Spectroscopic	Characteriza/on	of	Zinc	Phosphide	for	
Photovoltaic	Applica/ons
1.	Laboratory	of	Semiconductor	Materials,	Ins9tute	of	Materials,	École	Polytechnique	Fédérale	de	Lausanne,	Lausanne,	Switzerland.

2.	Xiong	Qihua	Group,	Division	of	Physics	and	Applied	Physics,	School	of	Physical	and	Mathema9cal	Sciences,	Nanyang	Technological	University,	Singapore.	
Nicolas	Humblot	1,2			
Supervision:	Elias	Stutz1
,	Anna	Fontcuberta	i	Morral1
,	Xiong	Qihua2
Mo/va/on	
The	terrestrial	abundance	of	materials	used	in	photovoltaic	(PV)	devices	is	of	great	interest	as	a	more	sustainable	way	of	genera9ng	power.	Zinc	phosphide	(Zn3P2)	is	an	earth	abundant	
material	with	a	direct	bandgap	that	is	promising	for	solar	cells	applica9ons.	Understanding	the	recombina9on	processes	and	finding	ways	to	reduce	non-radia9ve	pathways	are	crucial	for	
the	design	of	efficient	Zn3P2-based	solar	cells.		
	 Thesis	goal		
The	long	term	goal	is	to	produce	stable	and	efficient	solar	cells	with	Zn3P2	as	an	ac9ve	material.	The	aim	of	this	thesis	is	to	characterize	Zn3P2—grown	at	EPFL	using	molecular	beam	epitaxy	
on	graphene	(Van	der	Waals)	and	indium	phosphide	(InP)	substrates—by	means	of	photoluminescence	(PL)	and	Raman	spectroscopy*.	The	effects	of	passiva9ng	the	surface	with	an	
aluminum	oxide	layer	is	another	ongoing	inves9ga9on.	
Mo/va/on	&	thesis	goal	
The	 recombina9on	 processes	 in	 Zn3P2	 are	
dominated	 by	 surface	 recombina9ons.	 One	
Zn3P2	 grown	 on	 graphene	 sample	 was	 coated	
with	5	nm	of	aluminum	oxide	(Al2O3)	by	atomic	
layer	deposi9on	(ALD).	No	etching	of	the	na9ve	
oxide	 was	 performed.	 The	 ALD	 passiva9on	
increases	the	average	PL	intensity	by	a	factor	of	
2.3	(Fig.	6),	likely	reducing	one	source	of	non-
radia9ve	recombina9on.		
Time-resolved	PL	of	coated	and	uncoated	Zn3P2	
samples	 will	 help	 to	 iden9fy	 the	 bulk	 and	
surface	recombina9on	contribu9ons.		
Effects	of	a	passiva/on	layer	
• Op9mize	the	passiva9on	of	surfaces	with	adequate	oxide	and	remove	of	the	na9ve	one.	
• Master	the	growth	of	high	quality	Zn3P2	crystals	(e.g.	NWs)	and	study	their	defects	by	
means	of	PL	and	unusual	Raman	peaks.	
• Create	heterojunc9ons	using	zinc	nitride	(Zn3N2)	as	n-type	layer	and	Zn3P2	as	p-type.		
Outlook	
The	PL	spectra	described	here	were	acquired	with	532	or	633	nm	illumina9ons.		
• Room	Temperature	(RT)	
PL	 emissions	 from	 two	 Zn3P2	 samples	 grown	
on	graphene	show	two	dis9nct	spectra	(Fig.	1)	
at	RT.	The	sample	with	a	smaller	growth	9me	
(red	 curve)	 emits	 at	 1.53	 eV	 with	 a	 lower	
energy	shoulder	around	1.43	eV.	The	posi9on	
of	 the	 emission	 is	 unchanged	 with	 increasing	
laser	 power	 and	 the	 power	 law	 k=log(PL	
intensity)/log(Laser	 intensity)>1	 suggests	 an	
excitonic	 origin[1]	 of	 the	 signal.	 The	 longer	
growth	9me	sample	(blue	curve)	broadly	emits	
around	1.45	eV,	but	increasing	the	laser	power	
redshijs	 the	 emission,	 sugges9ng	 several	 electronic	 transi9ons	 to	 happen.	 The	 crystal	
quality	may	be	lower	for	this	sample	(perhaps	containing	more	grain	boundaries).		
