronic atten u ato r concept, then fu rth er research can
b e conducted to get the G aA s:C P W -P C S to do the sam e.
O f course, the C P W -P C S m ust p erfo rm as a m icrow ave tran sm issio n line in
th e off-state, and w e have show n that this has been achieved. U nfortunately, there is a
high level o f insertion loss for the fabricated devices ( -3 dB for the G aA s:C P W -P C S
an d - 6 .5 dB for the S i:C P W -P C S).
"While this is a p roblem for integration o f these
dev ices into a real system , it does not affect the resu lts presented in C h ap ter 6 . since
84
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
all o f the m easurem ents in this w ork are relative o n es, an d th erefo re o ff-state parasitics
are not a a so u rce o f m easurem ent error. In ad d itio n , im proved fab ricatio n techniques
an d /o r electro p latin g w ill greatly reduce the off-state insertion loss. F in ally , th e device
len g th s are very long to perm it ease o f ex p erim en tatio n ; thus, w hen all th ese factors
a re taken into account, no insertion loss problem s fo r the opto electro n ic atten u a to r are
foreseen.
85
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Chapter 5
GaAs:CPW -PCS Characterization
L aser dio d es (L D s) are ideal co m p act light so u rces
ap plications such as the optoelectronic attenuator.
for PC
sw itching
Since th e m o st h ighly d eveloped
type o f LD is the A IG aA s quantum well LD, as discussed earlier in C h ap ter 3. the
interaction o f A IG aA s L D s w ith the G aA s:C P W -P C S m ust be w ell ch aracterized .
In
C h ap ter 4. w e saw that the laser photon energy plays a key role in d eterm in in g w hat
kind o f pho to n ic ab so rp tio n takes place w ith in the PC sw itch.
S ince A IG aA s LD s
operate n ear the b and-gap energy o f the G aA s:C P W -P C S (i.e.. h v % £ e ) the ab sorption
m echanism m ay be eith er intrinsic, extrinsic, o r m ost probably a co m b in atio n o f both
types.
T his ch ap ter presents the results o f an ex ten siv e in v estig atio n into the
interaction betw een A IG aA s LD s and G aA s:C P W -P C S s.
T he in teractio n w as
m easured electrically, by m easuring the PC sw itc h on-state resistance, Ron. First, R on
was m easured using the 3 00-pm broad-arca LD d escribed in C h ap ter 3.
R on w as
m easured not o nly as a function o f LD pow er, but also as a function o f PC sw itch
tem perature and ap p lied dc bias (i.e., electric field).
The tem p eratu re w as v aried to
determ ine if the effectiv e band-gap energy, E„. could be tu n ed [75] to o p tim ize the
L D /PC sw itch interaction: as w ill be seen in Section 5.1. this w as indeed the case.
Since the electric field across the PC sw itch co nduction region can g reatly affect the
device's p erform ance, the dependence w ith electric field w as also studied.
For
exam ple, field -d ep en d en t carrier velocity o v ersh o o t is w ell know n in G aA s [110] and
the F ran z-K eld y sh E ffect m ay also play a role by changing the ab so rp tio n ed g e w ithin
the crystal [111 ],[ 112].
86
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T o verify the m easurem ents m ade w ith LD activ atio n , a T i:sap p h ire laser w as
used to m easure the dependence o f the PC sw itch R on as a function o f ph o to n energy,
w hile sim ultaneously checking the tem perature and electric lield d ependence.
A
substantial increase in the m agnitude o f Ron (and a decrease in th e PC conductivity)
w as o bserved for electric fields greater than 500 V /cm . T his increase in th e m agnitude
o f R on proved to be dep en d en t on the laser pho to n energy, /iv, since th is increase w as
only observed w hen h v w as ju s t below Eg. A num erical m odel has b een d eveloped
that, in the last section o f this chapter, is show n to be in relatively goo d ag reem en t
w ith experim ent.
T he
T i:sapphirc
m easurem ents
w ere
repeated
w ith
G aA s:C P W -P C S s
fabricated on three different sem i-insulatin g substrates; in all cases, the increase in the
m agnitude o f R on fo r h v < £ g w as observed, indicating that this is m ost likely a
geom etrical effect an d not due to the m aterial pro p erties itself.
§5.1 L a s e r D iode M e a s u re m e n ts
U sing the 3 0 0 -|im LD described in C h ap ter 3 (see L/I curve and spectrum .
Figures 3.8 and 3.9), the on-state resistance, /?on, w as m easu red as a function o f sw itch
tem perature and applied electric field. Figure 5.1 show s the ex perim ental setup used
to perform the LD m easurem ents.
The LD output w as first co llim ated , and then
focused, onto the 10-pm gap o f the G aA s:C P W -P C S . T h e sw itch input w as connected
to a dc bias through a 58-Q reference resistor, and the o u tp u t connected d irectly to a 1G H z bandw idth (T ektronix 7104) oscilloscope using a high-speed 5 0 -Q plug in. T he
sw itch w as m ounted on a P eltier cell and the tem perature con tro lled using an ILX
87
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
C orp. M odel L D T -5412 tem perature controller.
A 10-kQ th erm istan cc w as used to
p erm it accurate tem perature control near room tem perature.
Figure 5.2 sh o w s the experim ental results.
N ote that the m a g n itu d e o f R on
decreases w ith b oth increasing electrical bias and sw itch tem p eratu re.
T h at 7?on
d ecreases w ith increasing tem perature is expected sin ce the effectiv e b an d -g ap energy
o f G aA s also decreases w hen the tem perature goes up [75] (th is po in t w ill b e further
em p hasized in S ection 5.3). C o ncerning the dep en d en ce o f Ron versus electric field,
clearly application o f a fairly strong d rift field im p ro v es the sw itch p erfo rm an ce w hen
LD ac tiv atio n is em p lo y ed (i.e.. Ron decreases w ith increasing bias).
88
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
X = 880 nm
300-um LD
Y
W
C ollim ating lens
100ns
PRF = 1 kHz
F igure 5.1 E xperim ental setup used to m easure R on w ith a 3 0 0 -p m LD.
T he G aA s:C P W -P C S is m ounted on a T E cell to control its tem p eratu re.
2.5
V =10 V
2.0
20 V
a
&
30 V
0.5
0.0
10
20
30
40
50
60
70
80
T ( ° C)
F igure 5.2 M easured R on versus tem perature. T. and electric field. E.
N ote that R on d ec reases w ith both increasing E and T.
= 600 m W (peak).
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§5.2 Ti:Sapphirc Laser M easurements
To characterize Ron o f the G aA s:C P W -P C S versus p h o to n energy, /iv, w e used
a S chw artz E lectro-O ptics. Inc.. T i:sap p h ire laser (T itan -C W Series). T his T iisapphirc
laser co u ld be continuously tuned from approxim ately 700 to 900 nm . w here the band­
w idth o f the cav ity m irrors determ ines the actual tuning range.
T h erefo re, if
tem p eratu re tuning o f the G aA s:C P W -P C S has indeed m anifested its e lf as a sh ift in
th e effectiv e band-gap energy, E g. as indicated in Figure 5.2 for a fixed LD
w av elen g th , then fixing the tem perature and scanning the w av elen g th through
should confirm this hypothesis.
T he experim ental setup is show n in F ig u re 5.3. T he variable atten u a to r was
used to level the incident pow er to a constant v alu e as a function o f w avelength.
10x m icroscope objective was used to focus the laser beam
onto
the
A
lO-pm
G aA s:C P W -P C sw itch gap (spot size = 1 3 .5 p m ).
B ecause the laser w as operated in continuous-w ave (cw ) m ode, w e could use a
d c m easurem ent technique to m easure R on [42]. R on w as directly calcu lated from the
m easured voltage drop across the sw itch.
Since a h ig h-im pedance v o ltm eter (Fluke
M odel 75) w as used (Zm > 10 M Q ). the parasitic d c resistance o f the cen ter con d u cto r
can be neglected in the analysis. T he sw itch w as m ounted on a P eltier p late so th at the
sw itch tem perature could be controlled.
W e used tw o different app ro ach es to m easure the dep en d en ce o f R on with
respect to electric field.
O ne se t o f m easurem ents w as perform ed w ith a constant
electric field m aintained across the sw itch gap (i.e., we used the change in supply
voltage to calculate Ron). The o th e r technique w as to keep th e supply vo ltag e fixed
w hile m onitoring the field across the switch.
T he second tech n iq u e pro v ed to be the
90
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
m o st accu rate since the m easured d ata contained less noise; therefore, results using the
later te ch n iq u e
w ere used to characterize R on(h v, T. E) o f the G aA s:C P W -P C S .
F igure 5.4 show s the experim ental results used to d eterm in e Ron(h v) versus
G aA s:C P W -P C S tem perature.
T he incident pow er w as 100 m W and th e supply
vo ltag e w as 2 V (for an electric field o f 2 kV /cm ). N ote that the ab so rp tio n edge is
clea rly visib le in these resistance plots, and m oves to lo w er photon en erg ies w ith
increasin g tem perature, as expected [75]. In particular, n otice that a 20°C ch an g e in
tem p eratu re
results
in
a
decrease
in
Ron
o f m ore
than
a
factor o f
three
for h v = 1.405 eV .
F igure 5.5 show s the dependence o f Ron( h v ) for several supply voltages. Vc.
F or
F'c = 2 V
(E
= 2 kV /cm ). the variation o f R on w ith h v follow s the ab so rp tio n
pro file for sem i-in su latin g G aA s [113]; nam ely, the resistan ce is high for values
o f h v < Eq (i.e.. below band-gap radiation) and decreases for h v > Eg (i.e.. ab o v e bandgap radiation). T he m inim um resistance occurs w hen h v = Eg.
F or applied biases o f 4 and 8 V, F igure 5.5 sh o w s that this “ex p ected "
relatio n sh ip betw een h v and Eg is no longer observed; there is an increase in the
m ag n itu d e o f Ron at p hoton energies ju st below the band-gap energy, and this
resistan ce increase is seen to be strongly bias dependent. T hese m easu rem en ts w ere
repeated w ith G aA s:C P W -P C S devices m ade on three d ifferen t sem i-in su latin g
su b strates so that any m aterial effects could be elim inated as the "c au se " for this
o b serv ed bias dependency. In all three m aterials, the sam e resu lt w as o b tain ed , thus
indicatin g th a t this is eith er a sem i-insulating G aA s m aterial effect that is indep en d en t
o f w afer-to -w afer variation, or (m ost likely) a geom etrical effect.
91
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
V ariable Attenuator
T i:sapphire Laser
| |
Launcher
10X
Objective
spot
h M eter
O ptical Fiber
G aA s:CPW -PCS
TE C ooler
Figure 5.3 E xperim ental setup used to m easure £ on(/jv) v ersus electric field, £ , and
device tem perature, T. T he electrical circu it is as show n in F igure 5.1 w ith a
Fluke m ultim eter rep lacin g the o scillo sco p e.
8
T= 14.5 C
6
4
. 36.9 C
0
1.39
1.41
1.43
1.45
1.47
1.49
h v (eV)
Figure 5.4 M easured R on(hv) for 3 dev ice tem peratures. T. N o te for h v = 1.405 eV
R on(hv) decreased by a factor o f 3 for a 20°C increase in T. Vc = 2 V.
92
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T his field -d ep en d en t effec t is w avelength dependent.
T hus, one w ay to
inv estig ate w h eth e r or not the m aterial its e lf is a co n trib u tin g factor is to m easure the
m a te ria l's optical abso rp tio n coefficient, a . F igure 5.6 show s the m easured valu e o f a
ta k en w ith th e sam e sam ple used to fabricate the G aA s:C P W -P C S o f F igure 5.8.
N otice th a t th e absorption profile is relatively sm o o th and free o f any stru ctu re: w e can
con clu d e that th e m aterial is not providing any u n fo reseen effects th a t m ig h t cau se the
observed bias d ep e n d en t increase in Ron.
T here is one rem aining possibility; i f the observed increase in Ron is a function
o f p h o to n energy only, and not electric field, th en variation o f the o ptical p o w er should
have little effect o n o f Ron. H ow ever, if there is a large increase in R on as th e optical
p ow er is reduced, then this w ould indicate that th e increase in R on(h v, T, E) is d u e to
electric field p ro cesses for the follow ing reason: for a fixed sw itch bias, a red u ctio n in
the o p tical p o w er results in an increase in /?on w h ich , in turn, resu lts in an in crease in
the electric field across the PC sw itch.
F igure 5.7 show s the m easured data that co n firm s th at the increase in
Ron(hv. T, E) observed in F igure 5.5 is indeed a function o f optical p ow er (and hence
electric field).
N ote that, for an optical pow er o f 100 m W and Fc = 2 V, R on{hv) is
relativ ely flat - no resistance “ p eak" is o bserved. H ow ever, w hen the field increases,
as is the case for 25 m W o f o p tical pow er, then the peak b eh av io r is again observed.
T h u s. Figures 5.6 and 5.7 confirm that the o b serv ed b eh av io r o f F igure 5.5 is pro b ab ly
due to elec tric field effects and n o t optical ab so rp tio n -related m aterial effects. T h at is
to say. it is p ro b ab ly the C P W -P C S geom etry its e lf that p erm its the p eak in 7?on to be
observ ed for p h o to n energies ju s t belo w the band -g ap energy.
93
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
15
10
V =8 V
4 V
5
0
1.39
1.41
1.43
1.45
1.47
1.49
h v (eV)
F igu re 5.5 M easured Ron{h v ) for 3 bias v o ltag es. (rc. N ote th at at *1.405 eV
R on(hv) increases for Vc = 4 and 8 V. T = 10 °C .
150
130
10
90
70
50
1.380
1.385
1.390
1.395
1.400
1.405
hv (eV)
F igure 5.6 M easured a ( h v ) for G aA s sam ple used to fab ricate G aA s:C P W -P C S .
N o te that a (/iv ) is relatively sm ooth, in d icatin g that the m aterial does not
cause the p eak in Ron.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
20
15
Laser
= 25 mW
10
50 mW
5
100 mW
0
1.39
1.41
1.45
1.43
1.47
1.49
h v (eV)
F igure 5.7 M easured Ran(hv) for 3 laser pow ers. P |ascr. N ote th at at =1.405 eV
R on(hv) increases for /’laser = 25 and 50 m W . T = 10 °C.
95
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§5.3 GaAs:CPVV-PCS Conductivc-M ode Plasma M odel
T he interaction betw een the Ti:sapphirc laser an d the G aA s:C P W -P C S was
m odeled based on the schem atic representation show n in F ig u re 5.8.
T his m odel
assum es that the optical field incident on the sw itch is ap p ro x im ately uniform across
the sw itch gap: this is believed to be a valid ap p ro x im atio n since a p o rtio n o f the
incident beam covers the contacts and thus the G aussian profile o f the o ptical beam is
not fully enclosed in the sw itch gap. Even though below band -g ap radiation is at tim es
used, w e assum e that no bulk sw itching through the integral g ro u n d plane occu rs since
this contact is not ohm ic.
In addition, experim ental results w ith G aA s:C P W -P C S s
w ithout the integral ground plane w ere identical to those w ith the ground plane.
T he energy band-gap dependence on tem perature is given by [75]
(5.1)
W e will see shortly that this model o f the effective band-gap energy o f G aA s.
as a function o f tem perature, accurately predicts the resu lts that w ere m easured in
Section 5.2.
There ap p ear to be tw o primary transport m ech an ism s o ccu rrin g in the
G aA s:C P W -P C S
that reduce
the on-state
conductance:
carriers
arc
lost
to
recom bination (both at the surface and in the bulk), and carriers are sw ep t out o f the
sw itch gap on tim e scales shorter than the carrier reco m b in atio n lifetim e. O ur m odel
w as developed to investigate the effect o f these carrier loss m ech an ism s o n the overall
on-state conductance. We have developed a co nductive-m ode piasm a m odel to p red ict
the C P W -P C S o n-state resistance. Ron. as a function o f sw itch tem perature, incident
96
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
p h oton energy, and electric field. T he m odel calculates th e on-statc co n d u ctan ce. G on,
as a function o f all o f these param eters.
Since this m odel has been d iscussed in the
literature [42]. the final form is g iven below for the o n -state cond u ctan ce, Gon.
G.on
qn(l-^)a(/iv.D^c
C'hv
w here .v is the spatial coordinate indicatin g the distance from the sw itch surface. xr and
xt(x) are the PC carrier recom bination lifetim e in crystal and the PC carrier tran sit
tim e, respectively. In particular. t t(.r)=5(.v)2/ |i s Fb, w here S = [l + (2v/Z,)-tan(1.22X/Z,)].
n is the refractive index. p s is the PC carrier surface m obility, and Vh is the bias across
the sw itch.
X and v arc the pho to n w avelength and frequency, respectively.
The
m odel assum es that the field is con stan t in the sw itching gap and eq u al to o n e -h a lf o f
the circu it bias voltage. Fc.
The geom etrical factor D accounts for the diverg en ce o f the optical en erg y in
the substrate, p is the PC carrier bulk m obility, R is the optical reflectivity at the
G aA s-air interface, L is the PC sw itch gap length, and P jnc and h v arc the in cid en t
laser pow er and photon energy, respectively.
dep en d en t abso rp tio n
T he factor a ( h v . T) is the en erg y -
coefficient for sem i-insulating
G aA s
[107].
a
is also
tem perature d ependent since a change in tem perature cau ses a change in the b an d -g ap
energy. £ c. w hich im plies that a
m ust be shifted accordingly: this is h o w the
tem p erature-dependencc on the G aA s:C P W -P C S p erform ance is m odeled.
param eter in eq u atio n (5.2) is the substrate thickness. II.
T h e final
T he predicted o n -state
resistance. /?on, is the reciprocal o f equation (5.2).
T he m easured PC carrier lifetim es for the G aA s:C P W -P C S studied in this
chapter w ere given in Figure 4.8: t r surface = 5 ns. and Tr buik = 30 ns. T hese are the
97
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
v alu es used in the conductivc-m ode m odel (the m odel uses Tr>surfacc for .v < 1 p m . an d
Tr.bulk f° r 1
< x< H ).
T he theoretical value o f Ron(T) is w ithin ten p ercent o f th e ex p e rim en tal d ata,
a s sh o w n in F igure 5.9; this is w ith in experim ental erro r o f o u r m easu rem en t
ap p aratu s.
Thus it is believed that eq u a tio n 5.1 is an accu rate p red ictio n o f the
tem peratu re-d ep en d en ce o n Ron for the G aA s:C P W -P C S .
T he G aA s:C P W -P C S co n d u ctiv e-m o d e p lasm a m odel p red ic tio n for d iffe ren t
bias voltages is show n in F igure 5.10, alo n g w ith the experim ental d ata. N o tice th at
ex c ellen t agreem ent has been achieved for all op eratin g co n d itio n s e x c e p t the elec tric
field-dependent resistance peak at 1.405 eV. T he difficulty in m o d e lin g the electric
field-dependent resistance peak m ay be attributed to the fact th a t the m odel d o es n ot
ac c o u n t for any electric field disto rtio n s in the sw itch gap.
Since a high level o f
ca rrie rs arc created in th is region, th is is obv io u sly a sh o rtco m in g o f the m odel.
H ow ever, since m odels o f this type requ ire a rig o ro u s solution o f P o isso n 's eq u a tio n
[114], the m odel m ust be perform ed in three d im en sio n s, w hich is n ot o n ly tim e
co n su m in g , but difficult. In addition, sin ce this is only a fraction o f the total research
effo rt, and considering dial all else h as been adequately ex p lain ed , m ore d etailed
m o d elin g o f the G aA s:C P W -P C S is n o t w arranted for the o p to electro n ic a tten u a to r
application.
98
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without perm ission.
Cr-Au Contacts
itacts
-I I
\
I-
' ' s p o t ' 113 um
1 I
n
SI:GaAs
Substrate
Cr-Au G round plane-
F igure 5.8 S chem atic view o f the G aA s:C P W -P C S sh o w in g the g eom etrical
param eters used in the cond u ctiv e-m o d e p lasm a m odel.
T~- 14.5 UC
e
Theory
■ Measured
Theory
□ Measured
Theory
• Measured
4
1.39
1.41
1.43
1.45
1.47
1.49
h v (eV)
F igure 5.9 C onductive-m ode p lasm a pred ictio n o f Ron as a fu n ctio n o f ph o to n energy.
hv, for various tem peratures. T. V'b = 1 V (theory) and VC = 2 V (ex p erim en t).
99
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
y
= 4 V (Theory)
^ = 2 V (Theory)
= I V (Tlieory)
10
* F = 8V (M easurcd)
□ h'c = 4V (M easurcd)
g
F = ’ V(Xfeasuivd) ,
a&
* ■ ■*
1.39
1.41
1.43
1.45
1.47
1.49
h v (eV)
F igure 5.10 C onductivc-m ode p lasm a prediction o f R on as a function o f pho to n
energy for various bias voltages. Vc. Fb is the co n stan t bias assu m ed in the m odel and
Fc is the experim ental bias value related to the d ata points show n. T - 24.5 °C.
100
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§5.4 Sum m ary
The 3 0 0 -g m laser diode from C h ap ter 3 has been used to ch a racterize the
G aA s:C P W -P C S on-statc resistance.
It w as show n that the resistan ce co uld be
decreased by a factor o f 3 w ith increasing sw itch tem p eratu re.
In ad d itio n , the
resistance w as observed to decrease for rather large applied biases (10 to 30 V ). T he
F ranz-K eldysh effect w as determ ined not to be a factor since the ab so rp tio n ed g e in
the G aA s su b strate did not appear to chan g e its position.
The dependence o f the sw itch on-state resistan ce w ith ap p lied electric Held has
been experim entally determ ined using both a bro ad -arca LD and a tunable T i:sap p h irc
laser. For LD activation. Figure 5.2 clearly sh o w s that app licatio n o f a fairly strong
d rift field (10 to 30 kV /cm ) im proves the sw itch perfo rm an ce (i.e.. Ron decreases w ith
increasing bias).
H ow ever, the dependen ce o f Ron using the T k sap p h irc laser with
slightly less electrical bias (0.5 to 8 kV /cm ) w as ju s t the opp o site: Ron in c rea se d w ith
increasing bias
w hen a T ksapphirc
laser w as used to activate the sw itch. The
m odel o f Rm {hv, T. E) presented in S ection 5.3 has sh o w n that the later dep en d en ce
w ith electric field is correct and thus the LD d ata is in question.
R on(E) (at a fixed
tem perature) w as rem easured with the broad-area L D for electric fields sim ila r to those
used during the T ksapphire laser m easurem ents (500 to 8 kV /cm ); the electric field
dependence w as found to be consistent w ith the T i:S ap p h ire results. It is still unclear
w hy Ron versus strong electric fields actually decreases w ith increasing bias: perhaps
the w ell-know n phenom ena o f velocity overshoot in G aA s [110] m ig h t o ffe r a clue.
U sing the T k sap p h irc laser, and for field m ag n itu d es less than 500 V /cm . the
m easured o n-statc resistance as a function o f p h o to n energy roughly follow s the
absorption profile for the m aterial, w hich w as also m easured.
H ow ever, for field
m agnitudes in excess o f 500 V/cm . the on-state resistan ce d isp lay s a peak for photon
101
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
en e rg ies ju s t below the band-gap energy o f the sem ico n d u cto r.
