You are on page 1of 6

Comment

www.advenergymat.de

Comment on “Introduction of a Novel Figure of Merit for


the Assessment of Transparent Conductive Electrodes in
Photovoltaics: Exact and Approximate Form”
Christopher P. Muzzillo

λgap
Anand et al. introduce an implicit equation for transparent conductive JSC,TCE = q ⋅ ∫ AM 1.5G ( λ ) ⋅ TTCE ( λ ) ⋅ d λ (4)
0

electrode (TCE) merit in photovoltaics (PV) by lumping series resistance in


λ gap
with the diode. However, an explicit equation has been previously derived by J 0,TCE = q ⋅ ∫ BB300K ( λ ) ⋅ TTCE ( λ ) ⋅ d λ (5)
0
lumping series resistance external to the diode. For TCEs of practical interest
for PV, the choice of where to lump series resistance does not affect merit, so   q ⋅ (V + J ⋅ R TCE ⋅ A )  
external lumping is preferred. On the other hand, monolith deadspace, metal J = JSC,TCE − J 0,TCE ⋅  exp   − 1 (6)
  kB ⋅ (300 K )  
grid shadowing, and grid resistance shift TCE merit away from the figure of
merit behavior published by Anand et al. Here, q is the elementary charge, λgap
is the band gap wavelength, AM1.5G(λ)
is the wavelength-dependent AM1.5G
spectral irradiance, BB300K(λ) is the
­
1. Where to Lump Series Resistance blackbody radiation photon flux density at 300 K, and kB is the
Boltzmann constant. Anand et al. compute their exact FOM
Anand et al. recently published a figure of merit (FOM) for (ϕAnand) by finding the maximum power (MP) density from
transparent conductive electrode (TCE) performance in solar Equation (6) (PMP = JMP · VMP) and dividing it by the same
cells.[1] Their FOM uses a detailed balance calculation for max- power density, except calculated at a wavelength-independent
imum power conversion efficiency of a solar cell with a given TTCE of 1 and Rsh,TCE of 10−4 Ω sq−1
band gap (Egap) absorbing the air mass 1.5 global (AM1.5G)
spectrum. They assume that series resistance in the TCE PMP ( λgap ; TTCE ( λ ) ; Rsh,TCE ; L )
φAnand = (7)
(RTCE) is[2,3] PMP ( λgap ; TTCE = 1; Rsh,TCE = 10 −4 Ω sq −1 ; L )

Rsh,TCE ⋅ L2 In other words, the approach of Anand et al. is to add the


R TCE = (1)
3⋅ A TCE series resistance into the one-diode model. The depend-
ence of their exact, implicit FOM (ϕAnand) on Rsh,TCE is repro-
Here, Rsh,TCE is the sheet resistance of the TCE, A is the cell duced in Figure 2 for a band gap of 1100 nm and a wavelength-
area, and L is the cell length (see schematic in Figure 1a). They independent TTCE,avg of 0.9, which is defined as[4]
add RTCE to the ideal diode equation, which is implicit and
λgap
solved iteratively. Finally, they scale the detailed balance limit AM 1.5G ( λ ) ⋅ TTCE ( λ ) ⋅ dλ
short-circuit current density (JSC,DB) and dark saturation current TTCE,avg =
∫ 0
(8)
λgap
density (J0,DB) by the wavelength-dependent transmittance spec- ∫ AM 1.5G ( λ ) ⋅ dλ
0
trum of the TCE (TTCE(λ)) to arrive at current densities for the
diode equation (JSC,TCE and J0,TCE) Anand et al. also define an approximate FOM (ϕAnand,approx)
that fits the exact form (ϕAnand) well at high and low Rsh,TCE
λgap
JSC,DB = q ⋅ ∫ AM 1.5G ( λ ) ⋅ d λ (2)
0
 3 ⋅VOC,DB
 TTCE,avg , Rsh,TCE <
λgap  4 ⋅ L2 ⋅ TTCE,avg ⋅ JSC,DB ⋅ FFDB
J 0,DB = q ⋅ ∫ BB300K ( λ ) ⋅ d λ (3) φAnand,approx = (9)
3 ⋅VOC,DB 3 ⋅VOC,DB
0  , Rsh,TCE >
 4·Rsh,TCE ⋅ L ⋅ JSC,DB ⋅ FFDB 4 ⋅ L ⋅ TTCE,avg ⋅ JSC,DB ⋅ FFDB
2 2

