EMPIR 14IND02 - Plana Cal
Guidelines o he design o calib a ion subs a es,
including he supp ession o pa asi ic modes o
equencies up o and including 325 GHz
Ma co Spi i o, TUD
Uwe A z, PTB
Gia Ngoc Phung, F anz Jose Schmückle, Wol gang Hein ich, FVB
Roge Loza , FhG
h ps://plana cal.p b.de
July 2018
2
This Guide has been p oduced wi hin he EURAMET p ojec en i led Mic owa e measu emen s o plana ci cui s
and componen s. Mo e in o ma ion abou his collabo a i e esea ch p ojec can be ound on he p ojec ’s websi e
h ps://plana cal.p b.de .
Disclaime
Any men ion o comme cial p oduc s wi hin his Guide is o in o ma ion only; i does no imply ecommenda ion o
endo semen by he pa ne s in his p ojec .
The iews exp essed in his Guide a e hose o he au ho s and o he EMPIR 14IND02 p ojec eam.
Acknowledgemen o unding
The p oduc ion o his Guide was unded by he Eu opean Me ology P og amme o Inno a ion and Resea ch
(EMPIR). The EMPIR ini ia i e is co- unded by he Eu opean Union’s Ho izon 2020 Resea ch and Inno a ion
P og amme and he EMPIR Pa icipa ing S a es.
Au ho ship
P epa a ion o his Guide was led by Ma co Spi i o o he Technical Uni e si y o Del (TUD), Del (Ne he lands)
wi h ex ensi e inpu om all he membe s in ol ed o he EMPIR 14IND02 p ojec eam. The discussion and inpu
o all he pa ne s in he p ojec and hei colleagues a e g ea ly app ecia ed.
Sugges ion o he quo a ion o he e e ences
Spi i o, Ma co ; A z, Uwe ; Phung, Gia Ngoc ; Schmückle, F anz Jose ; Hein ich, Wol gang ; Loza , Roge .
Guidelines o he design o calib a ion subs a es, including he supp ession o pa asi ic modes o equencies up
o and including 325 GHz : EMPIR 14IND02 – Plana Cal, 2018. Physikalisch-Technische Bundesans al (PTB).
DOI: h ps://doi.o g/10.7795/530.20190424A
This documen and all pa s con ained he ein a e p o ec ed by copy igh and a e subjec o he C ea i e Commons
use license CC by 4.0 (h ps://c ea i ecommons.o g/licenses/by/4.0/).
3
Guidelines o he design o calib a ion subs a es,
including he supp ession o pa asi ic modes o
equencies up o and including 325 GHz
Deli e able pa icipa ing Pa ne s
TU Del
Technische Uni e si ei Del
Ne he lands
PTB
Physikalisch-Technische Bundesans al
Ge many
FhG
F aunho e -Gesellscha zu Foe de ung de
angewand en Fo schung e.V.
Ge many
FVB
Fo schungs e bund Be lin e.V.
Ge many
4
Con en s
Con en s ................................................................................................................................................ 4
1. P e ace .............................................................................................................................................. 5
2. In oduc ion ........................................................................................................................................ 5
2.1. Mul i-mode p opaga ion .................................................................................................................. 7
2.2. Bounda y condi ions (bounded-unbounded) ................................................................................... 8
2.3. Cha ac e is ic impedance de e mina ion ......................................................................................... 9
3. Mode supp ession ............................................................................................................................ 10
3.1. S uc u e op imiza ion o coplana wa eguides............................................................................ 10
3.1.1 In luence o CPW g ound wid h .................................................................................................. 11
3.1.2 Impac o g ound- o-g ound spacing ........................................................................................... 13
3.2 P obe opology .............................................................................................................................. 14
3.3 Chuck opology ............................................................................................................................. 17
3.4 Design guidelines ela ed o he p obe cons uc ion o coplana wa eguides ............................... 19
3.5 Thin- ilm Mic os ip lines (TFMSL) ................................................................................................. 22
3.6 Design guidelines ela ed o hin- ilm mic os ips .......................................................................... 27
4. Example o cus om designed TRL ki ............................................................................................... 28
Acknowledgemen s ................................................................................................................................. 30
Bibliog aphy ............................................................................................................................................ 30
5
1. P e ace
The scope o his guideline is o make he eade awa e o he cons ain s a ising when a emp ing p obe
le el calib a ion a highe equencies, especially abo e 110GHz. The causes o gene a ion o un-wan ed
adia ion modes a e i s b ie ly desc ibed, and an app oach o iden i y hei impac on he calib a ion
quali y is p o ided o he eade wi h gene al i s o de guidelines. The epo hen desc ibes he
app oaches ha ha e been in es iga ed and quan i ied as iable me hods o educe he gene a ion o
unwan ed modes. Finally, an example o a cus om buil Th u Re lec Line (TRL) ki wi h sugges ions on
how o c ea e he lines and he e lec condi ions a e gi en o illus a i e pu poses.