T his p eak in the
resp o n se has been investigated and, to date, its ex act nature is still not u n d erstood. W e
can, h o w ever, exclude the possibility that it is a m aterial-d ep en d en t p h en o m en a related
to optical ab so rp tio n processes because the m easured peak is a function o f laser p ow er
(and hence electric field) and because the m easured ab so rp tio n
profile in the
sem ic o n d u cto r is relatively sm ooth.
It has been show n that tem perature tu n in g o f the G aA S :C P W -P C S can be
u tilized to o p tim ize the laser/sw itch interaction: a factor o f 3 d ecrease in the o n-statc
resistance w'as o btained for a 20°C change in sw itch tem p eratu re. T hus. 15 m eV o f
effec tiv e b and-gap tuning has been dem onstrated o v er this tem p eratu re range.
For
co m p ariso n , laser-diode tem perature tuning o v er th is sam e range w ould result in a
change o f the laser photon energy o f less than 6.7 m eV . T h erefo re. G aA s:C P W -P C S
tem p eratu re tuning is m ore efficient than laser d iode tuning.
T hus, any laser d iode
o p eratin g near the absorption edge o f a G aA s:C P W -P C S . as w ell as o th e r h igh-speed
G aA s PC sw itches, can be m atched to th e sw itch by u sin g relativ ely sim ple
tem peraturc-control techniques.
Finally,
a
conductive-m ode
model
for
the
G aA s:C P W -P C S
has
been
developed. T his m odel accurately predicts th e response o f the G aA s:C P W -P C S as a
function o f photon energy, tem perature, an d electric field (fo r E <500 V /cm ).
C a lc u latio n s indicate that b oth the F ranz-K eldysh E ffect and v elo city o v ersh o o t effects
in G aA s are not responsible for this resistance peak. The m odel m ust still be refined
to acco u n t for the peak in the resistance profile th a t w as o b serv ed for electric fields
greater than 500 V /cm at photon energies ju st b elo w the ab so rp tio n edge.
102
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Chapter 6
O ptoelectronic Attenuator Experim ents
T he use o f o p to electronic techniq u es to control m icro w av e circu its and system s
co ntinues to be an are a o f intense research and d ev elo p m en t [3 7 ],[1 1 5 ].[1 16], B esides
the inherent speed advantages o f this approach, use o f a la ser to control m ultiple
m icrow ave circuits perm its both a high deg ree o f electrical iso latio n b etw een the
control signal and the m icrow ave circuit, and tim in g precision th a t can easily be in the
picosecond regim e [3],
T his ch ap ter describes a hybrid opto electro n ic atten u ato r sch em e that is
suitable
for rem otely co ntrolling
m icrow ave
in teg rated circ u its.
B y o p tically
illum inating the gap(s) betw een the cen ter and o u te r co n d u c to r(s) o f a co p lan ar
m icrow ave tran sm issio n line fabricated on a silicon su b strate (i.e .. the S i:C P W -P C S o f
C h ap ter 4), we have dem onstrated up to 45 dB o f m icro w av e atten u atio n at 1.7 G H z
using only 143 m W o f laser diode (L D ) pow er. T his is the h ig h e st level o f atten u atio n
reported to date for such an attenuator schem e [4 3 ].[5 9 ].[1 17]. E arlier w ork by Platte
and S auerer [117] used a sim ilar techniq u e with C W optical illu m in atio n ; how ever,
they achieved a m axim um attenuation o f 1.8 dB.
In C h a p te r 4 w e discussed how h igh-speed PC sw itch es can be fabricated to
support m icrow ave signal propagation.
U sing the S i:C P W -P C S d escrib ed in that
chapter, w e used a com m ercially av ailab le fibcr-pigtailed laser d io d e to o p tically vary
the attenuation o f b oth r f and m icrow av e signals th ro u g h this h ig h -sp eed PC sw itch.
In C hapter 2, tw o possible attenuation m ech an ism s w ere p resen ted .
T he first w as a
retlective atten u atio n m echanism w hereby the r f p ow er is reflec te d by an optically
induced im pedance m ism atch in the high-speed PC sw itch.
T he seco n d w as an
103
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
a b so rp tiv e atten u atio n m echanism w hereby the r f signal is ab so rb ed by a solid-state
p lasm a that has also been optically created. Since eith er atten u atio n m echanism m ay
be responsible for an y observed r f attenuation, w e co n d u cted several ex p erim en ts to
d eterm in e w hich m echanism correctly accounts for any o p tically induced attenuation
in th e S i:C P W -P C S.
In addition, how w ell the optoelectronic atten u a to r perfo rm s, as a function o f r f
an d /o r m icrow ave
frequency, LD pow er, and LD beam
profile, m ust also be
d eterm in ed so that the atten u a to r perform ance can be o p tim ized ;
results o f hybrid
o ptoelectronic attenuator experim ents perform ed w ith various laser diode sources, as a
function o f these control param eters, are presented in this chapter.
The initial
exp erim en ts w ere conducted w ith an r f generator o p eratin g at 500 M H z.
W e used
several laser so u rces to attenuate the 500-M H z signal: a h ig h -p o w er m ode-locked
N d :g lass laser system [84], the 300-pm LD d escribed earlier in C h ap ter 3. a 1-cm
lin ear A IG aA s LD array [37], and. finally, a fiber-pigtailed LD [118],
A m axim um
atten u atio n o f the 5 00-M H z signal o f 30 dB w as ach iev ed w ith the fiber-pigtailed laser
for an optical p o w er o f 375 m W [43],
H ow ever, since w e co u ld m easure only the
signal attenuation through, an d not the reflected pow er from , the S i:C P W -P C S . the
e x a ct nature o f the absorption m echanism rem ained unclear.
U sing a V ector N etw ork A nalyzer (V N A ) [119], w e determ ined th at the
m echanism responsible for the attenuation o f m icrow ave sig n als in the op to electro n ic
atten u a to r w as absorptive; up to 45 dB o f atten u atio n w as achieved with a fiber­
pigtailed LD p ow er o f as little as 143 mW w ithout an increase in the pow er reflected
from the S i:C P W -P C S [36],
D uring
the
experim ents
involving the
fiber-pigtailed
LD. an im portant
observ atio n w as m ade: the m easured attenuation saturated at around 10 dB (at 500
M H z) w hen diffraction-lim ited spherical optics w ere used to focus the laser beam onto
104
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
the PC sw itch gap. W hen the LD beam profile w as ex panded in o n e d im en sio n (i.e.,
parallel to the S i:C P W -P C S structure), the level o f atten u atio n increased by 20 dB
a b o v e th is saturated value [36]. T he details o f all these ex p erim en ts are p resented in
this
chapter.
T he
first
sectio n
d iscusses
the
m o de-locked
N d tg lass
laser
m easurem ents, w hile the second section contains a d escription o f th e m easured
atten u atio n obtained with v ario u s LD sources. The 5 00-M H z p u lsed rf atten u a to r d ata
(m easu red w ith a high-speed analog oscilloscope), an d cw m icrow ave ex p erim ental
results (m easured w ith a V N A ), both using the fiber-pigtailed LD , are g iven in the
rem a in in g tw o sections.
§6.1 Nd:Glass Laser M easurem ents
W e perfo rm ed the first o ptoelectro n ic attenuator m easurem ents u sin g the sam e
N d tg la ss laser sy stem [84] used in b oth Section 3.3 to o ptically D -sw itch a tw o-section
b road-area LD , and in Section 4.4 to m easure the G aA s:C PW -P C S PC ca rrie r lifetim e.
F igure 6.1 show s the experim ental setup for the hybrid optoelectronic atten u ato r; the
N d :g lass laser w as operated a t a 500-H z pulse repetition rate w ith a pulse output
energy o f 20 p J and a F W H M o f 10 ps. A m icroscope objective (1 0 x ) w as used to
focus the N dtglass laser beam onto the 10-pm gap o f the S i:C P W -P C S. T he r f source
frequency and
pow er for these m easurem ents w ere 500 M H z a n d +18 d B m .
respectively.
Figure 6.2 show s a sam ple o f the S i:C P W -P C S o u tp u t w av efo rm m easured
w ith a 1-GH z band w idth o scilloscope (T ektronix 7104), u n d er these co n d itio n s. The
attenuation. Sjji, in decibels (dB ), is defined in C h ap ter 2. equation (2.3), for an
arb itra ry tw o-port netw ork, as
105
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
S 2l (d B ) = 2 0 L o g , 0 l %
j
( 6 . 1)
H ere, F2 is the m easured rf voltage w h en the sw itch is illum inated, and V, is th e rf
voltage in the off-state. U sing this relationship and the setup sh o w n in F igure 6 .1 , we
found that the m axim um
m easured
atten u atio n
o f the
5 0 0 -M H z
signal in
the
S i:C P W -P C S w a s - -1 1 dB.
B ecause o f the rath er short optical pulse w idth and
long m e asu rem en t
tim e (10 ps com pared w ith 1 |is), the actual attenuation is m o st likely g reater th an this
value. H ow ever, since our goal is to provide attenuation on longer tim e scales th an 10
ps (i.e.. - I ns) this result is n o t o f m uch practical interest to us. In ad d itio n , sin ce we
w ish to use L D s as the hybrid optoelectro n ic atten u ato r optical so u rce (so th at an allsem ico n d u cto r attenuator is dem onstrated ), the im portance o f th is resu lt is th at it is the
first step tow ards o u r goal o f dem o n stratin g an all sem ico n d u cto r-b ased o p to electro n ic
attenuator. T he next section show s that this goal has been ach iev ed w ith the 300-|.im stripe b road-arca LD o f C h ap ter 3.
106
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Si:CPW -PCS
scope
I0X
N d:glass
Laser
PRF = 500 Hz
F igure 6.1 N d'.glass laser experim en tal setup used to m easu re St | o f hy b rid
o pto elctro n ic attenuator. L aser pulse energy = 20 p J ./J f = 500 M H z at + 18 dB m .
F igure 6.2 O ptoelectronic attenuator d ata taken w ith N d :g lass laser sy stem o f
F igure 6.1. V\ - 140 m V . Ho = 40 m V , => [S i] 1= 10.9 dB.
107
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
§6.2 Broad-Arca Laser Diode M easurements
W e m easured the optoelectronic atten u ato r perform ance using the 3 0 0 -p m stripe b ro ad -area LD d escribed earlier in C h ap ter 3 as the laser so u rce in the hy b rid
o p to electro n ic attenuator (F igure 6.1).
W hile w e perform ed the sam e ex p erim en t
d escribed in the last section, the optically induced r f atten u atio n o f a 500-M H z signal
prop ag atin g through the S i:C P W -P C S w as m easured.
A h ig h -v o ltag e 5 0 -Q p u lser
[120] w as used to driv e the 3 0 0 -p m LD. F igure 6.3 sh o w s a typical exam ple o f the
m easured
attenuation,
w here
both
the
5 0 0 -M H z
w aveform
(top
trace)
and
co rresp o n d in g LD current pulse (bottom trace) are d isp lay ed ; an attenuation o f 7 .9 dB
is show n.
N ote that the atten u ato r response roughly follow s th e LD pulse; th u s, the
S i:C P W -P C S m ay also be suitable as an o p tically co n tro lled r f m odulator.
F igure 6.4 show s the m easured atten u atio n p lo tted as a function o f LD peak
pow er. N otice that the attenuation appears to level o f f at aro u n d 6 dB for a LD p ow er
near 50 m W . U nfortunately, the m axim um atten u atio n m easured w ith the LD w as not
m uch b etter than that m easured w ith the N d:glass laser system , w h ich was a ro u n d 10
dB ; h o w ev er, a sign ifican t reduction in ttie laser d riv er (in term s o f cost, co m p lex ity ,
and size) has been achieved w ith the use o f a sin g le b road-area LD operating at m o d est
o u tp u t pow ers.
U sing the optical O -sw itching tech n iq u e d escrib ed earlier in Section 3.4 [38].
w e o p tically O -sw itched the sam e 300-p m . tw o -sectio n LD in an attem pt to increase
the atten u a tio n by increasing the LD o u tp u t pow er.
U n fo rtu n ately , the m easu red
atten u a tio n w as less than 1 dB w ith this tech n iq u e ( IS21 I = 0-65 dB ). This p o o r result
is, again, partly due to the short optical pulse w idth ( - 6 0 ps) co m p ared w ith th e long
m e asu rem en t tim e (1 p s), w hich is a d ifferen ce o f six o rders o f m agnitude. T h u s, the
actual attenuation is m o st likely m uch larger than 0.65 dB.
A lthough th is resu lt is
108
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
unim pressive, it is the first reported PC sw itch m easu rem en t on such a ^ -s w itc h e d
laser diode system .
In an effort to im prove this result, w ork is co ntinuing in our
laboratory to develop an all-electrical O -sw itching sch em e to im prove this resu lt for
m ore general PC sw itch in g applications.
The next LD to be used in the o p to electro n ic atten u ato r w as a 1-cm linear
A lG aA s LD array provided by A ryc R osen o f D avid S a m o ff R esearch C enter.
P rinceton, N J [37], T he array peak o u tp u t p o w er w as 38 W at a w av elen g th o f 804
nm .
U nfortunately, the laser driver m ust d eliv er ap p ro x im ately 40 A o f c u rre n t to
ach iev e this output p o w er level.
Using this LD array and a su itab le voltage p u lse r
[120] in p lace o f the laser source o f Figure 6.1. w e ach iev ed 26 dB o f r f (500 M H z)
attenuation for an optical pow er o f 5.35 W (see Figure 6.5). T he LD pulse w id th for
these m easurem ents w as 3 p s. The m easured atten u atio n as a function o f LD array
peak pow er is show n in Figure 6.6. A gain, the r f p ow er and frequency w ere +18 dB m
and 500 M H z, respectively. N otice that o n ce again the m easured atten u atio n appears
to level o f f as the optical pow er is increased.
Indeed, 20 dB o f the m ax im u m
attenuation o f 26 dB w as achieved w ith ap p ro x im ately I W o f LD pow er.
T h ese results can be greatly im proved since the sp o t size on the S i:C P \V -PC S
w as approxim ately 100 p m x 1 cm . and thus a sig n ifican t fraction o f the laser p o w er
w as w asted on both the sw itch contacts and on the su b strate o u tsid e o f the sw itch in g
gap area. H ow ever, both gaps in the S i:C P W -P C S stru ctu re w ere illum inated w ith this
LD array an d . since these gaps are electrically co n n ected in parallel, the effectiv e onstatc resistance m ay be quite low.
T hus, this type o f illum ination m ay be the m ost
suitable for the reflective attenuation schem e; how ever, uniform illu m in atio n o f both
PC sw itch gaps has been used by others [117] w ith rather d isap p o in tin g results (less
than 2 dB o f attenuation w as reported).
109
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Figure 6.3 O ptoelectronic attenuator d ata w ith 3 0 0 -g m broad-area LD.
T op trace: 500 -M H z signal envelope sh o w in g 7.9 dB o f attenuation.
B ottom trace: LD current pulse w aveform . N ote: response appro x im ately follow s
laser pulse shape.
12
10
8
6
4
i
0
0
50
100
150
P
200
250
300
(mW)
LD
Figure 6.4 M easured attenuatio n versus incident LD p o w er using
300-p.m broad-area LD o f C hapter 3. N ote that atten u atio n levels o f f at ~ 1 1 dB.
110
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
O ne
im portant question
rem ains:
W hat co n trib u tio n to
atten u atio n d o es the r f source pow er itself m ake?
the
m easured
F ig u re 6 .7 show s th e o p tically
induced atten u atio n for a fixed LD pow er o f 5.35 W (i.e., m ax im u m o p tical pow er o f
F igure 6.6) a s a function o f r f pow er. N ote that the p erfo rm an ce is ro ughly linear w ith
a total v ariatio n o f less than 7 dB for an r f p o w er v aria tio n o f th ree o rders o f
m agnitude.
T his show s that, although the r f p ow er level affects the atten u ato r
p erform ance, the contribution is m inim al com pared w ith th a t o f the laser source.
In
addition, for a m icrow ave system operating w ith a d ecad e variation in pow er, the
o p to electro n ic atten u a to r perform ance variation w ith r f p o w er w ould be less than 2 dB.
Ill
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Figure 6.5 1-cm LD array o p toelectron ic atten u ato r d ata sh o w in g 5 0 0 -M H z signal
en v elo p e durin g laser activation. D ata sh o w an atten u atio n , 1 j I . o f 26 dB.
30
25
20
CQ
-a
-2 1
15
< -1
—
10
5
0
0
1
2
3
4
5
6
^ (W )
Figure 6.6 M easured attenuation versus in cid en t LD p o w er using 1-cm LD array.
N ote atten u atio n levels o ff a t - 2 6 dB for - 5 W o f LD p o w er; r f p o w er = +18 dB m .
112
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
30
25
20
15
10
0
10
o
0
10
15
20
/ 5rf (dBm)
F igure 6.7 M easured attenuation versus r f p o w er using 1-cm L D array.
N ote that atten u atio n is approxim ately linear w ith r f pow er. O ptical p o w er = 5.35 W.
113
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
§6.3 Fibcr-Pigtailcd Laser Diode Experiments - RF Voltage M easurem ents
In the last section w e saw how sem ico n d u cto r LD s can be used to o ptically
induce 26 dB o f r f attenuation w ith a laser peak p o w er o f ~5 W. A s w as o bserved, the
1-cm LD array output energy w as spread o v er such a large area (100 p m x 1 cm ) th at
m u ch o f the av a ila b le energy did not illu m in ate the PC sw itch gaps an d thus w as
w asted.
In the next tw o sections w e show how a co m m ercially availab le fiber-
p igtailcd LD can be used to greatly im prove the hybrid atten u a to r p erfo rm an ce w hile
sim u ltan eo u sly reducing the LD peak p ow er req u irem en t from several w atts to less
than 150 m W .
T his section p resents r f voltage m easurem ents m ade w ith a 1-G H z
bandw idth oscillo sco p e, w hich are identical m easurem ents to those m ade earlier.
In
the next section, attenuator m easurem en ts m ade w ith a m icrow ave V ecto r N etw ork
A n aly zer (V N A ) are described. U sing the V N A . w e d eterm ine the e x a ct nature o f the
attenuation m echanism (either reflectiv e, absorptive, or possibly a co m b in atio n o f
both).
Figure 6.8 show s the experim ental setup used to m easure the S i:C P W -P C S
h ybrid
o ptoelectronic
attenuator perform ance
with
the
fib er-p ig tailcd
LD.
A
com m ercially available fiber-pigtailed LD [118] has replaced the o th e r laser sources
show n in F igure 6.1 an d described in Section 6.2. For these r f voltage m easu rem en ts
the LD w as o perated in pulsed m ode w ith a pulse w idth and rep etitio n frequency o f
200 ps and 500 Hz. respectively. U sing a 10x m icroscope o b jectiv e to achieve a 10p m spot size on a single S i:C P W -P C S gap. w e once ag ain o b serv ed a leveling o ff o f
the attenuation as a function o f LD pow er, as show n in Figure 6.9.
T h e r f source
frequency and pow er w ere 500 M H z and +18 d B m . respectively.
N otice that the perform ance is still lim ited to about 10 dB even for this
im proved LD beam sp o t size, o nly this tim e w e have reduced the req u ired LD p ow er
114
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
to less than h a lf a w att (375 m W ).
I f the perform ance is effectiv ely "satu ratin g "
because o f th e high optical intensity in cid en t on th e sw itch, then e x p a n sio n o f the
beam w ith resp ect to the C P W -P C S geo m etry (i.e.. lengthw ise alo n g th e co p lan ar
contacts) should show an increased level o f attenuation.
A cylindrical lens w as used to expand the beam to 10 pm x 2 m m . an d the
resulting attenuation did in fact increase, as is also sh o w n in F igure 6.9.
The
m axim um atten u a tio n for th is optical arrangem ent was ~ 30 dB . w hich is a 20 dB
im provem ent ov er the case w here only the m icroscope o b jectiv e w as used. T h u s u'e
have proven th e optoelectronic atten u ato r concept u sin g all sem ico n d u cto r-b ased
devices; it rem ains to see w hich attenuatio n m ode is involved.
B efore th is point is discussed, the afo rem en tio n ed saturation effect is w orth
com m enting on.
Since the rf frequency is 500 M H z,
the guide w av elen g th
(Xj, = )Jn, w here n = 3.42 is the refractive index o f silico n ) w ithin the S i:C P W -P C S is
approxim ately 17.5 cm . S ince this is m u ch larger than the S i:C P W -P C S its e lf (1.6 cm
total length) it is hard to im agine that the increased atten u atio n m easu red w ith the
expanded beam is due to any distributed resistance effects. I f the co rrec t atten u atio n
m echanism is p lasm a absorption, then it is also difficu lt to see ho w in c reasin g the
plasm a length w hile sim ultaneously d rastically reducing the p lasm a d en sity could
yield an increase in
(the LD po w er w as kept constant). O ne o th e r p o ssib ility is a
decrease in the photovoltaic response w h en the LD beam is expanded; a lth o u g h som e
evidence in the literature suggests that this m ay be a p o ssib ility for o p tically co n tro lled
M E SFE T s [63]. we have n o t attem pted to incorporate this effect into o u r device
m odel. In addition, the lack o f a reverse-biased ju n ctio n in the S i:C P W -P C S (th ere is
a slig h t p ^ / p ju n c tio n due to th e ion im planted region) im plies th at th is satu ratio n
effect is m ost likely not due to photovoltaic effects.
115
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
It is o b v io u s from eq uation (2.4) th at the satu ratio n b eh av io r is m ost lik e ly not
a c irc u it effect; if Ron tends to zero, S ^ \ goes to m in u s infinity.
C learly, this is not
indicative o f a saturation effect, w hich w ould cau se S i] to ap p ro ach a co n stan t value.
H ow ever, the S i:C P W -P C S has a finite value o f c o n tac t resistance R c, w hich w o u ld
lim it the m in im u m value o f Ron to Rc ; thus S ji w ould approach a co n stan t valu e in
this
case
and
the observed
saturation
effect
m easu rem en ts on a cu rvc-tracer indicate th at
m easu rem en ts m ust be perform ed to
w o u ld be ex p ected .
R c is n eg lig ib le,
d eterm in e the
A lth o u g h
m ore
actual value.
accu rate
Itisth erefo re
b elieved that the observed saturation effect is d u e to a lim itin g o f the m in im u m
perm issib le o n -state resistance, or conductance, by the finite S i:C P W -P C S Rc.
U sing the sam e laser setup
as Figure 6.8
and p erfo rm in g an id en tical
m e asu rem en t to those m ade in S ections 5.1 and 5.2 to m easure th e G aA s:C P W -P C S
o n -state resistan ce (?on, we m easured the v ariatio n o f A ) as a fu n ctio n o f /?on. F ig u re
6.10 co m pares the m easured value o f S i | v ersu s Ron along w ith the th eo retical
pred ictio n d escribed in S ection 2.2 for the reflectiv e atten u ato r schem e.
N o tic e th at
the m easured attenuation is approxim ately 5 dB g reater than th e pred icted v a lu e over
m ost o f the ran g e o f R on. C learly, either the reflective atten u ato r m odel is in n eed o f
som e refinem ent, o r the absorptive atten u ato r sch em e is valid.
Since w e could not m easure the reflected p o w er accurately to d eterm ine w hich
m o d e o f o p eratio n is correct, a V N A w as used to do the m easu rem en ts (as w ill be
d escrib ed
in section 6.4).