C. P. Muzzillo
National Renewable Energy Laboratory Here, VOC,DB and FFDB are the detailed balance open-cir-
15013 Denver W Pkwy cuit voltage and fill factor, respectively. While Equation (9) is
Golden, CO 80401, USA explicit, it does not match ϕAnand near the piecewise transition
E-mail: christopher.muzzillo@nrel.gov (Figure 2).
The ORCID identification number(s) for the author(s) of this article On the other hand, previous work has shown that lumping
can be found under https://doi.org/10.1002/aenm.202103119.
series resistance externally yields an exact, explicit equation for
DOI: 10.1002/aenm.202103119 TCE merit.[5–7] For example, a uniform TCE applied to a solar

Adv. Energy Mater. 2022, 12, 2103119 2103119 (1 of 6) © Published 2022. This article is a U.S. Government work and is in the public domain in the USA
16146840, 2022, 23, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202103119 by Cochrane Portugal, Wiley Online Library on [07/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advenergymat.de

Figure 1. Schematic showing the architecture and dimensions of a) the uncontacted TCE of ϕAnand and FOMTCE, b) the TCE with monolith deadspace
of FOMTCE,mono, and c) the TCE with monoliths and a grid of FOMTCE,mono,grid. Cell length (L), deadspace (d), grid metal height (hmetal), grid wire width
(w), and grid wire spacing (s) are defined.

cell operating at its MP point with current density (JMP) and J MP,TCE ⋅VMP,TCE  Rsh,TCE ⋅ L2 ⋅ J MP 
voltage (VMP) would reduce its current and voltage to JMP,TCE FOM TCE = = TTCE,avg ⋅  1 −  (12)
J MP ⋅VMP  3 ⋅VMP
and VMP,TCE
Equation (12) itself has not been previously published, but its
J MP,TCE = J MP ⋅ TTCE,avg (10) series resistance dependence was first evaluated at JSC/VOC in
1981,[9] and since then similar forms have been used to assess
Rsh,TCE ⋅ L2 ⋅ J MP transparent conductor merit: Jacobs et al.’s “efficiency factor”
VMP,TCE = VMP − R TCE ⋅ A ⋅ J MP = VMP − (11) (called FOMTCE in this Comment) simplifies to Equation (12)
3
for an interconnection width of 0.[5] Rowell and McGehee addi-
Equations (10) and (11) assume that JMP,TCE is only affected tionally used the empirical 0.38JSC/VOC and considered voltage
by TTCE,avg and VMP,TCE is only affected by RTCE.[8] The solution effects on JMP.[7] Equation (12) is shown as a function of Rsh,TCE
for an FOM is then in Figure 2. FOMTCE and ϕAnand are very similar for high perfor-
mance TCEs, but they diverge for TCEs with high Rsh,TCE.
Many authors have used analytical and numerical methods
to analyze how distributed series resistance affects PV perfor-
mance.[3,10] However, these models require time, expertise, and
specific knowledge of the materials and architecture, making
series resistance lumping an attractive general approach
for screening TCE materials. Koishiyev and Sites modeled the
effect of RTCE on FF for Cu(In,Ga)Se2 (CIGS) and amorphous
silicon (a-Si) by simulating distributed series resistance.[3] To
test the accuracy of the lumped series resistance approaches,
Figure 3 compares the normalized FF (FF/FF0) literature data to
ϕAnand and FOMTCE. For ϕAnand, the band gap energy (Egap) was
chosen so that JSC,DB/VOC,DB was equal to the literature micro-
cell JSC/VOC (0.05406 Ω−1 cm−2 and 0.003720 Ω−1 cm−2 for CIGS
and a-Si, respectively), and the corresponding JMP,DB/VMP,DB
were used for the FOMTCE calculations. Figure 3 shows that
ϕAnand and FOMTCE are almost indistinguishable, and both
show excellent fit to the data. In conclusion, for reasonable
FOM values (>0.6), both approaches to lumping series resist-
Figure 2. Figures of merit versus TCE sheet resistance (Rsh,TCE): ance yield equivalent results, although external lumping has
Anand et al. exact FOM (ϕAnand; black), Anand et al. approximate FOM
the advantage of an explicit equation.
(ϕAnand,approx; gray dashed), and FOMTCE (teal). Calculations use L = 0.5 cm,
TTCE,avg = 0.9, a band gap of 1100 nm for ϕAnand, JSC,DB = 43.52 mA cm−2, The approach used by Anand et al. and the one shown here
VOC,DB = 883.2 mV, FFDB = 0.8705 for ϕAnand,approx, and JMP,DB = 42.15 mA cm−2 yield similar results because they both approximate the effect of
and VMP,DB = 793.9 mV for FOMTCE. adding a TCE with Rsh,TCE and TTCE(λ) to an arbitrary solar cell

