30

Chapter 4

Fabricating Graphene by Chemical Exfoliation of

Graphite and Combined Experimental and Theoretical

Study on rGO/p-Si Heterojunction Solar Cell

Combined experimental and theoretical investigations on the heterojunctions of chemically

derived graphene with Si have been presented. The stability study of graphene oxide (GO)

and reduced GO (rGO) in aqueous medium were performed by visual observation and

surface charge measurement. The detailed characterizations by FT-IR, UV-Vis and Raman

exhibited the formation of rGO with a high optical band gap of 3.6 eV. The rGO was spin-

coated on the p-Si substrate for fabrication of a heterojunction device, with the structure of

rGO/p-Si. In the fabricated device, incident light was transmitted through the thin rGO film

to reach the junction interface, generating photoexciton, and thereby a photo-conversion

efficiency of 0.02% was achieved experimentally and its (rGO/p-Si heterojunction device)

theoretical simulation using SCAPS 1-D tool showed the efficiency of 1.32%. Such large

deviations in efficiency between experiment and theory have been discussed in details. In

addition to the material stability test, the device stability has also been verified both

experimentally and theoretically.

The contents of this chapter have been published in Carbon and Journal of Nanoscience and

Nanotechnology:

S. K. Behura, S. Nayak, I. Mukhopadhyay, O. Jani, “Junction characteristics of chemically-

derived graphene/p-Si hetero-junction solar cell,” Carbon, Vol. 67, p. 766-774 (2014).

K. Batra, S. Nayak, S. K. Behura, O. Jani, “Optimizing Performance Parameters of

Chemically-Derived Graphene/p-Si Heterojunction Solar Cell,” Journal of Nanoscience and

Nanotechnology, doi:10.1166/jnn.2014.9818 (2014).

31

4.1 Introduction

Graphene due to its high room temperature carrier mobility (15000 cm2/V.s) [1], tunable

band gap [50], ballistic transport with a mean free path of 300-500 nm [51] and transparency

of 97.7% [52], makes it a super candidate material for next generation electronic applications.

Among the various potential applications, graphene especially has shown great potential for

creating photovoltaic solar devices. Several attempts have already been made to incorporate

graphene into various solar cells as transparent electrodes [53], electron acceptors [54], hole

acceptors [55], counter electrodes [56] and photoactive promoters [57]. The first prediction

by Tongay et al. [58] on single-layer graphene contacted to a semiconductor substrate and

later, the formation of graphene/semiconductor Schottky barriers was experimentally

verified, opening the opportunity for graphene/p-Si heterojunction solar cells [27].

The fabrication of graphene/p-Si heterojunction solar cell from chemically-derived

graphene at room temperature is an excellent solution to the problems related to high-

temperature junction formation and graphene can be used as the junction emitter, as the

passivation layer, and anti-reflective layer. The simple planar structure of the graphene/p-Si

heterojunction device also helps in controlling the processing costs. Herein, graphene film not

only serves as a transparent electrode for light transmission on semiconductor photovoltaic

device [59, 60], but also as an active layer for electron/hole separation and hole transporting

medium. A maximum efficiency of 10.30% has been recently achieved using doped few-

layer graphene/silicon nanoarray configuration [61]. It was found that surface charge

recombination as well as graphene conductivity along with work function played important

roles in determining the solar cell performance. Nitric acid (HNO3) has been widely used to

dope graphene film to enhance the cell performance [61, 62]. This is due to the p-type

chemical doping effect of HNO3 which increases the work function, the carrier density of

graphene (decreasing the series resistance) and the built-in potential (increasing the open

circuit voltage).

An understanding of the physical limits for conversion of radiation into electrical

power of semiconductors is important for designing electronic devices and for understanding

their function and performance. Application of graphene in this field is a very open question

and requires investigation on different aspects of the material and the devices. The recent

works based on graphene-on-semiconductors heterojunction solar cells have been

32

summarized in Table 4.1. Most of the past reports demonstrate the heterojunction device

fabricated from a transferred CVD and mechanically exfoliated graphene film on Si [63, 64].