If the
reflective atten u ato r sc h em e
isthe co rrec t
in terp retatio n o f the o p to electronic atten u ato r resu lts presented in this section, th en as
the o p tic ally induced attenuation increases, the p o w er reflected from the o p to electro n ic
atten u a to r should also increase by roughly an eq u iv alen t am o u n t;
sim u lta n eo u sly m easure
I S o il a °d
i^ lM
use o f a V N A to
sh o u ld indicate i f this in terp retatio n is
correct.
116
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Si:C P W -PC S
sco p e
10X <
C ollim atin g lens
SDL fiberpigtailed LD
t
M u lti-m od e fiber (1 m)
. ■= 2 00 us
pulse
PRF = 500 H z
X
= 800 nm
F igure 6.8 F ibcr-pigtailed LD experim en tal setu p used to m easure S t j o f hybrid
optoeletronic attenuator; r f frequency = 500 M H z at + 18 dBm .
,l spot = l 0 p m x 2 m m
10 pm x 10 pm
0
50
100
150
200
250
300
350
400
/^ (m W )
F igure 6 .9 M easured attenuation versus in cid en t LD p ow er w ith the fiber-pigtailed
LD. N ote; IS t J
= 10 dB fo rz lspot = 10 p m x 10 p m . and increases to 30 dB w hen
the spot is expanded to 10 p m x 2 m m ; / rp = 500 M Hz.
117
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
35
30
10 p m X 2 mm
10 pm X 10 pm
25
Theory
—
20
15
10
0
0
100
200
300
R
400
500
600
(Q )
on
F igure 6.10 T heoretical and experim ental co m p ariso n for the reflective atten u atio n
m echanism . N ote: m easured S^i > prediction => reflection m ode m ay be incorrect.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
§6.4 Fibcr-Pigtailcd Laser Diode Experiments - VNA M easurements
T he lim its o f the r f transm ission m easu rem en ts d escribed in the previous
section can be sum m arized w ith tw o observations: first, the atten u ato r w as desig n ed to
o p erate at 3 G H z. not 500 M H z. and. second, the reflected p ow er from th e S i:C P W PCS has yet to be accurately m easured as a function o f LD pow er.
U sing a V ector
N etw ork A nalyzer (V N A ) [119] loaned to us by A n ritsu /W iltro n , Inc., we have
characterized the perform ance o f the optoelectronic atten u ato r. T hese m easu rem en ts
show , once and for all, that the correct atten u atio n m echanism to d escrib e the
o p toelectronic attenuator is absorptive (i.e.. the o p tically induced so lid -state plasm a
attenuates the p ropagating m icrow ave signal).
Figure 6.11 and 6.12 show the V N A d ata taken w ith th e experim ental setup o f
the last section (F igure 6.8) w here the r f source and o scillo sco p e have b een replaced
w ith p o n s 1 and 2 o f the V N A . respectively. The lOx m icro sco p e o b jectiv e w as again
used to form a 10-pm laser spot on a single gap o f the S i:C P W -P C S , as d escribed in
S ection 6.3.
T he fiber-pigtailcd LD w as ag ain used as the optical source, only this
tim e it w as operated in C W m ode to perm it VNA m easu rem en ts to be m ade. The first
m easurem ents m ade show ed, for identical LD peak pow ers, th at the m easured pulsed
rf voltage and C W m icrow ave values o f S’t ]. both m easured at 500 M H z, were
identical.
In addition, a spectrum an aly zer w as used to m easure the level o f
attenuation, w ith the result also being in agreem ent w ith the pulsed an d C W S^i
values. F o r com parison, a plot o f the m easured attenuation versus incident LD pow er,
at 500 M H z and m easured w ith the V N A . is show n in F igure 6.13. A gain, the L D spot
w as expanded by the cylindrical lens, and the observed attenuation increased as before.
N ote that the m easured attenuation o f Figure 6.11 is greater in th e m icrow ave
frequency range ( J > 1 G H z) than it w as in the V H F range (500 M H z). Indeed.
119
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
20.1 dB o f atten u a tio n w as m easured at the d esig n frequency o f 3 G H z w ith th e lOx
m icroscope objectiv e. M ore im portantly, the m easured reflected pow er. Sj [. actu ally
im proves (becom es m ore negative) as the L D p o w er is increased, as can be seen in
F igure 6.12. T h is tells u s that the atten u atio n schem e is. in fact, ab so rp tiv e, since an
increase in the reflected pow er is not observed as the level o f atten u a tio n w as
increased.
As the beam w as spread by the sam e cylindrical lens as before, tw o effects
w ere observed. First, the attenuation, as w ith the rf voltage m easu rem en ts m ad e w ith
the sam e laser in the pulsed m ode, increased as the beam w as ex p an d ed (see Figure
6.14 and 6.15).
T he second effect is that th e atten u atio n b eco m es narro w -b an d (i.e..
peaks at a particu lar frequency), w hich is the expected resu lt for an atten u atio n th at is
based on a p lasm a ab so rp tio n m echanism [122],[123]. T h is resu lt occu rs b ecause the
atten u atio n is m a x im u m at the m icrow ave frequency th at is equal to the plasm a
frequency, cop. T h u s, we have a d irect m easure o f cop from o u r d ata. S ince the p eak in
the absorption m o v e s to low er frequencies as the LD p ow er is increased (and hen ce the
plasm a density is increased), w e can control the ab sorption level and m ain tain eith er a
w ide-band atten u a tio n characteristic o r create a m o re n arro w -b an d response. It should
be noted that a sh ift in the absorption peak w ith plasm a d en sity is also expected
[122].[123] and is therefore not surprising.
In o rd er to m easu re the m axim um atten u atio n th a t co u ld be ach iev ed w ith the
10 pm x 3.5 m m spot o f Figure 6.15. we increased the laser p o w er to 143 m W and a
peak a tten u a tio n o f 45.6 dB w as m easured a t 1.7 G H z (see F ig u re 6.16). T h ese data
show that w e have truly dem o n strated a very useful m icro w av e control sch em e during
th is research, especially since the attenu atio n is absorptive, so th a t m atch in g schem es
for lim iting the deleterio u s effect o f pow er reflections b ack to the source can be
avoided.
120
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
W hat rem ains at this point is to rigorously m odel the V N A ex p erim en tal
results so th a t w e can achieve a clear und erstan d in g o f ho w the o p to electro n ic
atten u a to r w orks. T o date, on ly qualitativ e ag reem en t w ith th eo ry has been achieved
(fo r exam ple, see F igure 2.5); a m ore co m p lete theoretical d ev elo p m en t is clearly
required if th e opto electro n ic attenuator is to b e im proved an d /o r inserted into a
m icro w av e system .
121
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Marker - [T]
Fieq = 500 MHz
S ,j
= -1 1 .2 dB
M aik cr- d
Freq = 1 GHz
S ,, = -1 2 .7 dB
-18
^LD
= 270 mW
^
..........f j
r
M arker - d
S„. = -2 0 .1 dB
F igure 6.11 V N A optoelectronic attenuato r d ata w ith fiber-pigtailed LD (C W m ode).
/l Spot = 10 p m x 10 p m , LD p ow er as indicated. Note: Sn\ co m p arab le to pulsed
oscilloscope data. M axim um 1Sti I = 21.1 dB a t 3 G H z.
Maikcr -
SyslcmCalibration Error
[Tj
Fieq = 5 0 0 MHz
S
1
° mWJ
oa
T3
M a rk e r-d
7 m W ^
= -16
Frcq = 1 GHz
S
= -1 0 .4 dB
P. „ = 270 m w
M ariter- d
3
Freq
S .,
/(G H z)
= -1 1 .7 dB
= 3 GHz
= -1 3 .6 dB
F igure 6.12 V N A optoelectronic attenuato r d ata w ith fiber-pigtailed LD (C W m ode).
M easure o f reflected pow er is
| . N ote 15] | I = 10.4 dB at 1 G H z (d a rk value)
and im proves w ith increasing optical intensity. Spot size sam e as Fig. 6.8 .
122
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
30
25
A
= 10 p m x 3 .5 mm
20
CQ
-a
15
10
10 (m ix 10 pm
0
0
50
150
100
P
200
250
300
(raW)
LD
Figure 6.13 V N A m easured atten u atio n at 500 M H z versus in cid en t LD p o w er w ith
the fiber-pigtailed LD. N ote: I
I = 10 dB for 10 p m x 10 p m laser sp o t, and
increases to 22 dB w hen the spot is expanded to 10 p m x 3.5 m m . N ote d ata sim ilar to
Fig. 6.9 and show s that C W LD V N A m easurem ent ap p ro x im ately equal the p ulsed
LD r f voltage m easurem ent.
123
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Marker - [j]
-10
...........
: -------13
-14
ly m W 7^
-18
Frcq = 500 MHz
S
^
ST -22 ............■—
3
ri -26
= -2 2 .0 dB
Marker - (2]
C1*-J
O
Frcq = 1 G Hz
S
^ r ^ 5 6
-34
/ >LD = 227 mW
"■
0.5
1.5
= -28.9 dB
mW
98 mW :
M arker- [3]
-38
Frcq = 3 GHz
= -1 5 .9 dB
/(G H z)
Figure 6.14 V N A O ptoelectronic attenuato r d ata w ith fiber-pigtailed LD (C W m ode),
•'•spot = '0 gm
2.5 m m . LD pow er as indicated. N ote: 5 i | has increased vs.
10 x 10 p m spot. M axim um l-Sni I = 28.9 dB at 1 GHz.
M arker- jj]
-1 0
20 mW
-15
-20
ffl
-o
"a
U
Frcq = 5 0 0 M Hz
5
= -22.3 dB
36 mW
-25
Marker - [2]
-30
Frcq = 1 GHz
-35
5
........../> = 143 m W A
' LD
\
-40
**N^/* 65 m w **
-45
0
0.5
1
2.5
1.5
M arker- [|]
Frcq = 3 GHz
S ..
/(G H z)
= -29.4 dB
= -1 8 .1 dB
F igure 6.15 V N A O ptoelectronic attenuator d ata w ith fiber-pigtailed LD (C W m ode),
••spot = 10 p m x 3.5 m m spot size used. LD pow er as indicated. N ote: 1
i
has
r* I
increased further. M axim um liS-ii
?2 I !I = 29.4 dB at 1 G H z.
124
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
CM 4 - BM
W F . PlW E
0 .0 0 0 0 Ml
521 FORUflRO TRflNSniSSION
LOG tWO.
►R£F'-31.000dB
nflRKER 2
b . ooojb / o iu
1.700034 GHz
-45.590 <fi
flflRKER TO HflX
►nflRKER TO m u
1
I
0 .5 0 0 0 1 0 Ota
- 2 1 .8 2 3 JB
8
3 .0 0 0 0 8 0 Ota
0.100002
F igure 6.16 V N A O ptoelectronic atten u ato r d ata w ith fiber-pigtailcd LD (C W m ode).
/l Sp0t = 10 p m x 3.5 m m , LD p o w er = 143 m W .
M axim um |S->i I = 45.5 dB at 1.7 G H z.
125
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
§6.5 Sum m ary
A n optoelectronic tech n iq u e suitable for the co n tro l o f m icro w av e integrated
c irc u its has been d em onstrated. U sing a c o p lan a r w av eg u id e-p h o to co n d u ctiv e sw itch
fabricated on silicon (S i:C P W -P C S ), we h av e been able to v ary the atten u atio n o f
m icro w av e signals by up to 45 dB for a laser p o w er o f less th an 150 m W .
An
o b serv ed satu ratio n o f the m icrow ave atten u atio n , as a fu nction o f laser intensity, w as
observed .
P roper LD
beam
profile m atch in g , u sin g
optical
beam
ex p an sio n
te ch n iq u es, w as su ccessfully used to optim ize the atten u a to r p erform ance. In ad dition,
a 300-(.im -stripc b road-area laser diode, fab ricated in o u r laboratory an d d isc u sse d in
C h a p te r 3. w as also used to successfully activ ate the atten u ato r; up to 10 dB o f rf
a tten u a tio n at 500 M H z w as achieved.
S ilicon w as found to be the m ost su itab le m aterial for the o p to electro n ic
atten u a to r because its long PC carrier lifetim e [67] w hich p erm its a PC gain [99] to be
realized in this m aterial, as described in C hapter 4.
o p to electro n ic attenuator, d ata in the literature
For an all-G aA s based
[45].[100]
in d icate th at sim ilar
p erfo rm an ce to the S i:C P W -P C S can be achieved w ith the G aA s:C P W -P C S by
su p e rla ttice techniques.
126
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
C hapter 7
Silicon Carbide PC Switch Development
S ilicon a n d G aA s are not the o nly sem ic o n d u cto r m aterials su itab le for the
optoelectronic attenuator. S ilicon C arbid e (SiC) is also p o ten tially suitable since this
is a sem ic o n d u cto r m aterial w ith a w id e band-gap en erg y o f ~3 eV . A s such. SiC is
suitab le for tw o applications w here G aA s and Si ca n not co m p ete - h igh tem p eratu re
and high pow-er. Due to th e m aterial’s large therm al co n d u ctiv ity and h igh break d o w n
elec tric field, w hich are bo th approxim ately ten tim e s greater th an that o f G aA s. this
m aterial m ay be ideally suited for the o p to electro n ic atten u ato r application.
It w as
w ith this in m in d that we conducted ex ten siv e research to determ ine the feasibility o f
using the 6 H S ilicon-C arbidc (6 H -SiC ) polyphase for PC sw itch in g applications.
A lthough the optoelectronic potential o f silic o n carb id e (SiC ) has been know n
sin ce 1907 [124], only recently has electro n ic d evice grade SiC been available [125],
In this chap ter we describe how the opto electro n ic p ro p erties o f 6 H -SiC w ere
investigated u sin g lateral and vertical photo co n d u ctiv c sw itches. W e rep o rt the
m easurem ent o f photovoltaic and p h oto co n d u ctiv c effects for both g eo m etries and at
several w avelengths near the 6 H -SiC absorption edge. T he PC carrier lifetim e in ptype 6 H -SiC w as m easured, w ith surface and b ulk PC carrier lifetim es d eterm in ed to
be 4 0 and 200 n s. respectively [50]. A lthough th e dev ices po ssess dark resistan ces on
the order o f 10 Q , the sw itching efficien cy o f th e vertical sw itch es appro ach ed 32
percent, w hile th e resistance o f the lateral devices co uld b e red u ced by 50 p ercen t w ith
200 (iJ o f laser radiation a t X - 337 nm.
In ad d itio n , w e m easured photo co n d u ctiv ity in the vertical sw itches w ith a
d ev ice static po w er dissipation exceedin g 11 W. A lth o u g h th e sem ico n d u cto r g low ed
127
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
from the high level o f dc pow er being dissipated, only the sw itch m ount w as dam aged.
T his is further p ro o f that 6 H -SiC is indeed a high-tem perature o p to electro n ic m aterial.
In S ection 7.3.1, the results o f a detailed stu d y o f the photo v o ltaic (PV )
properties o f 6 H -S iC are presented. T he p h o tovoltage response, as a function o f laserbeam position and laser w avelength (near the 6 H -SiC ab so rp tio n edge), w ere
m easured. A n efficien t PV effect (a 40-m V signal w as generated using a 2 0 0 -p J pulse
at 337 nm ) w as m easured using a nitrog en laser w ith a 7-ns p u lse w idth. In addition,
this m aterial d isplayed a high-speed PV response to pico seco n d laser excitations:
m easurem ent-lim ited sub-nanosecond PV response tim es w ere observed for laser
photon energ ies less than the 6 H -SiC band gap energy. The photo v o ltaic resp o n se, as
a function o f laser w avelength and beam spatial p o sitio n w ithin the sw itching gap, was
m easured, alo n g w ith the PC carrier lifetim e and optical ab sorption coefficient.
The
data show that the m easured photovoltage is a sensitive function o f both spatial
positio n and optical absorption depth.
H ypothetical arg u m en ts are p resented that
qualitatively ex p lain the observed PV effects.
The sw itch response to picosecond laser pu lses in the U V . violet, green and red
spectral regions are show n to have su b -nanosecond ph o to v o ltag e response tim es.
Finally, since the optical absorption coefficient has not been well estab lish ed for
dev ice-g rad e 6 H -SiC . the optical absorption co efficien t near the 6 H -S iC band gap
energy ( £ g =3 eV ) w as also m easured.
For these sam ples the m easured value o f
w as ap p ro x im ately 3.1 eV.
T he PC m easurem ents w ere m ade w ith both d evice g eom etries w ith the
sw itches placed in a low -im pedance dc circuit, w hich w as desig n ed for m axim um
sensitivity.
T h ese m easurem ents are d escribed in section 7.4.
U sing lateral PC
sw itches and above-band-gap radiation, the PC carrier lifetim e at the SiC crystal
surface w as m easured and found to be approxim ately 40 ns. U sing vertical sw itches
128
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
an d belo w b and gap radiation, the bulk SiC PC ca rrier lifetim e w as d eterm in e d to be
ap p ro x im ately 200 ns for this m aterial.
In both cases, th e PC response sh o w ed a
double-ex p o n en tial decay, with both a fast co m p o n en t and a slo w er co m p o n en t, w hich
w e attrib u te to direct electron-hole recom bination an d re-em issio n o f trap p ed charge,
respectively. T his sam e behavior w as observed by O k u m u ra et a l [126] in 3 C -S iC .
T he prim ary technological barrier to the use o f SiC as a PC sw itch m aterial is
the low su bstrate and epitaxial layer resistivity v alues attainable to date. U sin g deeplevel tran sien t spectroscopy (D L T S), and p-n ju n c tio n diodes provided by N A SA
L ew is R esearch C enter, m easurem ents w ere m ade at O ld D om inion U n iv ersity in an
attem p t to identify both shallow and deep-level electronic im purities in 6 H -S iC .
To
date, w c have identified w hat we believe to be a fairly d eep-level im p u rity w ith an
activ atio n energy o f 0.58 eV.
§7,1 M aterial Properties of 6H-SiC
A lth o u g h 6 H -S iC is a relatively new m aterial as far as clcctro n ic-g rad e
substrates are concerned, considerable research into th e m aterial co n stan ts and grow th
o f silicon carbide substrates and epitaxial film s hav e been reported in th e literature
[ 127]—[ 130], In particular, invited review articles by D avis et a l [46] and T rew el al
[47]
discu ss
m ost o f the
relevant
issues
relatin g
to
this
n ew
an d
exciting
se m ico n d u cto r m aterial.
A s a consequence o f this extensive coverage in the literature, w e w ill not dw ell
on the num erous m aterials issues, but rather w e w ill sim ply com p are th e relevant
properties o f SiC w ith those o f G aA s and silico n , and p o in t o u t th e im portant
129
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
differences. T ab le 7.1 below com pares the relevant m aterials properties o f Si, G aA s
and 6 H -SiC .
T a b ic 7.1 C o m p a ris o n o f S iC P ro p e rtie s to Si a n d G aA s
P r o p e r ty
Si
G aA s
6 H -S iC
B and-gap en erg y (eV)
1.1
1.4
3.0
M axim um o p era tin g tem p (K)
600
760
1.580
M elting p o in t (K )
1,690
1.510
2 . 100+
B and-gap type
Indirect
D irect
In d irect
E lectron m o b ility (cnvVV-s)
1.400
8.800
400
H ole m obility (cm : /V-s)
600
400
40
B reakdow n field ( 10s V/cm )
J
4
4 0 -6 0
T herm al conductivity (W /cnvK )
1.5
0.5
5
Sat. electron d rift v e lo c ity (10 7 cm /s)
1
2
2.5
D ielectric constant
11.8
12.8
10.0
+ sublim ation o c c u rs at this tem perature.
T he m o st im p o rtan t m aterial
properties o f S iC . G aA s. and Si
for the
o p to e lec tro n ic atten u a to r application are the electron and hole m o b ilities an d satu rated
d rift velocity. From T ab le 7.1 we see that 6 H -SiC has lo w er carrier m o b ility th an Si
or G aA s. but a higher saturated d rift velocity, w hich w o u ld no rm ally be a d isa d v an tag e
for an y se m ic o n d u cto r m aterial. H ow ever, since the m ax im u m op eratin g elec tric field
for 6 H -SiC is appro x im ately ten tim es g reater than fo r G aA s. the lo w er m o b ility
values are offset by th e higher saturated velocity.
130
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
A s far as therm al b eh a v io r is concern ed , clearly 6 H -S iC is su p erio r to both
G aA s and Si, with a therm al conductivity o f m o re than 3.3 tim es Si an d 10 tim es
G aA s.
T his poin t is also ev id en t from th e greatly increased m ax im u m o p eratin g
tem p eratu re o f 6 H -SiC . T h u s, for high-povver and ultra-fast ap p lica tio n s w h ere high
electric field reg io n s are p rese n t in PC sw itches, 6 H -S iC has an o b v io u s ad v an tag e
over both Si a n d G aA s. It w as w ith these p o in ts in m ind that w e set out to ex p lo re the
su itab ility o f 6 H -S iC as a high -sp eed and p ow er PC sw itching m aterial.
§7.2 SiC Photoconductivc Sw itch Design and Fabrication
P -type 6 II-S iC su b strates w ere selected for this inv estig atio n . T h ese su b strates
w ere not sem i-in su latin g , sin c e the bulk resistiv ity w as only 0.5 Q -cm . T h is p resen ts a
serious pro b lem for p h o toconductive sw itch in g d ue to the low value o f d ark resistance.
To circ u m v en t this problem , h ig h resistivity (22 Q -cm ) p -d o p ed ep i-lay ers w ere grow n
on a 1-inch w afer (see Fig. 7.1) [140], T h e epi-layer th ick n ess w as 20 jam. A highly
doped p ++
> 1 0 19 c m '3) lay er w as then grow n on top o f th e p ep i-lay er to achieve
oh m ic co n tac t to the m aterial. T he /?++ ep i-lay er w as th en reactive ion etch ed [131] in
/VF3 (w ith A1 used as the etch a n t m ask) so that the sim p le g ap o f the lateral sw itch
could be elec trica lly isolated. A l-Ti alloy co n tacts [132] w ere then form ed on top o f
the p ++ m esas w ith a sm a lle r (lO -p m undersized) co n tact pattern. A fter th e ohm ic
co n tacts w ere annealed, gold overlayers w ere dep o sited on to p and pattern ed w ith the
sam e m ask.
C urrent-voltage (/- V) m easurem ents m ade on a curve tracer d isplayed
o h m ic co n tact behavior.
T h e lateral sw itch dark resistance varied from 13 to 20 Q
across the w afer.
131
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T he lateral device contact geom etry used w as a m icrostrip line w ith a sim ple
gap, L , o f 10-pm , 0.5-m m o r 3-mm, as show n in F igure 7.1.
A g ro u n d p lan e w as
dep o sited to form a 5 0 -Q characteristic im pedance: how ever, this contact w as not
connected to circuit g round during these experim ents sin ce the device w as placed on a
piece o f M ylar film. T his w as done so th at the photocarriers w ould flow b etw een the
tw o m icrostrip contacts a n d not be shunted to ground b y the lo w -im pedance su b strate.
T he m icrostrip line w id th w as 465 pm .
For the vertical PC sw itches, a 170-f2-cm ep itax ial layer w as g ro w n on th e SiC
su bstrate (see Fig. 7.2). The epi-layer thickness w as 20 pm .
H ighly d o p ed p ++
(,VA > 1 x 1019 c m '3) layers w ere then grow n on top o f these ep i-lay ers to ach iev e
o h m ic contact to the m aterial.
A l/Ti alloy contacts w ere then pattern ed in identical
fash io n as the lateral sw itches, w ith the exception o f the co n tact geo m etry .