Adv. Energy Mater. 2022, 12, 2103119 2103119 (2 of 6) © Published 2022. This article is a U.S. Government work and is in the public domain in the USA
16146840, 2022, 23, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202103119 by Cochrane Portugal, Wiley Online Library on [07/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advenergymat.de

Figure 3. Simulated results from the literature for distributed series resistance’s effect on normalized FF (FF/FF0) for a) CIGS (JSC/VOC = 0.05406 Ω−1 cm−2)
and b) a-Si (JSC/VOC = 0.003720 Ω−1 cm−2), compared with ϕAnand and FOMTCE. For CIGS, ϕAnand uses Egap = 1.070 eV (JSC,DB/VOC,DB = 0.05406 Ω−1 cm−2)
and FOMTCE uses the corresponding MP values for Egap = 1.070 eV (JMP,DB/VMP,DB = 0.05842 Ω−1 cm−2). For a-Si, ϕAnand uses Egap = 2.387 eV (JSC,DB/VOC,DB =
0.003720 Ω−1 cm−2) and FOMTCE uses the corresponding MP values for Eg = 2.387 eV (JMP,DB/VMP,DB = 0.003882 Ω−1 cm−2).

harvesting AM1.5G sunlight. Although it is named “exact” and more of an effect than where series resistance is lumped or
uses a detailed balance, ϕAnand is still an approximation because which J/V ratio is used.
the exact distributed series resistance must be simulated.[3,10] ϕAnand uses a single nominal detailed balance band gap, whereas
Whereas the derivation of ϕAnand starts from inside the solar FOMTCE can readily use experimental JMP and VMP values. For
cell and FOMTCE starts from outside, they end with similar instance, a reference solar cell with maximum power current den-
functions: the reciprocal of the limiting value of ϕAnand,approx sity JMP,ref and a TCE with transmittance TTCE,ref(λ) can be used to
(4Rsh,TCE ⋅ L2 ⋅ JSC,DB ⋅ FFDB ⋅ 3−1 ⋅VOC,DB
−1
) closely resembles the predict the current density that same reference cell would have if it
dimensionless series resistance[9] that appears in FOMTCE were fabricated with a new TCE of transmittance TTCE(λ)
(Rsh,TCE ⋅ L2 ⋅ J MP ⋅ 3−1 ⋅VMP
−1
).
λgap
The sensitivity of ϕAnand and FOMTCE to changes in TTCE,avg
and Rsh,TCE is shown in Figure 4. A 1% change to TTCE,avg and a J MP,TCE = J MP,ref + q ⋅ ∫ AM1.5G ( λ ) ⋅ (T ( λ ) − T
0
TCE TCE,ref ( λ )) ⋅ dλ (13a)
25% change to Rsh,TCE both shift merit by ≈1%, while switching
where series resistance is lumped (from ϕAnand to FOMTCE) If the internal quantum efficiency of the reference cell is
only shifts merit by 0.4%. Figure 4 also shows that evaluating known (QEint,ref ), then a more accurate form of Equation (13a) is
FOMTCE at JSC/VOC instead of the less commonly available
JMP/VMP only shifts merit by 0.3%. Therefore, when analyzing ∞