Finally, i is wo h o men ion ha since he unde s anding, modelling and de ini ion o solu ions o mm-
wa e calib a ion is a ela i ely new esea ch opic, his guide p esen s he de i ed knowledge o an
ongoing wo k in he la ge scien i ic communi y.
2. In oduc ion
The accu acy o S-pa ame e measu emen s o any de ice unde es (DUT) is se , a he i s o de , by
he quali y o he Vec o Ne wo k Analyze (VNA) calib a ion. This is ypically pe o med by measu ing a
ce ain numbe o known de ices (i.e., he calib a ion s anda ds). Depending on he speci ic calib a ion
echnique employed, he quali y o he calib a ion is dependen on he accu acy wi h which he calib a ion
s anda ds a e known/modelled.
Fo he in e es ed eade a comp ehensi e desc ip ion o calib a ion echniques can be ound in [1].
When conside ing plana de ices, o which wa e -p obes need o be employed, i is common p ac ice o
pe o m a p obe-le el calib a ion ( i s - ie ) using a low-loss subs a e (i.e., alumina o used silica), which
can hen be ans e ed o he en i onmen in which he DUT is embedded.
To inc ease he DUT measu emen accu acy, he es ix u e (i.e., pad and in e connec lines) whe e he
de ice is embedded can be u he emo ed wi h a de-embedding s ep o a di ec de ice le el calib a ion
[2].
Calib a ion echniques in which li le knowledge o he s anda ds is equi ed, like Line Re lec Ma ch
(LRM) [3] and TRL [4], end o be p e e ed when ope a ing in he mm-wa e equency ange (i.e., in he
con ex o his documen especially abo e 110GHz), since his educes he calib a ion e o s a ising om
he inaccu acy in he s anda d models.
Bo h TRL and LRM calib a ions de ine he e e ence plane a he cen e o he (non-ze o) h u s anda d,
and no a he p obe ips, o his eason when e e ing o a i s - ie (p obe le el) calib a ion a back shi
on he calib a ion e e ence plane needs o be applied (no e, ha his will employ he knowledge o he
ansmission line elec ical pa ame e s).
The well-known ela ion be ween equency and wa eleng h (), i.e., = / , whe e is he phase eloci y
in he medium (c/√ ), highligh s he impo ance o e e ing o he elec ical leng h o a medium a he
han i s mechanical one.
In simple wo ds, as he equency inc eases, assuming a ixed ma e ial hickness, a la ge po ion o he
wa e can be included in he dielec ic slab, as shown in he ske ch o Fig. 1 a). Mo eo e , also as he
pe mi i i y o he dielec ic inc eases (la ge  ), o a gi en equency, a la ge po ion o he wa e can

6
be included in he subs a e Fig. 1 b). Bo h he men ioned cases (i.e., highe equency and highe  ) can
lead o wa e adia ion/ e lec ion phenomena.
a) b)
Fig. 1: Simpli ied ske ch o he e ec o wa eleng h educ ion wi h equency o a gi en subs a e pe mi i i y a),
and o he same se o equencies o a subs a e wi h an highe ( ou imes) pe mi i i y b).
Comme cially a ailable calib a ion ki s, employ only homogenous Coplana Wa eguides (CPW)
ansmission lines s uc u es o pe o m he calib a ion p ocedu e, since a wid h s ep in he line, would
esul in a discon inui y (modelled as a se ies eac ance, as shown in Fig. 2 [5]), inc easing he e o
when de ining he e e ence plane a he p obe ips, o equi ing ex a co ec ion s eps o emo e he pad
and s ep wid h e ec s.
Fig. 2: a) op iew o he CPW line ske ch p esen ing a s ep wid h be ween he pad and he line, b) simula ion se up
o he h u employing a single shun capaci ance o modelling he pad, and he lumped model in oduced in [6] o
he pad- o-line ansi ion.
Mo eo e , he p obe landing a ea, o en e e ed o as he sc a ch ma k a e ouchdown and ska ing o
he p obe o each he equi ed (ad ised by he manu ac u es) o e - a el o ob ain a good (low) ohmic
con ac , does no scale wi h he ope a ional equency o he p obe. In Fig. 3 he landing a ea o he
same p obe model om company #1 a e shown o he WR10, WR5 and WR3 e sion, in Fig. 3 a), b)
and c) espec i ely. Fig. 3 d) and e) p esen s he landing a ea o he WR3 model o p obes p oduced
om company #2 and company #3, espec i ely.