T he
co n tacts consisted o f a grid d ed top electro d e w ith a so lid bottom electro d e (F ig . 7.2).
T he gridded to p electrode w as designed for m axim um optical co u p lin g efficien cy ,
w h ich im plies a m etal-to-open-area ratio o f 1:1.
T h e solid b o tto m elec tro d e w as
recessed from the SiC ch ip edges to reduce leakage current.
A lthough only the top
electrode w as fabricated on a highly doped (p + + ) cpi-laycr. the lo w su b strate
resistivity yielded ohm ic contact, as indicated by I-V m easurem ents on a T ek tro n ix
576 curve tracer. T he d ark resistance varied from 8 to 10 Q for the vertical devices.
132
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
p ‘ (p = 22 £2-cm)
< 1 urn
6 H -S iC S u b s tr a te
/
p + (p = 0.5 £2 cm)
P ++ (N a = 1019)
Ground plane
Figure. 7.1 S chem atic view o f lateral PC sw itch design w ith sim p le sw itch in g gap o f
length L. T he overall device length is 1 cm .
Gridded top electrode
P ' e p i— •
(p = 170 £2-cm)
Solid bottom
electrode
6 H -S iC S u b s tr a te
F igure 7.2 Schem atic view o f vertical PC sw itch design sh o w in g the g rid d ed top
electrode. S olid bottom electrode w as p laced at the center o f the device to su p p ress
leakage currents o n the chip edges. D evice size 1 cm x 1 cm .
133
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
§7.3 Photovoltaic Experim ents Using Lateral Switches
T his section is divided into three su b -sectio n s due to the d iv erse nature o f the
m easurem ents that w ere perform ed to assess the PV response o f 6 H -SiC . B oth '‘near
b an d -g ap " and “ h igh-speed" picosecond m easu rem en ts w ere perfo rm ed u sin g two
d iffe ren t UV laser system s; hence, these ex p erim en ts are d iscu ssed in S ectio n s 7.3.1
and 7.3.2, respectively. D ue to the rath er unusual nature o f th e o b serv ed PV response,
a third section (7.3.3) has been included for d iscussion o f the PV data. S ince tim e did
not perm it experim ental verification o f the an aly sis that is p resen ted , it sho u ld be kept
in m ind that o u r analysis is only o f a q u alitativ e nature and is intended to form the
basis o f a m ore detailed analysis that is p lanned for the future.
7.3. / N ea r B am l-G ap M easurem ents
T he experim ental setup used to d eterm in e the PV resp o n se near the 6 H -SiC
band gap energy is show n in F igure 7.3. A U V /v io let d ye (E x cito n . Inc.. N o. PB B O )
w as pum ped w ith an X eC l excim er laser so that the laser w av elen g th could be tuned
through the 6 H -S iC absorption edge (385 nm
<
< 430 nm).
For referen ce, the
p u blished band-gap energy o f this m aterial is 3 eV . w hich co rresp o n d s to a photon
w avelen g th o f 414 nm . W e used the X eC l (A. = 308 nm ) laser o u tp u t to d irectly assess
the sw itch behavior w ell above band gap. T h e X eC l laser o u tp u t en erg y w as 5:18 pJ.
w ith a 15-ns pulse w idth and a 1-Hz rep etitio n rate.
T h e dye o u tp u t en erg y w as
m aintained at - 1 6 p J ± 10%, w ith a 5-ns pulse w id th and 1-Hz rep etitio n rate. The
d ye w avelength could be m easured to w ith in 1 nm.
F igure 7.4 show s the m easured PV response from a 10-pm gap lateral PC
sw itch w hen activated by the N 2 laser. A d ep en d en ce o f the PV response on th e laser
134
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
beam position w as observed.
Since the 1-mm sp o t size w as m uch g reater th an the
sw itch gap, the spot was positioned in eith er one o f three positions: co v erin g th e gap
(p o sitio n 2), or a t the edge o f one o f the m ctalized m icro strip lines (p o sitio n s 1 an d 3).
Inspection o f th e data show s that the PV response o f the 10-pm SiC PC sw itch is fairly
fast; the PV decay tim e for positio n s 1 and 3 is ap p ro x im ately 10 ns, as in d icated in
th e figure.
T h is data suggests th a t 6 H -SiC m ay be su itab le for h ig h -sp eed UV
d etectio n ap plications.
In o rd er to fully understand this h ig h -sp eed PV response, the sp o t size
rela tio n sh ip w ith respect to the sw itchin g gap m u st be b etter controlled.
U sin g the
sy m m etrical circu it o f Figure 7.5. a 3-m m gap lateral PC sw itch w as illu m in ated w ith
th e X eC l laser d irectly (X = 308 nm ) to assess the ab ove band gap (i.e..
/i V |a s c r
> E„ )
d ev ice perform ance. The laser spot size w as on th e o rd er o f 0.5 m m . A PV effect w as
observ ed w ith th e polarity o f the detected p h o to v o ltag e being d ependent on th e laser
beam position w ith respect to the sw itch gap and sco p e con n ectio n , as sh o w n in F igure
7.6. N ote that the PV signal polarity for laser beam p o sitio n s at X = 0 and 3 m m are
reversed. W hen the positions o f the m atched 5 0 -Q cable an d scope term in atio n w ere
reversed, the b eh av io r observed w as identical to that show n in the figure: th u s, the
sw itc h behavior is sym m etrical.
A lso sh o w n in F igure 7.6 is the PV resp o n se as a function o f laser p o sitio n
m easured w ith belo w band-gap radiation (X = 431 nm ). In this case, w e see th at there
are tw o m ajor d ifferences betw een th is d ata and th e 308 nm data: First, th e PV signal
never reaches zero value anyw here w ithin the sw itc h gap.
S econd, the PV resp o n se
co n tain s m ore structure, w ith both positive and n eg ativ e p o larity signal co m p o n en ts
present.
We w ill discuss som e proposed ex p lan atio n s for these o b serv atio n s in
S ectio n 7.3.3.
135
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Lateral PC Switch
Z = SOD.
(50 D )
F igure 7.3 S ym m etrical circuit used to m easure 6 H -SiC PV response. L aser sp o t size
s i m m . Beam positio n s 1, 2, an d 3 co rresp o n d to m easured v alu es in Fig. 7.4.
and R i are the oscillo sco p e and m atched loads, respectively.
40
Beam Position 1
20
Beam Position 2
0
-20
Beam Position 3
-40
0
25
50
75
100
T im e (ns)
F igure 7.4 PV resp o n se o f a 10-pm gap lateral sw itch. T he laser pulse w idth w as
7.7 ns (X = 337 nm ). Pulse decay tim e =10 ns. Beam positions d efin ed in Fig. 7.3.
136
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
/IV
L ateral PC Switch
scope
]
° "El
"scope
(SO Q )<
(SO Q)
z „ = so n
F igure 7.5 S ym m etrical circuit configuratio n used to m easure 6 H -S iC PV response.
L aser spot size w as <1 m m . X denotes the beam po sitio n w ithin the gap.
L eft e le c tro d e
R ig h t e le c tro d e
(0.000)
/
(0050)
(1.500)
q qso)
(3.000)
L iter
X ^*'P O < »l£C
( ~ 500 um )
^
I1 ( 0 3 4 5 )
(1.730)
430 nm
(2 905)
0
(o .o o n i
o.ooo)
L
X (m m )
3 000 )
F igure 7.6 PV signal w aveform dependence as a function o f laser beam po sitio n .
R esponse at both 308 nm (i.e.. h v >Eg) and 450 nm (i.e., h v <A'g) are show n. The
laser beam position in the sw itching gap is indicated in parentheses.
137
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T he peak p hotovoltage w as m easured as a function o f laser w av elen g th and is
show n in F igure 7.7. Due to the low signal am plitude, ten w aveform s w ere m easured,
then averaged to com plete this Figure. T he laser beam p o sitio n w as at A' = 0.0 m m
w ith a 0.5-m m spot size.
From the figure w e observe that the peak PV response
occurs at about 401 nm , w hile the peak optical output o f blue SiC L E D s occu rs at
approxim ately 472 nm [133], T his apparent difference b etw een the peak PV response
and the peak spectral output o f blue LE D s m ay be d u e to the differen t m aterial
structures used to fabricate the respective devices; p-n ju n c tio n dio d es for blue L E D s,
and p-p+ ju n c tio n s
here for the
^•absorption cocf. < ^ e m is s io n 's
lateral and vertical PC sw itches.
In addition,
typically observed w ith L E D s, w ith w hich o u r resu lts are
consistent.
T he optical absorption coefficient, a . o f p-type 6H -S iC w as also m easu red (sec
Fig. 7.8) using the X cC I/dye laser system via an optical transm ission m easurem ent.
The absorption coefficient w as calculated assum ing a refractive index o f 2.58. N ote
that a depends strongly on A. and thus the absorption d epth. T h is indicates that carrier
generation for photon energies above and b elo w E„ m ay occur in d ifferen t ep i-lay er
regions o f the device, due to respective changes in the optical p en etratio n d ep th s at
these different w avelengths. A s a result, there sho u ld b e a p o sitional d ep en d en ce o f
the PV response at w avelengths above and belo w £ g due to the device structure, and
Figure 7.6 appears to support this conclusion. A m ore d etailed discu ssio n o f h o w all
o f the SiC PV response data can be used to draw co n clu sio n s pertain in g to device
perform ance is presented later in Section 7.3.3.
138
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
7.3.2 H igh-Speed Picosecond M easurements
T he high-speed PV m easurem ents w ere m ade on a 3-m m g ap lateral PC sw itch
using the sam e setup as Figure 7.5, excep t that a pico seco n d laser sy stem w as used in
place o f the X eC l /d y e laser system [134],
Several p icosecond m easu rem en ts w ere
m ade a t the follow ing w avelength regim es:
UV (7. = 266 nm ), v io let (7. = 450 nm ).
green (7. = 532 nm ) an d red (7. = 705 nm). T he laser pulse w idths for the U V , green
and red w avelengths w ere =10 ps, w ith the green pulse w id th b eing =150 ps.
The
energ ies w ere =10, 20, 200 and 4 0 p J for the UV, violet, green and red w avelengths,
respectively. The laser spot w as centered at A' = 0.0 m m (sp o t size for all cases <1.5
m m ).
F igure 7.9 show s the resu ltin g PV response for these p icosecond laser
excitations.
T he observed photovoltage und er violet (7. = 450 nm ) e x c ita tio n is sh o w n in
F igure 7.9 (b).
N ote that there is a su b-nanosecond PV rise tim e follow ed by two
d istin ct decay tim es; a fast sub-nanosecond decay tim e follow ed by a slo w er 4-ns
decay tim e. For p h oton energies w ell below E z (i.e., for the green. Fig. 7.9 (c). and
red. Fig. 7.9 (d). laser excitations) the PV response rise tim e w as id en tical, w h ile only
the
su b-nanosecond
fall tim e
w as observed.
It sh o u ld
be
n o ted
th at
m easurem ents w ere lim ited by the 1-GH z b andw idth o f th e analog o scillo sco p e.
these
For
the case o f the p h o to n energy greater than £ „ (i.e., the U V ex citatio n ), the response
had the sam e sub-nanosecond rise tim e, but o n ly th e slo w decay tim e w as o bserved,
w hich w as =6 ns.
139
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
£
•&
>
-10
-20
380
390
410
400
420
430
X (nm)
Figure 7.7 A verage peak PV resp o n se o f lateral d ev ice versus X.
L aser beam p o sition: X = 0.0 m m . N ote: Peak PV resp o n se at 401 nm .
400
300
200
100
0
380
390
400
410
420
430
X (nm)
F ig u re 7.8 M easured optical absorp tio n co efficien t, a . as a fu nction o f photon
energy. M easu rem en ts m ade w ith X cC l-pum pcd U V /v io Iet P B B O dye.
140
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
X = 532 nm, E-mc ~ 200 p J
X = 705 nm , £ j nc * 40 |iJ
F igure 7.9 Picosecond PV response o f a 3-m m gap lateral 6 H -S iC PC sw itch.
V isible responses (c) an d (d) oscilloscope-lim ited; U V (a) and v io le t (b) rise-tim e
oscilloscope-lim ited, fall-tim e » 6 ns. Beam po sitio n (Fig. 7.6): X - 0.
141
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
7.3.3 Photovoltaic Experimental Discussion
O ur success in m easuring a significant high-speed PV effect in 6 H -SiC
indicates that efficient UV photo d etectio n at high tem peratures is p o ssib le [135]. The
very fast ca rrie r decay tim es (tdecay - I ns f ° r belo w band gap radiation) o b serv ed
during these experim ents indicate that th e U V detection is also high-speed.
The
central q uestion therefore is: w hat are the m ech an ism s responsible for th e o b serv ed
variation in photovoltage as a function o f laser w avelength and beam p o sitio n ?
For
insight, w e offer the follow ing qualitativ e argum ents. T hese argum ents m ust sh o w a
correlation betw een the dependence o f the PV response on carrier generation, both in
term s o f the spatial position o f the laser beam w ith in the gap and in term s o f the
optical absorption d ep th o f the laser p hoto n s w ithin the sem iconductor. T h erefo re, any
m odel o f the PV response m ust show q u alitativ e ag reem en t betw een the m aterial
structure and the device geometry'.
F igure 7.10. an SEM im age o f the contact region, show s th at the p ++ ep i-lay er
p rotrudes *5 pm from the A l-Ti allo y contact edge. F igure 7.11 show s the p ro p o sed
energy band diagram for the lateral devices.
U pon illum ination o f this region w ith
above band gap radiation, designated by h v \ in th e Figure 7.11, carriers are g en erated
near the A l-Ti alloy//?+ + intcrface. T his m etal-sem ico n d u cto r interface is m odeled as
a S chottky contact. A lthough curve tracer data show o h m ic-lik e contact b ehavior, the
ju n c tio n behavior in the presence o f optically-injected electrons results in a S ch o ttk y
con tact region (also the l-V m easurem en ts arc dom inated by the lo w resistiv ity
substrate, not the p ++ epi-layer; this also m akes th e co n tact look “oh m ic like").
T herefore, carriers generated at this ju n c tio n w ill b e separated by the built-in S ch o ttk y
potential, w ith electrons entering th e m etai and holes b eing sw ep t into th e bulk.
142
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
M easurem ent o f a v oltage pulse w ith a n egative p o larity (see F igure 7.6, 308 nm ,
A' = 0.0 m m ) supports this hypothesis.
T he polarity o f the m easured photo v o ltag e is a function o f w here one m easu res
the PV w ith respect to the position o f the laser spot, as the d ata in F ig u res 7 .4 and 7.6
indicate. T his can be explained using a sim ple charge neu trality arg u m en t, w h ereb y if
one assu m es that the sw itch gap m ust rem ain electrically n eutral, then an y ch arg e that
is sw ept o u t o f the sw itch gap and into the m etal m ust be co m p en sated by in jecting th e
sam e polarity charge at the oth er contact.
I f a negative v o ltag e p ulse is m easu red at
the left (i.e.. o scilloscope) contact, this im plies that a po sitiv e p u lse sh o u ld be
m easured at the other contact, since this contact m ust in ject electro n s for the d evice to
rem ain neutral:
therefore, a net positiv e charge is detected at that co n tac t d u rin g
electron injection. C areful inspection o f F igures 7.4 and 7.6 sh o w th at th is is. indeed,
the case and thus the charge neutrality arg u m en t appears to be valid.
W hen below -band-gap radiation is used, denoted by /iv2 in the F ig u re 7.11. the
effect o f the Schottky ju n c tio n is n eglig ib le due to the increased p en e tratio n d epth at
this w avelength: how ever, the p Jr+!p interface, and the p / p + interface m u st now be
taken into account. In this case the polarity o f the ob serv ed p h o to v o ltag e, due to the
built-in field o f the p ++!p interface, is reversed since electro n s w ill be sw ep t into the
bulk and holes will be sw ept into the m etal.
T he voltage pulse sh ap e from th is
interface w ill tend to follow the laser pulse since the carriers are g en erated o n ly 1 p m
from the m etal/ p ++ interface. Since the ab so rp tio n d ep th for /iv2 is fairly deep (sec
Figs 7.8 and. 7.11), the built-in potential o f the p lp + interface also co n trib u tes to the
observed photovoltaic response at these w avelengths.
F or th is ju n c tio n , the signal
polarity w ill be the sam e as that produced by the m etal/p ++ interface, w ith on e
prim ary exception.
S ince the generated carriers are now 20 p m from the co llectin g
m etal contact, a significant broadening o f the PV pulse shape w o u ld be expected.
143
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
S in ce the h o le m obility in SiC is on the order o f 40 c m 2/V -s, this co rresp o n d s to a
ca rrie r d rift tim e o f =1 p s (assum ing a b u ilt-in field o f 100 m V is d ro p p ed ac ro ss the
20 p m epi-layer).
Inspection o f the 430 nm w av efo rm at X = 0.0 m m in F igure 7.6
show s that th is ex p lan atio n is in qualitativ e ag reem en t w ith ex p erim en t; the long decay
tim e is on th e order o f a m icrosecond.
T he high-speed response can best be su m m ed up as follow s: sin ce a sig n ifican t
resp o n se w as m easured for photon energ ies w ell below the 6 H -SiC band gap energy,
one m ust conclude th a t the sem ico n d u cto r co n tain s m any im purity centers.
T hese
ce n te rs o b viously perm it efficient carrier generation to occur, and the lifetim es o f
p h o to -in d u c ed carriers th a t are related to these im purity cen ters arc ex tre m ely sh o rt. It
m u st be n o ted that, unlike silicon, the free carrier co n cen tratio n o f S iC m ay be m uch
less than 4 0 percent o f the actual d o pin g co n cen tratio n [136],
T h erefo re, th e fast
b eh a v io r m ay w ell be caused by the increased ionization o f acceptors (here A1 for ptypc d o p in g ) since an increase in m ajority carriers (i.e.. h o les) is cau sed by the optical
input.
U pon rem oval o f the optical ex citatio n , eq u ilib riu m is re-estab lish ed on
pico seco n d tim e scales. T his can be seen by lo o k in g at the hole cap tu re tim e co n stan t
o f the ionized acceptors, w hich is given by
tc = [ ' V lha c] - '.
(7.1)
w here ;VA is the doping concentration, vt(l is th e therm al v elo city and a c is the cap tu re
cro ss
then
section.
A ssum ing iVA = 1017 c m - 3 ,
= 107 cm /s, and ctc = 10- 1 3 cm -,
t c = 10—11 s. w h ich indicates that pico seco n d b eh av io r is p ossible. T h erefo re a
fast resp o n se tim e ca n be expected assu m in g that b elo w band g ap ab so rp tio n is
m ediated by ionized im purities w ithin th e 6H -S iC m aterial. The lo n g er PV resp o n se
co rresp o n d s to direct clcctron-hole recom bination, w h ich h as been m easu red fo r these
144
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
dev ices to be on the o rd er o f nanoseconds [49]. Thus, th e resp o n se at 4 5 0 nm is m o st
likely a su p erp o sitio n o f these tw o effects. It is interesting th at the d evices resp o n d to
p ho to n en e rg ies that arc so m uch low er than £ g.
145
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
I
lOkU
X 7.300
i t *ri
22222
F igure 7.10 SEM im age o f the m ctal/sem ico n d u cto r interface sh o w in g a
5-pm p ++ ep i-lay er protrusion.
h v,
'W * -
Figure 7.11 Proposed energy band diagram for the co n tact region show n in Fig. 7.10.
Photon en ergies above and below £ c are indicated by /iV| and h v 2. respectively, alo n g
w ith corresponding electro n d rift directions.
146
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§7.4 Photoconductivc SiC Switch Experiments
Since the d ark resistance o f both sw itch ty p es w as less than 20 Q , a lowim pcdance circuit w as used for the photoconductivity m easurem ents.
T he circuit
consisted o f a sim ple d c bias, charging resistor, and load, as sh o w n in Figure 7.12. A
carbon-filled resistor w as used to m inim ize unw anted stray inductance.
T he load
resistance indicated is actually the oscilloscope input im pedance, w hich w as ac
coupled to im prove the m easurem ent sensitivity.
A UV dye (C o u m arin 440) w as pum ped w ith a L aser S cien ce, Inc.. N 2 laser
(M odel #V S L -337N D )
so that the laser w av elen g th could b e tuned aro u n d the
ab so rp tio n edge o f 6 H -SiC .
T he band-gap en erg y o f this m aterial is 3 eV . w hich
co rresponds to a w avelen g th o f 414 nm.
W e used the N; laser o utput to assess the
sw itch b eh av io r w ell above band gap. The N2 laser output en erg y w as =200 p J, w ith a
pulse w idth o f 7.7 ns and a 3-H z repetition rate. T he d ye o u tp u t energy w as =10 p j.
with the sam e pulse w idth and repetition rate.
T h e m easured PC response o f a 10-pm g ap lateral PC sw itch is show n in
F igure 7.13 as a function o f bias voltage.
typically 8 Q .
The calcu lated o n -state resistance was
N ote that the dark resistance w as reduced by 50 percent u sin g the
av ailable laser illum ination.
F o r the vertical devices, sim ilar results w ere observed, w ith two exceptions.
First, sin ce the sw itching gap is the substrate itself, a p o la rity change in th e PV
response w ith respect to the beam position across th e g rid d ed electrode w as not
expected or observed.
Second, since the vertical device dark
resistance was
approxim ately h alf o f the lateral sw itch dark resistance, co n sid erab le device heating
occurred for high bias values. A lso, the vertical sw itch es w ere placed on an insulating
surface a n d w ere therefore poorly heat sinked. A s a co n seq u en ce, for a device static
147
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
p ow er d issip atio n exceed in g 11 W , the vertical sw itch g rew to be so h ot th a t it glow ed
until the sw itch m o u n t failed.
T he device w as rem o u n ted and found to be still
o p erational. T he PC sw itch response is show n in F igure 7.14. Ron w as 1.9, 2.2, and
1.3 £2 for Fc = 5, 10, an d 15 V , respectively.
T h ese effects w ere observed in a vertical sw itch w ith laser rad iatio n that w as
w ell ab o ve the band gap energy o f 6 H -SiC .
W e then p u m p ed som e U V and violet
d yes w ith the N j laser so that the sw itch response for rad ia tio n ju s t ab o v e and below
the b and-gap energy could be m easured. The band-gap en erg y o f 6 H -S iC corresponds
to a w avelen g th o f = 414 nm , and the dyes w ere ch o sen to b e near th is w avelength.
T he sw itches responded as anticipated: sm aller p h o to co n d u ctiv c
v oltages w ere
observed for photon energies w ell below the band-gap ed g e, with v alu es co m parable
to those in Fig. 7.13 observed for photon energies at an d ab ove the ab so rp tio n edge.
T he vertical sw itch efficiency, r). is defined as the ratio o f th e sw itch ed -o u t voltage
pulse am plitude, Fs. to the supply voltage. Vc . W ith 337-nm rad iatio n and a bias o f 5
V, the vertical SiC sw itch r| w as 32 percent, and an increasing efficiency w as observed
for increasing sw itch bias.
F igure 7.15 show s the data used to d eterm ine th e b u lk carrier lifetim e w ith
b elow b an d gap (A. = 431 nm ) radiation. The p-type 6 H -SiC d isp lay e d a do u b leexpon en tial PC ca rrie r lifetim e decay behavior, w hich w as o b serv ed w ith the lateral
sw itches as w ell.
A s indicated in the figure, a fast decay o f =200 ns w as observed.