J MP,TCE = J MP,ref + q ⋅ ∫QEint,ref ⋅ AM 1.5G ( λ ) ⋅ (TTCE ( λ ) − TTCE,ref ( λ )) ⋅ dλ (13b)


TCE merit, the experimental error in TTCE,avg and Rsh,TCE has 0

Figure 4. Sensitivity analysis considering how TTCE,avg ± 1% and Rsh,TCE ± 25% change ϕAnand, FOMTCE evaluated at JMP/VMP, and FOMTCE evaluated at
JSC/VOC. Calculations use L = 0.5 cm, TTCE,avg = 0.9, Rsh,TCE = 10 Ω sq−1, a band gap of 1100 nm for ϕAnand, JSC,DB = 43.52 mA cm−2, VOC,DB = 883.2 mV,
JMP,DB = 42.15 mA cm−2, and VMP,DB = 793.9 mV for FOMTCE.

Adv. Energy Mater. 2022, 12, 2103119 2103119 (3 of 6) © Published 2022. This article is a U.S. Government work and is in the public domain in the USA
16146840, 2022, 23, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202103119 by Cochrane Portugal, Wiley Online Library on [07/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advenergymat.de

monoliths is then Tmono = L/(L + d) ), which reduces current


density to[5,7]

J MP ⋅ TTCE,avg ⋅ L
J MP,TCE,mono = J MP ⋅ TTCE,avg ⋅ Tmono = (14)
L+d

TTCE,avg ⋅ L  Rsh,TCE ⋅ L2 ⋅ J MP 
FOM TCE,mono =  1−  (15)
L+d  3 ⋅VMP

Equation (15) assumes that monoliths reduce current den-


sity but not voltage. The impacts of monolith contact resistance
and shunt conductance can additionally be computed[11] using
the previously reported functions for FF.[13] Figure 5 shows
that for T TCE,avg = 0.9 and Rsh,TCE = 10 Ω sq−1, FOMTCE,mono has
an optimal L of 0.4 cm, which is close to the L used in real
thin-film modules.[12] On the other hand, FOMTCE and ϕAnand
give the misleading impression that smaller L are always
better.
Figure 5. Figures of merit versus cell length (L): Anand et al. exact FOM
In practice, modules are fabricated with the optimal L. At
(ϕAnand; black), FOMTCE (teal), FOMTCE,mono (blue dashed), FOMTCE,grid
(orange), and FOMTCE,mono,grid (pink dashed). A band gap of 1100 nm was a given TTCE,avg, d, Rsh,TCE, JMP and VMP, FOMTCE,mono can be
used for ϕAnand, corresponding to JMP,DB = 42.15 mA cm−2 and VMP,DB = numerically maximized with respect to L to yield another figure
793.9 mV for the other models. Calculations use L = 0.5 cm, TTCE,avg = 0.9, of merit with more practical relevance, FOMTCE,mono,opt. The
d = 250 µm, ρmetal = 10−5 Ω cm, w = 56 µm, and hmetal = 13 µm. merit of TCEs with transmittance spectra (TTCE(350–800 nm))
and Rsh,TCE from the literature is computed in Table 1. As
Equation (13) allows current density losses to be referred expected, there is little difference between ϕAnand and FOMTCE
to experimental results for TCEs of arbitrary thickness using in Table 1. However, accounting for realistic monolith dead-
Fresnel calculations.[11] Moreover, experimental T TCE(λ) and space in FOMTCE,mono reveals that ϕAnand overestimates merit
T TCE,ref(λ) data from glass/TCE structures can be used, pro- for all TCEs. On the other hand, for graphene/poly(3,4-ethyl-
vided the corresponding experimental JMP,ref is known. Sim- enedioxy-thiophene) poly(styrene sulfonate) (PEDOT:PSS)/
ilar to FOMTCE, ϕAnand can also be adapted to experimental ZnO with Rsh,TCE = 230 Ω sq−1, ϕAnand underestimates merit