The illus a ions in Fig. 3 a e only quali a i e and a e no p o ided o ep esen he nominal esponse o
he p obes, bu p o ide a isual jus i ica ion o he ixed dimension o he signal line o he CPWs used in
mos o he comme cially a ailable calib a ion ki s.
As a esul , he high dielec ic cons an employed in comme cially a ailable subs a es, wi h he la ge
gaps in he CPW design ( equi ed o balance he L p o ided by he ixed line wid h and p o ide a 50Ohm
nominal Z0 o he line) leads o a ious unwan ed modes being p esen in he lines used o p obe le el
calib a ion in he equency ange abo e 110 GHz.

02 04 0
4*
02 04 0
LSa
CPad
Po 1
Z=50 Ohm
LSb
CS
50 µm
50 µm
W
LSa
CPad Po 2
Z=50 Ohm
LSb
CS
CPW line
a)
b)
7
Fig. 3: Landing a ea o p obe model om company #1 o he WR10 a), WR5 b) and WR3 c) equency band, p obe
model om company #2 o he WR3 d) and p obe model om company #3 o he WR3 e).
2.1. Mul i-mode p opaga ion
The di e en p opaga ing modes suppo ed by a CPW a e quali a i ely ske ched in Fig. 4. The CPW
mode cha ac e ized by opposi e di ec ion o he ields ac oss he slo s, ep esen s he in ended
p opaga ion mode and is o en e e ed o as CPW di e en ial mode. The CPW mode cha ac e ized by
Fig. 4: C oss sec ion o a CPW wi h ini e g ound planes, and ske ches o he E ield dis ibu ions o he i s
p opaga ing modes suppo ed.
in-phase di ec ion o he ield ac oss he slo s ep esen s an unwan ed adia ing mode and is o en
e e ed o as CPW common mode.
The TMn and TEn modes a e su ace wa es p opaga ing along he dielec ic slab, hei cu o equency
is a unc ion o he heigh , dielec ic cons an o he subs a e and he bounding condi ions. A he
equencies in which he lowes dispe si e mode exhibi s a p opaga ion cons an highe han he
dominan CPW mode, he la e is no bounded anymo e and ene gy leakage can occu om he nominal
CPW mode o he un-wan ed adia ion mode. This c i ical equency will depend on a ious pa ame e s:
subs a e pe mi i i y, heigh and bounda y condi ions. The bounda y condi ions will de ine he lowe
o de dispe sion mode (i.e., su ace wa e) ha can p opaga e.
In he p ac ical case o calib a ion subs a es, wo condi ions a e conside ed:
1) Me al chuck: his condi ion p o ides e ec i ely a g ound back plane o he CPW which will
cancel he TE0 mode o his con igu a ion, he TM0 is hen he i s dispe si e mode. The c i ical
equency o his con igu a ion was shown in [8], and is epo ed in eq. 1.
00
a c an( )
2 ( 1)
Me al
h

   
=     −
(1)
Whe e,

ep esen s he subs a e pe mi i i y, and h i s heigh .
WS
WGap
WGND WGND
CPW common mode
CPW di e en ial mode TM0 su ace wa e mode
TE1 su ace wa e mode
d
WGap
8
2) Dielec ic chuck: when he calib a ion subs a e is placed o e a dielec ic chuck p esen ing he
same pe mi i i y, he i s unwan ed mode o p opaga e is he TE0 which will c ea e leakage
e ec s a equencies abo e Dielec ic, see eq. 2, as desc ibed in [8].
00
2
2 ( 1)
Dielec ic
W
  
=   −
(2)
Whe e W is he o al wid h o he CPW line, including he g ound planes. Bo h eq. 1 and eq. 2 show ha
he c i ical equencies allowing leakage e ec s depends on he subs a e pe mi i i y ( ). In bo h cases
he usage o a high pe mi i i y ma e ial will lead o a lowe alue o he c i ical equency whe e ene gy
leakage will s a o occu . In he bounded case eq. 1, he usage o hin subs a es also shi s o highe
equencies he s a o he leakage e ec . I is impo an o no e ha eq. 2 ep esen s an idealized case
whe e no dielec ic discon inui ies exis , lea ing he ene gy ha has leaked o he unwan ed mode o low
away om he s uc u e.
