a novel wrap-around metal contact optimized for radial p-n ... · p+nn+ s corporates ors (b155 f or...

1

Supporting Information

A novel wrap-around metal contact optimized for

radial p-n junction wire solar cells

Sun-Mi Shin, Jin-Young Jung, Kwang-Tae Park, Han-Don Um, Sang-Won Jee,

Yoon-Ho Nam and Jung-Ho Lee*

Department of Chemical Engineering, Hanyang University, 55 Hanyangdaehak-ro, Sangnok-

gu, Ansan, Kyeonggi-do 426-791, Republic of Korea

Email: [email protected]

Department of Chemical Engineering

Electronic Supplementary Material (ESI) for Energy & Environmental ScienceThis journal is © The Royal Society of Chemistry 2013

Fig. S1

wire co

By em

realizat

cell siz

contact

samples

because

affected

with fin

factor b

the sma

(sample

increasi

(finger/

IEEE P

note tha

irrespec

1 J-V curve

ntacts. Diff

mploying t

ion of robu

zes and/or p

with conve

s which we

e the fill f

d by top-co

nger/bus ba

behaviors du

all one was

e size: 11

ing the sam

/bus bar) ne

Photovoltaic

at the I-V cu

ctive of the

es for solar

ferent cell si

the wrap-ar

st, reliable

pattern desi

entional me

ere difficult

factors for

ontact struct

ar electrode

ue to differ

cut (for m

cm2) had

mple sizes.

eed to be o

c Specialists

urves for th

cell sizes (s

cells with

izes are com

round cont

collection o

ign of met

etal grids, w

to fabricate

these smal

ture. For ex

es (Fig. S1a

ent cell are

easurement

been measu

In case of

optimized a

s Conferenc

he wrap-aro

see Fig. S1b

2

(a) conven

mpared for e

tact design

of charge ca

al grids. To

we actually

e metal grid

ll cells can

xample, foll

a) demonstr

eas although

t with differ

ured. In ge

convention

according to

ce, Washing

ound contact

b).

ntional plan

each scheme

n, we have

arriers which

o specifical

y thought th

ds on their

nnot suffici

lowing I-V

rate the com

h they are id

rent grid de

eneral, the

nal top elec

o the cell s

ton, D.C., U

t cells are o

ar grids, an

e.

e intended

h is irrespec

lly compare

hat too sma

tops were r

iently refle

curves of p

mpletely dif

dentically f

esign) right

fill factors

ctrodes, the

sizes (Ref.

USA, 1978)

observed to

nd (b) wrap

to emphas

ctive of var

re our wrap

all area (

Fig. S2

embedd

contacts

S2(a)] d

removin

was neg

was obs

Fig. S3

wavelen

wavelen

2 (a) A dig

ded in PDM

s (right), an

Our light

detached fro

ng from the

gligible [Fi

served from

Absorption

ngths from

ngths were

gital photog

MS, Si micro

nd (b) transm

absorptance

om the sub

e substrate b

g. S2(b)]. A

m the wire ar

n spectra of

400 nm and

notable to r

graph comp

owires only

mittance spe

e (Fig. 1c)

strate. How

because we

After remov

rrays contai

f PDMS. Lo

d 1100 nm,

reveal high

3

paring tran

(left), Si m

ectra of the

data were

wever, the el

have confir

ving from t

ining Ag ele

ow average

, but two pe

absorption

nsparency o

microwires s

sample (rig

measured b

lectrical pro

rmed that li

the substrat

ectrodes.

absorption

eaks at shor

(>10%) of P

of two Si m

urrounded b

ght) in (a).

by using th

operties wer

ight absorpt

e, low tran

of ~0.8% w

rt (1100)


Fig. S4

microw

~80% w

our wor

of a mic

wire sur

which t

4 External

wire sample

The reason

was originat

rk, despite

crowire-arra

The wire a

rfaces in wh

then resulted

Quantum

which is me

ning why th

ted mostly

the remarka

ayed-absorb

arrays were

hich the RIE

d in degrada

Efficiency

easured in F

he short circ

from the fa

ably increa

ber removed

fabricated b

E proceeded

ation (~0.7

4

(EQE) re

Fig. 2b.

cuit current

act that there

sed surface

d from the s

by RIE proc

d for long t

μs) in mino

esult of the

t was poore

e was no su

e-to-volume

substrate.

cess causing

ime in orde

ority carrier

e wrap-arou

r than the li

urface passi

ratio becau

g severe pla

er to pattern

life time.

ound Ag co

light absorp

ivation proc

use of the p

asma damag

n 60-μm-lon

ontacted

ptance of

cesses in

presence

ge to the

ng wires,


Fig. S5

solar ce

Step A:

Step B:

5 Schematic

ell with wrap

: A periodic

An 1.5-μm

wafer usin

The substr

mixture of

Periodic S

etching pr

adopted to

The wire p

determined

Park et al.