cells using two band gaps: A spectral band gap for the TTCE,avg relative to FOMTCE,mono,opt and FOMTCE,mono,grid,opt. Evaluating at
integral and a JSC/VOC-matching band gap for the JSC,DB and the optimal cell length (FOMTCE,mono,opt) reveals that TCEs with
J0,DB integrals. In summary, FOMTCE and ϕAnand yield practi- high Rsh,TCE (PEDOT:PSS + H2SO4 and graphene/PEDOT:PSS/
cally identical TCE merit for detailed balance solar cells, but ZnO in Table 1) warrant more merit than expected from
FOMTCE is explicit. FOMTCE,mono at L = 0.5 cm.
In addition to uniform TCEs, Anand et al. consider the trans-
mittance-sheet resistance tradeoffs of patterned metal grids
2. TCE Merit Is Not Universal because It Depends (called “AgNW2, CuNW2, Ag metal mesh, and nanotroughs”
in their work).[1,14] However, metal grids have nonuniformity
on Module Architecture
that induces resistance in the underlying semiconductor, so
Anand et al. consider monolithically integrated modules their merit deviates from that of uniform TCEs.[6] Therefore,
(Figure S1, Supporting Information[1]) with cell lengths of L = coupling a uniform TCE with a metal grid changes TCE merit.
0.5 cm. However, they do not consider the optical deadspace For example, the metal grating in Figure 1c has a transmittance
of real monoliths, so their cells cannot be stacked side by side of Tgrid = s/(s + w), and the sheet resistance of a grating with
to form a module and their results do not scale to arbitrary ρmetal resistivity and hmetal height is Rsh,grid = ρmetal · (s + w)/(w ·
areas. Real monoliths have optical deadspace, d (=250 µm in hmetal). Such a grid changes resistance in the TCE to RTCE,grid
typical modules;[12] see Figure 1b). The transmittance of the while adding resistance in the grid (Rgrid)[6]

Table 1. Comparison of TCE merit for TCEs with transmittance spectra and sheet resistance values from the literature.[1]

TCEa) Rsh,TCE [Ω sq−1] TTCE,avg ϕAnand FOMTCE FOMTCE,mono FOMTCE,mono,opt FOMTCE,mono,grid,opt


ITO 10 0.859 0.846 0.843 0.803 0.803 0.823
PEDOT:PSS + H2SO4 46.1 0.874 0.809 0.799 0.761 0.783 0.832
Graphene/PEDOT:PSS/ZnO 230 0.867 0.555 0.496 0.472 0.719 0.813
C60-surfactant/Ag 1.7 0.176 0.180 0.176 0.167 0.170 0.169

a)An ideal absorber harvesting 350–800 nm was used for ϕ −2


Anand, corresponding to JMP,DB = 26.32 mA cm and VMP,DB = 1178 mV for the other models. Calculations use L =
0.5 cm, d = 250 µm, ρmetal = 10−5 Ω cm, w = 56 µm, and hmetal = 13 µm.

Adv. Energy Mater. 2022, 12, 2103119 2103119 (4 of 6) © Published 2022. This article is a U.S. Government work and is in the public domain in the USA
16146840, 2022, 23, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202103119 by Cochrane Portugal, Wiley Online Library on [07/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advenergymat.de

Rsh,TCE ⋅ s 2 Rsh,TCE ⋅ Tgrid ⋅ w


2 2

R TCE,grid = = (16)
12 ⋅ A 12 ⋅ A ⋅ (1 − Tgrid )
2

Rsh,grid ⋅ L2 ρmetal ⋅ L2 ρ metal ⋅ L2 ⋅ ( s + w )


R grid = = = (17)
3⋅ A 3 ⋅ A ⋅ hmetal ⋅ (1 − Tgrid ) 3 ⋅ A ⋅ hmetal ⋅ w
Neglecting contact resistance and edge effects, MP current
density, MP voltage, and TCE merit are then[6]