•	From	300K	to	200K		
With	decreasing	temperature,	on	the	lower	growth	9me	
sample,	the	1.53	eV	transi9on	quenches	while	the	lower	
energy	 one	 increases	 (grey	 arrows	 Fig.	 2).	 A	 similar	
behavior	was	observed	on	a	previous	study[2].	The	origins	
of	the	peaks	we	observe	are	likely	to	be	comparable	to	the	
direct	 (1.50	 eV)	 and	 indirect	 (1.38	 eV)	 interband	
transi9ons	that	the	authors	reported	for	Zn3P2.	
The	1.41	eV—well	defined—peak	at	200K	can	be	related	
to	 an	 indirect	 transi9on	 in	 Zn3P2.	 This	 PL	 peak	 does	 not	
shij	with	increasing	laser	power	and	its	high	energy	tail	
suggests	band-to-band	recombina9ons.	The	power	law	k	=	
log(PL	 intensity)/log(Laser	 intensity)	 >	 1	 hints	 an	 exciton	
line[1].		
• From	150K	to	10K	
The	 1.4	 eV	 peak	 at	 150K
— p r e s e n t	 i n	 b o t h	
samples—either	 redshijs	
in	one	sample	(red	arrow	
Fig.	 3)	 as	 in	 a	 previous	
study[3]	 or	 shows	 a	
c h a n g e	 i n	 t h e	
temperature	 coefficient	
(red/blue	 arrows	 Fig.	 4)	
upon	 cooling.	 In	 this	
laner	case,	a	combina9on	
of	 several	 peaks	 with	
posi9ve	 and	 nega9ve	
temperature	 coefficients	
can	be	responsible	for	the	change[3].	A	low	energy	(1.25	eV)	defect	peak	rises	below	150K	
and	suggests	a	lower	crystal	quality	of	this	longer	growth	9me	sample.		
•	10K		
At	 low	 temperature,	 a	 phonon	 replica	 is	
suspected	to	contribute	to	the	low	energy	tail	of	
the	 asymmetric	 emission	 peak	 (Fig.	 5).	 The	
fiong	with	two	gaussian	func9ons	leads	to	a	45	
meV	energy	of	the	phonon,	close	to	the	43	meV	
reported	 in	 the	 literature[2,3].	 Power	 series	
suggests	 that	 the	 origin	 of	 the	 emission	 is	 no	
longer	 excitonic	 but	 rather	 donor-acceptor	
transi9ons.	The	blueshij	(blue	arrow)	could	be	
an	 indicator	 of	 slow	 recombina9on	 processes	
through	 the	 indirect	 bandgap	 of	 Zn3P2	 and	 be	
used	to	study	the	material	quality.		
Photoluminescence	from	Zn3P2	grown	on	graphene	
900875850825800775
Wavelength [nm]
1.41.451.51.551.6
Energy [eV]
0
2
4
6
8
10
PLintensity[arb.u.]
Uncoated
ALD coated
900875850825800775750
Wavelength [nm]
1.351.41.451.51.551.61.65
Energy [eV]
PLintensity[arb.u.]
Sample A
Sample C
1	μm
1	μm
(1)
950900850800
Wavelength [nm]
1.31.41.51.6
Energy [eV]
PLintensity[arb.u.]
200K
220K
240K
260K
280K
300K
1000950900850
Wavelength [nm]
1.21.251.31.351.41.451.5
Energy [eV]
0
2
4
6
8
10
PLintensity[arb.u.]
12 W
53 W
129 W
194 W
391 W
1.17 mW
(5)
1000950900850
Wavelength [nm]
x3
x2
1.21.31.41.5
Energy [eV]
PLintensity[arb.u.]
6.8K
21K
34K
39.9K
50.5K
60K
80K
110K
150K
1000950900850
Wavelength [nm]
1.21.31.41.5
Energy [eV]
PLintensity[arb.u.]
12.1K
22K
31K
40.8K
50.3K
60K
70K
80K
100K
120K
150K
(4)1	μm 1	μm
Raman	spectrum	of	Zn3P2	was	acquired	at	
10K,	391	μW	at	633	nm	(Fig.	8).	The	arrows	
iden9fy	the	peaks	ascribed	to	different	
irreducible	modes	(A1g,	B1g,	B2g	or	Eg)[3].	The	
good	correla9on	of	the	11	observed	peaks	
with	previous	studies	supports	the	presence	
of	the	desired	phase	(α-Zn3P2)	in	our	
material.	In	addi9on	to	the	zone-centered	
Raman	ac9ve	phonons,	the	two	∗	annotated	
peaks	are	unusual	addi9onal	A1g	modes	that	may	be	related	to	phosphorus	inters99als	
ac9ng	as	acceptors	in	Zn3P2	(intrinsically	p-type).	Using	different	laser	excita9ons	for	the	
study	if	these	peaks	could	be	a	useful	tool	for	understanding	the	material	quality.		
Raman	spectroscopy	of	Zn3P2	grown	on	graphene	
*
*
200 250 300 350 400 450
Raman shift [cm-1
]
0
2
4
6
8
10
Ramanintensity[arb.u.]
(8)
1	μm
1	μm
1	μm
(6)
Zn3P2	 thin	 films	 grown	 on	 InP	 cover	 the	
whole	 substrate.	 The	 PL	 from	 Zn3P2	 thin	
films	 and	 nanowires	 grown	 on	 InP	 is	
dominated	by	emission	from	the	substrate.		
InP	has	three	characteris9c	peaks	that	we	
all	 observe	 (inset	 Fig.	 7).	 We	 addi9onally	
observed	 a	 shoulder	 (dashed	 line)	 on	 the	
defect	 InP	 peak	 (full	 line).	 It	 was	 first	
suggested	 that	 Zn3P2	 contribute	 to	 this	
shoulder.	 However,	 this	 hypothesis	 was	
discredited	 by	 the	 observa9on	 of	 the	
phonon	replica	of	the	defect	InP	peak	at	1.34	eV	(inset	Fig.	7).	This	InP	phonon	replica	
shows	the	same	shoulder	and	behavior	with	increasing	power.	The	shoulder	is	thus	due	
to	a	transi9on	happening	in	InP	(most	likely	a	donor-acceptor	pair	transi9on).		
905.0901.9899.1892.0
Wavelength [nm]
1.371.37481.37891.39
Energy [eV]
0
2
4
6
8
10
PLintensity[arb.u.]
5.55 W
20.8 W
40.8 W
145 W
928899876
1.341.381.42
PLint.[a.u.]
Photoluminescence	from	Zn3P2	grown	on	InP	
(7)


1.	Schmidt,	T.	et	al.,	Physical	Review	B	45,	8989	(1992).	
2.	Kimball,	G.	et	al.,	APL	95,	112103	(2009).
3.	Briones,	F.	et	al.,	APL	39,	805–807	(1981).		
4.	Pangilinan,	G.	et	al.,	Physical	Review	B	44,	2582	
References	
*	Both	spectroscopic	techniques	are	widely	known	and	not	described	in	this	poster
(3)

Poster master thesis

  • 1.
    Spectroscopic Characteriza/on of Zinc Phosphide for Photovoltaic Applica/ons 1. Laboratory of Semiconductor Materials, Ins9tute of Materials, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland.
 2. Xiong Qihua Group, Division of Physics and Applied Physics, School of Physical and Mathema9cal Sciences, Nanyang Technological University, Singapore. Nicolas Humblot 1,2 Supervision: Elias Stutz1 , Anna Fontcuberta i Morral1 , Xiong Qihua2 Mo/va/on The terrestrial abundance of materials used in photovoltaic (PV) devices is of great interest as a more sustainable way of genera9ng power. Zinc phosphide (Zn3P2) is an earth abundant material with a direct bandgap that is promising for solar cells applica9ons. Understanding the recombina9on processes and finding ways to reduce non-radia9ve pathways are crucial for the design of efficient Zn3P2-based solar cells. Thesis goal The long term goal is to produce stable and efficient solar cells with Zn3P2 as an ac9ve material. The aim of this thesis is to characterize Zn3P2—grown at EPFL using molecular beam epitaxy on graphene (Van der Waals) and indium phosphide (InP) substrates—by means of photoluminescence (PL) and Raman spectroscopy*. The effects of passiva9ng the surface with an aluminum oxide layer is another ongoing inves9ga9on. Mo/va/on & thesis goal The recombina9on processes in Zn3P2 are dominated by surface recombina9ons. One Zn3P2 grown on graphene sample was coated with 5 nm of aluminum oxide (Al2O3) by atomic layer deposi9on (ALD). No etching of the na9ve oxide was performed. The ALD passiva9on increases the average PL intensity by a factor of 2.3 (Fig. 6), likely reducing one source of non- radia9ve recombina9on. Time-resolved PL of coated and uncoated Zn3P2 samples will help to iden9fy the bulk and surface recombina9on contribu9ons. Effects of a passiva/on layer • Op9mize the passiva9on of surfaces with adequate oxide and remove of the na9ve one. • Master the growth of high quality Zn3P2 crystals (e.g. NWs) and study their defects by means of PL and unusual Raman peaks. • Create heterojunc9ons using zinc nitride (Zn3N2) as n-type layer and Zn3P2 as p-type. Outlook The PL spectra described here were acquired with 532 or 633 nm illumina9ons. • Room Temperature (RT) PL emissions from two Zn3P2 samples grown on graphene show two dis9nct spectra (Fig. 1) at RT. The sample with a smaller growth 9me (red curve) emits at 1.53 eV with a lower energy shoulder around 1.43 eV. The posi9on of the emission is unchanged with increasing laser power and the power law k=log(PL intensity)/log(Laser intensity)>1 suggests an excitonic origin[1] of the signal. The longer growth 9me sample (blue curve) broadly emits around 1.45 eV, but increasing the laser power redshijs the emission, sugges9ng several electronic transi9ons to happen. The crystal quality may be lower for this sample (perhaps containing more grain boundaries). • From 300K to 200K With decreasing temperature, on the lower growth 9me sample, the 1.53 eV transi9on quenches while the lower energy one increases (grey arrows Fig. 2). A similar behavior was observed on a previous study[2]. The origins of the peaks we observe are likely to be comparable to the direct (1.50 eV) and indirect (1.38 eV) interband transi9ons that the authors reported for Zn3P2. The 1.41 eV—well defined—peak at 200K can be related to an indirect transi9on in Zn3P2. This PL peak does not shij with increasing laser power and its high energy tail suggests band-to-band recombina9ons. The power law k = log(PL intensity)/log(Laser intensity) > 1 hints an exciton line[1]. • From 150K to 10K The 1.4 eV peak at 150K — p r e s e n t i n b o t h samples—either redshijs in one sample (red arrow Fig. 3) as in a previous study[3] or shows a c h a n g e i n t h e temperature coefficient (red/blue arrows Fig. 4) upon cooling. In this laner case, a combina9on of several peaks with posi9ve and nega9ve temperature coefficients can be responsible for the change[3]. A low energy (1.25 eV) defect peak rises below 150K and suggests a lower crystal quality of this longer growth 9me sample. • 10K At low temperature, a phonon replica is suspected to contribute to the low energy tail of the asymmetric emission peak (Fig. 5). The fiong with two gaussian func9ons leads to a 45 meV energy of the phonon, close to the 43 meV reported in the literature[2,3]. Power series suggests that the origin of the emission is no longer excitonic but rather donor-acceptor transi9ons. The blueshij (blue arrow) could be an indicator of slow recombina9on processes through the indirect bandgap of Zn3P2 and be used to study the material quality. Photoluminescence from Zn3P2 grown on graphene 900875850825800775 Wavelength [nm] 1.41.451.51.551.6 Energy [eV] 0 2 4 6 8 10 PLintensity[arb.u.] Uncoated ALD coated 900875850825800775750 Wavelength [nm] 1.351.41.451.51.551.61.65 Energy [eV] PLintensity[arb.u.] Sample A Sample C 1 μm 1 μm (1) 950900850800 Wavelength [nm] 1.31.41.51.6 Energy [eV] PLintensity[arb.u.] 200K 220K 240K 260K 280K 300K 1000950900850 Wavelength [nm] 1.21.251.31.351.41.451.5 Energy [eV] 0 2 4 6 8 10 PLintensity[arb.u.] 12 W 53 W 129 W 194 W 391 W 1.17 mW (5) 1000950900850 Wavelength [nm] x3 x2 1.21.31.41.5 Energy [eV] PLintensity[arb.u.] 6.8K 21K 34K 39.9K 50.5K 60K 80K 110K 150K 1000950900850 Wavelength [nm] 1.21.31.41.5 Energy [eV] PLintensity[arb.u.] 12.1K 22K 31K 40.8K 50.3K 60K 70K 80K 100K 120K 150K (4)1 μm 1 μm Raman spectrum of Zn3P2 was acquired at 10K, 391 μW at 633 nm (Fig. 8). The arrows iden9fy the peaks ascribed to different irreducible modes (A1g, B1g, B2g or Eg)[3]. The good correla9on of the 11 observed peaks with previous studies supports the presence of the desired phase (α-Zn3P2) in our material. In addi9on to the zone-centered Raman ac9ve phonons, the two ∗ annotated peaks are unusual addi9onal A1g modes that may be related to phosphorus inters99als ac9ng as acceptors in Zn3P2 (intrinsically p-type). Using different laser excita9ons for the study if these peaks could be a useful tool for understanding the material quality. Raman spectroscopy of Zn3P2 grown on graphene * * 200 250 300 350 400 450 Raman shift [cm-1 ] 0 2 4 6 8 10 Ramanintensity[arb.u.] (8) 1 μm 1 μm 1 μm (6) Zn3P2 thin films grown on InP cover the whole substrate. The PL from Zn3P2 thin films and nanowires grown on InP is dominated by emission from the substrate. InP has three characteris9c peaks that we all observe (inset Fig. 7). We addi9onally observed a shoulder (dashed line) on the defect InP peak (full line). It was first suggested that Zn3P2 contribute to this shoulder. However, this hypothesis was discredited by the observa9on of the phonon replica of the defect InP peak at 1.34 eV (inset Fig. 7). This InP phonon replica shows the same shoulder and behavior with increasing power. The shoulder is thus due to a transi9on happening in InP (most likely a donor-acceptor pair transi9on). 905.0901.9899.1892.0 Wavelength [nm] 1.371.37481.37891.39 Energy [eV] 0 2 4 6 8 10 PLintensity[arb.u.] 5.55 W 20.8 W 40.8 W 145 W 928899876 1.341.381.42 PLint.[a.u.] Photoluminescence from Zn3P2 grown on InP (7) 
 1. Schmidt, T. et al., Physical Review B 45, 8989 (1992). 2. Kimball, G. et al., APL 95, 112103 (2009). 3. Briones, F. et al., APL 39, 805–807 (1981). 4. Pangilinan, G. et al., Physical Review B 44, 2582 References * Both spectroscopic techniques are widely known and not described in this poster (3)