: Formation

A cost-effi

bottom sid

This appr

direction o

a front side

First, boro

cs showing

p around Ag

c array of sil

m-thick SiO

ng high-dens

rate was pat

f CF4 and C

SiO2 dot arr

rocess, inc

o define the

pitch (a cent

d by the pa10 described

ns of a radia

ficient, conf

des of a waf

roach produ

of a wafer, w

e and a BSF

on and phos

the overal

g contacts (

licon micro

O2 layer wa

sity plasma

tterned usin

HF3 plasma

rays (2 μm

luding etch

high aspect

ter-to-cente

atterned siz

d more detai

al p–n juncti

formal dopi

fer using dif

uces a high

which conco

F on backsid

phorus silic

5

ll process s

(without det

owires

as deposited

chemical v

ng low-level

a.

in diamete

hing (SF6)

t ratio Si mi

er distance in

ze of a phot

ils about thi

ion and bac

ing techniqu

fferent dopa

h-efficiency

omitantly in

de.

cate precurs

sequences f

taching wire

d on a 8-in

vapor depos

l lithograph

er) were use

and polym

icrowire arr

n between m

to mask. Pr

is procedure

k surface fi

ue was emp

ants in one t

y p+nn+ s

ncorporates

sors (B155 f

for fabricati

es from the

ch n-doped

ition

hy and react

ed as etch m

merization

rays.

microwires)

revious exp

e.

eld (BSF)

ployed to co

thermal cyc

tructure ac

a radial p–n

for B and P

ing a Si m

substrate).

d (5 Ωcm),

tive ion etch

masks. The

(C4F8) ste

) and diame

perimental w

o-dope the

cle.

cross the th

–n junction M

P509 for P, s

icrowire

Si(100)

hing in a

e plasma

ps, was

eter were

work by

top and

hickness

MWs on

supplied


6

by Filmtronics) were spin-coated on each dummy wafer.

Then, the SOD dummy wafers were baked at 150 °C for 20 min. To form the BSF, a

phosphorus-coated dummy wafer was directly contacted to the bottom side of a

target wafer so as to degenerately dope boron into the backside.

The boron-coated dummy wafer, however, was located in proximity mode on the

front side of the target wafer to tailor the doping profile of boron diffusion.

Simultaneous diffusion doping was carried out in a tube furnace under a mixed

ambient of N2 and O2 at 900°C for 5 min. Previous work by Jung et al.14 reported

more details about this procedure.

Step C: Spin-coating of PDMS.

A mixed solution (10:1 wt %) of PDMS base (sylgard 184) and curing agent was

spin-coated on the wire arrays at 4000 rpm for 480 sec.

To maintain the height of PDMS uniform, a solution should be stirred enough to

cover whole sample surface prior to dropping for spin-coating.

Then, the coated sample was cured at 80°C for 8 hrs. The vacuum oven was

necessary to remove air bubbles originated from the mixing process of PDMS base

with curing agent.

A drastic change in temperature should be avoided because the PDMS film can be

cracked due to the difference in thermal expansion coefficients. Previous work

performed by Putnam et al.4 reported more details about this procedure.

Step D: PDMS etching

Spin-coated PDMS was etched using a solution of tetrabutylammonium fluoride

(TBAF) mixed with dimethyl fluoride (DMF) in a 1:1 volume ratio. The etch rate

was 10 μm/min, but was normally sensitive to temperature.

A height of PDMS was variable by adjusting dipping time. After dipping in the

solution, the sample was rinsed in DMF solvent. This process was to remove the

PDMS residues on wire arrays.

Then, cleaning using deionized (DI) water was conducted. This cleaning process was

important because the PDMS residue was likely remained in between wire arrays.