JMP ⋅ TTCE,avg ⋅ L ⋅ s
J MP,TCE,mono,grid = J MP ⋅ TTCE,avg ⋅ Tmono ⋅ Tgrid = (18)
(L + d ) ⋅ ( s + w )

VMP,TCE,mono,grid = VMP − (R TCE,grid + Rgrid ) ⋅ A ⋅ J MP


 Rsh,TCE ⋅ Tgrid
2
⋅ w2 ρmetal ⋅ L2  (19)
= VMP −  2 +  ⋅ J MP
 12 ⋅ (1 − Tgrid ) 3 ⋅ hmetal ⋅ (1 − Tgrid ) 
Figure 6. Figures of merit versus TCE sheet resistance (Rsh,TCE): FOMTCE
(solid teal), FOMTCE with L = 0.7 cm (dotted teal), FOMTCE with a band
FOM TCE,mono,grid = TTCE,avg ⋅ Tmono ⋅ Tgrid ⋅ gap of 900 nm (JMP,DB = 32.91 mA cm−2 and VMP,DB = 1021 mV; dashed
 R  J  teal), FOMTCE,grid,opt (solid purple), FOMTCE,mono,grid,opt (solid pink), and
sh,TCE ⋅ Tgrid ⋅ w ρmetal ⋅ L2
2 2
1− 
MP
 FOMTCE,mono,opt (blue). Calculations use L = 0.5 cm, TTCE,avg = 0.9, a band
2 + ⋅
  12 ⋅ (1 − Tgrid ) 3 ⋅ hmetal ⋅ (1 − Tgrid )  VMP  (20) gap of 1100 nm (JMP,DB = 42.15 mA cm−2 and VMP,DB = 793.9 mV), d =
250 µm, ρmetal = 10−5 Ω cm, w = 56 µm, and hmetal = 13 µm unless noted
TTCE,avg ⋅ L ⋅ s   Rsh,TCE ⋅ s ρmetal ⋅ L ⋅ ( s + w )  J MP 
2 2
or optimized.
= ⋅1−  + ⋅
(L + d ) ⋅ ( s + w )   12 3 ⋅ hmetal ⋅ w  VMP 

