A morphological and electronic study of ultrathin rear passivated

: The effects of introducing a passivation layer at the rear of ultrathin Copper Indium Gallium di-Selenide Cu(In,Ga)Se 2 (CIGS) solar cells is studied. Point contact structures have been created on 25 nm Al 2 O 3 layer using e-beam lithography. Reference solar cells with ultrathin CIGS layers provide devices with average values of light to power conversion efficiency of 8.1 % while for passivated cells values reached 9.5 %. Electronic properties of passivated cells have been studied before, but the influence of growing the CIGS on Al 2 O 3 with point contacts was still unknown from a structural and morphological point of view. Scanning Electron Microscopy, X-ray Diffraction and Raman spectroscopy measurements were performed. These measurements revealed no significant morphological or structural differences in the CIGS layer for the passivated samples compared with reference samples. These results are in agreement with the similar values of carrier density (~8x1016 cm-3) and depletion region (~160 nm) extracted using electrical measurements. A detailed comparison between both sample types in terms of current-voltage, external quantum efficiency and photoluminescence measurements show very different optoelectronic behaviour which is indicative of a successful passivation. SCAPS simulations are done to explain the observed results in view of passivation of the rear interface.


Introduction
Thin film solar cells based on CIGS absorber layers have recenrtly recorded a light to power conversion efficiency of 22.9% [1].CIGS has thereby been increasing its great potential to be used as a promising absorber material for thin film solar cells.CIGS is a direct band-gap material having light absorption properties far larger than indirect bandgap materials such as Si.Hence, Si absorbers need to be sufficiently thick (~300 μm) to absorb incoming photons.To reduce this thickness value, for economic reasons, advanced Si solar cell designs include a combination of an adequate rear surface passivation with micron-sized local point contacts and a rear reflector [2].Application of this technology improves significantly the rear surface passivation and rear internal reflection, while enabling the use of thinner wafers from more than 300 µm to less than 200 µm [3].For p-type Si solar cells, aluminium

A C C E P T E D
M A N U S C R I P T 2 oxide (Al 2 O 3 ), silicon oxide (SiO 2 ) and silicon nitride (SiN x ) are the typical passivation layers used [4,5].Industrial Laser technology is implemented in Si solar cells for the creation of the micron-sized openings on passivation layers [6][7][8].We note that these concepts are still very incipient in CIGS technology.Techniques like formation and subsequent removal of spherical nano-particles in chemical bath deposition of CdS [3,9] and e-beam lithography [10][11][12] have been implemented in CIGS solar cells for the creation of nano-contact openings on rear passivation layers.Generally, high efficiency CIGS solar cells have absorber thickness ranging from 1.5 µm to 3 µm, and there are significant materials savings that can be achieved if this value is reduced to values lower than 500 nm.However, lowering of the absorber thickness causes degradation of solar cell performance and this has been systematically and experimentally studied by different groups since it is a topic of industrial and scientific relevance [3,[9][10][11][12][13][14][15][16][17][18][19][20][21][22].There are two main problems that occur when the absorber thickness is greatly reduced: (i) increase in the rear interface recombination and (ii) optical absorption losses.To compensate for the electrical degrading effects, local nano-sized point contacts are generated on the rear surface passivation layer [3,[9][10][11][12]16,18,21,23,24].In this way, rear contact recombination on the Mo-CIGS interface is significantly reduced due to sufficient coverage of the interface with the passivation layer, while the nano-contacts allow for charge carrier flow.Assuming short minority carrier lifetimes and thus short diffusion lengths, L n -between 0.75 µm and 1.5 µm [3,9,25,26], for thin film CIGS solar cells the diameter of the contact openings are kept approximately to 200 nm while the pitch (the distance between two consecutive nano-contacts) is kept to 2 µm [3,9,11,12].The passivation layer effectively lowers the rear interface recombination by lowering the number of interface defects (chemical passivation) and by the creation of a built-in electrical field (field-effect passivation) that repels minority carriers [27][28][29].With regards to optical losses, several approaches are under study [18,21,23,24].
In this work, we provide detailed description of how the nanopatterning of the passivation layer is accomplished, but most importantly we focus on the effects of the incorporation of the passivation layer in the structural and morphological properties of the CIGS layer itself by X-ray diffraction (XRD) and Raman spectroscopy measurements.Additionally we also present the results of capacitance-voltage (C-V), external quantum efficiency (EQE) and current-voltage (J-V) measurements.Along with the experimental results, the rear passivation effects are simulated by SCAPS with a special emphasis on the influence of the effects of the rear interface recombination velocity in the cell performance.