The metallic characteristics of the CVD and exfoliated graphene hinder the possibility of

fabricating p-n heterojunction based devices. Keeping in view the above mentioned

shortfalls, the chemically-derived rGO with high optical band gap can be suitable for a

heterojunction device fabrication. In addition, this will also avoid high-cost deposition

techniques and complicated processing, which are essential for Si-based p-n and p-i-n type of

devices. Therefore, an attempt has been taken to study the simplest heterojunction made of

rGO-on-p-Si without any doping or other configurations.

In this work, we conducted a comprehensive study on chemically-derived graphene-

on-Si-based Schottky junction solar cells using both experimental and computational

techniques. Chemically-derived graphene was synthesized using modified Hummers method,

subsequently reduced using NaBH4 and spin-coated on p-Si substrates. The present results

suggest great potential of the graphene/Si as high-efficiency and low-cost photovoltaic

devices. SCAPS (A Solar cell Capacitance Simulation) program was used to simulate the

model rGO/p-Si device of the present study using experimental results for chemically-derived

graphene.

Table 4.1: List of graphene-on-semiconductor based solar cells in the literature.

Sr. Device Type VOC (V) JSC (mA/cm2) FF η (%) References

1 Graphene-P3HT/C60 0.43 3.5 0.41 0.61 [65]

2 GS/n-Si 0.48 6.5 0.56 1.70 [63]

3 Graphene/Si 0.517 13.2 0.58 3.93 [66]

4 FLG/Graphite - - 4.35 [67]

5 Graphene/n-SiNWs 0.462 9.2 0.30 1.25 [68]

6 G/Si Pillar Array 0.487 16.03 0.45 3.55 [69]

7 SLG/SiNWs 0.19 0.154 0.25 2.15 [70]

8 SLG/n-Si 0.54 25.3 0.63 8.6 [71]

9 rGO/n-Si 0.254 4.28 0.23 0.25 [72]

10 Graphene/Si 0.51 24.28 0.6 7.5 [73]

11 FLG/P3HT/SiNH 0.48 38.86 0.552 10.3 [61, 62]

12 TiO2/Graphene/Si 0.612 32.7 0.72 14.5 [74]

33

4.2 Experimental Methodology

For characterization and device fabrications, the experimental studies began with the

synthesis of graphene oxide (GO) from pure graphite (purity ~99.9%) by modified Hummers

method [8, 75]. The schematic of the whole chemical synthesis process is demonstrated in

figure 4.1. GO contains a range of reactive oxygen functional groups, which renders it a good

candidate for use in the electronic device applications. Therefore, it is very much essential to

reduce the highly resistive GO. It was chemically reduced by sodium borohydride (NaBH4), a

strongly reducing agent.

The aqueous solutions of GO and reduced GO (rGO) were prepared in DI water with

a concentration of 1 mg/ml. The solutions were spin-coated on p-Si and glass substrates for

device fabrication and characterization. The morphology of graphite flakes, GO and rGO

were characterized using scanning electron microscopy (SEM) (JEOL, JSM-6010LV) and

atomic force microscopy (AFM) (Pico-IC). The transmission electron microscopy (TEM)

images were recorded using FEB Tecnai G2 20, instrument operated at 200 kV (The

Netherland) to observe the nanoscale structures. Raman spectrometer (Invia Reflex/514,

Incoterm, UK) was used to confirm GO and its reduction. Raman spectroscopy was carried

out with laser excitation energy of 514 nm. The optical characterizations were carried out

using Fourier transform infrared spectroscopy (Varian 800, Japan) and UV-Vis (UV-2600,

Shimadzu, Japan). The thermo-gravimetric analysis (TGA) of the materials was performed

under continuous argon atmosphere on a PerkinElmer (USA) Pyris 1 analyser. The samples

were scanned from 50 oC to 1000

oC at a heating rate of 10

oC/min. GO samples were heated

from 50 oC to 500

oC at 1

oC/min to avoid thermal expansion due to rapid heating. Particle

charge detector (Mutek) was used to determine the specific surface charges of GO in different

solvents. Current-voltage (I-V) characterizations of the fabricated devices were done with the

Photo Emission Tech. solar simulator in dark and under air mass (AM) 1.5 simulated solar

radiations. The computational study was carried out using SCAPS simulation software, which

is one dimensional simulation program developed at University of Ghent. This program was

designed basically for the simulation and studying the properties of photovoltaic devices.