Figure 3 shows FOMTCE,mono,grid as a function of L. At a given FOMTCE,grid is shown in Figure 5. In practice, Rsh,TCE can
JMP, VMP, d, ρmetal, w, and hmetal, FOMTCE,mono,grid can be numeri- affect band bending and lateral conduction within silicon het-
cally optimized with respect to L and Tgrid (or s) to arrive at erojunction solar cells’ p–n junctions,[16] leading to series resist-
FOMTCE,mono,grid,opt (Table 1). Adding a low temperature (220 °C) ance that is nonlinear in Rsh,TCE and requires simulation.[17]
screen-printed grating with ρmetal of 10−5 Ω cm, w of 56 µm, and There is no single ideal situation for which TCE merit
hmetal of 13 µm[15] shifts TCE merit to Rsh,TCE that are more than should be calculated: Anand et al. discussed absorber band
an order of magnitude higher (FOMTCE versus FOMTCE,mono,grid gap and cell length, both of which change the x-intercept of
in Figure 5). FOMTCE in Figure 6. Perhaps for brevity, Anand et al. did not
Figure 5 demonstrates that an indium tin oxide (ITO)-like TCE discuss module architecture, which changes the x-intercept by
(TTCE,avg = 0.9; Rsh,TCE = 10 Ω sq−1) applied to an 1100 nm band even more and also changes the y-intercept (FOMTCE,grid,opt,
gap absorber performs better the shorter the cell length for ϕAnand FOMTCE,mono,grid,opt, and FOMTCE,mono,opt in Figure 6).
and FOMTCE. When integrated with monoliths (d = 250 µm),
such a TCE has an optimal cell length of 0.4 cm and current den-
sity is reduced by a factor of Tmono = 0.941. When integrated with 3. Conclusions
monoliths and a screen-printed metal grid, such a TCE optimally
conducts over 0.15 cm (fingers spaced 0.3 cm apart), has a cell Anand et al. introduce an implicit equation for merit of trans-
length of 1.2 cm, and current density reduced by factors of Tmono = parent conductive electrodes (TCEs) in photovoltaics (PV) by
0.979 and Tgrid = 0.982 (Tmono⋅Tgrid = 0.961). These examples show lumping series resistance in with the diode. Previous work
how module architecture builds in optical losses and specifies the lumped series resistance external to the diode to derive an
optimal distance over which the TCE conducts current, both of explicit equation for merit. Since both approaches are good
which have pronounced effects on TCE merit. approximations for literature data, the explicit figure of merit
For applying TCEs to silicon heterojunctions, low-temper- is preferred. Although not discussed by Anand et al., module
ature screen-printed metal grids are used without monolith architecture deserves consideration because it changes the
deadspace, leading to a simpler version of Equation (20) FOM function. For example, monolith deadspace changes cell
length dependence, and metal grids impart more merit to TCEs
FOM TCE,grid = TTCE,avg ⋅ Tgrid ⋅ with high sheet resistance.
 R  J 
sh,TCE ⋅ Tgrid ⋅ w ρ metal ⋅ L2
2 2
1−  
MP
2 + ⋅
  12 ⋅ (1 − Tgrid ) 3 ⋅ hmetal ⋅ (1 − Tgrid )  VMP  (21) Acknowledgements
This work was authored by the National Renewable Energy Laboratory,
TTCE,avg ⋅ s   Rsh,TCE ⋅ s 2 ρ metal ⋅ L2 ⋅ ( s + w )  J MP 
= ⋅ 1− + ⋅ operated by Alliance for Sustainable Energy, LLC, for the U.S. Department
( s + w )   12 3 ⋅ hmetal ⋅ w  VMP  of Energy (DOE) under Contract No. DE-AC36-08GO308. Funding was

Adv. Energy Mater. 2022, 12, 2103119 2103119 (5 of 6) © Published 2022. This article is a U.S. Government work and is in the public domain in the USA
16146840, 2022, 23, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aenm.202103119 by Cochrane Portugal, Wiley Online Library on [07/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advenergymat.de