Step E:

Step F:

Step G

: Ag deposi

Ag electro

wires were

metal thick

The reason

physically

between th

PDMS rem

PDMS coa

: Selective A

The Ag co

solution (m

etching so

etching tim

infiltrated

ition

odes were th

e embedded

knesses wer

n why we

y separate

he substrate

mained) was

ating and etc

Ag removal

oated on pro

methanol :

olution likel

me was requ

into the inte

hermally ev

d in PDMS r

re varied fro

coated Ag

Ag electro

e and Ag. In

s fixed at 10

chback

and D abo

protect a p

protruded

which the

by wet etc

l

otruded part

ammonia :

ly penetrate

uired to prec

erfacial reg

7

vaporated on

remained. T

om 100 to 1

on top of P

ode from th

n our work

0 m.

PDMS c

ove, were re

planar Ag la

Then, the

out of PDM

Ag in prot

ching.

ts of wires

hydrogen

ed along th

cisely limit

ion of PDM

nto Si micro

The depositi

1300 nm by

PDMS rem

he substrat

k, the gap (c

coating and

epeated onc

ayer, not sid

e Ag-coated

MS [see left

truded parts

was remov

peroxide=4

he interface

in order to

MS/wire (see

owires in wh

ion rate was

adjusting d

mained at wi

te while co

correspondi

etchback s

ce again bec

dewall-coate

d wire top a

t figures, Fi

s could be s

ed for 1 mi

4:1:1, volum

between A

prevent the

e figures be

hich the bo

s ~10Å/sec,

deposition ti

wire bottoms

ontrolling

ing to the h

steps, i.e.,

cause we ne

ed Ag.

and sidewa

igs. S5(a) &

selectively r

in using Ag

me ratio). S

Ag and PDM

e overetchin

elow).

ttoms of

, and the

ime.

s was to

the gap

height of

steps C

eeded to

alls were

& (d)], in

removed

g etching

ince the

MS, the

ng of Ag


Step H

Fig. S6

coating

of PDM

formed

: Finally, th

6 SEM imag

and etchba

MS after etc

after remov

he bottom co

ges (a-c) de

ack processe

chback proc

val of PDM

ontact was t

escribing ho

es. (d) Ag-c

cess; (e) se

MS. (d) and (

8

thermally de

ow the cont

coated-top r

elective rem

(f) correspo

eposited on

tact height

regions of a

moval of Ag

nd to (a) an

to a back si

was adjuste

microwire

g; (f) wrap-

nd (c), respe

ide of the so

ed by using

were protru

-around Ag

ectively.

olar cell.

g PDMS

uded out

g contact


Contac

area com

where ρ

A circu

Ag/Si c

where r

Ag/Si w

For this

by ITO/

surroun

ct resistanc

We extrac

mpared to A

ρc is contact

ular area of

contact wher

The wire t

r is a wire ra

when the L i

s reason, a b

/Si, which c

We can al

nd top-regio

ITA

e of ITO/Si

cted the con

Ag/Si contac

t resistivity,

ITO/Si con

re L denote

tops are calc

adius of 1 μ

is 500 nm.

black dot co

corresponds

lso calculat

ons of micro

TO/Si A

i [Fig. 3(c)]

ntact resista

cts. Contact

, and A is c

ntact is dire

s a height o

culated by

topA

μm. Note th

500L nA

onnected w

s to the sam

e the Rc of

owires. The

ITO/SiA

500L nm

9

]

ance of ITO

t resistance

ccR A

ontact area

ectly compa

of a cylinder

p region

hat the area

2nm rL

with red dott

me area form

f ITO/Si wh

contact are

2A r

top-regA

2A

O/Si while

(Rc) is dete

. The ρc of

ared with a

r.

2r

of wire-top

500L nmL

ted lines [Fi

med by Ag/S

hen ITO at

a can be de

L

gion A

500LrL

maintaining

ermined by:

ITO/Si is 7

sidewall ar

p is equal to

ig. 3(c)] sta

i contact at

ttempts to t

scribed usin

0 2nm r

g the same

1.43×10-3 m

rea of a cy

o the contact

ands for a p

L of 500 nm

three-dimen

ng L as follo

r L

contact

(eq. 1)

mΩ cm2.

lindrical

(eq. 2)

t area of

(eq. 3)

planar Rc

m.

nsionally

ows:

(eq. 4)

(eq. 5)

(eq. 6)


where Δ

using th

To extr

necessa

ΔA and ΔL

he equations

ract the co

ary; then, th

L stand for

s above, we

R

ontact resist

e unit was c

_ ITO/ScR

the extende

e could sum

_ ITcR

_ ITO/SicR

tance per u

converted fr

i52 Lr

10

ed contact a

m up the cont

_TO/Si

I

c

A

_

5002c

Lr

unit cell ar

rom [Ω] into

_ ITO/Si

500 2c

nm r

area and co

tact resistan

_ ITO/Si

ITO/Si

_ ITO/Si

0 2nm r

rea, the inf

o [Ω cm2].