Experimental Details
The solar cell device structure with rear surface passivated Al 2 O 3 layer and ultrathin CIGS absorber (500 nm) is built up in the following way: SLG/Mo/Al 2 O 3 /CIGS/CdS/i:ZnO/ZnO:Al with Ni/Al/Ni front contacts.With the exception of the passivation layers and of the CIGS, all of the layers were produced according to the Ångstrom Base line and more information can be found in [3,[9][10][11][12]21].In addition, for the passivated cells a precursor Sodium fluoride (NaF) layer was evaporated just before the CIGS deposition.The NaF layer is not an addition to the final device structure as it used up in the CIGS during its deposition.The use and benefits of NaF is explained in the section 3.2 and also in the references [10][11][12][30][31][32][33].The deposition sequence of different layers (upto the CIGS absorber) for both reference and passivated samples is illustrated in Fig.These ungraded (flat evaporation rate) CIGS absorbers are advantageous as they have high reproducibility and thereby assist the investigation of the rear surface passivation qualities.Each substrate has 12 cells and after scribing, each solar cell has an area of 0.5 cm 2 .The presented solar cell results will be average of these 12 cells.
For the structural analysis of the CIGS in reference and rear passivated samples, x-ray diffraction (XRD) analysis have been done in Bragg-Brentano θ-2θ configuration with a PanAnalytical X-Pert PRO MRD system with a CuK line of 1.5406 Å.For the analysis of surface modification of CIGS on the reference and passivated samples Raman spectroscopy was performed.A Confocal Raman Microscope 300 R (WiTec) with green laser excitation wavelength of 532 nm, 1 mW of laser power and a Zeiss objective of 100x in the backscattering configuration was used.Light J-V at AM1.5 and EQE measurements were performed in home-built systems.Scanning electron microscopy (SEM) images were taken in high resolution by a FEI Nova Nano 650 system using acceleration voltages of 3 kV and 5 kV.Capacitance-voltage (C-V) measurements were done in dark in Agilent E4980A LCR system with bias voltage ranging from -0.5 V to 0.5 V at 10 kHz bias frequency and amplitude of input signal, V RMS = 20 mV.Photoluminescence measurements were performed in a Bruker Vertex 80 V spectrometer, equipped with an InGaAs detector.The excitation was done at the wavelength of 457.9 nm (spot diameter of ≈1 mm), with a laser power measured at the front of the cryostat window.Simulations of the solar cell were done using SCAPS 3.3 [3,11,[34][35][36][37][38] to better understand the dissimilarities between reference and passivated solar cells with ultrathin CIGS absorbers.