provided by U.S. Department of Energy Office of Energy Efficiency [5] D. A. Jacobs, K. R. Catchpole, F. J. Beck, T. P. White, J. Mater. Chem.
and Renewable Energy (EERE) Solar Energy Technologies Office A 2016, 4, 4490.
Award Numbers 33665 and 34352, as well as the Office of Technology [6] C. P. Muzzillo, Sol. Energy Mater. Sol. Cells 2017, 169, 68.
Transitions (OTT) Award Number 37248. The views expressed in [7] M. W. Rowell, M. D. McGehee, Energy Environ. Sci. 2011, 4, 131.
the article do not necessarily represent the views of the DOE or the [8] A. Mette, Dissertation, Verlag Dr. Hut 2007, 241.
U.S. Government. The U.S. Government retains and the publisher, [9] M. A. Green, Solid-State Electron. 1981, 24, 788.
by accepting the article for publication, acknowledges that the U.S. [10] a) G. L. Araujo, A. Cuevas, J. M. Ruiz, IEEE Trans. Electron Devices
Government retains a nonexclusive, paid-up, irrevocable, worldwide 1986, 33, 391; b) A. Cuevas, S. López-Romero, Sol. Cells 1984, 11, 163;
license to publish or reproduce the published form of this work, or allow
c) A. de Vos, Sol. Cells 1984, 12, 311; d) A. De Vos, P. De Visschere,
others to do so, for U.S. Government purposes.
Sol. Cells 1983, 10, 69; e) M. W. Denhoff, N. Drolet, Sol. Energy Mater.
Sol. Cells 2009, 93, 1499; f) I. L. Eisgruber, J. R. Sites, presented
at First World Conference on Photovoltaic Energy Conversion (Eds:
Conflict of Interest P. D. Moskowitz, H. Steinberger, W. Thumm) 5–9 December 1994;
g) A. W. Haas, J. R. Wilcox, J. L. Gray, R. J. Schwartz, presented
The author declares no conflict of interest. at 34th IEEE Photovoltaic Specialists Conference, 7–12 June 2009;
h) A. W. Haas, J. R. Wilcox, J. L. Gray, R. J. Schwartz, presented at 35th
IEEE Photovoltaic Specialists Conference, 20–25 June 2010; i) K.-S. Lee,
Data Availability Statement Prog. Photovoltaics 2013, 21, 195; j) L. D. Nielsen, IEEE Trans. Electron
Devices 1982, 29, 821; k) D. Pysch, A. Mette, S. W. Glunz, Sol. Energy
The data that support the findings of this study are available from the Mater. Sol. Cells 2007, 91, 1698; l) M. Seeland, H. Hoppe, Phys.
corresponding author upon reasonable request. Status Solidi A 2015, 212, 1991; m) A. Vishnoi, R. Gopal, R. Dwivedi,
S. K. Srivastava, IEE Proc. G: Circ., Devices Syst. 1993, 140, 155;
n) J.-M. Wagner, J. Carstensen, R. Adelung, Phys. Status Solidi A 2020,
Keywords 217, 1900612; o) J.-M. Wagner, S. Rißland, A. Schütt, J. Carstensen,
R. Adelung, Energy Procedia 2017, 124, 197; p) J.-M. Wagner,
figure of merit, photovoltaics, transparent conductive electrodes A. Schütt, J. Carstensen, R. Adelung, Energy Procedia 2016, 92, 255;
q) A. Luque, J. M. Ruiz, A. Cuevas, J. Eguren, M. G. Agost, presented
Received: October 7, 2021 at Photovoltaic Solar Energy Conference, Luxembourg, September 1977.
Revised: December 21, 2021 [11] C. P. Muzzillo, M. O. Reese, L. M. Mansfield, ACS Appl. Mater.
Published online: May 3, 2022 Interfaces 2020, 12, 25895.
[12] V. Bermudez, A. Perez-Rodriguez, Nat. Energy 2018, 3, 466.
[13] M. Burgelman, A. Niemegeers, Sol. Energy Mater. Sol. Cells 1998, 51,
129.
[1] A. Anand, M. M. Islam, R. Meitzner, U. S. Schubert, H. Hoppe, Adv. [14] K. D. M. Rao, C. Hunger, R. Gupta, G. U. Kulkarni, M. Thelakkat,
Energy Mater. 2021, 11, 2100875. Phys. Chem. Chem. Phys. 2014, 16, 15107.
[2] a) N. C. Wyeth, Solid-State Electron. 1977, 20, 629; b) M. A. Green, [15] D. Erath, M. Pospischil, R. Keding, M. Jahn, I. Lacmago Lontchi,
Solar Cells: Operating Principles, Technology, and System Applications, A. Lorenz, F. Clement, Energy Procedia 2017, 124, 869.
Prentice-Hall, Inc., New York 1982. [16] M. Bivour, S. Schröer, M. Hermle, S. W. Glunz, Sol. Energy Mater.
[3] G. T. Koishiyev, J. R. Sites, Sol. Energy Mater. Sol. Cells 2009, 93, 350. Sol. Cells 2014, 122, 120.
[4] T. M. Barnes, M. O. Reese, J. D. Bergeson, B. A. Larsen, [17] A. Cruz, E.-C. Wang, A. B. Morales-Vilches, D. Meza, S. Neubert,
J. L. Blackburn, M. C. Beard, J. Bult, J. van de Lagemaat, Adv. Energy B. Szyszka, R. Schlatmann, B. Stannowski, Sol. Energy Mater. Sol.
Mater. 2012, 2, 353. Cells 2019, 195, 339.

Adv. Energy Mater. 2022, 12, 2103119 2103119 (6 of 6) © Published 2022. This article is a U.S. Government work and is in the public domain in the USA

You might also like