Guideline #1
Compu e he c i ical equency ( c) o he calib a ion subs a e employed in he gi en con igu a ion (i.e.,
me allic o dielec ic chuck) and selec hose in which c is ou side he calib a ion equency o only
occu s in he uppe calib a ion ange, o a oid excessi e coupling o powe o unwan ed modes.
2.2. Bounda y condi ions (bounded-unbounded)
The e ec o (i.e., e o a ising om) he p esence o unwan ed modes can be, o a i s app oxima ion,
quan i ied by means o elec o-magne ic (EM) simula ions. The bounda y condi ions in a eal en i onmen
will p esen e ec comp ised be ween he me al chuck condi ion (i.e., ull e lec ion a he back side,
p o iding a wo s case condi ion Fig. 5 a) and he ma ch/ open bounda y condi ion (i.e., in ini e medium
a he back side, p o iding a bes case condi ion Fig. 5 b).
a) b)
Fig. 5: Alumina subs a e in ensi y o he E ield shown in dB scale, using ixed scale ange, a 200GHz, o he
ollowing lowe bounda y condi ion se ings: a) pe ec elec ical conduc o , b) adia ion open (ma ching laye ).
Fo his eason we can use he di e ence o he wo s case bound me ic [9], see eq. 3, in he wo cases
o p o ide a quali a i e indica ion o he accu acy ha he calib a ion on he gi en subs a e can achie e.
𝑊𝐶𝐵(𝑓)= max|𝑆𝑖𝑗
′(𝑓)− 𝑆𝑖𝑗(𝑓)| (3)
To p o ide some in o ma ion on he sensi i i y o he calib a ion accu acy o he bounding condi ions, we
conside he WCB de ia ion compu ed as he dis ance be ween he me ic in he wo bounding
condi ions. This de ia ion is shown o bo h he Alumina and Fused Silica case in Fig. 6 (le axis). The
peak o peak a ia ion o he cu e can be in e p e ed as he po en ial a ia ion o he WCB due o
bounda y condi ion a ia ions. While he me al chuck condi ion is an ex emely pessimis ic case, due o
he common p ac ice o e i e in e pose , Fig. 6 s ill p o ides in o ma ion on he highe se e i y, in he
calib a ion accu acy, o no pe ec e mina ion in high pe mi i i y subs a es.
Me al chuck condi ion (PEC bounda y)
9
No e, ha he bound in eq. 3, needs o use he same no maliza ion impedance o he S’ and S ma ices,
mo eo e he bound will also inco po a e he impac o he p opaga ion o e o h ough he chosen
calib a ion p ocedu e gi en he ac ha co ec ed da a a e compa ed.
When eading he igh axis o Fig. 6, he mean alue o he WCB o he bounda y ma ch case (ideal) is
p esen ed o bo h calib a ion subs a es. The inc eased (a e age) e o in TRL calib a ions on high
pe mi i i y subs a es can be associa ed o he ex emely complex modeling equi emen s ( hus p one o
e o s) o he line pa ame e s (i.e., Z0) o p ope ly accoun o he loss mechanism connec ed o highe
o de modes.
Fig. 6: Wo s case bound di e ences be ween he ma ch (dielec ic) and PEC (me al) bounda y condi ions o bo h
Alumina and Fused Silica subs a es.
Guideline #2
The WCB di e ence be ween he ideal (ma ch) and wo s case (me al) bounda y condi ion, ex ac ed
om he simula ed esponse o he calib a ion s uc u es, can be used o in o m on he sensi i i y o he
calib a ion subs a e o he eal bounding condi ion employed.
2.3. Cha ac e is ic impedance de e mina ion
To p ope ly desc ibe he beha iou o he ansmission lines used du ing he calib a ion p ocess in bo h
TRL and LRM p ocedu es he cha ac e is ic impedance o he line needs o be p ope ly de ined.
The gene alized equi alen ci cui o a ansmission line whe e mul i-mode p opaga ion is p esen can be
ep esen ed as shown in Fig. 7.
Fig. 7: Gene alized equi alen ci cui o a ansmission line wi h dis ibu ed adia ion, whe e R' and G' a e
espec i ely he se ies adia ion esis ance and he shun adia ion conduc ance pe uni leng h [7].
The p esence o he adia ion esis ance and conduc ance equi es he usage o accu a e modelling
s a egies o p ope ly ex ac he eal and imagina y pa o he cha ac e is ic impedance o lines used in
he calib a ion p ocess.
16
c)
d)
Fig. 15: Calib a ed esul s o he magni ude o S21 o a CPW o 500 µm (a and b) and 7400 µm (c and d) leng h;
le : Measu emen s wi h di e en p obes; igh : Simula ions wi h di e en exci a ions compa ed o model o [8]
(legend shown in op g aphs).
