Pulsed Laser Annealing of Ag-Paste on n-Doped Emitter

Pulsed laser sources are attractive on account of their spatial and temporal controllability at room temperature. Pulsed lasers, in visible (VIS) (300 – 515 nm) and infrared (IR) (900 – 1064 nm) spectral ranges, with pulse widths in micro to femtoseconds range, are used in a wide range of applications including doping, etching, texturing and deposition. In this study, an Nd-YAG dicing laser operating at 1064 nm wavelength with 200 nanosecond pulse duration has been employed to form silver ohmic contacts to an n-type emitter on a p-type silicon substrate. The laser beam was used to anneal screenprinted Ag polymer paste over a broad (~ 7 to 500 mJ/cm2) range of laser fluences. Computer numerical control software allowed fabrication of geometrical patterns with controllable diameters in 50-150-μm range. Contact resistance measurements were performed using the transmission line method (TLM). Contact resistivity exhibited fast decay from very large values to relatively constant as a function of laser fluence. This variation was attributed to laser energy below the threshold energy which no alloyed Ag/Si contact could be formed. The lowest contact resistivity at 200 mΩ.cm2 was measured at 35 mJ/cm2. This value was two orders of magnitude higher than the lowest value for thermally annealed contacts. For the laser parameters investigated here, optimum laser fluences were in 0.2-0.6 J/cm2 range. It may be possible to attain lower resistivity values trough post-laser annealing.


INTRODUCTION
Ohmic electrical contact formation in industrially produced n-doped, p-type crystalline silicon (Si) solar cells is based on simultaneous high temperature rapid thermal annealing of front surface silver (Ag) and rear surface aluminum (Al) metallic pastes (Soin & Majlis 2006). Screen printing of Ag and Al pastes is attractive due to its lower cost and high throughput in comparison with alternate expensive and slow techniques such as photolithography (Sepeai et al. 2011). The annealing process is carried out in conveyer belt type furnaces at temperatures up to 870 o C. For thinner wafers, particularly at high temperatures, (below 180 µm), thermal expansion mismatch between metal and Si results in excessive wafer bowing and stress that lower process yield (Popovich et al. 2013). It is therefore, desirable to develop alternate, ohmic metal/Si contact formation processes at either room temperature or lower (< 200 o C) temperatures (Sakib & Ahad 2018).
Pulsed lasers have been used in materials processing for over 50 years (Sugioka et al. 2010) due to their unique controllability in spatial and temporal domains. Pulsed lasers have also been applied in solar cell processing including doping, deposition, etching, and annealing. Most laser work is focused on formation of Al film or paste to p-type Si wafers (Wang et al. 2013). Relatively little attention has been focused on formation of Ag contact to n-doped Si substrates. In this work, pulsed laser annealing of polymer Ag paste to n-doped Si wafers has been investigated as an alternative to expensive and high temperature conventional furnace-based rapid thermal annealing processes.

OHMIC CONTACT
An ideal ohmic contact is defined as a metal -semiconductor contact with negligible resistance relative to the bulk or spreading resistance of the semiconductor (Vinod 2011). It can be formed with metals exhibiting smaller work function than the semiconductor. In general, Ag does not form an ohmic contact as the Ag work function, eφ Ag (~ 4.4 eV) is higher than Si electron affinity, χ si (~ 4.4 eV). However, with a highly-doped emitter (N D ≥ 10 18 atoms/cm 3 ) region in physical contact with metal, an ohmic contact can be formed. This is due two facts: (a) lowering of the barrier height that allows thermionic emission current and (b) presence of thin (≤ 100 A) barrier layer for current flow direct tunneling current. In real devices, the formation of ohmic contacts requires; (a) highly doped surface layer to enhance tunneling, (b) appropriate combination of φ m and χ i to reduce φ b , and (c) defect states within the tunneling barrier region (Fahrenbruch & Bube 1983).