Al 2 O 3 passivation Layer with Nano-contacts created by e-beam
The application of e-beam lithography process generated the elliptically shaped nano-sized dots (point contacts) on the Al 2 O 3 (25 nm) passivation layer.The top view of the wellordered pattern of the nano point contacts are shown in Fig. 2a.The pitch and the nanocontact dimensions are shown in Fig. 2b.The images were taken post etch and photoresist stripping.The shape of the dot follows the symmetry of the e-beam during the exposure.
The imperfection of the electron optics, in particular the astigmatism, leads to a beam elongated in a privileged direction.As a consequence, the imprinted dots are elliptic.The equivalent diameter is used as an estimation of the size of the dots, being obtained from the quadratic mean of the long axis and the short axis of the ellipses.The measurements were first taken from SEM images of a calibration sample after development with MIBK.T he ebeam process could produce openings with a much better circular aspect at the expense of exposure time.As each 5x5 cm sample already takes approximately 12 hours, the use of the non-perfect circles is a compromise between exposure time and shape of the contact dot.Also visible in the picture of Fig. 1b is the normal morphology of Mo films, even with the Al 2 O 3 layer deposited on top.Such fact means that the 25 nm Al 2 O 3 are deposited in a conformal way minimizing micro/nano shunts.X-ray diffraction analysis followed the SEM analysis to better understand the structural changes (if any) for the CIGS on reference and passivated samples.The used Brag-Brentano configuration is mostly a bulk measurement technique.In Fig. 4 (a & b) the XRD diffractograms are presented for the reference and passivated samples.It is seen in both diffractograms that the main CIGS XRD peak is present at 26.7 0 corresponding to the crystalline family planes (112).Other peaks are found for the crystalline planes (204) at 44.3 0 and (312/116) at 52.2 0 [39,40].Prominent peaks are also found for Mo at 40.5 0 and for ZnO at 34.3 0 corresponding to the crystalline planes (110) and (111) respectively.It is noted for both the sample types that CdS peaks do not appear in the XRD diffractograms, as expected, as the layer is too thin to be probed by this analysis and it might even be in an amorphous state [41].Also due to its very low thickness, the Al 2 O 3 layer in the passivated sample does not result in any diffraction peaks.Overall analysis reveals that there are no significant changes in the CIGS due to the presence of the Al 2 O 3 rear passivation layer.The intensity ratio values estimated for the Al 2 O 3 rear passivated samples shows similar values to the ones for the reference samples (see TABLE I), indicating that Al 2 O 3 has not changed the structural properties of the CIGS layer.Both ratios involving show higher values for our CIGS films than for the powders, the powder diffraction data are for samples with infinite thickness.For our thin samples, the 112 peak, due to its glancing angle will have a larger diffracted volume than the (204) and (312) peaks.So from the XRD point of view we can state that the CIGS in reference and passivated samples are similar.
To understand further if the deposition of the Al 2 O 3 rear passivation layer causes any surface modifications to the CIGS or not, Raman spectroscopy was performed on the reference and the passivated samples.The solar cells were HCl etched, so that the layers above the CIGS layer are removed and the CIGS layer is exposed.Previous literature studies have shown that the HCl etch does not affect the CIGS layer [29,42].So the Raman spectroscopy will provide information only about the structural changes to the CIGS layer down to a depth of 100 nm, which is very useful for ultra-thin devices with 500 nm absorber layer [43].Representative curves for the Raman spectroscopy analysis are shown in the Fig. 5.Both samples reveal the CIGS A 1 mode peak located at 175.1 cm -1 [29,[44][45][46].This figure also shows the peaks fitting which allows to extract the spectra parameters.For the most prominent peak, A 1 mode, the fitting results show similar results with FWHM close to 6 cm -1 for both reference and passivated samples.Hence, we can say from the Raman analysis results that the CIGS on the Al 2 O 3 passivated sample does not undergo any significant surface structural modifications in the probed region when compared with the reference sample.

A C C E P T E D
M A N U S C R I P T

Implementation of Al 2 O 3 Passivation Layer into Ultrathin Cu(In,Ga)Se 2 Solar Cells
The J-V analysis is summarized in Table II and representative J-V and EQE curves are shown in Fig. 6.An analysis of the implementation of the Al 2 O 3 rear passivation layer with the nano-contacts into ultrathin CIGS solar cells reveals an increase in open circuit voltage (V oc ) and short circuit current density (J sc ) compared with the reference device.The improvements in the V oc and J sc could be due to the modifications in the rear surface passivation and likely also due to an increase in internal reflection.It is also noticed that there is a decrease in the fill factor (FF) for the rear passivated cells (64.7 %) compared with the reference value of 66.5 %.Nevertheless, the values of efficiency are higher for the rear passivated cells due to the combined improvements in V oc and J sc .Both solar cells have a CIGS absorber layer thickness of 500 nm.Comparison of the J-V and EQE for the reference