Al hough he de ia ions be ween he p obes do no appea o be e y signi ican , pa icula ly o he sho
line, Fig. 15 e eals an in e es ing beha io . The same CPW measu ed wi h di e en p obes shows
di e en esul s, wi h de ia ions up o 0.2 dB o he long line (Fig. 15 c). The ocus o ou in es iga ion is
no p ima ily o de e mine he maximum de ia ion. I is o explain how he p ope ies o he p obe can
change he esul ing calib a ed da a. Fo a sho CPW line o 500 µm leng h, one no mally assumes ha
he cu e beha io would ollow a smoo h unc ion wi h a sligh inc ease a highe equencies
compa able o he model o [8] in Fig. 15 (b). Howe e , he measu ed calib a ed esul s (Fig. 15 a and c)
as well as he simula ed esul s wi h P obes 1 and 2 exhibi a wa e-like beha io which does no appea
a all in he model o [8] o he simula ion wi h b idge exci a ion (leas pa asi ics). The simula ion wi h
p obe exci a ion, on he o he hand, e eals he a ious pa asi ics o he p obe exci a ion, i.e., coupling
om p obe o subs a e modes, adia ion, and inging ields be ween he p obe and o he s uc u es.
This means ha he easons o he unphysical cu e beha io o he p opaga ion cons an s can be
a ibu ed o he p obes, in combina ion wi h he calib a ion p ocess.
a)
b)
020 40 60 80 100
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
S21 (dB)
F equency (GHz)
020 40 60 80 100
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
S21 (dB)
F equency (GHz)

17
c)
d)
Fig. 16: Field plo s: magni ude o he elec ic ield o a 700 µm long CPW exci ed wi h di e en p obes; le (a and
c): p obe 1; igh (b and d): p obe 2, op: a 50 GHz; bo om: a 100 GHz.
The ield plo s shown in Fig. 16 illus a e he in luence o he p obe cons uc ion (no e ha he le -hand
p obe is loca ed a he wa e edge). Fig. 16 shows ha , depending on he geome y o he p obe needles
and he abso be cons uc ion, he ields a ound he p obe ansi ion a e dis ibu ed di e en ly. The
elec ic ields wi hin he ai egion below he le -hand p obes di e om he ields below he igh -hand
side p obes. Also, he ields in he close icini y o he CPW and he p obe ips di e , as do he inging
ields be ween he p obes, which a e clea ly highe as in he case o he b idge exci a ion.
3.3 Chuck opology
Bo h phenomena ( adia ion depending on he c oss-sec ion dimensions o he CPW as well as he impac
o he p obe ype) a e ela ed o he gene a ion and p opaga ion o subs a e modes. The pa asi ic e ec s
due o he p opaga ion o he subs a e mode canno be comple ely a oided o his laye ed sys em. Bu
hey a e s ongly dependen on he pe mi i i ies o he chuck ma e ial and he wa e .
a)
b)
Fig. 17: Ve ical elec ic ield componen exci ed wi h p obe 2 a 100 GHz (c oss sec ion); a): on ce amic chuck
 ,chuck = 6.0; b): on chuck wi h same pe mi i i y as he wa e , i.e.,  ,chuck = 9.7. Due o ini e disc e iza ion cells used
in simula ion only a mode a e esolu ion o he ield pa e n can be displayed.
18
a)
b)
Fig. 18: Field plo s: magni ude o he elec ic ield a 100 GHz ( op iew) exci ed wi h p obe 2 o nominal CPW
g ound wid h; a): on ce amic chuck  ,chuck = 6.0; b): on chuck wi h same pe mi i i y as he wa e , i.e.,  ,chuck = 9.7.
E.g., i he wa e is placed on a chuck wi h same pe mi i i y as he wa e , he subs a e mode anishes
because hen wa e and chuck o m a homogeneous medium, which does no suppo a su ace wa e
mode any mo e. Fig. 17 and Fig. 18 suppo hese s a emen s, compa ing he e ical elec ic ield
componen o he CPW line (20400 µm) on di e en chuck ma e ials. I he chuck pe mi i i y is lowe
han ha o he wa e (he e, a ce amic chuck wi h  ,chuck = 6 is used), one obse es a supe posi ion o
adia ion e ec s and p opaga ion o a subs a e mode which is gene a ed a he p obe ips due o he
laye ed s uc u e. Acco dingly, in Fig. 18 (a), he elec ic ields sp ead o e he whole wa e and a e
e lec ed back a all discon inui ies and edges.