PULSED LASER SySTEM
For all the work reported, pulsed laser optical system based on a Q-switched Nd:yAG laser was employed. This pulsed laser was part of yMS-50D dicing system for Si wafers and solar cells. This laser operated at 1064 nm wavelength with pulse duration of 200 nanoseconds, modulating frequency of 50 kHz, and minimum spot size ≤ 0.05 mm. The laser optical system is integrated with CNC system consisting of computercontrolled xy-wafer stages operating at maximum speed of ~ 120 mm/s to allow formation of any desired pattern; laser system specifications have been summarized in Table 1.

LASER-MATERIAL INTERACTIONS
Pulsed lasers find wide range of material processing applications on account of their ability to spatially and temporally resolve and control material etching or deposition rate (Sugioka et al. 2010). These inherent advantages allowed pulsed lasers to become a manufacturing tool in photovoltaic (PV). Based on its wavelength, laser radiation is either fully or partially absorbed in the material. Laser-material interactions will lead either to local heating through electronic excitation or by electron-phonon coupling.
The physical response of a material under laser incidence is a function its optical absorption as described in Figure 1. From the plotted data of spectral absorption as a function of wavelength for metals and semiconductors, it is noted that for all metals the absorption depths are short and independent of wavelength. For semiconductors, absorption depths are similar to metals at short wavelengths (< 500 nm). At longer wavelengths, especially near the bandgap, absorption lengths increase by several orders of magnitude. Therefore, based on the choice of laser used especially for Si, absorption depths can vary from few nanometers to hundreds of micrometers. When a pulsed laser, with energy higher than material ablation threshold is incident on a material surface, it initiates melting, vaporization, and recrystallization in the interaction region ( Figure 2). This material removal also caused the dense vapor plume ejected that may contain solid and liquid clusters of material from the irradiated zone. Ablation threshold is the specific characteristic of the materials that usually depends on their melting and boiling points as well as the absorption coefficient at a particular wavelength. For crystalline Si, the absorption depth at 1064 nm laser radiation is over 100 µm it is also the laser employed in work reported here. The measurement of contact resistivity is generally obtained by transmission line method (TLM). In TLM, the total resistance, R T between two neighboring contact pads is measured as a function of increasing separation between them and plotted as a function of d. For the work reported here, a large TLM mask of pad dimensions 4 × 10 mm 2 with separation, d, varying started from 0.1 mm to 0.7 mm at an increment of 0.1 mm as seen in Figure 3 were used.

METHODOLOGy
In earlier work (Ahmad et al. 2017) detailed analysis of temperature variation on metallization quality has been reported. The rapid thermal annealing experiments were performed in conveyor belt furnace in which wafer moves through six temperatures at constant speed. Results from these experiments are for comparing with laser annealed contacts using the same TLM mask. For all the work reported here, polymer silver paste (Ferro LF33-750) was screen-printed on multicrystalline boron doped Si wafer with bulk resistivity of 0.5 -3.0 Ω.cm and thickness of 200 μm, please see Figure 4 for details. Front surface of Si wafers were doped by standard POCl 3 diffusion in a tube furnace to form an n-type emitter surface with a sheet resistance 50 Ω/sq. Screen-printed contacts were dried in an oven for 10 minutes of at 150 o C followed by laser-annealing for three different geometric patterns. Three different diameter configurations consisted of focused spots sizes with diameters of 0.05 mm, 0.10 mm, and 0.15 mm at periodicity of 1 mm. The resistivity Ag/Si contact was measured for all three configurations. Varied for all three cases Figure 5 illustrates TLM metallization pattern used for laser annealing of screen-printed polymer Ag paste. Several wafers were prepared with varying contact sizes while maintaining same range of pulsed energy. Contact size variation was achieved with the assistance of CNC software.