34

Figure 4.1: Scheme showing the chemical route for the synthesis of graphene. 1: Oxidation

of graphite to graphite oxide with oxygen function groups on the surface. 2: Exfoliation of

graphite oxide in water by sonication to obtain GO colloids that are stabilized by electrostatic

force of repulsion. 3: Controlled conversion of GO colloids to conducting graphene colloids

through deoxygenation by sodium borohydride reduction.

4.3 Characterization of Graphene

There are many reports which represent the stability of GO in aqueous medium rather than

organic medium [76]. So, first we tested the stability of the GO and rGO by measuring the

surface charge (zeta potential) of as-prepared GO sheets in aqueous as well as in organic

medium (ethanol, propanol and dimethyl formamide). The results show that these sheets are

highly negatively charged (40 C/g ± 20) when dispersed in water rather than organic medium

[Figure 4.2 (a)], as a consequence of ionization of the carboxylic acid and phenolic hydroxyl

groups that are already exist on the GO sheets [76]. This result suggests that the formation of

stable GO colloids should be attributed to electrostatic repulsion, rather than just the

hydrophilicity of GO as previously assumed. The inset of figure 2a shows the stability of GO

and rGO in water for 0 and 24h. After 24 h, the rGO was settled down, while GO was

dispersed with brownish colour. This is due to the fact that, carboxylic acid groups are likely

to be reduced by borohydride under the given reaction conditions [77], therefore these groups

35

are absent in the reduced product as confirmed by the FT-IR analysis [Figure 4.2 (b)]. The

oxygen containing functional groups of GO generated bands at 870 cm-1

(C=C, conjugation),

1090 cm-1

(C-O stretching), 1234 cm-1

(C-OH stretching), 1420 cm-1

(C-O-H deformation),

1635 cm-1

(C=C stretching) and 1750 cm-1

(C=O stretching) [78]. However, the oxygen

containing functional groups were almost entirely removed during reduction except few

groups, which further confirms the oxidation and reduction of graphite flakes.

Figure 4.2: (a) Surface charge with stability analysis (inset) and (b) FT-IR spectra of GO and

rGO.

Furthermore, the identification rGO formation was confirmed by Raman analysis,

which is the fingerprint for carbon materials identification. Figure 4.3 shows the Raman

studies of GO and rGO, presenting a disorder induced D-band at 1348 cm-1

and a graphitic G-

band at around 1574 cm-1

. The strong D-band is associated with vibrations of carbon atoms

with dangling bonds or formation of sp3 hybridization with oxidation. At higher wavenumber,

a small peak at 2692 cm-1

and a broader peak at 2905 cm-1

is observed corresponding to the

2D and D + G combinational mode, respectively. The D+G-band is the combination of D and

G mode [79, 80]. Further detailed physics of Raman modes are demonstrated pictorially in

Appendix II. The slight reduction in 2D peak and the presence of a broad D + G peak signify

well oxidation of the graphitic material. However, there is no significant difference of the

Raman spectra for the GO and rGO, rather only a small enhancement in the IG/ID ratio is

observed which is shown in the inset of Figure 4.3.

36

Figure 4.3: Raman spectra of graphite, GO and rGO taken at Raman excitation wavelength

of 514 nm. Inset indicates the reduction of D- and G-band intensity, which confirms the

formation of rGO.

As the morphology of the GO and rGO are known to govern their optical and

electrical properties, these synthesized sample were subjected to detailed morphological

characterization by SEM, TEM and AFM. Figure 4.4 (a) shows SEM image of starting

graphite powder. Flakes of 40-50 µm are clearly observed. The inset depicts a digital image

of graphite powder. Figure 4.4 (b) shows the morphology of the prepared GO sample. A

smooth and flat overlapping structure reveals that the thin sheets of graphene oxides were

stacked randomly, and few layers are folded. Figure 4.4 (c) and (d) shows the surface

morphology and TEM image of rGO which is transparent in nature. The inset of figure 4.4

(d) depicts selected area electron diffraction pattern, which is typical of few layer with

crystalline structure. These samples were very stable under electron beam.