N ot co m p letely sh o w n in the figure is the m easured slo w er decay o f = 8 0 0 ps. At X =
337 nm . the fast decay w as =330 ns, follow ed by a slo w e r decay o f =590 p s. In
com pariso n , the su rface lifetim e m easured w ith the 10-pm lateral sw itc h (Fig. 7.13)
w as = 40 ns.
148
with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
T w o distinct ca rrie r decay tim es, as show n in Figs 7.13 and 7.15, w ere
m easured in 3C -S iC by O k u m u ra et a l [126]. We believe th at th e ir ex p lan atio n fo r the
3C -SiC b e h a v io r is also valid for our ex perim ental resu lts in 6 H -S iC , nam ely th a t the
fast co m p o n en t is due to d irect clectron-h o le reco m b in atio n , w hile the long d ecay tim e
is caused by the rc-em ission o f carriers from traps that reside w ith in th e energy b an d gap. F u rth e r experim ents a re planned to investigate this hypothesis.
T he corresponding ca rrie r lifetim e at the sw itch surface is = 40 ns. T he v alu e
o f su rface lifetim e quoted is. how ever, an estim ate sin ce Fig. 7.13 show s a slig h t
variation in this value w ith applied voltage.
T he ap p lied v o ltag e also results in an
increase in the sw itch tem perature w hich m ay acco u n t for th is change.
C a rrie r
transport effects, such as carrier sw eep-out, m ay also acco u n t for th is behavior.
149
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
«c=sn
A/VPC Switch
N-, Laser
R
s
=son
337 nm
D ye
f
N-, Laser
431 nm
F igure 7.12 PC response experim ental setup show ing electrical circuit, optical
com ponents and sw itch (lateral o r vertical) position. R$ is the scope im pedance.
0.0
-0.5
1. 0
a
V = 10 V
1.5
-
2.0
-2.5
0
50
100
150
200
T im e (ns)
Figure 7.13 PC response o f a 10-pm lateral 6 H -S iC sw itch as a function o f circuit
bias. L aser param eters are identical to those indicated in Fig. 7.12. X = 337 nm .
150
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
0.0
-0.5
1.0
V =10 V
1.5
15 V
-
2.0
-2.5
0
100
50
200
150
T im e (ns)
F igure 7.14 V ertical PC sw itch response versus circuit bias. L aser param eters are
identical to those o f Fig. 7.12. k = 337 nm .
5
0
slow
-o
= 836 ps
-10
15
-20
-25
-30
last
= 203 ns
-35
0
100
200
300
400
T im e (ns)
Figure 7.15 PC carrier recom bination lifetim e m easured w ith vertical PC sw itch.
VQ = 10 V, k = 431 nm . T he double exponential decay tim es. Tfast and t sjow indicated.
151
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
§7.5 SiC Dcep-Lcvcl Transient Spectroscopy Measurements
O ne o f the key technological issues relating to PC sw itch es is the olT-state (i.e..
d ark) resistance.
At present, SiC device technology is lim ited by a lack o f high
resistivity substrate and epitaxial layers.
increasing a m a te ria l's resistivity.
T here are tw o prin cip al m eth o d s for
The first is through p u rificatio n o f the su b strate
m aterial, w hich is the m ethod used in silicon.
The seco n d is th rough d eep-level
com p en satio n o f free charge through the appropriate in tro d u ctio n o f deep -lev el
im purities, w hich is how G aA s is able to achieve sem i-in su latin g behavior.
I f one is to engineer deep-level im purities to co m p en sate a sem ico n d u cto r, then
the existing im purity levels m ust first be identified. T hese levels can be due eith er to
atom ic states th a t reside in the crystal lattice, such as those co rresp o n d in g to the
dopants used to m ake the m aterial either p o r n type, or intrinsic d efects due to crystal
im perfections, w hich is n orm ally the case fo r a binary sem ico n d u cto r, such as G aA s.
In the case o f G aA s, there are tw o such well know n levels, nam ely the EL2 an d EL5
[96], S ince SiC is also a binary sem iconductor, it is g en erally believ ed that sim ilar
native d efects are present and thus exploitable.
By introduction o f the appropriate d opant, these intrinsic levels can be used to
"com pensate" th e sem iconductor to achieve high resistiv ity (i.e.. sem i-in su latin g )
behavior, w hich is the case for chrom ium -doped GaA s.
In this sectio n , w e w ill see
h o w D LTS m easurem ents w ere m ade and how these m easu rem en ts can be used to
identify im purity levels in SiC p-n junctio n d iodes.
The D LTS technique is well understood and ex ten siv ely covered in the
literature [137],[138].
A ppendix A co n tain s an analysis sh o w in g how the D L TS
technique w as used to identify both accep to r and donor traps in w h at follow s.
The
152
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
im portant trap characteristics, such as trap activ atio n energy, cro ss sectio n , and
concentration, can be readily identified w ith the D L TS technique.
Tire D -center in 6 H -S iC is a boron-related d eep hole trap o b serv ed p rev io u sly
in L P E -grow n 6 H -SiC diodes. W e report D LTS m easu rem en ts in w h ic h the D -ccn ter
signature is observed in high-purity epitaxial layers form ed b y chem ical v ap o r
d eposition.
A n activation energy o f 0.58 eV
and a capture
cross se ctio n o f
betw een 1x 10—14 and 2 x l 0 -14 cm 2 w as determ ined for th is level.
E lectrically active d ee p levels in elcctro n ic-g rad e silicon carb id e m a teria ls is o f
substantial technological im portance.
For exam ple, o n e d eep -lev el defect asso cia te d
w ith v anadium is suspected o f acting as the m in o rity carrier lifetim e k iller in SiC
op to electro n ic
devices
[139],
behavio r
that
is
an alo g o u s
to
transition
m etal
c o n tam in atio n in gallium arsenide. D eep levels arc esp ecially im p o rtan t in h ig h -p u rity
m aterial, w here even tiny residual contam in atio n from im purities o r native d e fe c ts can
influence the resulting electronic properties o f the d evice.
W e have undertaken D L TS m easurem ents on p-n ju n c tio n d io d e s form ed from
new ly available high-purity 6 H -SiC epitaxial layers [52].
T his prelim inary study
reports the o b servation o f the boron-related “D cen ter" in these ep itax ial layers. T he
D -center has been described in at least tw o reports [53].[54] as b eing a co m p lex w ith
b oro n and som e other defect, m ost likely native [54]. Further, R ef. [54] su g g ests that
the D center acts like a donor, despite being a hole trap lying ap p ro x im ately 0.58 eV
ab ove the valence band o f 6 H -SiC .
T he 6 H -SiC epilayer structure w as grow n o n a 6 m m x 7.5 mm p iece cu t from
a com m ercially available n+ 6 H (0001) silicon-face S iC substrate p o lish ed 3° to 4° o f f
the (0001) SiC basal plane [140].
T he com plete e p ih y e r dev ice stru ctu re w as
accom p lish ed w ithin one continuous gro w th run, w ith the lig h tly -d o p ed n- and p-type
epilayers
being
produced
using
a
novel d o p an t
control
process
[141 ]—[ 143 ].
153
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Individual diodes w ere then defined b y saw ing the w afer into I m m x I m m squares
using a d ic in g saw . T hese d iodes form ed the “ 1561 -2 ” series.
In preparation for electrical characterizatio n , individual d io d es w ere m ounted
onto T O - 18 headers.
E lectrical co n tac t w as m ad e to the p + and n + cap layers w ith
silv er epoxy. C urren t-v o ltag e (I-V) ch aracterizatio n indicated that this arran g em en t
m ade satisfactory ohm ic contact. C ap acitan ce-v o ltag e (C -V) m e asu rem en ts w ere
p erform ed on 1561-2 series diodes. A ssu m in g eq u al back g ro u n d d ensity in b o th th e p
and n ep i-lay crs such th a t ;V,\ = iVD = .Vg, .VQ was calcu lated from the C -V
m easurem ent to be about 2.3 x 1015 cm - 3 . T his value is in ag reem en t w ith intended
do p in g levels o f 3 x 1015 cm -3 [143].
A ju n c tio n potential o f ab o u t 1 V w as also
calculated.
T he D L TS setup consists o f a I-M H z capacitan ce b rid g e d riving a lock-in
am p lifier designed according to the sug g estio n by Lang [137] to m aintain a constant
t |/ t i ratio as th e repetition ra te / = //A is adjusted to alter the rate w indow . A closedloop controlled therm oelectric co n tro ller is used to ram p the tem p eratu re o v e r a range
I0 °C to 80°C.
154
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
0.004
I 0 .5 8 eV
0 .0 0 3
7 non
0.002
0.001
0.000
-
0 .0 0 1
-20
0
20
40
80
60
T ( ° C)
Figure 7.16 T w o D LTS m ajority spectra f o r 6 H -SiC p-n d iode series 1561-2 show ing
m ajority-carrier trapping effects at 50 and 200 H z P R F. N ote th at at 50 H z th ere is a
d eeper level evident from a peak in th e spectra for T > 80°C.
0.000
0.020
~
-
<
-0 .0 4 0
0 .5 8 eV
Q O O I t)
-0 .0 6 0
10
30
50
70
T ( ° C)
F igure 7.17 T w o DLTS m inority spectra for 6 H -SiC p-n d iode series 1561-2 show ing
m inority-carrier trapping effects at 100 and 200 H z PRF. N ote th at there are no d eeper
levels ev id en t in this spectrum .
155
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
B oth m ajo rity carrier and m inority carrier sp ectra w ere co llected; sam ples o f
b oth types o f sp e ctra are show n in Figures 7.16 and 7.17. respectively. T he m ajority
and m inority ca rrie r spectra w as m easured using four different repetition rates, w hich
perm its a g raphical determ ination o f the trap activ atio n energy, £ A, density, My. and
capture cross-sectio n . crp. A ppendix A describes in detail how the m easured D LTS
spectra can be used to determ ine these param eters.
D uring these m easurem ents, the static reverse bias v oltage w as - 1 5 V since
soft b reakdow n w as typically observed beyond th is potential. D uring m ajority carrier
trap filling, the reverse bias w as reduced to - 1 .5 V. D uring m inority carrier injection,
the diode w as biased to 2.5 V. The m inority ca rrie r injection current w as ab o u t
5.5 mA.
T he m ajority carrier m easurem ent resulted in a w eak D LTS signature, as is
ev id en t from the noise show n on both spectra (PR F = 50 H z and 200 Hz) in Figure
7.16. W e have show n the D LTS spectra for two rate w indow s (th e rate w indow is set
autom atically by the instrum entation via the PR F setting).
In addition, earlier
m easurem ents, m ade w ithout the aid o f a closed-Ioop tem perature controller, indicated
the presence o f a very deep (-1 .3 eV ) level (this d ata w as inaccurate due to the errors
introduced in m easuring the actual device tem perature and hence the D LTS spectra
w ere re-m easured).
T his deeper level w as n ot fully resolvable over the lim ited
tem perature range o f the tem perature controller; how ever, evidence o f this level w as
also observed at a PRF o f 50 Hz (see Fig. 7.16).
In contrast, the D L TS signature o bserved afte r m inority carrier injection w as
m uch larger, w ith the sign o f the D LTS signal reversing and thus sig n alin g th e
presence o f a separate im purity level in the o p p o site epitaxial layer o f the p-n diode.
B ecause eith er the p or n layer can be supplying eith er D LTS signal, w e could n o t
im m ediately identify the nature (hole or electro n em ission) o f the respective D LTS
156
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited without perm ission.
signals.
H ow ever, a m ajority carrier DLTS signal w as not o b serv ed o v er this
tem p eratu re range w hen a diode from a different scries, 1286-8, w as tested. Since
1286-8 is a p+ -n abrupt ju n c tio n diode grow n sim ilarly to th e tw o -sid ed 1561-2, the
lack o f a m easurable m ajority carrier signal in 1286-8 su g g e sts that th e m ajo rity carrier
signal in 1561-2 arises from a trap in the p-typc epi layer.
T he A rrhenius plot in Figure 7.18 contains both d ata from the m ajo rity carrier
trap peaks and the m inority carrier trap peaks.
W ithin the ex p erim en tal erro r
associated w ith estim ating the position o f the peaks th em selv es, the d ata from both
m easu rem en ts virtually coincide.
T his indicates that both th e m ajo rity carrier and
m inority ca rrie r D L T S signatures result from the sam e
level.
A rrh en iu s plot
param eters (slope an d intercept) d eterm in ed by curve fitting to the m ajo rity carrier and
m inority c a rrie r data separately do not show statistically sig n ific an t d ifferen ces.
In
addition, w hen analyzed for their shapes the spectra th e m se lv es arc found not to differ
sig n ifican tly .
T hus w e hypothesize that a single level is resp o n sib le for both
signatures.
T he em ission param eter for a given trap level is a function o f tem p eratu re
acco rd in g to the follow ing w ell-know n expression d eriv e d from the p rin cip le o f
d etailed balance:
EP
=
G P
< v > ; V ve - £‘ M r,
(7.2)
w here ep is the em ission param eter o f a hole trap. crp is the capture cro ss section for
this trap, <v> is the therm al velocity in the m aterial, and /Vv is the effectiv e density o f
states o f th e valence band. E A is the therm al activ atio n en erg y o f the trap referred to
the valen ce band, T is the tem perature, and k is the B o ltzm an n constant.
157
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
IE-2
..— I
Spectrum
r—i •
Minority Carrier j
-----1
Spectrum________r
IE-3
2.75
3
3.25
3.5
3.75
looo/roc1)
F igure 7.18 A rrhenius plot o f both m in o rity and m ajo rity carrier peaks o f
Figs. 7.16 and 7.17. N ote that both plots sh o w roughly the sam e activation energy,
trap density and cross section. The line is a best fit o f Eq. (7.3) using both sets o f data.
158
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T he tem perature dependence o f th e capture cro ss section is a m atter for
o bservation. I f the captured species are attracted to the capturing ce n te r by C oulom bic
a ttractio n (as in the case o f an ionized accep to r ca p tu rin g a hole), then the capture
cro ss section can be expected to be co rresp o n d in g ly enhanced, an d . q u ite possibly
e x h ib it w eak tem p eratu re dependence.
In this case, we plot ep/T 2 v ersu s 1/7’ to
c o n stru c t the A rrh en iu s plot sh o w n in F ig u re 7.18. E q uation (7.2) b ecom es
e / T2 = a
m *-k ~~e ' E' ,ir
.
(7.3)
h
w here m * is the effectiv e m ass o f holes and h is P lanck’s constant. From m o b ility data
in 6 H -S iC we estim ated m * /m 0 = 0.25. T ab le 7.2 co n tain s
and o p estim ated by
fitting eq uation (7.3) to the em ission p aram eters calcu lated from the peaks o f both the
m ajo rity and m in o rity ca rrie r spectra individually an d as a w h o le.
T hese values
co m p are favorably w ith values reported for the boro n -related D -ccn ter by Suttrop at al
[54]. (T he analysis used to construct th e A rrhenius p lot o f F igure 7.18 is co n tain ed in
A pp en d ix A.)
B ecause o u r value o f capture cross section is in ferred from th e in tercep t o f the
A rrhenius plot, w e cannot com m ent o n the possible tem p eratu re d ep en d en ce o f that
variable.
F urther experim ents will d etec t a possible dependence by an a ly zin g the
ch ange in the D L TS peak a s a function o f bias p u lse w idth.
In th e m ean tim e, we
point out that if the cross section increases according to P-. as is som etim es assum ed,
the activ atio n energ y m ust be recalculated. T hese values are also p resented in T able
7.2. w h ich co m p are favorably w ith activ atio n energies calculated b y A nikin at al [53]
for the D -centcr.
Suttrop at a l [54] provide evidence for the link betw een th e D -cen ter and boron
doping. In th eir report they show a d irect relationship betw een b o ro n im plantation and
159
with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
the D -center concentration detected by D L TS over a range from less than 1 0 15 c m -3 to
g reater than 1017 cm - 3 . B ecause o f the sim ilarity o f th e shapes o f th e D L T S spectra
found in the literature, and the sim ilarities o f the p aram eters calcu lated from th em , we
conclude that o u r D LTS signatures result from the boron related D -ccn ter d esp ite the
fact that no boron w as intentionally introduced. A lthough w e co n sid ered the "E 3 /E 4 "
native defect [139], its ap parent associatio n w ith high -en erg y electro n irradiation
d isco u n ts this possibility.
In
this
m easurem ent,
we
d etected
a
m ajority
ca rrie r
trap
w ith a
concentration o f 3 x l 0 13 cm - 3 , and a m inority carrier trap w ith co n c en tra tio n 5 x l 0 14
cm - 3 .
W e have concluded that both trap s are the bo ro n -related D -cen ter delect,
therefore, the D -center concentration in o u r C V D g ro w n d io d e is ab o u t 20 tim es
g reater in one layer than the other.
T h is could be d ue to d ifferin g levels o f boron
co ntam ination in the p recu rso r gases, o r it is a p o ssib le in d icatio n o f d ifferen ces
betw een the p - and n -type layers (e.g., differen t co n cen tratio n s o f the p articu la r d elect
center-b o ro n co m plexes w hich form the D -center). B ecause m easu rem en ts on an
ab ru p t one-sided p+ -n d iode produced virtually no m ajority ca rrie r D L T S signal
w ith in the detection lim it, w e conclude that the m inority carrier sig n atu re arises from
hole cap tu re in the n-type cpi-layer follow ing m in o rity -carrier injection d u rin g the
forw ard-bias current pulse.
160
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Tabic 7.2. Com parison o f D-Ccntcr Parameters as M easured Here and Reported
in the Literature.
[cm - 3 ]
[eV]
[eV]
[cm 2]
0.59
0.65
(3 ± 2) x 1 0 - 14
5 x 1 0 '4
0.55
0.61
(8 ± 4) x 1 0 -'5
Both
-
0.58
0.63
(1.7 ± . 6) x 10->4
M ajority, p"
-
0.58
0.63
(1.0 ± .5) x 1 0 - '4
0.57
0.62
(5 ± 3) x I0-1S
-
0.63
1 x 1 0 - '4
0.71
3.6 x 1 0 " 14
M ajority
X
£a
M inority
E rlan g en 3
layer
Io ffeb
■r>
a ccT-
.Yy
Signal
1561-2
const.
o
UJ
ct =
M ajority, p"
layer
9.7 x 10'6
CTP
a W . S uttrop. G. Pensl. and P. L anig. A ppl. Phys. A 51. 231. (1990)
b M. M. A nikin. A. A. L ebedev. A. L. S yrkin. and A. V. S uvorov. Sov. Phys.
S em icond. 19. 69 (1985)
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited without perm ission.
§7.6 Sum m ary
T he m easurem ent o f efficient p h o toconductive sw itch in g in p -type 6 H -S iC
sw itches w as lim ited by the low dark resistivity o f the m aterial.
W ith h ig h e r
resistivity substrates, the sw itching efficiency should im prove d ram atically , alth o u g h
the valu e m easured w ith the vertical sw itch (rj = 32% ) w as n ot for the o p tim u m
circuit. The encouraging o b servation th at the vertical 6 H -S iC d ev ices can o p erate at
high tem p eratu res o b viously has im plicatio n s for p u lsed -p o w er ap p licatio n s.
T he
ability to fabricate large p h otoconducting devices for h ig h -v o ltag e ap p licatio n s is still,
how ever, uncertain, since we could not hold o ff large v o ltag es w ith th ese d ev ices
(/?dark = 2 <T2).
It is im p o rtan t to note that w ith the vertical structure, substantial
p hotoconductivity w as observ ed at 337 nm . even though this co rresp o n d s to a ph o to n
energy that is well ab ove the 6H -S iC band-gap energy.
W e believe that carrier
d iffu sio n and drift are responsible for this observed behavior.
O ur success in m easuring a sig n ifican t photovoltaic effect in 6 H -S iC . even
w ith th e observ atio n o f oh m ic contact b ehavior in the d ark state, in d icates that
efficien t UV photo detection at high tem peratures is possible.
T he very fast carrier
decay tim es (Tdccay ~10 n s) observed during these ex p erim en ts indicate that th e UV
d etec tio n m ay also be high speed.
T he identification o f several deep-level electronic im p u rities m ay ev en tu ally
lead to higher resistivity 6 H -SiC m aterials, w ith the u ltim ate goal o f sem i-in su latin g
behavior being achieved. O nce this can be accom plished. 6 H -S iC will then be su itab le
for b oth ultra-fast and high-pow cr PC sw itching ap p licatio n s, and. in p articu lar, the
optoelectronic attenuator schem e can be im plem ented u sin g SiC w ith a m uch
im proved therm al d issip atio n possible then w ith eith er G aA s o r silicon.
162
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Chapter 8
Conclusions and Future Research
§8.1 C onclusions
W ith the exception o f th e perform ance o f the G aA s:C P W -P C S d ev ices, this
research has proven to be en tirely successful.
W e have d em o n strated a w orking
optoelectro n ic m icrow ave a tten u a to r that can provide up to 45 dB o f atten u a tio n for a
m odest optical pow er o f less than 150 m W . O u r u nderstanding o f h o w this d ev ice, the
S i:C P W -P C S . operates is satisfacto ry for o u r application, and all the d esig n goals
(g re ate r than 20 dB attenuation for reasonable LD pow ers, an all-scm ico n d u cto r-b ased
system , an d LD turn-on tim es in the su b-nanosecond regim e) h av e been m o re than
ad equately m et.
In several areas our w o rk has p ro v en to be ev e n m o re fruitful than we
im agined. A goo d exam ple o f th is is the o p tical ^ -sw itc h e d LD research; not o n ly did
w e achieve state-of-the-art results (largest reported pulse energ y from a b ro ad -area
LD ), we have also discovered th a t this tech n iq u e can be used to th o ro u g h ly in v estig ate
the LD dynam ics, even on tim e scales shorter than the p h o to n ca v ity ro u n d -trip tim e
[38],[39]. W ork is continuing in o u r lab in th e hope th at sim ila r p erfo rm an ce can be
achieved w ith o u t the need for a m ode-lock ed , h ig h -p o w er N d:giass laser.
In term s o f the G aA s:C P W -P C S T i:sap p h ire la ser ch aracterizatio n o f the
d ev ice o n-state resistance, m uch w as learned fo r both this ap p licatio n and th e g eneral
area o f A lG aA s L D /G aA s high -sp eed PC sw itch interaction.
It is hoped th at others
w ill benefit from our w ork; this is altogeth er possible due to the lively d isc u ssio n s that
took place at a recent conference w here this w ork w as presen ted [42].
163
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
T h e SiC w ork is also new , particularly in th e PC sw itch in g area.
Indeed, this
w as the first system atic investigation undertaken o n PC sw itches in S iC : p rio r w o rk
concen trated on using PC techniques fo r ch aracterizatio n purposes [126].
T he
p ico seco n d laser m easurem ents o f the PV resp o n se o f th is m aterial w ere also new and,
provided that a w orking m odel can be developed, should h elp industry d esig n new and
im proved UV photodetectors and sensors.
Finally, the cataloging o f electro n ic
im p u rities in both bulk and epitaxial SiC m aterial is o f such p rim e im portance that
further im provem ents in SiC
m aterials
w ill occur only after these levels arc
u n derstood. O ur co nfirm ation o f p rior data from G erm any (U niversity o f E rlangen) is
a step in this direction. T his is especially tru e since m uch skepticism w as associated
w ith the “ D -ccnter" and, as a result, our p ro p er identification o f the ch aracteristics o f
this d e fe c t should perm it im provem ents to the m aterial. T his m ay turn out to be the
native defect in SiC that can be used to ach iev e sem i-insulating behavior, ju s t like the
EL2 and ELS in G aA s [96].