Fig. 19: Calib a ed simula ion esul s o he magni ude o S11 o a 20400 µm long CPW on di e en chuck
ma e ials.
020 40 60 80 100
0.00
0.01
0.02
0.03
0.04
0.05
|S11| (linea )
F equency (GHz)
Chuck wi h
 = 6.0
 = 9.7
19
Fig. 20: Calib a ed simula ion esul s o he magni ude o S21 o a 20400 µm long CPW on di e en chuck
ma e ials.
I he chuck has he same pe mi i i y as he wa e (Fig. 17 and Fig. 18 b), he subs a e modes a e
supp essed. The e o e, he la e al sp eading ields do no appea anymo e and also he e lec ion
coe icien (see Fig. 19) exhibi s lowe alues. This in u n leads o ewe ipples and a smoo he cu e
beha io in he ansmission coe icien S21 (see Fig. 20).
Guideline #5
Using o he chuck a ma e ial which has a pe mi i i y alue simila o he calib a ion subs a e educes
he e ec s which con ibu e o he deg ada ion o he accu acy o CPW mTRL calib a ions.
Fu he in es iga ions ha e shown ha his is ue also o a chuck ma e ial wi h a pe mi i i y la ge
han ha o he wa e , because such a laye ed s uc u e does no suppo su ace wa es ei he .
3.4 Design guidelines ela ed o he p obe cons uc ion o coplana wa eguides
In o de o cla i y he impac o p obes, his sec ion will ocus mo e on e ec s ela ed o he p obe and he
in e ac ion wi h i s icini y. In gene al, he e a e essen ially ou easons de ec able – he p obe, he line
s uc u es, he ci cui design (densi y, pa e n) and he wa e s uc u e (also chuck) and ma e ial. Many
e ec s a e ela ed o he p obe. I s size and shape, he GSG needle con igu a ion (leng h, dis ance,
opening o he coaxial line), he p obe slan backwa ds o he p obe head and he abso be ma e ial
con ibu e o he inging ields and hus gene a e modes. In o de o simpli y he s uc u es bu keep he
e ec s alid we designed a i icial p obes. Fo he mTRL calib a ion we used line leng hs o 550 µm
( h u), 667 µm, 900 µm, 1600 µm, 2685 µm, 3700 µm, 7115 µm and 10000 µm as well as an open and
sho wi h 800 µm access line leng h a a o al leng h o 3700 µm build upon GaAs ( = 12.9) wi h open
bounda y a he bo om side. Wi h his assump ion, a semi-in ini e hal space is emula ed.
20
a)
b)
c)
d)
e)
)
Fig. 21: The simpli ied p obe sys ems wi h p obes 1, 2 and 3 wi h he wid h and coaxial opening scaled om a o .
Uppe ow: Top iew. Lowe ow: Bo om side wi hou subs a e.
a)
b)
010 20 30 40 50 60 70
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
S21 (dB)
F equency (GHz)
b idge
CPW-model
c)
10 20 30 40 50 60 70
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
 (S21(b idge) - S21(CPW model)) * 1000
F equency (GHz)
b idge
Fig. 22: A se o calib a ion lines exci ed wi h lumped po s placed in a i icial b idges a he posi ion o he p obes.
a): Line 3. b): T ansmission S21. c): Di e ence o S21 compa ed o he CPW model o [8].
In o de o allow a basic simula ion o he “ ue” pe o mance o he CPW p ope ies, a lumped b idge
exci a ion is used. All he calib a ion s uc u es we e simula ed and his da a was hen p ocessed wi h he
mTRL calib a ion algo i hm. The ex ac ed esul s a e compa ed o he analy ical CPW model [8]. Fig. 22
shows ha he calib a ed da a o he b idge model coincide wi h he esul s o he analy ical CPW model
e y well. The e o e, one can s a e ha he b idge model can se e as a eliable e e ence o
compa ison.
21
a)
010 20 30 40 50 60 70
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
S21 (dB)
F equency (GHz)
p obe 1
p obe 2
p obe 3
CPW-model
b)
010 20 30 40 50 60 70
-4
-2
0
2
4
 (S21(p obe i) - S21(CPW model)) * 1000
F equency (GHz)
p obe 1
p obe 2
p obe 3
Fig. 23: The calib a ed esul s o line 3 (1600µm) o all 3 p obes. a): T ansmission S21. b): Di e ence o S21 o he
CPW model o [8].