37

Figure 4.4: SEM micrograph of (a) graphite powder with its digital image in the inset, (b)

GO, (c) rGO and TEM micrograph of (d) rGO with its SAED pattern in the inset.

To evaluate the size and thickness of the rGO flake, AFM analysis was subjected to

detail. Figure 4.5 (a) shows an AFM image of the rGO flake. Figure 4.5 (b) shows the

thickness profile image, presenting the thickness of the rGO flake as around 10 nm. Here, it

should be noted that a pure single layer graphene has a thickness of 0.34 nm, however, a rGO

sheet is ~1 nm thick with presence of functional groups, defects and absorbed water

molecules on rGO surfaces [81]. The AFM studies show that the lateral size of rGO flakes

are of the order of 1-5 μm, consisting of 10 layers of graphene. The thickness of the GO and

rGO thin film can be controlled with the concentration of the solution during spin coating,

spinning speed and spinning time.

38

Figure 4.5: (a) AFM topography image and (b) thickness profile of an rGO showing flake of

thickness ≈ 10 nm.

Thermal stability is an important criterion for the use of graphene-based devices under

high temperature conditions. For this purpose, graphite, GO and rGO were investigated using

TGA, as presented in figure 4.6. The TGA curves of pristine graphite show a very negligible

weight loss around 1.5% of its total weight. On the other hand, GO shows a weight loss of

10% at 120 oC and then 25% at 200

oC corresponding to the removal of physically adsorbed

water, moisture and COOH groups, respectively. Next, the fast weight loss of GO takes place

until 440 oC due to pyrolysis of oxygen bearing functional groups associated with GO. It

clearly shows that thermal stability of GO is very poor compared to graphite. Furthermore,

rGO shows very steady weight loss with temperature, which is around 30% of its total

weight. This apparently shows removal of oxygen bearing functional groups after reduction.

It also shows that thermal stability of rGO is better than GO. This observation also indicates

the reduction of GO to rGO by removing the carboxyl, epoxide and hydroxyl groups. DTA

curve (figure 6 (b) also shows a strong exothermic peak at around 600 oC in case of rGO

corresponding to the combustion of rGO molecule.

39

Figure 4.6: (a) TGA and (b) DTA plot for graphite, GO and rGO.

In addition to the thermal properties, optical properties, in particular bandgap, is the

key factor which governs the photo-conversation efficiency. GO due to its insulating nature

possesses high optical band gap, which can be optimized using the ratio of sp2 and sp

3

carbon. Since we focused on the heterojunction device based on rGO; the transmittance of

rGO-coated on glass substrates was measured (Figure 4.7). The transmittance increases

throughout the visible region and slightly decreases near infrared light region. The rGO-

coated glass shows a transmittance of 80.92% at 550 nm wavelength. The optical gap of rGO

can be obtained from the Tauc plot, using the relation αhν = (hν – Eg)1/2

, where α is the

absorption coefficient, hν is the photon energy and Eg is the optical gap. The inset depicts a

Tauc plot for the as-synthesized rGO. The optical band gap calculated to be 3.62 eV.

40

Figure 4.7: Transmittance of rGO on glass substrate with the inset shows the Tauc plot.

The main challenge in graphene for optoelectronics research is the optimized value of

transmittance (Tr) and sheet resistance (Rs’). Therefore, rGO films of various thicknesses

were spin-coated on SiO2/Si substrate and the Tr and Rs’ were measured as a function of

thickness and shown in Figure 4.8. Lowest sheet resistance of 1 kΩ/□, though at a low

transparency of ~ 72%, was achieved for rGO films with thickness of ~ 15 nm. The best

optoelectronic properties were obtained for lower thicknesses with a thin film of 10 nm

exhibiting sheet resistance of ~ 1.2 kΩ/□ at a transmittance of ~ 81%, while 5 nm thin films

exhibited transmittance of ~ 90% at a sheet resistance of ~ 20 kΩ/□ Since for single-layer

graphene, the transmittance value may approach 97%, while its sheet resistance is very high

of the order of several kΩ/□ For this reason, doping of the graphene film with HNO3, has

emerged for the high photo-conversion efficiency of these types of devices. Doping has

advantage of decreasing the Rs’ of the film.