RESULTS AND DISCUSSION
An extensive investigation of rapid thermal a of Ag/Si contacts has been reported elsewhere (Ahmad et al. 2017). Since same TLM mask dimensions were used, it is worthwhile to briefly review thermally annealed contact results. Figure  6 plots resistivity as a function of temperature in 600-920°C temperature range. Resistivity exhibits four orders of magnitude reduction as temperature is increased from ~ 600°C to 900°C. A well-defined minimum is observed at 870°C after which resistivity starts to increase. The lowest contact resistance is less than 1 mΩ·cm 2 for the six-zone conveyer belt IR furnace. Clearly, laser-annealed contact exhibits contact resistivity higher by more than two orders of magnitude.
For this laser, the pulse width is 200 nano second and maximum pulse energy is 10 microjoule (µJ) giving maximum power is 50 W (pulse energy/pulse width). The minimum threshold energy at which contact resistivity could be measured was approximately 1.25 µJ. Due to large laser spot dimension variations, it is important to plot resistivity as a function of laser fluence (J/cm 2 ) by simply dividing pulse energy by the surface area. Figure 7 plots resistivity as a function of surface area for all three contact configurations.  The solid lines in Figure 7 represent curve-fitting to the experimental data. For the smallest focused spot size of 0.05 mm, best fit was achieved with Lorentzian, 4 Parameter equation (f = y 0 + a/(1+((x -x 0 )/b) 2 ). While for both larger diameters (0.1-0.15 mm), best fit was achieved with exponential decay, modified single, 3-parameter equation (f = a·e (b/(x+c)) ).
Comparison of resistivity variation between thermal ( Figure 6) and laser (Figure 7) annealed contacts reveal the following facts:

High resistivity at low temperatures during thermal
annealing is similar to that of laser fluence below threshold value, 2. Thermal annealing contact response is linear whereas laser-based contact shows relatively invariant response once energy increases over the threshold value, and 3. Lowest laser-annealed resistivity at ~ 200 mΩ cm 2 is two orders of magnitude higher than thermally-annealed contact.

EFFECT OF LASER PARAMETERS
The range of laser fluences investigated here result in many effects including ablation and redeposit ion of the liquid metal/Si alloys onto the surface; referred to as liquid pile-up. At the end of the pulsed laser cycle, the re-deposited molten regions recrystallize instantaneously high cooling rates. Therefore, it is inherently difficult to compare thermal and laser annealing processes. Laser-material interactions are fundamentally different than the thermal annealing processes for contact formation. The absorbed laser energy on the Ag/Si interface raises temperature of both the paste and Si substrate. Below threshold energy, no alloyed contact is possible between Ag and Si, hence the very high resistivity values. This behavior is identical for laser and thermal annealing experiments. As laser fluence is increased, laser energy melts Ag/Si interface and penetrates into the Si substrate. The absorbed laser energy is not significantly absorbed at Si surface due to its weak Si absorption at 1064 nm wavelength. At the interface, complex mixture of solid, liquid, and vapor phases is formed that extends well below the interface into the bulk region of the sample (Figure 9). CONCLUSION Pulsed laser annealing of polymer Ag contact on n-type emitter has been investigated using TLM approach. Three laser pattern configurations were investigated with laser fluence varying from ~ 7 to 500 mJ/cm 2 . Lowest contact resistance at 200 mΩ.cm 2 was observed at laser fluence of ~ 35 mJ/cm 2 , which were approximately two orders of magnitude higher Critical features of the resistivity measurements are summarized below.
1. Resistivity decreases as laser fluence increases, 2. Lowest resistivity is observed for the largest pattern and vice versa, 3. Resistivity is highest for lowest laser fluence and decreases rapidly as laser fluence is increased, and 4. Optimal laser fluence for lowest contact resistivity is in 0.2-0.6 J/cm 2 range. than thermally-annealed contacts. High contact resistivities of laser-annealed contacts are attributed to the complex nature of liquid-vapor interface as well as the long absorption length of Nd-yAG laser in Si. Optimum configuration is observed between laser fluences in ~ 0.1-0.3 J/cm 2 range for the largest spot sizes. At larger fluences, loss of doped Si in emitter region is higher due to ablation and vaporization. Contact resistance may be reduced through post laser annealing in hydrogen ambient.