Fig. 23 indica es ha hough he calib a ion p ocess should deduc all e ec s o he p obe and i s
en i onmen lea ing only he S-pa ame e o he measu ed DUT he cu es show ha he p obe size
(and shape) s ill gene a es de ia ions. The di e ence o he linea S21 o he CPW model o [8] is shown
in Fig. 23 b. The maximum de ia ion is app oxima ely 0.4% and occu s wi h p obe 3.
a)
b)
c)
010 20 30 40 50 60 70
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
S21 (dB)
F equency (GHz)
p obe 1
p obe 2
p obe 3
CPW model
d)
010 20 30 40 50 60 70
-20
-15
-10
-5
0
5
10
15
20
 (S21(p obe i) - S21(CPW model)) * 1000
F equency (GHz)
p obe 1
p obe 2
p obe 3
Fig. 24: (a and b) P obe 2 wi h in-line neighbo (leng h = 1.66 mm, dis ance o DUT = 200 µm). The
calib a ed esul s o line 3 (1600µm) o all 3 p obes; (c) ansmission S21 and (d) di e ence o S21 o he
CPW model o [8].
In o de o include he p obe in e ac ion wi h i s icini y, a neighbo ing s uc u e is added in in-line
di ec ion o he DUT. Fig. 24 c and d show he compa ison o he calib a ed esul s o he con igu a ion
wi h a DUT and a neighbo ing s uc u e exci ed wi h he h ee di e en p obes. A he i s glance, one
can obse e a dip o esonance beha io in all he cu es due o he in e ac ion wi h he in-line
neighbo ing s uc u e. I is impo an o no e ha he s eng h o he dip beha io di e s depending on he
size o he p obes. P obe 3 wi h i s la ge dimensions shows he mos emphasized dip beha io which
yields a maximum de ia ion o 1.5 %. The e o e, he in luence o me al con igu a ions in he icini y o he
p obe needle egion can ha e s ong e ec on he DUT measu emen . One needs o sepa a e hese
in es iga ions in o wo cases, me al inside o ou side he p obe shadow, which is e e ed o as he a ea
below he p obes ha would be shadowed i he p obe would be illumina ed om he op. Fu he

22
in es iga ions conce ning he in luence o in-line and side-way neighbo ing s uc u es ha e been
pe o med.
Guideline #6
The esul s e eal ha one should keep sensi i e egions o he p obe shadow ee o s uc u es o
a oid p obe coupling o neighbo ing s uc u es, as shown in Fig. 25 (in es iga ions pe o med up
o 70 GHz).
Fig. 25: Design ecommenda ions ela ed o he p obe cons uc ion.
3.5 Thin- ilm Mic os ip lines (TFMSL)
While he si ua ion wi h coplana wa eguides has been discussed in he p e ious sec ions, he pu pose
o his sec ion is o add he espec i e esul s o hin- ilm mic os ip lines. In o de o pe o m he
in es iga ions, wa e s wi h di e en layou s we e designed and ab ica ed. Wha is s udied he e is he
impac o posi ioning o he calib a ion s uc u es and o he p obe on he calib a ed esul s o he di e en
DUTs. The e o e, wo se s o calib a ion lines on wo diced e icles a e used. Bo h se s we e ab ica ed
on he same wa e . The p ocess comp ises a hin- ilm mul ilaye s ack which is ealized on op o a
500 µm hick Bo o loa subs a e.
a)
b)
23
c)
d)
Fig. 26: a): Layou o e icle 1 (c) : pho og aph o a zoomed pa o e icle 1); b): Layou o e icle 2 d):
zoomed egion a ound he igh p obe ip.
The g ound laye is bu ied in he 18 µm hick Benzocyclobu ene (BCB) laye s ack, he signal s ip
me alliza ion le el (wi h a hickness = 5 µm) is loca ed on op o he BCB. An in e media e me al laye
le el can be added. This p ocess was de eloped a FhG IZM as mo he boa d o mic owa e modules bu
can be ea ed as an example o o he hin- ilm wa e opologies as well. The measu emen s o he
calib a ion s uc u es and DUTs a e pe o med on a chuck wi h a 1 cm hick ce amic suppo . Fig. 26 op
shows he layou s o he wo calib a ion se s. Each includes a comple e se o line elemen s, consis ing o
4 hin- ilm mic os ip (MS) lines (signal wid h w = 37 µm) o di e en leng hs (l = 900, 1800, 2700 and
5400 µm), as well as an open and a sho s uc u e as e lec s anda ds. Loads we e no implemen ed
because esis o s we e no a ailable on his speci ic wa e un (in gene al, he p ocess includes
esis o s). One h u line ailed ( e icle 2) so ha o all possible combina ions only one h u line could be
used. F om he s uc u es (see Fig. 26), we ob ained wo calib a ion se s: Cal 1, by using he se on
e icle 1, and Cal 2, by using he se on e icle 2 (he e wi h he h u line o e icle 1).
a)
b)
Fig. 27: E ec i e pe mi i i y (a) and a enua ion (b) o measu ed and simula ed calib a ion se s Cal 1 and Cal 2.