41

Figure 4.8: Transmittance and sheet resistance of rGO thin films as a function of thickness.

4.4 Fabrication and Characteristics of rGO/p-Si Heterojunction

Solar Cell

To investigate the potential application of chemically-derived graphene (CDG) in

optoelectronic devices, we have fabricated CDG-based heterojunction solar cell by spin-

coating solution processed graphene on p-Si. Spin-coating of CDG on p-Si has advantages

over other carbon nanostructures of being naturally compatible with thin film processing,

making large device areas. As it is described in the experimental methodology, the aqueous

solutions of rGO were prepared in DI water with a concentration of 1 mg/ml and spin-coated

on p-Si followd by baking in 80 oC for 10 minutes.

4.4.1 Current-Voltage Characteristics: Figure 4.9 (a)-(d) shows the schematic

diagram, the I-V curve both under dark and illumination of 1000 W/m2, the equivalent circuit

and the energy-band diagram of the fabricated rGO/p-Si heterojunction device, respectively.

I–V characteristic (Figure 4.9 b) in dark condition shows very good rectification with small

42

leakage current in the reverse bias. For a Schottky barrier diode with assumption that the

current is due to thermionic emission, the relation between the applied forward bias and

current can be expressed as [27]:

I = IS [exp (VD/nkBT)-1]...................................................................................................... (4.1)

Where I is the diode current, IS is the reverse bias saturation current, VD is the voltage across

the diode, kBT is the thermal voltage, (kB is the Boltzmann constant and T is the temperature

in Kelvin), and n is the ideality factor. IS can be extracted by extrapolating the straight line of

Ln (I) to intercept the axis at zero voltage:

IS = AA*T2 exp (-qφSBH/kBT).............................................................................................. (4.2)

Where A is the effective area of the device (0.25 cm2), A* is the Richardson constant (32

A/(cm2K

2) for p-Si substrates.

Under illumination, the fabricated device with rGO shows a photovoltaic action.

Photo excited electrons and holes are generated in the Si substrate which are separated and

collected by means of a built-in electric field at the heterojunction of rGO/Si. The device

characteristics are open-circuit voltage (Voc) of 0.27 V, short-circuit current (Isc) of 0.11 mA

and fill factor (FF) of 0.12 with power conversion efficiency (η) 0.02%. This efficiency value

is one order magnitude higher than the previously reported rGO/p-Si device efficiency by

Mohammed et al. [82]. Comparing with another device rGO/n-Si work [72], our device

shows a magnitude decrease in efficiency. This might be due to the partial reduction of GO as

it is confirmed from the measurement of the optical band gap (rGO) ≈ 3.62 eV. According to

G. Kalita et al. [72], their GO and partial reduced GO show an optical band gap of 3.6 and 2.8

eV, respectively. The lower conductivity of the synthesized rGO in our case, gave rise to

higher series resistance and ultimately lowers the Isc of the fabricated device.

43

Figure 4.9: (a) Schematic view of prototype rGO/p-Si heterojunction device, (b) I-V curve

under light and dark, (c) equivalent circuit model and (d) energy-band diagram.

The equivalent circuit shows that the current density is greatly influenced by the

series resistance (Rs) of the device. The shunt resistance (Rsh) takes into account all parallel

resistive losses across the photovoltaic device including leakage current. The RS is mainly due

to the bulk and contact resistance of the device. By neglecting the shunt resistance (Rsh) of the

device, the forward bias current is mainly affected by RS.

Rectification behaviour is observed at the interface of graphene/p-Si device with the

appearance of a barrier height, which is calculated experimentally below. Considering the

work function for graphene (ΦG) = 4.70 eV, and electron affinity for Si (χp-Si) = 4.05 eV, the

calculated barrier height (ΦBP) and built-in potential (Vbi) are given in Eq. 4.3 and 4.4.