A C C E P T E D M A N U S C R I P T
10 and passivated cells is shown in Fig. 6 (a & b).The comparison shows that: (1) there is an increase in V oc (+29 mV) due to the significant improvement in the rear passivation and (2) an increase in J sc (+2.3 mA/cm 2 ) due to the improvement of the rear internal reflection and/or passivation.This can also be seen in Fig. 6b, where we clearly see an increase in EQE for all wavelengths above 450 nm.The dark J-V plots for the reference and passivated shows no shunting while the light J-V plots shows shunting.So, the shunting effect is entirely light dependent.There is a cross-over effect on both reference and passivated samples, hence, this is not an effect of the introduction of the passivation layer.The cross -over effect on J-V curves is a continuous object of study in the literature and it is not yet fully understood.One possible explanation is that in the dark a large concentration of acceptors near the CIGS-CdS junction act as barrier.When illuminated these acceptors are fulfilled by the photogenerated carriers and the barrier is mitigated [47,48].The cross-over is present for both reference and passivated samples, thus, the space charge region might be the same for both and it is in agreement with our results.A low fill factor (-4 % (abs.)), is noted for the passivated sample which could be due to the fact that ultra-thin devices have several recombination patterns.This should be further investigated.Similar changes in fill factor for passivated samples can be found in [17] and for a detailed electronic analysis [49] can be looked upon.Also, from the plot of Fig. 6c we see a higher difference of EQE between passivated and reference samples.This fact suggests the improved J sc obtained for the passivated solar cells.The difference of EQE in the infrared region becomes lower which suggests that the poor response in the infrared region for both sample types is most probably due the rear collection issues and also due to the fact that ultra -thin CIGS absorbers suffers from incomplete photon absorption [12,13,18,19,[50][51][52].  C-V measurements were done for the reference and Al 2 O 3 rear passivated solar cells to obtain the width of the depletion region, ω (nm) at the CIGS-CdS interface and as well the net acceptors concentration, N CV (cm -3 ) in the depletion region.These values are important to understand if the assumption that the CIGS doping and that the CIGS/CdS interface is different between both samples or not.ω and N CV are extracted at 0 V without the effect of any external bias as commonly done [27,29,53,54], the resulting values are presented in

A C C E P T E D M A N U S C R I P T
Table III and representative curves are shown in Fig. 7.By analysing the results we see no significant changes in the electrical properties for the reference and passivated solar cells.The probed properties by this technique are strongly affected by the CIGS doping levels and by the CIGS/CdS interface.With regards to the CdS interface, as both samples were manufactured with the same conditions, a modification to the rear contact is very unlikely to be expected.However, due to the diffusion-blocking properties of the Al 2 O 3 layer, a difference in sodium (Na) concentration in both samples might occur.Nonetheless, as the results from the C-V measurements are very similar, we can say that the Na concentration in the CIGS layer, that ultimately drives the doping levels, should be very similar in both samples.This is of particular importance as in the reference cell, Na is introduced via indiffusion from the glass substrate, whereas for the passivated cell, a Sodium fluoride (NaF) precursor layer is used.Previous studies have shown that the CIGS has a threshold value of Na incorporation for what its properties are stable [30][31][32][33], which seems to be the point at what both samples are present.Hence, we can pinpoint even more that the modification in the J-V behaviour of both samples are very likely due to modifications of the rear passivation.
Fig. 8 shows the photoluminescence (PL) behaviour of both samples at 6 K with a laser excitation intensity of ~3 mW.For quantitative results, the PL analysis would need to be made with several dependencies like excitation power, temperature and wavelength.However, in this work we are using PL only to compare the samples qualitatively.We note that the luminescence in both samples have the same asymmetrical shape, common to highly doped semiconductor materials and also the passivated sample has a blue shift in its emission compared with the reference sample.These two facts are indications that the major recombination mechanism that dominates this emission is the same but that some passivation of optically active defects occurred in the passivated sample that allowed a blueshift of the emission [55][56][57].These results are in very good agreement with the already presented ones showing that the CIGS properties are likely the same and that the rear interface recombination is being supressed allowing for a blue shift of the emission translated in device terms into a higher V oc value.

Modelling of Ultrathin Cu(In,Ga)Se 2 Solar Cells with Effects of Nano-patterned Al 2 O 3 Rear Passivation Layer
SCAPS simulations [37,38] based on the model presented in the literatures [3,11,35,36] allow us to evaluate the degrading effects of lowering the thickness of the CIGS absorbers .The solar cell parameter results for CIGS solar cell with standard absorber thickness (1.5 µm) and with ultrathin CIGS (500 nm) are compared and shown in Table IV.The surface recombination velocity (S b ) is simulated in SCAPS for CIGS solar cells with 500 nm absorber thickness.As discussed above, the deposition of an Al 2 O 3 rear passivation layer can significantly reduce S b .We assumed as reference a value of 10 7 cm/s for the rear interface recombination velocity (S b ) [3,11,35].The simulation results indicate that the approach of Al 2 O 3 rear passivation layer for 500 nm CIGS solar cells has the potential to considerably increase the V oc and thereby the efficiency.In a first instance, the main effect of changing the recombination velocity is to increase the V oc , however after a significant decrease of S b , J sc starts to increase as well.The best solar cell (with NaF pre-treatment) with the nanopatterned Al 2 O 3 rear passivation has V oc 611 mV, J sc 24.5 mA/cm 2 , FF 65.3 % and efficiency of 9.8%.Without passivation, the V oc comes down to 582 mV, J sc to 22.2 mA/cm 2 and the