In e icle 1 (see Fig. 26 a), all he calib a ion s uc u es a e loca ed on he le -hand side o he e icle,
nex o he edges o he e icle whe eas in e icle 2 (Fig. 26 b) he calib a ion s uc u es a e placed on he
igh -hand side o he e icle. In he labeling o he p obe and he S-pa ame e se , he index 1 always
deno es he le and index 2 he igh -hand side. Also, all elemen s a e placed in a di e en en i onmen
ega ding he neighbo ing s uc u es. A e measu ing and simula ing all he calib a ion elemen s, he
TRL calib a ion p ocess was applied o he measu ed and simula ed da a o bo h calib a ion se s 1 and 2.
020 40 60 80 100 120
2.0
2.1
2.2
2.3
 e
F equency (GHz)
Measu emen
Cal 1
Cal 2
Simula ion
Cal 1
Cal 2
020 40 60 80 100 120
0.0
0.5
1.0
1.5
2.0
2.5
 (dB/cm)
F equency (GHz)
24
The i s esul s o he TRL calib a ion a e he line p ope ies which a e plo ed in Fig. 27. Compa ing he
e ec i e pe mi i i y and he a enua ion cons an one obse es al eady sligh di e ences be ween he
wo se s o bo h measu ed and simula ed esul s. The de ia ion o he measu ed da a a 40 GHz in  e is
abou 4% and in  abou 12%. This is pa ly clouded by he de ia ions be ween simula ion and
measu emen s, which a e p obably due o he ac ha he modeled p obe does no map he eal p obe
geome y comple ely and he limi ed knowledge o he ma e ial p ope ies o he wa e .
The ques ion is whe he and how hese admi edly ela i ely small de ia ions be ween he wo calib a ion
se s will in luence he calib a ion o a DUT. The e o e, he impac when changing be ween he calib a ion
se s (Cal 1 and Cal 2) on wo selec ed DUTs will be in es iga ed. These DUTs a e a daisy chain (DUT1)
and a coupled line s uc u e (DUT2).
a)
b)
c)
Fig. 28: DUT 1 is a daisy-chain con igu a ion using 50 ia ansi ions, a): geome y o simula ion, b): layou
con igu a ion, c): calib a ed esul s o measu emen s and simula ions using Cal 1 and 2.
020 40 60 80 100 120
-10
-8
-6
-4
-2
0Measu emen
Cal 1
Cal 2
Simula ion
Cal 1
Cal 2
S21 (dB)
F equency (GHz)
-20
-10
0
10
S11 (dB)
25
DUT 1 (Fig. 28) is a daisy-chain con igu a ion using 50 ia ansi ions placed a he lowe wa e side o
he e icle (Fig. 1 a). I is su ounded by neighbo ing s uc u es a he sides, such as meande line,
couple s, and ano he daisy chain.
DUT 1 is an example o a s uc u e wi h high ansmission and low e lec ion. I has no in-line neighbo s.
The esul s a e calib a ion a e p esen ed in Fig. 28. One inds good ag eemen be ween all 4 esul s
(measu ed and simula ed da a o Cal 1 and 2). The de ia ions be ween he wo calib a ion se s can be
neglec ed. The di e ences in e ec i e pe mi i i y become appa en only in he uppe equency ange
and de ia ions in a enua ion canno be seen a all, due o he ac ha all co ec ions by he calib a ion
p ocess ela e o e y sho s uc u es and hus he a enua ion has mino in luence. In gene al, we
obse e his low sensi i i y o calib a ion impai men s o all DUT ypes wi h low e lec i e p ope ies like
h u-lines, il e s and daisy chain s uc u es.
a)
b)
c)
d)
Fig. 29: DUT 2 is a coupled-line con igu a ion wi h 1800 µm coupling leng h (a): DUT2 placed on a wa e (b); (c) :
expec ed beha io o DUT2 and d): Measu ed beha io o DUT 2 compa ed o simula ion.
DUT 2 (Fig. 29 a) is a coupled-line con igu a ion wi h wo pa allel s ips in a dis ance o 100 µm and is
placed on he igh -hand side o e icle 1, wi h he igh e mina ion simila o he calib a ion elemen s on
e icle 2 (see Fig. 29 d). I is placed nex o a meande line and one o he calib a ion lines and has