(unit r L

ontact heigh

nce of ITO/S

L

formation o

cell area)

ht, respectiv

Si.

of a cell a

vely. By

(eq. 7)

(eq. 8)

area was

(eq. 9)


Fig. S7

adoptin

emitter

shallow

depth sh

provide

emitter

area, w

radial ju

top side

of wire

contacts

wire cel

7 Schematic

ng wrap-arou

Radial p-n

area, yieldi

wer the junc

hallower, p

e a conform

layer, with

we are curre

unction wir

e of wires, i

e arrays. Th

s will furth

lls.

cs of (a) con

und contact

n junction w

ing high Au

ction depth

lasma dopin

mally doped

a very abru

ntly develo

re solar cell

it is possibl

his selectiv

her improve

nventional r

ts

wire solar

uger recomb

while redu

ng techniqu

d, ultrashall

upt junction

oping the ad

ls. Since ou

le to remov

ve emitter c

cell conve

11

radial p-n ju

cells conve

bination; thu

ucing the e

ues23 would

low junctio

n doping pro

dvanced sel

ur contact is

ve the unnec

concept ad

ersion effici

unction, and

entionally r

us, we need

emitter surf

be highly r

on undernea

ofile. To fur

lective emit

s located cl

cessary emi

dopting plas

iencies repo

d (b) selecti

require a la

d to find out

face area. T

recommend

ath a degen

rther reduce

tter structur

ose to the w

tters positio

sma doping

orted to date

tive emitter

arge amoun

t how to eff

To make a j

dable becaus

nerately dop

e the emitter

re (see Fig.

wire bottom

oned at the

g and wrap

e in radial j

concept

nt of the

fectively

junction

se it can

ped thin

r surface

S7) for

ms, not a

top side

p-around

junction


Fig. S8

Experim

thickne

*At eac

8 Plots of

mentally ob

ss with thei

Met

thickn

(nm

100

200

300

400

700

100

130

ch condition

f FF value

btained, cha

ir standard d

tal

ess*

m)

FF

0

0

0

0

0

00

00

n, five samp

es and a

ampion and

deviations.

F@champio

29.69

29.56

37.85

51.86

61

65.4

64.42

ples were ide

12

table show

averaged F

on, %FF

entically fab

wing the d

FF data are

F@ average

%

27.8

28.4

36.7

37.3

57.9

62.5

63.3

bricated to

detailed sa

shown as a

e Sta

dev

1

1

1

3

2

1

1

evaluate the

ample infor

a function o

andard

viation

1.65

1.65

1.28

3.81

2.85

1.89

1.58

e fill factors

rmation.

of metal

s.


Fig. S9

1.

2.

A detailed

At Line 57

s oR =R

The influe

unchanged

determine

to only kn

The Rc val

(1) used fo

3. We ca

procedure d

7, Page 2, th

c m+R +R =

ence of Ro

d while var

ed by wafer

now the Rc a

lues can be

or plotting f

Rc

an also know

describing h

he following

o e=R +R w(base and

rying the m

r resistivity

and Rm at th

obtained as

figure 3c.

cothRWLT

□

w the Re val

13

how to extra

g equation i

where Re=(Rc

emitter res

metal thickn

and emitter

he same tim

s a function

)h(L/LT

Figure 3c

lues from a

act Rs from

is denoted,

+Rm), Ro=co

sistances) c

ness adopte

r sheet resis

me.

of contact h

cotWρR c □

right-hande

the variable

onstant

an be precl

ed for a fro

stance. To e

height (L) f

)ρRh(L

c

□

ed, y-axis of

es of Rc and

luded becau

ont electrod

extract Rs, w

from the equ

(1)

f Fig. 2a be

d Rm.

use it is

de. Ro is

we need

uation

low.


4.

Since the

below). A

of Rc and R

Re=Rc+Rm,

s a result, w

Rm.

Rm can be

we can final

14

Figure 2a

e obtained f

lly extract a

Figure 3d

from the val

a total serie

lues of Re a

es resistance

and Rc (see

e (Rs) as a

e Fig. 3d

function


a novel wrap-around metal contact optimized for radial p-n ... · p+nn+ s corporates ors (b155 f or...

Documents