§8.2 Future Research
T here arc several areas from w hich additional research can be co nducted,
in cluding im provem ents to the optoelectronic atten u ato r d evices as w ell as research to
d eterm in e w hether m onolithic integration can be achieved.
In this sectio n , we will
proceed w ith our discussion o f these potential research topics ch ro n o lo g ically , starting
w ith C h ap ter 3 (laser diodes) and ending w ith C hapter 7 (S iC ). T he last section will
be a d iscussion o f possible m onolithic integration schem es and issues for the
o ptoelectronic attenuator.
164
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
In the area o f LD technology, im p ro v ed high-speed packages should be
desig n ed an d fabricated, since the LD m ount vve desig n ed several years ago has proven
to be inadequate for high-speed LD operation. T h is new package should be able to
ac co m m o d ate h igh-speed m atching netw orks, both for the gain and loss sections, to
p erm it m ore efficien t operation as w ell as h ig h -sp eed m odulation. T he seco n d area for
research is in the 0 -sw itc h in g o f tw o-sectio n b ro ad-arca LD s. A lthough the o p tically
0 -sw itc h c d research proved to w ork quite w ell, elim in atio n o f the N d:glass laser from
the system w ould perm it 0 -sw itc h in g to be d o n e m uch m ore efficiently. R esearch that
is curren tly being conducted by S teve Yang in o u r laboratory should be con tin u ed :
M r. Y ang is trying to electrically 0 -sw itc h the sam e tw o-section LD using a steprecovery dio d e (SR D ) [104] to rapidly forw ard-bias the second (i.e.. m odulator)
section.
S ince the S R D can generate pico seco n d electrical pulses, the hope is to be
able to ach iev e sim ilar 0 -sw itc h in g perform ance as that d escribed in C h ap ter 3.
In the area o f high-speed PC sw itches, nam ely the C P W -P C S , the only
im p ro v em en ts w ould com e in the are a o f red u cin g the cen ler-co n d u cto r parasitic dc
resistance. T his can be accom plished in tw o w ays: first, reduce the sw itch length and.
second, plate-up the m etallization using stan d ard electroplating processes.
B oth o f
th ese techniques w ould solve the problem , and w hich is chosen is a m atter o f
convenience. The high insertion loss o f the S i:C P W -P C S w as caused by the use o f the
incorrect m ask w hen the devices w ere fabricated: this is an o bvious im p ro v em en t (Si
su bstrate thickness w as 10 m ils, w hile the m ask desig n w as for a 20-m il substrate). As
for the p o o r perform ance o f the G aA s:C P W -P C S . we erred in ch o o sin g a p ++ highly
d o p ed epitaxial layer on a sem i-insulatin g substrate: the sem i-in su latin g substrate
m aterial is slightly /i-type, and hence a p ++-n ju n c tio n w as inadvertently form ed
durin g the M B E grow th.
T his unw anted ju n c tio n w ould decrease th e current flow
un d er optical illum ination.
T his m ay ex p lain w hy Ron o f the G aA s:C P W -P C S
165
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
m easured in C hapter 5 w ere so m uch higher than those m easured w ith th e S i:C P W PCS in C h ap ter 6 . U se o f an «'H " highly doped ep itax ial layer, using a A u-G e/A u
ohm ic co n tac t geom etry, should solve this problem . The only additional co m p licatio n
is the need to therm ally anneal the contact (w h ich w as th e reason a p ++ layer w ith
C r/A u co n tact m etallization w as originally chosen).
H ow ever, recent w o rk by Eric
Funk in o u r laboratory on the M icro-B O SS [70] has show n th a t A u -G e/A u contacts
can
be
properly
patterned
and
annealed
on
sem i-in su latin g
G aA s;
thus,
no
technological problem s are expected for changing to this type o f ep i-lay er and contact.
T he initial observation o f a decrease in the G aA s:C P W -P C S
Ron w ith
increasing electrical bias, using a LD as the optical source, still rem ains a puzzle. A
m ore detailed look into the behavior o f the G aA s:C P W -P C S is n ecessary o n ly to
an sw er this question: Is the decrease in Ron observed w ith high-level b ias (10-30 V)
due to velocity overshoot o r to the F ranz-K eldysh effect? In ad d itio n , g iv en the high
level o f attenuation described in C h ap ter 6 using th e S i:C P W -P C S , p erh ap s a d etailed
characterization o f th is device sim ilar to that p erfo rm ed for the G aA s:C P W -P C S
should be conducted.
perform ance.
T his w ould help to further im prove th e S i:C P W -P C S
O f course, since A lG aA s/G aA s LD s op erate w ell above th e band gap
energy o f the S i:C P W -P C S , then tem perature tu n in g m ay be o f no benefit; how ever,
electric field effects arc relevant and should be w ell understood.
T he hybrid optoelectronic attenuator fabricated w ith the S i:C P \V -PC S w orked
very w ell - so w ell th a t greater than 45 dB o f atten u atio n at 1.7 G H z w as ach iev ed for
a laser p o w er as low as 143 m W .
A lthough the initial m odel for m icrow ave
attenuation via p lasm a absorption roughly predicted this result, a m o re d etailed
theoretical m odel for the S i:C P W -P C S is w arranted. Perhaps th is m odel w o u ld co n sist
o f a sim ple extension o f the m odel proposed by P latte [117], H ow ever, sin ce this is a
fairly
com plicated
geom etrical
m odel
(in v o lv in g
th e
S w artz-C h risto ffel
166
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
T ransform ation [144]) th a t assum es carrier confin em en t in very w ell-d efin ed regions,
this ap proach m ay be o f lim ited use, especially since the D eb y e length in the S i:C P W PCS is m ore than 100 p m , and it is d ifficu lt to see ju st how w ell the carriers can be
confined. Instead, a m o re general approach w ould be ju stifie d , co m b in in g the analysis
used to m odel th e Ron o f the G aA s:C P W -P C S , co u p led w ith the basic solid-state
plasm a analysis p resented in C hapter 2.
T he perfo rm an ce o f the G aA s:C P W -P C S o p to electro n ic atten u a to r m ay be
im proved in tw o w ays.
First, the epitaxial layer sh o u ld be o f the sam e ty p e as the
sem i-insulating substrate, w hich is slightly n-typc. T his w ould p erm it p ro p er ohm iclike current flow to o c c u r w ithin the PC sw itch gap. S eco n d , using the prin cip le o f
photoconductive gain m entioned earlier in C hapter 6 [99], q u an tu m -w ell structures
m ay be incorporated in to the G aA s su bstrate to achieve ca rrie r co n fin em en t and thus
increase the PC carrier recom bination lifetim e to values clo se to th o se fo r the S i:C P W PCS.
T herefore, since G aA s is a d irect band-gap m a teria l, the G aA s:C P W -P C S
perform ance should ex ceed that o f the S i:C P W -P C S sin ce silicon is an indirect bandgap m aterial and, thus, the quantum efficien cy o f G aA s sho u ld be h ig h er than for
silicon.
W e have fabricated G aA s:C P W -P C S s w ith o u t an y ep itax ial layers, and
m easurem ents o n these devices should indicate w h eth er the first im p ro v em en t will
w ork.
In the literature [45],[100],[109], there is stro n g evid en ce th at the PC carrier
lifetim e can be g reatly increased and thus the second im p ro v em en t, though adding
considerable com p lex ity to the o p to electro n ic attenuator, sh o u ld also w ork.
In the S iC PC sw itch area, m uch rem ain s to be d o n e. T here is a large gap in
the know ledge o f both the optically and electro n ically activ e trapping centers in the
SiC m aterial.
SiC P C sw itches cannot be im proved until the pro p erties o f these
centers are understood. U ntil that lim e, sem i-in su latin g m aterial w ill not be feasible,
and, hence, high -p o w er PC sw itching w ill be im p o ssib le.
In addition, the PV
167
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
m easurem ents indicate the b asis for the d ev elo p m en t o f very h ig h -sp eed p h o to sen so rs;
o u r picosecond m easurem ents indicate that the PV response m ay be even h ig h e r speed
than p rev io u sly thought.
T h is fact, coupled w ith th e fairly high PV efficiencies
m easured, show s th a t our dev ices m ay prove to be w o rth fu rth er investigation.
Since the PV response in the lateral 6 H -SiC PC sw itch es w as m easu red , Cree
R esearch, Inc. [140] has fabricated a n ew set o f d evices w ith one m a jo r contact
g eom etry difference; the p ++-p interface, w hich before w as ex p o sed , has b een covered
by the co n tac t m etallization.
T hus, m easu rem en ts to verify the q u alitativ e m odel o f
C h ap ter 7 can be m ade, and the PV resp o n se properly m o deled and u n d erstood. O nce
this has been accom plished, further dev elo p m en t o f h ig h -sp eed SiC PC sw itches for
U V sen so r ap p licatio n s can be conducted.
O ther new SiC PC sw itch es are presently under develo p m en t.
N A S A -L eR C
[142] has fabricated som e high resistiv ity m aterial (>300 Q -cm ). and initial lowvoltagc m easu rem en ts indicate that the d ark resistance o f th ese d ev ices can exceed
300 kQ .
N A S A -L eR C is presently fabricating a new set o f PC sw itch es using the
sam e m ask set that C ree used to fabricate th e lateral PC sw itch es o f C h ap ter 7. T hese
devices are large enough to perm it high-voltage operation. A gain, since the theoretical
voltage b reakdow n in 6 H -SiC is a factor o f 10 tim es g reater th an that o f G aA s. success
w ith these devices m ay yield m icrow ave pow ers th at are tw o orders o f m agnitude
greater than p o ssib le w ith a com parably sized G aA s device.
T he final area for future investigation has to do w ith th e m o n o lith ic integration
o f the o p to electro n ic attenuator. B asically, if a su itab le LD can be fabricated w ithin a
M M IC substrate that con tain s a planar transm ission line, then th e problem
is
essen tially solved (provided the G aA s:C P W -P C S p erfo rm an ce can b e im proved),
f ig u re 8.1 show s one p ossible m ono iith ic op to electro n ic atten u ato r co n fig u ratio n ,
p rovided that a LD can be incorporated into the M M IC substrate.
As m en tio n ed in
168
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
C h ap ter 4. a m icrostrip line is the preferred planar tran sm issio n stru ctu re since, in this
case, bulk optical abso rp tio n is desired and this g eom etry is ob v io u sly m ore suitable
than a surface geom etry, like the C P W -P C S.
W e can achieve fu rth er refin em en t by
desig n in g the group velocity o f the m icrow ave signal to be slo w er than th a t o f the
op tical signal; in this w ay, even faster atten u ato r tu m -o n tim es can be ach iev ed , since
the p lasm a w ould be generated prior to the m icrow ave signal arrival.
A lthough considerable progress on O E lC s has been m ad e in recen t years
[4],[5], the incorporation o f LD s into M M IC substrates is a d ifficu lt p ro p o sitio n . T his
is especially true since fairly high optical pow ers are required (although 143 m W
sounds m anageable, for a m onolithic LD. a few m W is typical [17]) to g en erate the
necessary plasm a. H ow ever, im provem ents in LD technology, and in the PC sw itch,
m ay perm it th is problem to be resolved. As we saw in C hapter 6 , co rrect u sag e o f the
con tact geom etry o f th e C PW -PC S is w hat perm itted 45 dB o f atten u atio n to be
achieved (recall that the beam was ex p an d ed to achieve this atten u atio n value); thus,
sim ilar issues m ay help to reduce the LD p o w er requirem ent in a m o n o lith ic
o p to electronic attenuator schem e.
A n o th er critical issue is how to isolate the LD bias from the rest o f th e M M IC
circuitry.
W e can do this w ith a bias-tee arrangem ent, w here the cap acito rs. C, and
inductor. L, form the bias-tee (the cap acito rs block d c, w hile the in d u cto r blo ck s high
frequencies). Since the LD off-state im pedance is quite high, n eg lig ib le lo ad in g o f the
transm ission line should occur during the atten u ato r off-state.
O bviously, w h en the
atten u ato r is functioning, any pow er absorbed by the no w lo w -im p cd an ce LD is
beneficial to the attenuator perform ance, provided th a t the LD is not dam aged.
A s m entioned in C hapter 3. a further refinem ent w ould be to co n n ect only the
second (i.e.. m odulator) section o f the LD to the transm ission line. T his w o u ld solve
m ost o f the above problem s relating to LD bias isolation, pro v id ed that su fficien t
169
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
electrical isolation betw een the LD sectio n s is m aintained.
At present, our devices
have an electrical isolation on the order o f 300 kQ . w hich m ay be su fficien t to
properly isolate the gain section from the m od u lato r section.
O f course, another possibility is to place the LD an d d riv er feed lines
com pletely o utside the M M IC circuitry, b u t still w ithin the sam e substrate.
In this
fashion, electrical isolation m ay be quite straightforw ard due to the sem i-in su latin g
b ehavior o f the G aA s substrate.
In addition, the LD o u tput ca n be routed to any
location on-chip using optical w aveguide technology, w hich in G aA s is the sam e
G aA s/A lG aA s system used to form both O E IC and M M IC co m p o n en ts (w e d iscu ssed
this early on in C hapter 1: please refer to section 1.1 for a review o f this point).
It therefore seem s plausible that th e optoelectronic atten u ato r m ay be suitable
for m onolithic integration to provide yet another O M M IC ch ip for the m icrow ave
system designer. In addition, SiC will continue to be an area o f intense research du e to
its highly desirable characteristics, as o u tlin ed in T able 7.1.
O nce SiC m aterials
becom e m ore com m on and w ell understood, it should be a relatively straig h tfo rw ard
proposition to fabricate a SiC hybrid op toelectronic attenuator, esp ecially in light o f
the high therm al conductivity o f this m aterial.
Finally, the D L T S w ork, it is hoped,
w ill lead to higher resistivity, and eventually, sem i-insulating m aterials.
O ur D LTS
studies should greatly help the overall know ledge o f the electro n ic im purity levels in
this w ide-band gap m aterial and help to m ak e this dream into reality.
Finally, to quote R ichard P. Feynm an, "F o r a successful technology, reality
m ust take precedence over public relations, for nature cannot be fooled” [145]. T hus,
it is hoped this w ork w ill fall into the category o f a successful technology, and that
nature w ill p erm it further enhancem ents to be successful.
170
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
F igure 8.1 Schem atic representation o f p ro p o sed m o n o lith ic o p to electro n ic attenuator.
T he inductor, L, and capacitors, C, co n stitu te a bias-tee for the LD d riv er signal.
N ote that the m icrow ave s ig n a l./rf-, and th e o ptical photons, hv. b o th p ro p ag ate in the
sam e m edia. T he PC sw itch location is d enoted by R on.
171
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Appendix A
SiC DLTS Calculations
In C h a p te r 7, the observ atio n o f the D -cen ter in 6 H -SiC , using d eep level
tran sien t sp ectro sco p y , o r D LTS, w as discussed. U sing eq u atio n s (7,2) and (7.3), the
activ atio n en e rg y , E , t r a p cross section, a p and trap d ensity, jV^, w here d eterm in ed .
W hat follow s th e rigorous step-by-step an aly sis used to d eterm in e these co n stan ts.
A.
M inority T rap A nalysis
T h e m easured d ata is in the form o f the tem p eratu re spectra show n in F igure 7.17.
T h e d a ta is sum m arized below for the m inority ca rrie r traps:
Table A .l M inority Trap DLTS Data
/
/
CP
Tpeak
Tpeak
U
AC
(H z)
(s- 1 )
(°C )
(K )
(V)
(pF)
50
113
26
299
-6.3
1.4
100
226
32
305
-6.4
1.4
150
339
38
311
-6.4
1.4
200
452
44.5
317.5
-6.4
1.4
= pulse repetition frequency, ep = em issio n param eter. Tpcak = tem p eratu re o f
spectrum peak o f am plitude U.
U sing th e D LTS system param eters and calib ratio n factors, all o f the d esired
trap constants c a n be com puted using the fo llow ing form ulae:
172
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
D LTS A nalysis Formulae
z p = 2 . 26 /
(A . 1)
w here X = .V/ = 300 m V , is the lock-in am p lifier sensitivity.
U = voltage o f spectral peak.
U / = 4 .7 V is th e lock-in peak calib ratio n voltage at A C = AC0. and
AC'0 = 1 pF is th e capacitance b rid g e calibration factor.
I.
D eterm ination o f N j
(A .2)
w here
= background dopant concentration = 2.3 x 1015 c m '3. and
C = 24.4 pF = ju n c tio n capacitance for the reverse bias o f —15 V applied
during D L T S m easurem ents.
S ubstitution o f the ab ove values into eq u atio n (A .2) yields
N - p - (1.4 p F /2 4 .4 pF )x4x2.3x 1015 cm-3 =>
N f = 5 .3 x l 0 15 c m "3
173
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
II.
D eterm ination o f £ .(
T o determ ine the ac tiv a tio n energy, EA. we tak e the natural log o f eq u atio n (7.2) and
perform a least squares fit (o f the form y = nix + b) to get
(A .3)
U sing the d ata from T able A .l, and sub stitu tin g these v alu es into eq u atio n (A .3 ), w e
obtain:
b = 14.9
m = -6430 K =>
E a = 0.55 eV
R 2 = 0.976
w here R 2 is the co rrectio n coefficient for the least sq u ares fit. T hus, the activ atio n
energy associated
w ith the m inority carrier trap o f d io d e
1561-2(1) has been
determ ined.
III.
D eterm ination o f a p
W hile the slo p e, m, o f the least squares fit d eterm in e s the value o f EA, the y
intercept, b. determ in es the value o f a p . Since b = l n { a p <v> N f T 2 ). then
,h
e
cr
(A .4)
174
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
To calculate a p w e need to k n o w if the trap is a hole trap or an electro n trap.
S ince both m inority and m ajority carrier signals w ere o b serv ed , we don not have direct
inform ation to answ er this question.
H ow ever, the literatu re strongly su g g ests the
D L TS spectra com e from the boron-related > '‘D -center" w h ich is b elieved, as stated in
C h ap ter 7. to be a hole trap. A ssum ing this to be true, then th e product o f the therm al
velocity, <v> tim es the effective density o f states in the v alence band ;Vv(SiC ). is
i/:
( v ) N v ( S iC )
=
,kT
3 /2
2 ir mh k T
h2
(A .5)
( v ) N v ( S i C ) / = , /7 2 71 m l k 2
/T 2
“V
/,3
w here /??/,* is the effective hole m ass in th e SiC crystal lattice and h = 6.63x1 O' 34 (J-s)
is P lan k 's constant.
S ince accurate d ata for <v>/Vv(SiC ) is u navailable for S iC . and
since there are published values for mt,*(SiC) [146] and /«/,*(G aA s) [113], then taking
the ratio o f <v>A''v(SiC )/72 w ith <v>/Vv(G aA s)7 2 is a useful w ay to d eterm ine a p :
( v ) N v(S iC ) / /
/T 2
( v ) N v(G aA s)
mh ( S iC )
m ‘
(G
c l -I s
_
)
0.25
(A .6 )
0.5
T'
S ince the electron and hole m obilities are nearly equal in 6 H -S iC . w e have assum ed
that w/e*(SiC ) ~ M/,*(SiC) in equation (A .6).
U sing equation (A .6 ), w e m ay no w estim ate the a p as follow s:
c v o /V ^ S iC y r 2 =
Vi
<v>/Vv.(GaAs)7'2 = 'A (107 c m /s ) (7 x l0 18 c m '3)/(300 K)2 or
<v>/Vl.(SiC)/72 = 3.9x1020 (s cm 2 K2) ' 1
(A .7)
175
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
T hus, u sing b = 14.9 from th e least sq u ares fit. and the ex p ressio n from eq u a tio n (A .7 ).
the hole cap tu re cross se ctio n m ay be com p u ted
_
e " ’ (a'AT',
C p~
3 .9 x 1 0 20( s c n r
K z y'
4 x l 0 ‘15 cm 2
(A .8)
T his co m p letes the analysis used to calcu late the m in o rity ca rrie r trap
coefficients presented in C h ap ter 7. T able 7.2. This analysis m a y then be rep eated for
the m ajo rity carrier trap data, show n b elo w in Tabic A.2.
B.
M qioritv T rap A n aly sis
T he m easured d ata is in the form o f the tem perature sp e ctra sh o w n in F igure
7.16.
T he data is sum m arized below for the m ajority carrier trap s:
T abic A.2 M ajority Trap DLTS Data
/
CP
Tpcuk
Tpcak
U
AC
(Hz)
( s ' 1)
(°C)
(K)
(V )
(pF)
50
113
26
299
3.2
0.068
76
172
31
304
3.1
0.066
100
226
33
306
3.5
0.078
150
339
41
314
3.0
0.064
200
452
42
315
3.1
0.066
/ = pulse repetition frequency, zp = em ission param eter. Tpcaf. = tem p eratu re o f
spectrum peak o f am p litu d e U.
176
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout p erm ission .
A gain, using the D L T S system p aram eters and calib ratio n factors, all o f the
d esire d trap constants can be com puted u sin g the equ atio n (A .l) . T h e o n ly d ifferen ce
in th is case is due to the w eak m ajority carrier signal, resu ltin g in a need to increase
th e lo c k -in sensitivity from X = 300 m V to X = 30 m V .
T h u s the m ajo rity carrier
p aram eters are:
A" = 30 m V , is the lo ck -in am plifier sensitivity,
A7 = 300 m V , is the lock-in peak calib ratio n voltage at A C = AC0, and
U = voltage o f spectral peak,
U i = 4.7 V is the peak voltage from th e capacitance b ridge, and
AC0 = 1 pF is the capacitan ce bridge calib ratio n factor.
I.
D eterm ination oJ'N-p
S u b stitu tio n o f the above v alu es into equatio n (A .2) yields:
N t = (0.068 pF /2 4 .4 p F ) x 4 x 2 .3 x l0 15 c m "3
II.
N t = 2 .6 x 10 13 c m "3
D eterm in a tio n o f EA
A s for the m inority carrier traps, we d eterm in e the ac tiv atio n en erg y . E A. u sin g
eq uation (A .3). and the d ata from T able A.2:
6 = 16.4
m = -6879 K =>
E a = 0.59 eV
£ - = 0.976
177
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
H ow ever, if a p x 7^. then the least squares fit o f eq u atio n (A .2) yields an
ac tiv atio n energy o f 0.65 eV ; th is is the origin o f the tw o v alues o f E ./ presented in
T able 7.2.
T hus, the activ atio n energy associated w ith the m ajority carrier trap o f
dio d e 1561-2(1) has been determ ined.
III. D eterm ination o f a p
As before for the m inority carrier trap, th e y intercept, b . in con ju n ctio n w ith
eq u a tio n (A .7) determ ines the value o f a p . T hus, using b = 14.9 from th e least squares
fit. and the ex p ressio n from equation (A .7), the h o le capture cross section is
.16.4
cr„ =
(-v a :-2 )\ - l
c \-i
3.9 x 10:o(.v c m ' K~)
= 3. 3 x 1 0 '14 cm 2
(A .9)
C. D L TS C alculation Sum m ary
In the case o f the ‘‘D -center,'’ both th e m inority and m ajority carrier traps are
o n e in the sam e, therefore the average o f som e o f the ab ove calcu lated param eters is
the m ost accurate representation o f these D L TS coefficients.
T hus, the final D L TS
an aly sis yields (w hich arc the values presented earlier in C h ap ter 7. T able 7.2):
(£
,[.