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T 14 efficiency to 8.9%.These experimental values provide for a difference in V oc of 29 mV and of J sc 2.3 mA/cm 2 after applying the passivation of the rear.With SCAPS simulations, we see that the V oc can increase in this order of magnitude if S b is expected to lower from 10 7 to values around 10 2 -10 4 cm/s.In the simulations we have obtained very similar values for the figures of merit in comparison to the experimentally obtained results.Only the fill factor showed higher values and also thereby the efficiency since we are very far from the actual experimental conditions.This difference between the experimental and simulated values are expected, due to the fact that the model used in the simulation does not include all the loss mechanisms which are present in the real device.Nonetheless, it is possible to identify that there is still margin to progress with the quality of the passivation since a higher increase of the V oc and of the J sc is theoretically possible.

Conclusion and Outlook
From our careful analysis we summarize that the nano-patterned Al 2 O 3 rear-surface passivation layer created by e-beam lithography technology can be implemented for performance enhancement of CIGS solar cells and that its introduction does not change the CIGS crystalline properties.Ultrathin CIGS absorber layers with the rear passivation show an improvement in all the solar cell parameters with respect to unpassivated reference cells and without undergoing any significant structural or morphological changes.Additionally, the cells with ultrathin CIGS absorbers have been modelled with SCAPS to understand the effects of rear passivation.SCAPS modelling results indicates a performance degradation with lowering the absorber thickness to 500 nm, but implementation of the rear passivation effects by reducing the charge carrier recombination velocity an improvement in performance is noticed.As no significant structural and morphological changes are noticed in the CIGS absorber, nano-patterns of different dimensions along with different thickness of the passivation layer can be tested in the future.

5 Fig. 2 .
Fig. 2. SEM images of nano-contacts on the Al 2 O 3 rear passivation layer: ; (a) shows the well-ordered pattern of the nano-contacts; (b) shows the pitch, the consecutive distance between two nano-contacts and the approximate longer and shorter dimensions of the elliptic nano-contacts are shown

Fig. 4 .
Fig. 4. (a) and (b) shows the XRD diffractograms of reference and passivated solar cells , respectively.The relative intensity peaks of CIGS related peaks for both the reference and the passivated solar cells remains unchanged.

Fig. 5 .
Fig. 5. Representative Raman spectra for reference and passivated samples along with fitting results .

Fig. 6 .
Fig. 6.(a) J-V and (b) external quantum efficiency (EQE) plots for reference and Al 2 O 3 rear passivated CIGS solar cells.(c) Absolute difference of EQE (in %) between passivated and reference sola r cells.CIGS absorber layer thickness for both types of cells is approximately 500 nm.

Fig. 7 .Fig. 8 .
Fig. 7. Variation of the charge carrier concentration along the width of the depletion region for the (a) reference and (b) passivated solar cells.The red spots in the graphs shows the values for ω and N extracted at 0 V bias voltage.
U S C R I P T

TABLE I
RATIO OF XRD INTENSITIES FOR BOTH REFERENVE AND PASSIVATED SAMPLES AND FOR A POWDER SAMPLE

TABLE III AVERAGE
AND STANDARD DEVIATION VALUES OF ω AND M EXTRACTE D BY C-V MEASUREMENTS FOR THE REFERENCE AND FOR THE Al 2 O 3 REAR PASSIVATED CELLS

TABLE IV SOLAR
CELL PARAMETER RESULTS FROM SCAPS SHOWING THE EFFECT OF LOWERING THE THICKNESS OF THE ABSORBER AND OF CHANGING THE RECOMBINATION VELOCITY (S b )