BP = Eg - G + χp-Si = 0.47 eV ..…………………..………………………………......(4.3)

The difference in potential between Fermi level and valance band edge (Vp) = 0.12 eV

Now, Vbi = Bp – Vp= 0.47 – 0.12 = 0.35 eV...………………………………......………. (4.4)

44

4.4.2 Capacitance-Voltage Characteristics: A capacitance-voltage (C-V)

measurement was employed to probe the rGO/p-Si heterojunction system for the knowledge

on associated changes in the density of ionized donors (ND) and built-in-potential (Vbi) which

is presented in Eq. 4.4. Now, the solution to the Poisson’s equation for a specific charge

distribution in the depletion region gives rise to a relationship (Eq. 4.5),

1/C2 = 2 (Vbi + V)/(qNDεs)…………………………………………………………………(4.5)

Here, the parameter q is the charge of electron and εs is the dielectric constant of the

semiconductor. Figure 4.10 displays C-V and 1/C2-V plots at 100 kHZ for a rGO/p-Si

heterojunction at room temperature. The resulting 1/C2 analysis confirmed the Si to be p-type

with a hole concentration of 2 x 1014

cm-3

. The built-in potential was extracted to be

approximately ≈ 0.3 eV, which is consistent with the band bending that would be required for

the Fermi levels of the graphene and Si to align.

Figure 4.10: C vs. V and 1/C2 vs. V (inset) plot of rGO/p-Si at 100 kHz and 300 K.

45

4.5 Computational Methodology and Simulations

After evaluating the structural, optical, stability and electrical parameters of chemically-

derived graphene, the possibility of implementation as a heterojunction solar cell and its

stability was verified. The simulation work was carried out using SCAPS program. This

simulation package is equipped with advanced tools for I–V, QE (quantum efficiency) and

carrier transport simulations [83]. Simulation models are generated by digital description of

physical parameters of each structure layer, including contacts. Program SCAPS is constantly

developed since 1990 and available free of charge for scientific research. SCAPS uses

simultaneous of Poisson’s equation, the continuity equation and the boundary condition to

model one-dimensional solar cells. The program output includes simulated capacitance

values.

4.5.1 Simulation and Stability of rGO/p-Si Heterojunction Solar Cell: To

understand our experimental results, a thorough computational study was undertaken to

investigate the influence of rGO thickness on the efficiency of this simplest device. An

efficiency of 1.32% was found for rGO/p-Si device, while considering our experimental

results such as: Si thickness of 500 µm and rGO thickness of 10 nm with transparency of

80.92%, which were about 2 orders of magnitude higher than our experimental value. This

discrepancy may be due to the partial reduction of GO, which hinders its intrinsic property. A

maximum efficiency of 6.74% was achieved, while considering a single-layer graphene.

Figure 4.11 (a) represents the J-V curve and (b) shows the variation of the efficiency with the

graphene thickness.

Figure 4.11: (a) J-V characteristic plot of rGO/p-Si heterojunction and (b) Variation of the

simulation efficiency vs. rGO thickness.

46

Moreover, the effect of temperature on the performance of rGO/p-Si heterojunction

solar cell has been tested with an aim to study its stability. Figure 4.12 (a) represents the J-V

curve and (b) shows the variation of the efficiency with the temperature. As it can be clearly

understood from figure 4.12 (b), the stability of the device needs to be improved as there is

sudden fall in efficiency with increasing temperature from room temperature to 350 K.

Figure 4.12: (a) J-V characteristic plot of rGO/p-Si heterojunction and (b) Variation of the

simulation efficiency vs. temperature.

4. 6 Summary

To conclude, we have demonstrated the fabrication of a solid state heterojunction with

chemically-derived graphene-on-Si configuration. High transparency in the visible and near

infrared light was obtained by coating the rGO layer, which allowed light to reach the Si

interface. In the fabricated device, a built-in electric potential was created at the junction, by

which photoexcited electrons and holes were transported and collected to cause a

photovoltaic action. The simple fabrication technique of the rGO/Si heterojunction device can

be exploited for other applications replacing high cost fabrication techniques. The fabricated

heterojunction device showed an efficiency of 0.02%, which can be enhanced by optimizing

our device structure. The computational analysis predicted efficiency of 1.32% using the

derived experimental value. According to the computational study, a better and efficient

device (efficiency of 6.74%) can be achieved using single-layer and high-quality rGO. This

unique combination of experimental and simulation study opens up new direction for

graphene-based low-cost photovoltaic cells.