( £ ., .
(vVr ) . .
' ' 'minority
5 x 1014 c m '1
0v v r• )'majority
• ,
3 x 1013 c m -}
CTp
X
CTp X
T ) majority - minority
=
0.63 eV
( A .10)
co n s t.)majlirily + minoriI, = 0.58 eV
^ p ) majority + minoriiy
2 x 1 0 ’ 14 cm 2
178
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
References
[1 ]
S. Jayaram an and C . H. Lee, "O bserv atio n o f T w o-P hoton C o n d u c tiv ity in G aA s
w ith N anosecond and P icosecond L ight P u lses,” A ppl. Phys. Lett. 20, 3 9 2 -3 9 4 ,
1972.
[2]
D. H. A uston, U ltrashort L aser P u lse s a n d A p p lica tio n s. S pringer-V erlag. 1988.
[3]
C. H. Lee, "P icosecond O ptics an d M icrow ave T echnology,” IE E E Trans.
M icrow ave Theory Tech. 38 (5). pp. 5 9 6 -6 0 7 , M ay 1990.
[4]
N. B. C haim , I. Ury. and A. Y ariv, "In teg rated O p toelectronics," IE E E S p ectru m
1 9 (5 ), pp. 3 8 -4 5 , M ay 1982.
[5]
S. R. Forrest. "O ptical D etectors: T hree C o n ten d ers," IE E E S p ectru m 2 3 , (5),
pp. 7 6 -8 4 , M ay 1986.
[6 ]
R. A. Pucell (editor). M onolithic M icro w a ve In teg ra ted C ircu its. IE E E P ress.
N ew Y ork. N Y . 1985.
[7]
G. E. B rehm , "M ultifunction M M IC H istory
from a P rocess T ech n o lo g y
P erspective." IEE E Trans. M icrow ave T heory Tech. 38 (9), pp. 1 1 6 4 -1 1 7 0 .
S eptem ber 1990.
[8 ]
E. D. C ohen, "M M IC from the D epartm ent o f D efense P ersp ectiv e," IE E E
Trans. M icrow ave Theory Tech. 38 (9 ), pp. 1 1 7 1 -1 1 7 4 . S eptem ber 1990.
[9]
P. R. H erczfeld, S p ecia l Issue on A p p lica tio n s o f L ightw ave Technology’ to
M icrow ave
Devices.
Circuits,
a n d S ystem s,
G u est
Editor.
IE E E
Trans.
M icro w a ve Theory Tech.. 38 (5). special issue. M ay 1990.
[10] R. N. S im ons. O ptical C ontrol o f M icro w a ve D evices. A rtech H ouse. B o sto n .
M A. 1990, C hapter 1.
[11] A. G. Stove, “ A utom obile R adar," Appl. M icrow ave M agazine. S p rin g 1993. pp.
1 0 2 -1 1 5 , and A. K im , S. E. S addow , R. Y oum ans. L. J. Jasper, M. W einer, a n d
179
R e p ro du ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
C.
H.
Lee.
C haracterizatio n ,"
“ P hotoconductivc
IE E E
LEOS
M o nolithic
S u m m er
W ideband
Topicals:
T ran sm itter
O p tica l-M icro w a ve
Interactions, W 1 .4 , C onference P roceedings, S anta B arbara, C A , Ju ly 19, 1993.
[12]
C. Y. C hen, A. Y. C ho, C. G. B ethea, P. A. G arbinski, Y. M. Pang. B. F. L evine,
and K. O gaw a, "U ltrah ig h Speed M o d u latio n -D o p ed H etero stru ctu re F ieldE ffect P hotodetectors,” Appl. Phys. Lett. 42, Ju n e 1983, pp. 1 0 4 0 -1 0 4 2 .
[13]
T. U m cda, Y. Cho, and A. S hibatom i. "P ico seco n d H E M T P h o to d etecto r."
Ja p a n J. A p p lie d P hysics, V ol. 25. O cto b er 1986. pp. L 8 0 1 -L 8 0 3 .
[14]
R.
N.
S im ons.
"M icrow ave
P erform ance
of
an
O p tically
C o n tro lled
A lG aA s/G aA s H igh E lectron M obility T ran sisto r and G aA s M E S F E T ," IE E E
Trans. M icrow ave Theory Tech.. 35 (12), D ecem ber 1987.
[15]
T. R. C hen. K. U taka. Y. Z huang. Y. Y. Liu. and A. Y ariv, "A V ertical
M onolithic C om bination o f an InG aA sP /lnP L aser and a H ctcro stru ctu re B ipolar
T ran sisto r," IE E E J . Q uantum E lectron.. 23 ( 6), Ju n e 1987.
[16]
A. Y ariv, Q uantum E lectronics, T hird Ed.. Jo h n W iley & Sons, N ew Y ork, N Y .
1989.
[17]
I. Ury. K. Y. Lau. N. B. C haim , and A. Y ariv, "V ery H igh F requency G aA lA s
L aser-F icld E ffect T ransistor M o n o lith ic-In teg rated C ircu it." A ppl. Phys. Lett.
41 (2), Ju ly 1982.
[18]
C. P. Lee. S. M argalit, I. Ury, and A . Y ariv. "In teg ratio n o f an In jectio n L aser
w ith a G unn O scillator on a S em i-In su latin g G aA s S u b strate," Appl. Phys. Lett.
32 (12). S ep tem b er 1987.
[19]
T. Sucta and M . Izutsu. “Integrated O ptic D evices for M icrow ave A p p licatio n s,"
IE E E Trans. M icrow ave Theory Tech. 38 95), pp. 4 4 7 -4 8 2 , M ay 1990.
180
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[20]
K. U om i. T. M ishim a, and N. C hinonc, “ U ltraliigh R elaxation O scillation
Frequency (up to 30 G H z) o f H ighly p -doped G aA s/G aA lA s M u ltip le Q u an tu m
W ell L asers." A p p l. Phys. Lett. 51 (2), pp. 7 8 -8 0 , Ju ly 1987.
[21]
B. J. V. Z eghbroeck, W. Patrick, J. M . H albout, and P. V cttin g er. "1 0 5 -G H z
B andw idth M etal-S em iconductor-M ctal P h o to d io d e," IE E E E lectro n Dev. Lett.
9, (10). pp. 5 2 7 -5 2 9 , O ctober 1988.
[22]
R. G . W alker, "B roadband (6 G H z) G aA s/A lG aA s E lectro -O p tic M odulator
w ith L ow D rive Pow er," A ppl. Phys. Lett. 54 (17), pp. 1 6 1 3 -1 6 1 5 . A pril 1989.
[23]
S. Y. Y ang, S. H. Lin. and Y. M. H oung, "G a A s T rav elin g -W av e P olarization
M odulator w ith B andw idth in Excess o f 20 G H z a t 1.3 p m ." A ppl. Phys. Lett. 51
(2), pp. 8 3 -8 5 , July 1987.
[24]
R. N . S im ons, O ptical C o n tro l o f M icrow ave D evices. A rtech H ouse. B oston.
VIA, 1990, C hapter 5.
[25]
W. B. S cott. " A ir Force F unding Jo in t S tu d ies to D ev elo p S m art Skin A v io n ics."
A viation W eek a n d Space T echnology 1 2 8 (1 6 ). pp. 6 5 -6 9 . A p ril 18, 1988.
[26]
A. S. D aryoush, et al. "H igh-S peed F iber-O ptic Links for D istribution o f
S atellite T raffic," IE E E Trans. M icrow ave T h eo ry Tech. 38 (5), pp. 5 1 0 -5 1 7 .
M ay 1990.
[27]
I. Bahl and P. Bhartia, M icrow a ve S o lid S ta te C ircu it D esig n . John W iley &
Sons, Inc.. N ew Y ork. N Y , pp. 6 2 6 -6 5 9 . 1988 (see also ref. [122]).
[28]
P. B h artia an d I. Bahl. M illim eter W ave E n g in e erin g a n d A p p lic a tio n s. John
W iley & Sons. Inc., N ew Y ork. N Y , C h ap ter 1. 1984.
[29]
D. P. N eikirk. P. C heung, M. S. Islam , an d T. Itoh, "O p ticaily -C co n tro lled
C oplanar W aveguide Phase S hifters." M icrow ave J .. pp. 7 7 -8 8 , D ecem b er 1989.
[30]
J. L. V orhaus. et al. “M onolithic D ual-G ate FET D igital F h ase S h ifters,"
IE E E
Trans. M icrow ave T heory Tech.. 30. pp. 9 9 1 -9 9 2 , 1982.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
[31]
I. Bahl and P. B hartia, M icrow ave S o lid State C ircu it D esig n , John W iley &
Sons, Inc., N ew Y ork, N Y , pp. 6 0 4 -6 2 5 . 1988.
[32]
H ew lett-P ackard, “ A n A tten u ato r D esign U sing p-i-n D iodes.” A p p licatio n N ote
N o. 912.
[33]
Y.
T ajim a,
et
al,
"G aA s
M onolithic
W ideband
( 2 -1 8
G H z)
V ariable
A ttenuators,” IE E E M T T -S Int. M icrow ave Sym p. D ig est., 1982. pp. 4 7 9 -4 8 1 .
[34]
A. S. D aryoush, “ O ptical S ynchronization o f M illim eter-W av e O scillato rs for
D istributed A rc h itectu res,” IE E E Trans. M icro w a ve Theory T ech.. 38, (5). M ay
1990.
[35]
E. E. Funk, E. A. C hauchard. M. J. Rhee, and C. H. Lee, “ 80-kW Inductive
Pulsed P o w er S ystem w ith a P hotoconductivc S em ico n d u cto r S w itch ." IE E E
P h o to n ics Technol. Lett. 3 (6 ), June 1991.
[36]
S. E. S ad d o w an d C. H. Lee. “V N A M easu rem en ts on an O ptoelectronic
A tten u ato r for M icrow ave C ircuit C o n tro l," IE E E 1994 In tern atio n al M icrow ave
C onference. San D iego, C A .. subm itted for publication.
[37]
A . R osen, e l al, “O ptically A chieved p-i-n D iode S w itch U tilizin g a Tw oD im ensional L aser A rray al 808 nm as an O ptical S o u rce.” IE E E Trans.
E lectron D evices, 26 (2), pp. 4 5 5 -4 5 6 . F ebruary 1989.
[38]
B. J. T hedrez. S. E. Saddow . Y. Q. L iu and C. H. Lee, "E x p erim en tal and
T heoretical In vestigation o f Large O u tp u t Broad A rea A lG aA s S em iconductor
L aser D io d es," IE E E P hotonics Tech. Lett. 5 ( 1 ), pp. 1 9 -23 . Jan u ary 1993.
[39]
S. H. Y ang, B. J. T hedrez. S. E. S addow . Y. Q. L iu and C. H. Lee. "C rossC orrelation M easurem ent o f the T urn-on D elay and P u lsew id th o f a (9-sw itched
T w o -sectio n
S em iconductor
L aser,"
IE E E
P h o to n ic s
Tech.
Lett.
5
(12).
D ecem ber 1993.
182
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[40]
N. M oore and K. Y. Lau, “U ltrahigh E fficiency M icrow ave S ignal T ran sm issio n
U sing T andem -C ontact Single Q uantum W ell G aA lA s L asers," Appl. Phys. Lett.
55 (4). S eptem ber 1989.
[41]
R. S. T ucker, "H igh-S peed M odulation o f S em ico n d u cto r L asers." (invited)
IE E E J . L ightw ave Teclm ol., 3 (6 ), D ecem ber 1985.
[42]
S. E. S addow , B. J. Thedrez, S. L. H uang. T. J. M erm agen, and C. II. Lee. “An
Investigation o f th e T em perature and Electric Field D ependence o f a G aA s
M icrow ave P hotoconductivc S w itch ," S P IE -O E /L A S E '93: O p tically A ctivated
S w itching, conference proceedings, 1873-31. Los A ngeles, C A . Jan u ary 1993.
[43]
S. E. S addow . B. J. T hedrez, and C. H. Lee, “ A n O p to electro n ic A tten u ato r for
the C ontrol o f M icrow ave C ircuits." IE E E M icro w a ve & G u id e d W ave Lett.
3 ( 1 1 ) . pp. 3 6 1 -3 6 2 . Oct. 1993.
[44]
P. J. Stabile and A. Rosen. “A Silicon T ech n o lo g y
for M illim eter-W av e
M onolithic C ircu its" RCA R eview , 45, pp. 5 8 7 -6 0 5 . D ecem b er 1984.
[45]
R. K. A hrcnkiel, e t al, “ U ltralong M in o rity -C arrier L ifetim e E pitaxial G aA s by
P hoton R ecycling" Appl. Phys. Lett. 55 ( 1 1), pp. 1 0 8 8 -1 0 9 0 . S ep tem b er 1989.
[46]
R. F. D avis. G. K elner, M. S hur, J. W. Palm our. and J. A. E dm ond, “ T hin Film
D eposition and M icroelectronic and O ptoelectronic D evice F abrication and
C h aracterization in M onocrystalline A lpha A nd B eta S ilico n C arb id e." Proc.
IE E E . 79 (5). pp. 6 7 7 -7 0 1 . M ay 1991.
[47]
R. J. Trew . J-B. Yan, and P. M. M ock, "T he Potential o f D iam ond and SiC
E lectronic D evices for M icrow ave and M illim eter-W ave P o w er A p p licatio n s."
Proc. IE E E , 79 (5), pp. 5 9 8 -6 2 0 . M ay 1991.
[48]
S. E. S addow . P. S. C ho. J. G oldhar. J. P alm o u r and C H. Lee. "H ig h -S p eed
P hotovoltaic
R esponse o f P-Type 6 H -S iC ," N inth A n n u a l IE E E S a r n o ff
Sym p o siu m P roceedings, Princeton. NJ, M arch 26, 1993.
183
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[49]
S. E. Saddow , P. S. Cho, J. G oldhar, F. B arry M cLean, J. P alm o u r a n d C H. Lee,
"L ateral and V ertical P -Type 6H -S iC P hotoconductivc Sw itch
International
C onference
on
SiC
and
Related
R esponse."
M aterials-/G SC7?A /
'93,
C onference D ig est, W ashington, DC, N ovem ber 1993.
[50]
S. E. S addow . P. S. Cho, J. G oldhar, J. P alm our and C H. Lee, "P h o to co n d u ctiv e
M easurem ents o n P-Type 6H -S iC ,” S P IE -O E /L A S E '93\ O p tica lly A c tiv a te d
Sw itching, conference proceedings, 1873-33, Los A ngeles. CA, January 1993.
[51]
S. E. S addow . P. S. Cho. J. G oldhar, J. P alm our and C . H. Lee, "P-Type 6 H -SiC
P hotoconductive S w itches," 20th IEEE International C o n feren ce on Plasm a
S cience (IC O P S ’93), V ancouver, B.C.. Canada, June 1993.
[52]
M. S. M azzola. S. E. Saddow . P. N eudeck, et a l. "O b serv atio n o f the D -C en ter
in
C V D -G row n
6H -SiC
p-n
D iodes,"
A ppl.
Phys.
L ett., su b m itted
for
publication.
[53]
M. M. A nikin. A . A. L ebedev, A. L. Syrkin. and A . V. S uvorov. Sov. Phys.
Sem icond. 19 (69). 1985.
[54]
W. Suttrop, G . Pensl. and P. Lanig. A ppl. Phys. A . 51 (2 3 1 ), 1990.
[55]
C. H. Lee. P ic o sec o n d O ptoelectronic D evices. A cadem ic Press. Inc.. C h a p te r 5.
1984.
[56]
I. Bahl and P. B hartia. M icrow ave S o lid S ta te C ircuit D esig n . John W iley &
Sons, Inc.. N ew Y ork. NY. C hapter 12, 1988.
[57]
1. Bahl and P. B hartia. M icrow ave S o lid S ta le C ircuit D esig n , John W iley &
Sons. Inc.. N ew Y ork. NY. pp. 4 7 2 -4 7 3 , 1988.
[58]
E. A. W olf and R. Kaul. M icrow ave E ngineering a n d System s A p p lica tio n s.
John W iley & Sons. Inc., C hapter 6 , 1988.
[59]
N. W eber, "N e w R eflection-T ype O ptoelectronic M icro w av e Sw itch for CW
A pplications." IE E Electron. Lett. 26 (7). pp. 4 5 5 -4 5 6 .
184
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[60]
E. A. W olf and R. K aul. M icrow ave E n g in eerin g a n d S ystem s A p p lic a tio n s.
John W iley & Sons, Inc., C hapter 10, 1988.
[61]
L. M. D evlin, B. J. Buck. J. C. C lifton, A. W. D ean, an d A. P. L ong. “ A 2.4
G H z S ingle C hip T ran sceiv er.” IE E E M icrow ave a n d M illim eter-W a ve C ircu its
Sym p. D igest. A tlanta, G A . pp. 2 3 -2 6 . Ju n e 1 4 -1 5 , 1993.
[62]
A. K im , S. E. S addow . R. Y oum ans, L. J. Jasper, M. W einer, and C . II. Lee,
"P h o to co n d u ctiv e M onolithic W ideband T ran sm itter C h a ra cteriza tio n ," IE E E
L E O S Sum m er Topicals: O ptical-M icrow ave In tera ctio n s. YV1.4, C o n feren ce
P roceedings, S anta B arbara, C A . July 19, 1993.
[63]
A.
Paoclla,
"V ariatio n s
in
the
G ain
and
B andw idth
of
the
M ESFET
P hotoresponsc w ith B ias and O ptical Intensity,” IE E E L E O S S u m m er Topicals:
O ptical-M icrow ave Interactions. M3.4. C onference P ro ceed in g s. S an ta B arbara.
C A . pp. 2 5 -2 6 , Ju ly 19. 1993.
[64]
C. H entschel. H ew lett-P a c ka rd F iber O ptics H andbook. T h ird E d itio n , H ew lettP ackard G m bH . B oeblingcn Instrum ents D ivision. G erm an y , p. 24. 1989.
[65]
A. Y ariv. Q uantum E lectronics. John W iley & Sons. Inc., pp. 2 5 5 -2 6 0 , 1989.
[6 6 ] S.
R am o.
J.
R.
W hinnery.
and
T.
van
D uzer.
F ield s
and
IVaves
in
C om m unication E lectronics. John W iley & Sons., C h ap ter 13. 1984.
[67]
R. S. M uller and T. I. K am ins, D evice E lectronics f o r In teg ra ted C ircuits.
S econd E dition. John W iley & Sons. N ew Y ork, N Y , pp. 2 2 6 -2 2 8 , 1986.
[6 8 ] J. D. Jackson, C la ssic a l E lectrodynam ics, John W iley & Sons. Inc., N ew Y ork.
N Y . S econd E dition. C hapter 10, 1975.
[69]
S. M. Sze. P hysics o f S em icondu cto r D evices. John W iley & Sons. Inc., N ew
Y ork. N Y . S econd E dition, p. 7 4 4 -7 4 5 . 1981.
[70]
M. A. R ichardson, M. S. M azzola. and S. S pence, "D e sig n o f B ulk O ptically
C o ntrolled S em iconductor S w itches for M icro electro n ic A p p licatio n s," IE E E
185
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
In tern a tio n a l Conf. on P lasm a Scien ce - IC O P S '93. C onf. D igest, V an co u v er,
B .C ., C anada. 5C 8, p. 195. June 7 - 9 1993.
[71] J. E. Bovvers (G uest E ditor), ‘"Special Issue on S em ico n d u cto r D io d e L ase rs.”
IE E E J . Q uantum E lectron. 25 (6), Ju n e 1989.
[72]
Y. A rakaw a, A. L arsson. J. Palaski, and A. Y ariv, “ A ctive Q -sw itch in g in a
G aA s/A lG aA s M ultiquantum W ell L aser W ith an Intracavity M o n o iith ic Loss
M od u lato r,” Appl. Phys. Lett. 48, pp. 5 6 1 -5 6 3 , 1986.
[73]
P. R. C laisse and G . W. T aylor. “ Internal Q uantum E fficiency o f L aser D io d es."
E lectron. Lett. 25 (21), pp. 1 9 9 1 -1 9 9 2 . 1992.
[74]
G . P. Agravval and N . K. D utta. L ong-lV avelength S em ic o n d u cto r L a sers. V an
N o stran d R einhold. N ew Y ork. N Y , 1986.
[75]
H. C. C asey. Jr., and M. B. P anish, H eterostructure Lasers. P arts A an d B.
A cadem ic Press. N ew Y ork. N Y . 1978.
[76]
Y. A rakaw a and A. Y ariv. “T heory o f G ain, M odulation R esponse, and S pectral
L inew idth in A lgaas Q uantum W ell L ase rs,” IE E E J . Q ua n tu m E lectron. 21,
pp. 1 6 6 6 -1 6 7 4 , 1985.
[77]
A. Paradisi and I. M ontrossct. “ N u m erical M odeling o f B istable L aser D io d es
w ith S aturable A b so rb ers," I E E E J . Q uantum E lectron. 27 (3), pp. 8 1 7 -8 2 3 .
1991.
[78]
B. J. T hedrez and C. H. Lee, “A R eassessm en t o f S tandard R ate E q u atio n s for
L ow Facet R eflectivity S em icond u cto r Lasers usin g T raveling W ave R ate
E quations." IE E E J . Q uantum Electron. 28 (12). pp. 2 7 0 6 -2 7 1 3 . 1992.
[79]
D. A. B. M iller, D. S. C helm a. T. C . D am en. A. C. G o ssard . W . W ieg m an n . T.
H. W ood, and C. A. B um s, “ B an d -ed g e E lectroabsorption in Q u an tu m W ell
S tructures: The Q uantum C onfined S tark E ffect," Phys. Rev. Lett. 5 3. pp. 2 1 7 3 2176, June 1984.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[80] J. B erthold, A. F. J. L evi. T. T anbun-E k. R. A. Logan, and S. N. G. C hu.
"V o ltag e-C o n tro lled ^ -s w itc h in g in Ingaas/Inp S ingle Q u an tu m W ell L asers,"
A ppl. Phys. Lett. 55 (19), N ovem ber 1989.
[81]
P. V a sil'e v . I. S. G oldobin. A. A. D idzhyulis, M. I. E fim chik, S. V. E frem ov, B.
I. L evitas. A . B. Sergeev, and E. V. S hatkovskii, “ F orm ation o f E lectrical Pulses
w ith a R ise T im e S horter T han 10 Ps by an O p toelectronic S w itch C o n tro lled by
an Injection Laser,” Sov. Tech. Phys. Lett. 14 ( 8 ), 1988.
[82]
P. V a sil'e v . “ Picosecond O ptoelectro n ics." Sov. J. Q uantum E lectron. 20 (3). pp.
2 09 -227,
1990, and P. P. V asil'e v , "P ico seco n d Injection Laser: A N ew
T ech n iq u e for U ltrafast O -S w itch in g ." IE E E ,J. Q uantum E lectron. 24 (12).
D ecem ber 1988.
[83]
J. O 'G o rm a n . A. F. L evi, and W. S. H obson. “H igh P o w er S w itching o f
M u ltielectro d c Broad A rea L asers.” Electron. Lett. 27 (1). p. 13, 1991.
[84]
L. Y an. P. T . Ho, C. H. Lee, and G. L. B urdge, “G eneration o f U ltrash o rt Pulses
from a N eodym ium G lass L aser S y stem ," IE E E J. Q uantum E lectron. 25. pp.
2 4 3 1 -2 4 4 0 . 1989.
[85]
A. F. L evi. J. O 'G o rm an . D. D ykaar, W. S. H obson. T. T an b u n -E k . and R. A.
L ogan. "U ltrashort Pulse G eneration using Intracavity-M odulated Q uantum W ell L asers." Conf. on L asers and E lectro-O ptics -
CLEO
'91. C T u L l.
B altim ore. M D . 1991.
[86] D. R. D ykaar, K. B erthold. A. F. J. Levi, T. T an b u n -E k . and R. A. L ogan.
“ L arge-signal Picosecond R esponse o f In G aA s/ln P Q uantum W ell L asers w ith
an Intracavity Loss M odulator," Appl. Phys. Lett. 56. pp. 1 6 2 9 -1 6 3 1 , 1990.
[87]
E. A. W o lf and R. K aul, M icrow ave E n g in eerin g a n d S ystem s A p p lica tio n s.
Joh n W iley &. Sons. Inc., C hapter 5, 1988.
187
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[8 8 ]
M.
C.
R.
C arvalho
and
W.
M argulis.
“ L aser
D iode
Pum ping
w ith
a
T ran sm issio n Line T ran sfo rm er," IEE E M icro w a ve a m i G u id ed W ave Lett. 1
(12), 1991.
[89]
M. C. R. C arvalho, W. M argulis, and J. R. S ouza. “A N ew , S m all-S ized
T ran sm issio n Line Im pedance T ransform er, w ith A p p licatio n s in H igh-Speed
O ptoelectronics,'" IE E E M icrow ave a n d G u id ed W ave L ett., 2 (11), 1992,
[90]
A C A D E M Y ® is a registered product o f E E s o f Inc.
[91]
D U R O ID 6010® is a registered trade nam e o f R ogers C orp.. C handler, AZ.
[92]
N. G. P aulter. A. J. G ibbs, and D. N. S inha, "F ab ric atio n o f H igh-Speed G aA s
P h otoconductive Pulse G enerators and S am p lin g G ates by Ion Im p lan tatio n ,"
IE E E Trans, on E lectron D evices. 35 (12). pp. 2 3 4 3 -2 3 4 8 , D ecem ber 1988.
[93]
D. H. A u sto n , "Im pulse R esponse o f P h o to co n d u cto rs in T ran sm issio n L ines."
IE E E J . Q u a n tu m E lectron. 19. pp. 1 9 8 -2 2 0 , 1988.
[94]
R.
H.
B ube.
P hotoelectronic
P roperties
o f S em ico n d u cto rs,
C am b rid g e
U niversity Press, G reat Britain. C hapter 1. 1992.
[95]
E. A. C hau ch ard . C. C. K ung. C. H. Lee. M . J. R hee, an d V. D iadiuk. "A New'
M ethod to G enerate S quare Pulses: O p to electro n ic S w itch in g in a C urrent
C harged T ran sm issio n L ine." IE E E Trans. P la sm a Sci. 15 ( I) , Feb. 1987.
[96]
G. M. M artin, A. M ilonneau, and A. M ircea, “E lectro n T raps in Bulk and
E pitaxial G aA s C ry stals," E lectronics Lett. 13 (7). pp. 1 9 1 -1 9 3 . M arch 1987.
[97]
R. A. R oush, D. C. S toudt, and M. S. M azzola, “C o m p en satio n o f shallow
silicon d o n o rs by deep copper acceptors in g alliu m arsen id e." A ppl. Phys. Lett.
62 (21). pp. 2 6 7 0 -2 6 7 2 . 24 M ay 1993.
[98]
H. A. Z arem , et al, “ E ffect o f Al M ole F ractio n o n C arrier D iffusion L engths
and L ifetim es in A IxG a |_ xA s," Appl. Phys. Lett. 55 (25), pp. 2 6 2 2 -2 6 2 4 , 18
D ecem ber 1989.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
[99]
C. H. Lee, P ico seco n d O ptoelectronic D evices, A cad em ic Press, N ew Y ork, N Y ,
pp. 1 3 1 -1 3 2 , 1984.
[100] K. Secger, Sem ico n d u cto r Physics: A n Introduction, Fourth E dition, S p rin g er
Scries in Solid-State S ciences, Vol. 40, S pringer-V crlag, Berlin, p p. 4 1 6 -4 2 4 .
1988.
[ 101 ] J. R ivera and T. Itoh, U niversity o f T exas Lab. R eport 79-1. A ugust 1979.
also presented in M icrow ave E ngin eerin g a n d S ystem s A p p lica tio n s, by E. A.
W o lff an d R. K aul, John W iley & S ons. Inc.. N ew Y ork. N Y , pp. 1 9 4 -1 9 5 .
1988.
[102] T. Itoh (editor). P la n a r Transm ission L ine S tru ctu res, IE E E Press B o o k s, 1979.
[103] I. Bahl and P. B hartia, M icrow ave S o lid S ta te C ircuit D esign, Jo h n W iley &
Sons, N ew Y ork. NY. 1988, C h ap ter 2.
[104] S. L. H uang. O nA Vafer O ptoelectronic T echniques f o r M M IC A p p lica tio n s.
Ph.D . D issertation, U niversity o f M aryland, A ugust 1993.
[105] E E s o f T O U C H ST O N E ® U ser's M anual, Pub. 1.4-01186, January 1986.
[106] D. L. M azzoni, K. C ho, S. E. S ad d o w and C. C. D avis, " A H ybrid C o h eren t
Fiber-O ptic Probe for R em ote S ensing o f E lectro-O ptic E ffects in G aA s."
Optical F iber Sensor C onference O F S -8 . M onterey. CA . Jan. 29. 1992.
[107] M. D. S turge, "O ptical A bsorption o f G allium A rsen id e betw een 0 .6 and 2.75
eV ," Phys. Rev. 127 (3). A ugust 1962.
[108] S. M. Sze. P hysics o f Sem iconducto r D evices. John W iley & S o n s, Inc.. N ew
Y ork, N Y . second edition, p. 750. 1981.
[109] L. W est, T. C. H asenbcrg, A . K ost. M. M atloubian. G . C. V alley, and J. O.
W hite, "M odulation o f M illim eter W aves by O ptically Illu m in atin g n-i-p-i
S tructures," IE E E L E O S Sum m er Topicals: O p tical-M icrow ave In tera ctio n s.
M4.4, C on feren ce P roceedings. S anta B arbara. C A . July 19, 1993.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
11 10] J. Xu and M . Shur. "V elocity-F icld D ependence in G aA s," IE E E Trcms.
E lectron D evices 34 (8), A ugust 1987.
[111] W.
Franz,
“ EinfluB
eines
E lektrischen
F eldes
auf
eine
O p tisch e
A b so rp tio n sk an te," Z. N aturforsch. 13a, p.p. 484 (1958).
[112] L. V. K eldysh, “T he E ffect o f a S trong E lectric Field on th e O ptical P roperties
o f Insulating C ry sta ls," Sov.
Phys. JE T P 34 (7), pp. 7 8 8 -7 9 0 ,
N o v em b er 1957.
[1 1 3 JJ.
I. P ankove, O p tica l P rocesses in S em ico n d u cto rs.
P rentice-H all.
Inc..
E nglew ood C liffs, N J, 1971.
[114] W. T. W hite, C. G. D ease. M. D. Pocha, and G. II. K h an k a. "M o d elin g G aA s
H igh-V oltage.
S ubnanosecond
P hoto co n d u ctiv e
S w itch es
in
O ne
Spatial
D im ension," IE E E Trans. E lectron D evices. 37 (12). pp. 2 5 3 2 -2 5 4 1 , D ecem b er
1990.
[115] A. J.
S eeds
and
A.
A. A. de
S alles, "O ptical C o n tro l
of
M icro w av e
S em ic o n d u cto r D evices," IE E E Trans. M icrow ave T heory Tech. 38 (5).
pp. 5 7 7 -5 8 5 . M ay 1990.
[116] P. R. H erczfeld. and A. S. D aryoush. A. R osen. A. S harm a.
an d V. M.
C ontarino. “Indirect S ubharm onic O ptical Injection L o ck in g o f M illim eter W ave
IM P A T T O scillato r." IE E E Trans. M icrow ave T h eo ty Tech. 34 (12).
pp. 1 3 7 1 -1 3 7 6 , D ecem ber 1986.
[117] W. P latte an d B. S auerer, "O ptically C W -ln d u ced L o sses in S em ico n d u cto r
C oplanar W aveguides." IE E E Trans. M icrow ave T h eo ry Tech. 3 7 (1),
pp. 1 3 9 -1 4 9 , January 1989.
[118] S pectra D iode Labs N o. S D L -2372-P 3,
[119] W iitron C orp. M odel 360 V ector N etw o rk A nalyzer,
[120] A V T E C H C orp. M odel A V L -2C hig h -v o ltag e pulser.
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
[121] A nalog M odules, Inc. M odel 778M O D high vo ltag e pulser.
[122] A. M. Y urek, C. D. S triffler, and C. H. Lee,
"O p to e le ctro n ic D evices for
M illim eter W aves." in In fra red a n d M illim eter W aves, Vol. 14. M illim eter
C o m p o n en ts a n d Techniques, K. J. B utton, Ed. O rlan d o , FL: A cad em ic Press.
1985.
[123] A. M. Y urek. M illim eter-W ave T echnique f o r C h a ra cte rizin g the B eh a vio r o f
O ptically In d u ce d P lasm as in Sem ico n d u cto r W aveguides, Ph.D . D issertation,
U niversity o f M aryland, 1984.
[124] P. A. G laso w . "6H -S iC Studies an d D evelopm ents at th e C o rp o rate R esearch
L aboratory o f S iem ens AG and the Institute for A p p lie d P h y sics o f the
U niversity in E rlangen (F R G )," from
A m o rp h o u s a n d C rysta llin e Silico n
C arbide a n d R ela ted M aterials. S p rin g er-V erlag P ro ceed in g s in P hysics. 1988,
34. pp. 1 3 -3 3 .
[125] J. A. P ow ell and L. G. M atus, “ R ecent D ev elo p m en ts in SiC (U S A )." from
A m o rp h o u s a n d C rystalline S ilico n C arbide a n d R e la te d M a teria ls. S pringerV erlag P ro ceedings in P hysics, V ol. 34, pp. 2 - 1 2 , 1988.
[126] H. O k u m u ra. K. Endo, S. M isaw a, E. S akum a, and S. Y oshida, “ S tu d ies on
C arrier L ifetim e and D eep L evels in C V D -G ro w n 3C -S iC . ” from
A m o rp h o u s
a n d C rystalline Silico n C arbide a n d R ela ted M a teria ls II, S p rin g er-V erlag
P roceedings in Physics. 1989. Vol. 43, pp. 9 4 -9 9 .
[127] P. A. Ivanov and V. E. C heleno k o v . “ R ecen t D ev elo p m en ts in SiC SingleC rystal E lectronics." Sem icond. Sci. Technol. 1, pp. 8 6 3 - 8 8 0 (R ev iew A rticle),
(1992).
[128] M. B hatnagar and B. J. Baliga, “ C o m p ariso n o f 6 H -S iC , 3 C -S iC , and Si for
P ow er D evices," IE E E Trans, on E lectro n D evices. 40 (3 ), pp. 6 4 5 -6 5 5 ,
M arch. 1993.
191
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
[129] V. I. G avrilenko, et al. “ Energy B and S tructure and O ptical P ro p erties o f
W urtzite-S tructure Silicon C arbide C ry stals." Phys. Stat. Sol. (h). 162. p. 447
(1990).
[130] V. B. Shields, K. Fekadc, and M. G. Spencer. “N ear-E q u ilib riu m G ro w th o f
Thick. H igh Q uality B eta-S iC by S u b lim atio n ." Appl. Phys. Lett. 62. (16).
pp. 1 919-1921. 1993.
[131] J. W. Palm our. R. F. D avis, P. A stell-B urt. and P. B. B lackborow . "E ffects o f
C athode M aterials and G as Species on the Surface C h aracteristics o f D ry E tched
M onocrystalline B eta-SiC T hin F ilm s.” C era m ic Trans. 2. pp. 4 9 1 -5 0 0 . 1989.
[132] J. C rofron. P. A . Barnes. J. R. W illiam s, and J. A. E dm ond. “C o n tact R esistance
M easurem ents o n P-Type 6H -S iC ," Appl. Phys. Lett. 62 (4). pp. 3 8 4 -3 9 6 . 25
January 1993.
[133] J. A. Edm ond. IT. S. Kong, and C. IT. Carter. "H ig h T em p eratu re R ectifiers. UV
P hotodiodes, an d B lue L E D s." from
A m o rp h o u s a n d C rystalline Silicon
C arbide a n d R e la te d M aterials IV. S pringer-V erlag P roceedings in P hysics. Vol.
7 1 . pp. 3 4 4 -3 5 1 . 1992.
[134] D. L. ITallen, Y. C ui. W. T. H ill. T. M ikes and J. G o ld h ar. “G en eratio n o f
Intense 10-ps. 193-nm Pulses U sing Sim ple D istributed F eedback D ye Lasers
and an A rF* A m p lifier." A ppl. Opt. 31 (33). 20 N o v em b er 1992.
[135] D. M. Brow n, e t al. "S iC U V P hoto d etccto rs," IE E E Trans. Nuc. Sci. 3 9 (6). pp.
1 9 7 4 -1 9 8 1 . D ecem ber 1992.
[136] J. M. M cG arrity. F. B. M cLean. W. M . D elancey. J. P alm our. C. C arter. J.
E dm ond and R . E. O akley, "S ilicon C arbide JF E T R adiation R esponse," IE E E
Trans. E lectron D evices. 40 (2). pp. 3 2 5 -3 3 3 . February. 1993.
[137] D. V. Lang. “D eep-level transient spectroscopy: A N ew M ethod to C haracterize
T raps in S e m ic o n d u c t o r s J. Appl. P hys. 45 (7). pp. 3 0 2 3 -3 0 3 2 . 25 Ju ly 1974.
192
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
[138] D. C. Look, E lectrical C haracterization o f G aA s M aterials a n il D evices. John
W iley & Sons. N ew Y ork, N Y , C hapter 4, 1989.
[139] J. S chneider and K. M aier. “ P oint D efects in Silicon C a rb id e," P h ysica B 185
(1), pp. 199, 1993.
[140] C ree R esearch. Inc., 2810 M eridian P arkw ay, Suite 176, D urham , N C . 27713.
[141] J. A . Pow ell, D. A. Larkin, J. B. Petit, and L. G. M atus. “A dv an ces in Silicon
C arbide
C hem ical
F abrication,"
V apor
D eposition
in P roceedings
o f the
(C V D )
for
S em ico n d u cto r
Ist In tern a tio n a l H igh
D evice
T em perature
E lectronics C onference. A lbuquerque. N M , pp. 1 9 2 -1 9 7 , 1991.
[142] J. A. Pow ell, D. A. Larkin, J. B. Petit, and J. H. Edgar. "G ro w th and
C haracterization o f 3C -S iC and 6H -S iC Film s on 6H -S iC W afers." A m o rp h o u s
a n d C rystalline Silicon C arbide IV, edited by M. M. R. C. Y. Y ang an d G. L.
H arris. S pringer Proceedings in Physics V ol. 71, pp. 2 3 -3 0 , 1992.
[143] D. A. L arkin, P. G. N eudeck. J. A. Pow ell, and L. G. M atus. "In v estig atio n o f
the G row th o f 3C -S iC and 6H -S iC Film s o n L ow -T ilt-A ngle V icinal (0 0 0 1 ) 6HSiC
W afers," International C onference
on Silicon
C arbide
and
R ela ted
M aterials (IC SC R M '93), C onference P roceedings, W ashington, D C . N ovem ber.
1993.
[144] W. K. H. Panofsky and M. P hillips, C la ssica l E lectricity a n d M agnetism .
S econd E dition. A ddison-W esley Publishing Co.. R eading. M A , C h ap ter 4.
1962.
[145] J. G leck, The Life a n d Science o f R ich a rd F eynm an. P anthenon B ooks. N ew
York. N Y , p. 428. in R eport o f the P resid en tia l C o m m issio n o n th e Space
S huttle C hallenger A ccident. 1986.
R ep ro d u ced with perm ission o f th e copyright ow ner. Further reproduction prohibited without p erm ission.
[146]
J. W. Faust, Jr., “T ables o f d ata o n silicon carb id e," S ilico n C arb id e-1 9 7 3 , ed. R.
C.
M arshall, J. W. Faust, Jr.. C. E. R yan, U niv. o f S o u th C a ro lin a Press,
C olum bia, SC , pp. 6 6 8 -6 7 6 . 1973.
194
R ep ro d u ced with p erm ission o f th e copyright ow ner. Further reproduction prohibited w ithout perm ission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Curriculum Vitae
N am e:
S tephen E dw ard Saddow .
P erm an en t address:
6863 H ap p y h eart Lane.
C olum bia, M aryland, U nited S tates o f A m erica.
D egree an d date conferred:
D octor o f P hilo so p h y , 1993.
D ate o f birth:
N o v em b er 7, 1961.
Place o f birth:
Jersey C ity. N ew Jersey, U n ited States o f A m erica.
S eco n d ary education:
L ivingston H igh School, L iv in g sto n , N ew Jersey
C o lleg iate institutions attended:
Institution
Dates Attended
Degree
Date o f Degree
W estern N ew E ngland C ollege
S p rin gfield. M assachusetts
9/79 - 5/83
B.S.E.E.
M ay 1983
P o ly tech n ic U niversity
B rooklyn. N ew Y ork
9/85 - 5/88
M .S.E .E .
January 1988
U niv ersity o f M aryland
C o lleg e Park, M aryland
9 /8 8 - 1 2 / 9 3
Ph.D.
D ecem ber 1993
M ajor:
M inors:
E lectrophysics in E lectrical E ngineering.
M icroelectronics and C ontrol S ystem s in E lectrical E ngineering.
P rofessio n al publications:
J o u r n a l p a p ers:
1.
S. E. Saddow , B. J. T hedrez, and C. H. Lee, “A n O ptoelectronic A tte n u ato r for
the C ontrol o f M icrow ave C ircuits.” IE E E M icro w a ve G u id ed W ave Lett. 3 (11).
pp. 3 6 1 -3 6 2 , O cto b er 1993.
2.
B. J. T hedrez, S. E. S addow . Y. Q. Liu and C . H. Lee, “ E xperim ental and
T h eo retical Investigation o f Large O u tp u t B road A rea A lG aA s S em ico n d u cto r
L ase r D iodes, " IE E E P hotonics Tech. Leii. 5 ( 1 ), pp. 1 9 -2 3 , January 1993.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission .
3.
S. H. Y ang, B. J. T hedrez, S. E. S addow , Y. Q . L iu and C. H. Lee. "C ro ssC o rrelatio n M easurem ent o f the T urn-on D elay and P ulsew idth o f a ^ -s w itc h e d
T w o -sectio n S em iconductor Laser,” IE E E P h o to n ic s Tech. Lett. 5 (12).
D ecem ber 1993.
4.
M . S. M azzola, S. E. S addow . P. N cudeck. and V . K. L akdaw ala, "O b serv atio n
o f the D -C en ter in C V D -G row n p-n D io d es,” A ppl. Phys. L ett., su b m itted for
publication.
C onference p apers:
1.
S. E. S addow , B. J. T hedrez, A. B alekdjian. and C. H. Lee. "A n O p tically
C o n tro lled M icrow ave A ttenuator,” IE E E L E O S S u m m er Topicals: O pticalM icrow ave Interactions, M 1.4. C onference P ro ceed in g s. S anta B arbara, C A , Ju ly
19, 1993.
2.
A. K im . S. E. Saddow , R. Y oum ans, L. J. Jasp er. M. W einer, and C. H. Lee.
"P h o to co n d u ctiv e M onolithic W ideband T ra n sm itter C h aracterizatio n ." IE E E
L E O S S u m m er Topicals: O ptical-M icrow ave In tera ctio n s. W 1 .4 , C o n feren ce
P roceedings. S anta B arbara, C A , July 19. 1993.
3.
S. E. S addow , P. S. Cho. J. G oldhar, F. B arry M cL ean. J. W. P alm o u r and C H.
Lee, "L ateral and V ertical P -T ype 6H -S iC P h o to co n d u ctiv e S w itch R esp o n se,"
International C onference on S iC and R elated M aterials-/G S C J?A /'9 J, C o n feren ce
D igest. S pringer-V erlag, W ashington, D C , N o v em b er 1993.
4.
S. E. S addow , B. J. Thedrez, S. L. H uang, T. J. M erm agen. and C. H. Lee. "A n
Investigation o f the T em perature and E lectric Field D ependence o f a G aA s
M icrow ave P hotoconductive S w itch," S P IE -O E /L siS E '93: O p tically A ctivated
S w itching, conference proceedings. 1873-31, Los A ngeles, CA , Jan u ary 1993.
5.
S. E. S addow , P. S. C ho, J. G oldhar, J. W. P alm our an d C H. Lee.
"P h o to co n d u ctiv e M easurem ents on P -T ype 6 H -S iC .” S P IE -O E /L A S E '93:
O ptically A ctivated S w itching, conference p ro ceed in g s, 1873-33, Los A ngeles.
CA , January 1993.
6.
S. E. Saddow , P. S. Cho, J. G oldhar, J. W. P alm o u r and C H. L ee, "H ig h -S p eed
P hotovoltaic Response o f P-Type 6 H -S iC ," N in th A n n u a l IE E E S a r n o ff
Sym p o siu m P roceedings, Princeton. NJ. M arch 26, 1993.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited without p erm ission.
7.
S. E. S addow . P. S. Cho. J. G oldhar, J. W. P alm o u r and C. IT. Lee. ” P -T ype 6ITSiC P hotoconductive S w itches," 20th IE E E In tern a tio n a l C o n feren ce on P lasm a
S cience, V ancouver, B.C. (C anada), 7 - 9 June 1993.
8.
D. L. M azzoni, K. Cho, S. E. S addow and C. C. D avis, “ A H ybrid C o h eren t
F iber-O ptic Probe for R em ote Sensing o f E lectro-O ptic E ffects in G aA s."
O ptical F iber S en so r C onference O F S -8 , M onterey, CA, Jan u ary 1992.
9.
K.. A. B oulais, J. Y. Choe, S. E. Saddow , S. T. C hun. K. Irw in an d M. J. Rliee,
“ A nalysis o f a High C urrent M icro-C hannel E lectron S o u rce." A n n u a l TriService/N A SA C o ld C athode W orkshop, G rccnbelt, M D , M ar 1992.
10. K. A. B oulais, J. Y. Choe, S. T C hung, A. K rall. K. Irw in, S. S ad d o w and M . J.
Rhee. "T h e N SW C M icro-channel E lectron S ource P ro g ram ," B E A M S '92,
W ashington. D C ., M ay 1992.
P rofessio n a l p o sitio n s held:
7/88 - present:
E lectronics E ngineer. U .S. A rm y R esearch L aboratory,
W eapons T echnology D irectorate, A delphi, M aryland.
6/83 - 6/88
E lectronics E ngineer. U .S. A rm y A rm am en t R esearch &
D evelopm ent C enter, P recision M unitions B ranch,
Picatinny A rsenal, N ew Jersey.
5/82 - 5/83
P roduct E ngineer, E lectronic C o ils, inc.. P ro d u ct D ev elo p m en t
Lab.. S pringfield. M assachusetts.
R ep ro d u ced with p erm ission o f the copyright ow ner. Further reproduction prohibited w ithout p erm ission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
6 740 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа