Excimer

Laser Assisted Photoresist Removal Process in Gaseous Media by Response Surface Methodology and Artificial Neural N Investigation of the performance of 248 nm excimer laser assisted
photoresist removal process in gaseous media by response surface methodology and artificial neural network
James Jacoba, b, ⁎

jamesm@barc.gov.in

P. Shanmugavelub

R. Balasubramaniamb

a
Homi Bhabha National Institute (HBNI), Anushakti nagar, Mumbai, 400094, India

b
Chemical Technology Group, Bhabha Atomic Research Centre, Mumbai, 400 085, India

⁎
Corresponding author at: Homi Bhabha National Institute, Anushakthi Nagar, Mumbai, 400094, India.

Abstract

Photoresist removal by Excimer laser in different gaseous media viz., hydrogen, helium, argon, nitrogen and air is studied under different parameters viz., laser pulse energy, pulse repetition frequency (PRF) and gas flow rate. Totally 180 experiments (36

experiments for each gaseous medium) were conducted based on the full factorial design. Statistical analysis of data by 3way ANOVA (ANalysis Of VAriance) was carried out to investigate the main effects on photoresist removal process. Photoresist removal

rate is observed to be significantly higher in case of hydrogen gas medium compared to other gaseous media studied. Hence, another 48 set of experiments (based on full factorial design) was carried out in hydrogen gas medium. These experimental data

were used for modeling the photoresist removal process. Three key operating parameters (i.e. laser energy, pulse repetition rate and hydrogen gas flow rate) were selected as independent variables, and the amount/depth of photoresist removed was

considered as the dependent variable (response). Response Surface Methodology (RSM) and Articial Neural Network (ANN) models were developed based on the experimental design approach to predict etch depth. The predictive capacity of the two models

was compared. Analysis of the results showed that the operating parameters can also inuence each other’s optimum value. Higher photoresist removal rate observed in hydrogen medium may be attributed to possible Laser Assisted Chemical Reaction

(LACR) of hydrogen gas with photoresist polymer. A mechanism has been proposed to explain the interaction of hydrogen molecule with photoresist during the laser irradiance, resulting in generation of hydrogen radicals, which may subsequently undergo

additional bond-breaking process of olefinic functional groups (having π bond), along with laser bond breaking process. To the best of the author’s knowledge, this is the first study providing information of LACR of hydrogen gas with photoresist polymer,

resulting in higher photoresist removal rate. The LACR of hydrogen gas with photoresist polymer, being a cold process, has potential application in semiconductor industry, where researchers are trying to exploit the process of photoresist removal using

hydrogen radical generated at 2000 °C by tungsten hot wire.

Keywords: Excimer laser; Photoresist removal; Gas media; Artificial neural network (ANN); Response Surface Msurface methodology (RSM)

1 INTRODUCTIONntroduction
In semiconductor industries, fabrication of microsystems is accomplished by adopting a series of steps that includes thin-lm deposition, polymer resist patterning and etching. PMMA, which is a positive-tone, and SU-8 which is a negative-tone resist, are widely

employed as film resists for the fabrication. There is great demand from many industries for high throughput manufacturing of MEMS devices using SU-8. Removal of photoresists are often performed by wet cleaning process using chemicals such as sulfuric

acid–hydrogen peroxide mixture and organic amine solutions. These chemicals are expensive and are not environmental friendly. Researchers are making various attempts to develop an alternate environmental friendly polymer resist cleaning process to reduce the cost

and environmental damage. Attempts have been made to develop dry cleaning process using oxygen plasma and ozone gas [1–3]. This method has its inherent demerits due to the oxidation of the substrate, apart from degradation of device properties due to the

bombardment of high-energy ions on the substrate. Oxygen/ozone plasma ashing may also cause oxidation of substrates and metal wiring due to its operating process temperature of higher than 250 °C [4–,5]. To overcome the problems associated with plasma cleaning

process, researchers have studied an alternate resist removal method using hydrogen radicals. The main disadvantage of this process is the requirement of high temperature of the order of 2000 °C to generate hydrogen atoms using hot tungsten filament [6–10].

The present study is focused to explore the possibility of developing a new technique for photoresist removal process without using hazardous chemicals or employing higher temperature. This new method is a dry process by utilizing excimer laser ablation under

room temperature in the presence of hydrogen gas. The authors also studied the effect of different gaseous medium like air, nitrogen, argon and helium on the ablation process. The results obtained with hydrogen gas showed remarkably superior photoresist removal rate

compared to other gases. The authors also put forward a mechanism for photoresist removal by laser in hydrogen gas medium. As a part of this study, we also investigated individual and interaction effects of three key operating parameters (i.e. laser energy, pulse
repetition rate and gas flow rate) on the laser assisted photoresist removal in hydrogen gas medium. Two approaches, Response Surface Methodology (RSM) and Artificial Neural Network (ANN) were employed to model the process and their prediction capability for

photoresist removal was compared. Details of the experiments along with results are presented in this paper.

2 EXPERIMENTAL SETUPxperimental Setup

2.1 Excimer laser system
An Excimer laser ablation system, Model number WS 300 (M/s Optec, Belgium), comprising of KrF laser source, beam delivery unit (consisting of attenuator, homogenizer, photo-mask, projection lenses), CNC controlled X-Y table with vacuum chuck and pulse

energy monitor, was used for this study. The experimental setup is shown in Fig. 1.The specification of system used is given in Table 1. Excimer laser pulse with a wavelength of 248 nm and pulse duration of 25 ns (FWHM) was generated for the study. Pulse repetition

frequency in the range of 10 - –50 Hz was used in the experiments. Mask projection technique of excimer laser micromachining was employed in our study. A circular mask of 20  mm diameter was used to produce the 5 mm diameter plain hole (with demagnification of 4X

for the projection lens) on the polymeric workpiece. A gas handling setup as shown in Fig. 2 was fabricated for admitting gas at a predefined flow rate over the surface of polymer during the photoresist removal process. The set up consisting of a cylindrical chamber, made

of stainless steel 304 grade material, having an effective volume of about 100 ml, was used for photoresist removal study. The chamber has inlet and outlet connections fitted with valves. The chamber is connected to a rotary vane vacuum pump for evacuation through

isolation valves. The evacuation arrangement was used to study the process under vacuum environment. Further, the setup has arrangement for admitting gas at variable flow rate. A calcium fluoride window of 50  mm diameter is fixed over ablation chamber, normal to

laser beam axis.

Fig. 1 Excimer laser experimental setup.

alt-text: Fig. 1

Table 1 Specification of Excimer laser system.

 alt-text: Table 1

Excimer Laser type Laser MediaKrFKrF

Laser Media

Wavelength (nm) 248

Maximum Pulse Energy (mJ) at source 1000

Max. Average Power (W) 50

Maximum repetition rate (Hz) 50

Pulse Duration (FWHM, ns) 25

Spot size on mask plane 30 mm x 30 mm

X-Y Stage Travel & repeatability 250 mm x 150 mm; ± 1 µm

Z-Focus Stage range & repeatability 200 mm; ± 1 µm

Fume Extraction HEPA filter based suction

Fig. 2 Gas handling setup used for Photoresist removal.

alt-text: Fig. 2

2.2 Photoresist coating on metal substrate

A commercially available photoresist polymer E1020 (supplied by M/s Cadmosil Chemical Pvt. Ltd, India) is being used by us, for the manufacture of micro grooved metallic parts using the two stage process of excimer laser lithography of photoresist coated

metallic workpiece followed by chemical micromachining. This polymer was used for our experimental study. Stainless steel cylindrical cups (50  mm dia. × 5 mm depth) were used as metal substrate, and the photoresist was filled and cured. The filling, was then air dried

and subsequently cured at 110 °C for about 1 hour in vacuum at about 10-−2 mbar absolute pressure. This was done to achieve uniform coating thickness of the polymer which is free from entrapped gases in the film. A film thickness of about 5000  µm was used as a

precautionary measure in our experiment in such a way that laser beam does not reach the metal surface. The test samples were placed inside the ablation chamber of the gas handling system and mounted on the X-Y stage of laser system for photoresist study.

3 DESIGN OF EXPERIMENTSesign of experiments

3.1 Preliminary photoresist etch study in gaseous media
The gases used for study include hydrogen, helium, nitrogen, argon and air. These five gases were selected for study based on their thermal conductivity values, as shown in Table 2, covering a range between 17 × 10-−3 W/m.K and 287 × 10-−3 W/m.K [11]. In order

to compare the photoresist removal process under various gaseous media, the photoresist was exposed to 2000 laser shots under various experimental conditions. For each gaseous medium, totally 36 numbers of experiments were carried out under different parameters
as shown below.

• levels for Laser pulse energy: 30, 40, 50, 60 mJ

• 3 levels for Pulse repetition rate: 20, 30, 40 Hz

• 3 levels for Gas flow rate: 0.8, 1, 1.2 lpm.

Table 2 Thermal conductivity of gases [11].

alt-text: Table 2

Gas Thermal conductivity at 300 K (W/mK x 10-−3)

Argon 17.9

Nitrogen 26

Air 26.2

Oxygen 26.6

Helium 156.7

Hydrogen 186.9

As can be seen from Fig. 3, mean mass of photoresist removed under different gaseous media of air, nitrogen, argon and helium did not show significant change in mass of photoresist removed, while that under hydrogen media is quite different. From these

experiments, it was observed that hydrogen gas environment provided remarkably higher material removal rate compared to all other gases studied. Hence, further study was done only in hydrogen environment.

Fig. 3 Comparison of Mean mass (±SD) of photoresist removed in different gaseous media.

alt-text: Fig. 3

3.2 Full factorial design for hydrogen gas medium

The Full Factorial experimental design was employed for the design of the experiments to evaluate the effect of three operating parameters, i.e., gas flow rate (l/min), pulse repetition rate (Hz) and laser energy (mJ). These selected parameters were considered

as the independent variables and the amount of photoresist etched (µm) after 500 laser pulse of irradiation was selected as the dependent variable (response). The experimental ranges were obtained based on previous preliminary experiments. Accordingly, 48 tests were

dened for these three independent variables in four levels for PRF and flow rate and three levels for energy, as shown in below.

• Gas flow rate, A: 0.6, 0.8, 1, 1.2 l/min (4 levels)

 • Pulse repetition rate, B: 10,20,30, 40 Hz (4 levels)

• Laser pulse energy, C: 40, 50, 60 mJ (3 levels)

The response (dependent variable) of photoresist mass removed and in turn the etch depth, for 500 laser shots were calculated for all the experiments. Etch depth is computed by gravimetrically from the relation, Etch depth  = Mass of polymer removed/(area of

laser spot x density of polymer). In all experimental cases, the laser pulse energy was set to a value which is above the threshold value for the polymeric material. The experimental results are shown in Table 3.The data is subjected to statistical analysis using ANOVA

(ANalysis Of VAriance) methodology to establish the effect of gaseous media on the photoresist removal process by Excimer laser. These data sets were then used to develop the RSM and ANN models.

Table 3 Experimental data for hydrogen gas medium along with predicted values by RSM and ANN models.
alt-text: Table 3

Run# Input parameters for experiments Experimental value of Etch Depth (µm) Mean*± SD Predicted etch depth by models

RSM ANN

Flow rate (lpm) Laser pulse repetition (Hz) Laser Energy (mJ) Depth RD % Depth RD%

1. 0.6 10 40 76.6 ± 7.7 81.8 -6.878.32.22−6.8 78.3 2.2

2. 0.6 10 50 98.4 ± 0.4 99.4 -1.097.6-0.83−1.0 97.6 −0.8

3. 0.6 10 60 116.4 ± 2.9 110.7 4.9 117.1 0.6

4. 0.6 20 40 40.3 ± 3.3 41.2 -2.438.8-3.65−2.4 38.8 −3.6

5. 0.6 20 50 51.0 ± 2.5 55.0 -8.049.5-2.86−8.0 49.5 −2.8

6. 0.6 20 60 57.3 ± 1.6 62.6 -9.358.01.27−9.3 58.0 1.2

7. 0.6 30 40 24.5 ± 2.6 21.5 12.1 28.0 14.5

8. 0.6 30 50 35.4 ± 1.5 31.6 10.9 35.8 1.0

9. 0.6 30 60 41.0 ± 2.2 35.3 13.8 38.3 -6.510−6.5

10. 0.6 40 40 23.2 ± 2.2 22.7 2.3 21.7 -6.311−6.3

11. 0.6 40 50 26.8 ± 1.3 29.0 -8.128.66.912−8.1 28.6 6.9

12. 0.6 40 60 28.0 ± 3.3 29.0 -3.628.51.913−3.6 28.5 1.9

13. 0.8 10 40 87.3 ± 4.4 84.2 3.5 83.9 -3.914−3.9

14. 0.8 10 50 101.2 ± 0.7 102.2 -1.0101.70.515−1.0 101.7 0.5

15. 0.8 10 60 118.7 ± 2.6 114.0 4.0 118.5 -0.116−0.1

16. 0.8 20 40 42.8 ± 3.3 42.5 0.6 42.5 -0.617−0.6

17. 0.8 20 50 57.1 ± 3.1 56.8 0.6 52.8 -7.518−7.5

18. 0.8 20 60 55.6 ± 1.3 64.8 -16.558.55.219−16.5 58.5 5.2

19. 0.8 30 40 26.8 ± 1.1 21.7 19.1 29.3 9.4

20. 0.8 30 50 35.4 ± 0.4 32.2 9.1 37.7 6.5

21. 0.8 30 60 38.7 ± 1.1 36.4 5.7 38.6 -0.322−0.3

22. 0.8 40 40 21.8 ± 3.4 21.8 0.3 20.6 -5.723−5.7

23. 0.8 40 50 26.4 ± 1.8 28.5 -8.128.99.424−8.1 28.9 9.4

 24. 0.8 40 60 25.6 ± 2.7 29.0 -−13.0 29.2 13.9

25. 1 10 40 86.5 ± 5.8 86.3 0.3 88.2 1.9

26. 1 10 50 105.9 ± 1.6 104.8 1.0 105.8 -0.127−0.1

27. 1 10 60 122.7 ± 0.4 117.1 4.6 119.9 -2.328−2.3

28. 1 20 40 39.7 ± 5.4 43.5 -9.643.710.029−9.6 43.7 10.0

29. 1 20 50 54.8 ± 1.7 58.3 -6.454.90.230−6.4 54.9 0.2

30. 1 20 60 59.3 ± 1.4 66.8 -12.659.2-0.231−12.6 59.2 −0.2

31. 1 30 40 26.9 ± 3.1 21.6 19.8 27.8 3.4

32. 1 30 50 36.3 ± 1.3 32.6 10.3 37.9 4.5

33. 1 30 60 39.5 ± 1.9 37.3 5.7 39.6 0.2

34. 1 40 40 13.4 ± 0.8 20.6 -−53.1 17.6 31.4

35. 1 40 50 28.8 ± 2.0 27.8 3.7 27.9 -3.036−3.0

36. 1 40 60 28.0 ± 2.0 28.7 -2.730.79.537−2.7 30.7 9.5

37. 1.2 10 40 88.4 ± 2.3 88.2 0.3 89.7 1.5

38. 1.2 10 50 108.8 ± 2.3 107.2 1.5 108.7 -0.139−0.1

39. 1.2 10 60 120.1 ± 0.5 119.9 0.1 121.7 1.4

40. 1.2 20 40 47.4 ± 4.0 44.3 6.7 42.2 -11.041−11.0

41. 1.2 20 50 52.0 ± 0.5 59.5 -14.555.56.742−14.5 55.5 6.7

42. 1.2 20 60 60.9 ± 3.8 68.5 -12.461.00.143−12.4 61.0 0.1

43. 1.2 30 40 21.1 ± 0.4 21.3 -0.625.018.344−0.6 25.0 18.3

44. 1.2 30 50 38.1 ± 0.4 32.7 14.2 37.3 -2.145−2.1

45. 1.2 30 60 43.9 ± 3.4 37.9 13.7 41.5 -5.446−5.4

46. 1.2 40 40 15.3 ± 0.3 19.1 -24.614.8-3.047−24.6 14.8 −3.0

47. 1.2 40 50 28.0 ± 1.8 26.8 4.2 27.0 -3.648−3.6

48. 1.2 40 60 30.5 ± 3.0 28.2 7.4 32.4 6.2

*
The values shown here is an average value of triplicate experiments conducted for 500 laser shots.

4 MODELING METHODS4.1Response Surface Modeling Methods

4.1 Response surface methodology
A second-order polynomial equation as shown below was used to model the mass removed/etch depth of photoresist, as a function of the three independent variables mentioned above.

Y=β0+β1A+β2B+β3C+β4AB + β5AC+β6BC+β7A2+β8B2+β9C2

Where,

• Y is the etch depth produced under hydrogen gas medium (response)

 • β0represents the intercept,

• β1, β2andβ3are the coefficients of the independent variables,

• β7, β8 and β9are the quadratic terms, and

• β4, β5, β6, are the interaction coefficients.

• A, B and C are gas flow rate, pulse repetition rate and laser energy respectively.

Moreover, the analysis of variance (ANOVA) and optimization of the photoresist removal process were also performed.

4.2 Artificial Neural Nneural network

For ANN, a three-layer feed-forward neural network was developed and trained by back-propagation gradient-descendent algorithm. The experimental data were randomly classi ed into two sets containing 32 and 16 data to be employed for training and

validation respectively. Training data sets were used for updating weights and biases via Levenberg–Marquardt algorithm and the test data were utilized to evaluate the generalization ability of the trained network. A hyperbolic tangent sigmoid function and two linear

functions were used as transfer functions for the neurons in the hidden layer and the neurons in the input and output layers respectively. The input and output layers had 3 and 1 neurons respectively. To determine the optimal number of neurons in the hidden layer,

different topologies were examined during which the number of neurons varied between 3 and 15. Based on the values of R 2 and RMSE, 5 number of neurons were observed to be providing satisfactory results. Accordingly, ANN model for hydrogen gas medium was

developed with 3 inputs- 5 nodes mono hidden layer-1 output concept, as shown in Fig.4.5RESULTS AND DISCUSSIONS 4.

Fig. 4 Artificial Neural Network training structure.

alt-text: Fig. 4

5 Results and discussions

Photoresist materials are hydrocarbon polymers with phenyl-, carboxy, keto, olefenic functional groups. The interaction of laser photon with photoresists differ with the type of functional groups that are present. The present study is focused to understand the

interaction of laser photon with photoresist, in the presence of inert (Helium, Argon, Nitrogen), oxidizing (air) and reducing (hydrogen) gaseous environment. Photoresist removal rate by excimer laser beam is associated with both photochemical (cold ablation by bond

breaking) and photo-thermal mechanisms [12]. The material removal process is characterized by pulse-by-pulse removal of small amount of material at submicron depth, with minimal damage to the surrounding areas. In photo-thermal process, the deposited energy is

converted to heat, leading to thermal decomposition [12]. It has been generally accepted that material removal proceeds by a combination of these two mechanisms. In general, a photon of short-wavelength excimer laser, has energy ranging from 4 to 10 eV, which covers

the bond energy of most of the organic polymeric materials as shown in Table 4. KrF excimer laser having a photon energy of 5.2 eV, is sufficient to break most of the covalent bonds of organic polymeric materials. Even though the photochemical action dominates over

the photothermal process, the photothermal process still has a significant role in the photoresist removal process, which may have detrimental effects on the ablated surface characteristics. Being an inert and good thermal conducting medium, helium gas is expected to

minimize the photothermal contribution part in the net photoresist removal rate. Among other gases, hydrogen gas is of special interest due to its higher thermal conductivity and reduction property. So, it is expected to play dual role, as coolant and reducing agent under

appropriate conditions. It is expected to minimize the photothermal contribution part in the net photoresist removal process, due to its higher thermal conductivity. It is also expected to undergo chemical reduction reaction with unsaturated/olefinic/phenyl functional group of

polymer, provided hydrogen radical is generated due to laser irradiation. In any case, its influence during the photoresist removal process using excimer laser will be unique.

Table 4 Bond dissociation energy [12].

alt-text: Table 4

Type of Covalent Bond Bondingenergy(eV)

O - O1.47N- N1.69C–Br2.95C-N3.16C–Cl3.43C–C3.60H- N4.03C–H4.25N==N4.33H–H4.52O- H4.81C–F5.08O = O5.15N≡O 1.47

N N 1.69
 C Br 2.95

C N 3.16

C Cl 3.43

C C 3.60

H N 4.03

C H 4.25

N N 4.33

H H 4.52

O H 4.81

C F 5.08

O O 5.15

N N 9.79

5.1 Effect of gaseous media

Mean mass of photoresist removed under different gaseous media, as shown in Fig. 3, provide information that air, nitrogen, argon and helium did not show significant change in mass of photoresist removed. The standard deviation of the mean value for these

gases are more or less same. On the contrary, hydrogen gas showed different behavior compared to other gaseous media studied. From Turkey ’s mean comparison plot (Fig. 5) and comparison of main effects plots (Fig.6a- 6a–c), it is obvious that photoresist removal in

hydrogen gas medium falls under different category compared to the rest of the gases studied. Despite its higher thermal conductivity, hydrogen gas showed remarkable increase in ablation rate, thus proving to be an assisting chemical agent during the photoresist

removal by excimer laser. Its standard deviation on the mean value is also higher compared to other gases. This wide spread standard deviation data may be attributed to the possible influence of main input parameters viz., laser energy, PRF and hydrogen gas flow rate

on the etch depth produced. Photographs of 5 mm dia. circular pattern etched on the photoresist under different gaseous media is shown in Fig.7 7. It is interesting to note that the etched surface for all the gases, except hydrogen, showed a pattern of irregular etches all

along the circumference and black coloration. Formation of black colour on the surface may be explained as detailed below. Laser assisted burning of polymeric material with traces of oxygen present as a contaminant in the gaseous medium might be the reason for

producing carbonaceous deposits on the surface [13]. It was also observed that the black colour on the surface was more prominent in central region of the ablated area than the circumference. This was an obvious observation as the central region of laser beam has

higher hot spot, resulting in higher rate of burning of the polymeric material with oxygen. In contrast to other gases studied, hydrogen gas showed good quality of surface contour without any black marks. Like other gases, hydrogen gas used for the study might also have

traces of oxygen as a contaminant. But, surface blackening was not observed in this case due to the possible laser assisted chemical reaction of oxygen with hydrogen rather than polymeric material.
Fig. 5 Turkey’s Means comparison plot for the five gases studied.

alt-text: Fig. 5

Fig. 6 (a) Comparison of main effect (gas flow rate) on photo etching in different gas media. Fig. 6b(b) Comparison of main effect (pulse repetition frequency) on photo etching in different gas media. Fig. 6c(c) Comparison of main effect (energy) on photo etching in different gas media.

alt-text: Fig. 6
Fig. 7 Photograph of 5 mm dia. circular pattern etched on the photoresist under different gaseous mediaFig. 7Fig. 8a.

alt-text: Fig. 7

Since laser assisted photoresist etching process under hydrogen gas medium has showed distinct characteristics with respect to etch rate and surface quality, three-way ANOVA analysis of data obtained for hydrogen gas medium was carried out to understand

the main effects and interaction effects of all the three variable parameters. Main effects are illustrated in Fig.8a- 8a–c, whereas the interaction effects are illustrated in Fig. 8d. While the response of etch depth showed increasing trend with respect to laser energy and flow

rate, it showed decreasing trend with respect to PRF, as shown in these plots. While etch depth was observed to be increasing logarithmically with energy (slope  = 47; R2 = 0.98), it was observed to be decreasing logarithmically with PRF (slope = -57; R2 = 0.97). Etch

depth was also observed to be increasing logarithmically with hydrogen gas flow (slope = 4; R2 = 0.96). As far as the two factors interaction effects were concerned, flow rate showed profound effects combined with both energy and PRF. But, combined effects of energy

and PRF did not show drastic change on etch depth, which may be attributed to their opposing logarithmic trends with almost equal slopes, as seen in their main effects plots ( Fig.8a- 8a–c).
Fig. 8 (a) Effect of laser energy on etch depth in hydrogen medium. Fig. 8b(b) Effect of PRF on etch depth in hydrogen medium. Fig. 8c(c) Effect of hydrogen flow rate on etch depth in hydrogen medium. Fig. 8d(d) Interaction Effects in hydrogen mediumFig. 8a.

alt-text: Fig. 8

5.2 Modelling by Response Surface Mresponse surface methodology

Based on RSM methodology, a mathematical quadratic equation as shown below was obtained.

Y=-11.06 + 12.18 A - 5.34 B + 4.81 C - 0.55AB + 0.23AC - 0.037 BC-3.078 A2+0.104B2 -0.0313 C2

where Y represents the photoresist etch depth (µm) and A, B, and C are hydrogen gas flow rate (l/min.), pulse repetition rate (Hz) and laser energy (mJ) respectively. The predicted values for the photoresist etch depth and the relative percentage deviation from

the experimental values derived from the RSM model are presented in Table 3. Predicted values by RSM model were in good agreement with the experimental values. The predicted values for etch depth by RSM model were within 10% variation for most of the cases as

can be seen from Table.3. RSM model fit was satisfactory as its regression coefficient (R2) is 0.98. Moreover, in the normal probability plot, as shown in Fig. 9, the residuals for etch depth fell closer to the straight line, confirming the integrity of developed RSM model. The

deviation between experimental and predicted values was less than 10% for most of the cases. Three-dimensional response surface plots ( Fig.10a- 10a–c), predicted on the quadratic model, were drawn to study the effect of the input parameters on photoresist etch

removal rate. These plots provide supplementary assessment of the relationship between the response of etch depth and the three independent variables studied. Fig. 10a–c confirmed that PRF has got a negative effect on etch depth while the other two parameters,

namely, pulse energy and flow rate are having positive effects on the etch depth.
Fig. 9 Normal probability plot for etch depth by RSM model.

alt-text: Fig. 9

Fig. 10 (a) 3D Response surface plots showing interaction effects of energy and flow on photoresist etch depth in hydrogen medium. Fig. 10b(b) 3D Response surface plots showing interaction effects of energy and PRF on photoresist etch depth in hydrogen medium. Fig. 10c(c) 3D Response surface plots showing interaction

effects of flow and PRF on photoresist etch depth in hydrogen medium.

alt-text: Fig. 10

5.3 Modelling by Artificial Neural Nartificial neural network

Having understood the main and interaction effects of input parameters viz., laser energy, PRF and hydrogen gas flow rate on the Material Removal Rate (MRR) of photoresist, ANN model is constructed. For this modeling purpose, 48 experiments each in

triplicate were carried out under different input parameters, for a fixed number of laser shots of 500, as shown in Table. 3. Each of these input parameters are introduced to ANN structure by an input layer neuron. Experimental data sets are used to train the ANN to make

the general understanding of the model. Among the data sets, 2/3 of the data is used for training ANN, while 1/3 is used for validation purpose. While training set is used to adjust the network weights and errors in each iteration, validation sets are used to optimize the

ANN architecture with respect to number of neurons for the hidden layer. Based on the least mean square error as evaluation criterion, a structure of neural network with single hidden layer with five nodes is observed to be appropriate one for ANN modeling as shown in

Fig. 4. Correlation of training and validation data sets with the predicted values, as shown in Fig.11a- 11a–b, showed correlation coefficient > 0.99, indicating that the single layer five neural nodes ANN model is sufficient to explain the result satisfactorily.
Fig. 11 (a). Regression graph of actual and predicted values for training sets (R2 = 0.997). Fig. 11 b(b). Regression graph of actual and predicted values for Validation sets (R2 = 0.96).

alt-text: Fig. 11

The model equation for the prediction of etch depth is given by,

Etch depth (µm)= [106.71 – (17.63 * H1) – (8.04 * H2) - (100.08 * H3) – (49.20 * H4) + (17.47 * H5)]

where H1, H2, H3, H4 and H5 are five hidden nodes, which are governed by the following relations; H1= tanh [0.5 * ((6.59 – (0.29 *Flow) – (0.038* PRF) – (0.13 *Energy))]

H2= tanh [0.5 * ((–9.35 + (4.94 * Flow) + (0.075 *PRF) + (0.05 *Energy))]

H3=tanh [0.5 * ((0.48 – (0.136 *Flow) + (0.18 *PRF) – (0.027 *Energy))]

H4= tanh[0.5 * ((–0.56 + (0.26 *Flow) + (0.043*PRF) – (0.028 * Energy))]

H5= tanh [0.5 * ((1.09 + (3.34 * Flow) + (0.061 *PRF) – (0.102 * Energy))]

Optimization of input parameters to derive maximum etch depth is also carried out. As shown in Fig. 12, flow rate of 1.2 lpm, PRF of 10 Hz and energy of 60 mJ are computed to be optimum input parameters. Under this optimum conditions, a maximum etch depth

of 120 µm was achievable for 500 laser shots. Prediction capability of the ANN model was tested using the data sets and the predicted values are shown in Table 3. The predicted values were in good agreement with that of experimental data, within 10% variation for most

of the cases as can be seen from Table. 3

Fig. 12 Profile showing optimum input parameters for maximum etch depth under the experimental conditions.

alt-text: Fig. 12

5.4 Comparison of RSM and ANN models

The predicted values obtained from the ANN and RSM models are compared in Table 3. Prediction performance of both the models are compared based on the correlation of experimental versus prediction values ( Fig. 13) and percentage of relative deviation of

predicted values from the experimental values (Fig. 14). The results suggest that both models are suitable to predict this photoresist removal process by excimer laser. However, the ANN model appears to be superior for both training as well as validation of model. This

may be attributed to the fact that RSM is limited by a second order polynomial, whereas ANN is a more reliable modeling technique to represent the non- linearity of this system.
Fig. 13 Comparison of RSM and ANN models based on correlation graph between experimental and predicted values.

alt-text: Fig. 13

Fig. 14 Comparison of RSM and ANN models based on their relative deviation%.

alt-text: Fig. 14

5.5 Role of hydrogen gas on photoresist removal by excimer laser

Photoresist removal rate, in general, is sum of the rates attributed to photochemical (PC) and photothermal (PT) processes [ 14]. Depending upon the conditions in which the removal process is conducted, the extent of contribution by PC and PT may vary. Much

above the threshold limit of laser fluence, photothermal process contributes significantly to the net ablation rate. Based on their thermal conductivity values, hydrogen and helium gases should minimize the photothermal part of ablation. The present study is first of its kind

which provides an insight into the role of hydrogen gas, during excimer laser photoresist removal process. Among the gas media studied, thermal conductivity was highest for hydrogen gas medium. Being a higher thermal conductivity medium, hydrogen gas was

expected to minimize photo-thermal contribution for the net photoresist removal rate. This might reduce the net photoresist removal rate in hydrogen gas medium. On the contrary, hydrogen gas showed remarkable increase in ablation rate, thus proving to be an assisting

chemical agent during the photoresist removal process. In order to confirm if such laser assisted chemical reaction was happening or not, experiment was carried out under vacuum (10 -−2 mbar) and also at 600 mbar (absolute) hydrogen gas medium. The mass ablated in

hydrogen medium was seen to be higher than that obtained in vacuum condition, as shown in Fig. 15. The photoresist removal in hydrogen medium is observed to be increasing with energy, reaching a maximum value and then showing a decreasing trend for higher

energy. However, the photoresist removal rate in hydrogen medium is higher than that of vacuum condition for the entire energy range.
Fig. 15 Comparison of photoresist removal for 3000 laser shots in vacuum (10-−2 mbar) and hydrogen gas (600 mbar).

alt-text: Fig. 15

FTIR spectrum of experimental photoresist is compared with vapor phase produced with laser irradiance in vacuum and hydrogen environment is shown in Fig.16a. It may be noted that the absorption of experimental photoresist is very high compared to vapor

phase product generated after laser irradiance as can be seen from Fig. 16a. Therefore comparison of only vapor phase product formed under vacuum and hydrogen gas media is shown separately in Fig.16b. From this figure it can be observed that spectral bands in three

regions showed change in peak heights. It may be noted that the spectral region in the range 3100 -3000 cm-1is due to C-–3000 cm−1is due to C–H stretching adjacent to double bond or aromatic ring. 3300-3400 cm-–3400 cm−1 spectral region is due to N

-H stretching. It is obvious from the IR spectra that C-–H stretching region is more prominent than N

-H stretching region for vacuum experiment. On the contrary, N

-H stretching region is more prominent than C-–H stretching region for hydrogen gas experiment. This might be attributed to hydrogenation of olefin functional group. A broad spectral band in the range of 3800 -–3600 is also observed for hydrogen experiment.

This is due to -−OH of hydrogenated/opened epoxy group. From these observations, we may infer that there might be a Laser Assisted Chemical Reaction (LACR) of hydrogen gas with photoresist polymer. It is well known that atomic hydrogen has the effect of etching

sp2 bonding (unsaturated olefinic bonds, π bonds) fractions in the photoresist polymer films [15,16].
Fig. 16 a(a). Comparison of FTIR spectrum of experimental photoresist with vapour phase produced during laser irradiance in vacuum and hydrogen gas environments. Fig. 16 b(b). Comparison of FTIR spectrum of vapour phase produced during laser irradiance of experimental photoresist in vacuum and hydrogen gas

environments.

alt-text: Fig. 16

5.6 Proposed mechanism for photoresist removal by laser in hydrogen gas medium
Researchers [17,18] have studied photo degradation of polymers in oxygen medium and proposed a mechanism for photo oxidation of polymer. Since oxygen molecule absorbs UV photon, it is electronically excited to singlet state, which exhibits several specific

reactions and the one that has been most often invoked in the photo-oxidation of polymers is the formation of a hydro-peroxide by oxidation of an olefin containing an allylic hydrogen, which could further decompose and lead to additional chain scission.

In our study, hydrogen gas, though doesn't absorb UV photon, is showing remarkable increase in photoresist removal rate under laser irradiance. Hydrogen gas, even though transparent to UV photon, may still play some role in enhancing the photoresist
removal rate. A mechanism has been proposed here to explain the key role that hydrogen gas plays during the process of photoresist removal by laser. When a hydrogen molecule approaches the polymer surface, as indicated in Fig.17 a-a–b, it will first feel the van der

wals force of interaction, which is a weak interaction due to polarization effects, resulting in physical adsorption. At this stage, the molecular orbitals of hydrogen molecule are interacting with the delocalized π electron cloud of polymeric benzene group or olefinic groups,

leading to lowering and broadening of the bonding and antibonding orbitals of hydrogen molecule. When hydrogen approaches even closer or at polymer-gas interface, there may be an exchange between π electrons cloud of polymer and hydrogen molecular orbitals,

leading to associative chemisorption. This interaction results in weakening of the internal bonding of the hydrogen molecule due to the filling of the otherwise antibonding (empty σ* state) orbital. The resultant bond breaking leads to generation of hydrogen free radicals.

Compared to the restricted movement of polymeric macro radicals in the polymeric material the hydrogen free radical, being very reactive, unstable and mobile, can diffuse and migrate along the polymer chain attacking SP 2 and SP hybridization carbon center

(unsaturated π bonds around carbon). This is an additional pre-processing step, which converts the polymer into saturated macromolecules, i.e π bond is converted into σ bond due to hydrogenation. This process alters the physical and chemical property of polymeric

surface, which aids for easy bond scission by laser photon interaction with polymer. So, it can be easily concluded that increase in photoresist removal rate in hydrogen environment is due to gas assisted chemisorption of hydrogen molecule on polymeric surface resulting

in generation of highly reactive hydrogen radicals.

Fig. 17 a(a). Showing polymer-gas interaction leading to hydrogen molecule bond breaking by π electron cloud of polymer. Fig. 17 b(b). Schematic potential energy diagram for a hydrogen molecule approaching an electron rich surface.

alt-text: Fig. 17

Researchers (19-–24) have explored the possibility of employing hydrogen radical as a work horse for photoresist removal as an alternate method to the existing conventional method such as chemical method, oxygen plasma and ozone gas. They produced

hydrogen radical by the catalytic decomposition of hydrogen molecules in a mixture of hydrogen and nitrogen gas using a tungsten hot-wire catalyzer at elevated temperature of about 2000  °C. Our is the first report showing the effect of hydrogen radicals in excimer laser

ablation of photoresist polymers. The proposed mechanism and its reasoning is described in detail the following sections.

In order to validate our proposed mechanism of LACR of photoresist with hydrogen gas under laser irradiance, we have compared the photoresist removal rate of poly(methylmeth-acrylate (PMMA) with our experimental polymer E1020. It may be noted that the

photoresist, PMMA is an aliphatic class of polymer, not having unsaturated/olefinic functional groups, while our experimental polymer is an aromatic class polymer having phenyl and carbonyl functional group. Photoresist removal rate of these two polymers were studied

in vacuum and hydrogen environments. Comparison of photoresist removal rates for these photoresists as a function of laser source energy in vacuum and hydrogen environments are shown in Figs.18a-. 18a–b. In vacuum condition, the aliphatic class polymer showed

higher ablation rate compared to aromatic class polymer. But, in hydrogen environment, aromatic type polymer showed higher ablation rate than the aliphatic type. These observations can be easily explained based on the polymeric structure. Polymer photoresists, which

are widely used in electronic industries/laser micromachining process are shown in Fig. 19.These polymers are classified as aliphatic and aromatic photoresists. Majority of the polymer photoresists, possessing phenyl group functional group, are termed as aromatic class,
while others are aliphatic class photoresists. The main reactions observed during irradiance of polymers with laser are bond scission, chain cross linking and oxidative degradation. Irradiation of polymers by UV lasers leads to the formation of radicals whose nature is

dependent on type of reactive functional groups of polymers. The concentration of the arising macro radicals depends on many factors such as polymeric structure, temperature, concentration and duration of action of free radical and specific area of samples.

Fig. 18 a(a). Comparison of aromatic and aliphatic class photoresist removal in vacuum. Fig. 18 b(b). Comparison of aromatic and aliphatic class photoresist removal in hydrogen gas (600 mbar).

alt-text: Fig. 18
Fig. 19 Chemical structures of various polymer photoresists.

alt-text: Fig. 19

5.6.1 Laser photoresist removal in vacuum

In vacuum condition, where no other external molecular interaction is expected to happen, laser light only plays the role in ablation process either by photolytic or photo-thermal or combination of both processes depending on the laser energy level. In both processes, polymer

undergoes degradation /decomposition into many gaseous products including monomers and other molecular fragments. The major difference between aliphatic and aromatic class polymers is the type of bonds between carbon backbone only. While aromatic class polymer has conjugated

double bonded (

-C = C-C

) carbon as backbone due to the presence of benzene rings, aliphatic class polymer has single bonded (

-C-C-C

) carbon chain as backbone. The energy required for breaking single bonded carbon (C

-C bond energy is 347 kJ/mol) is lower than that required for breaking of double bond (

-C = C

C bond energy is 614 kJ/mol). In other words, for given laser energy, the ablation rate shall be higher for aliphatic class than aromatic class polymer. Our experimental results obtained for photoresist removal in vacuum condition, as shown in Fig. 18a, is supporting our proposed

mechanism.

5.6.2 Laser photoresist removal in hydrogen gas

Comparison of photoresist removal rate as a function of laser energy in hydrogen gas environment for both the aromatic and aliphatic class polymers is shown in Fig. 18b. The photoresist removal rate in hydrogen gas environment is significantly higher compared to that in vacuum and

air environment. Moreover, aromatic class polymer showed higher photoresist removal rate at lower laser energy level than aliphatic class polymer. It is interesting to note that both class polymers showed same peak maximum ablation rate in hydrogen environment. Aromatic class polymer is

observed to show peak maximum ablation rate at about 500 mJ laser energy, while aliphatic class showed peak maximum at 1000 mJ. This indicates that aromatic class polymer needs lower laser energy compared to aliphatic class polymer for optimal removal rate. This enhanced ablation rate

may be attributed to hydrogen gas assisted excimer laser resist removal process. This observation is in accordance with our proposed mechanism of LACR of polymer with hydrogen gas under laser irradiance.

5.7 Potential area of application

Being a cold process, the proposed hydrogen assisted excimer laser ablation can be exploited as an alternative method to existing conventional processes of photoresist removal process to overcome their disadvantages. Our finding of hydrogen assisted

excimer laser ablation of polymer has potential application in semiconductor industry. In manufacturing of semiconductor devices, ion-implanted photoresists are widely employed in certain stages of manufacturing. The term 'ion-implanted photoresist' means, a photoresist

implanted with ions of Boron/Phosphorous etc. depending on the application. In general these ion implanted photoresists are difficult to remove and one of the methods used for its removal is by using atomic hydrogen generated at higher temperature [ 19–24]. Our finding

of hydrogen gas assisted excimer laser based polymer removal process may find a potential use in semiconductor industry as an alternate process to the tungsten hot-wire method wherein atomic hydrogen is used to remove the photoresist.

6 CONCLUSIONSonclusions
Photoresist removal using excimer laser under the flowing medium of air, nitrogen, argon and helium, produced similar and lesser ablation rate whereas hydrogen gas environment showed substantially higher ablation rate. This remarkably higher ablation rate in

presence of hydrogen gas proves that hydrogen acts as an assisting chemical agent during the laser irradiance of photoresist. Considering the profound main and interaction effects of laser pulse energy, pulse repletion rate and gas flow rate on the photoresist removal

under hydrogen gas medium, one can conclude that hydrogen gas gets involved in the laser assisted chemical reaction with photoresist. This is responsible for the higher photoresist removal observed, compared to other gases studied. The authors have put forward a

mechanism explaining this observed phenomenon. As for the modeling of laser assisted photoresist removal under hydrogen gas medium is considered, the prediction capability of ANN model is superior than RSM model. Being a cold process, hydrogen gas assisted

excimer laser photoresist removal process can be exploited as an alternative method to existing conventional processes to overcome their disadvantages. A higher throughput per pulse, resulting from the enhanced rate of mass removal associated with excimer laser

micromachining in hydrogen gas environment, could provide economic advantage for the process, considering the high cost of the consumable pre-mix gas used in the excimer laser system.
Acknowledgment
The authors wish to express their sincere thanks to Professor. Ramesh K. Singh (Department of Mechanical Engineering, Indian Institute of Technology, Mumbai, India) for his guidance and support for this work. The authors also express their

gratitude to Mr. Rajashekhara Biradara, for his assistance in carrying out the experiments.

References
[1] Mi-Hee Jung and Ho-Suk Choi, ’’Photoresist etching using Ar/O2 and He/O2 atmospheric pressure plasma, ’’Thin Solid Films 515, 2006, 2295–2302.

[2] C.H. Yi, C.H. Jeong, Y.H. Lee, Y.W. Ko and G.Y. Yeom, Oxide surface cleaning by an atmospheric pressure plasma, Surf. Coat. Technol. Coat Technol 711, 2004, 177–178.

[3] C.H. Yi, Y.H. Lee and G.Y. Yeom, ’’ The study of atmospheric pressure plasma for surface cleaning, ’’Surf. Coat. Technol.Surf Coat Technol 171, 2003, 237–240.

[4] H. Morinaga, M. Suyama and T. Ohmi, Mechanism of metallic particle growth and metal-induced pitting on Si wafer surface in wet chemical processing, Journal of the Electrochemical Society Electrochem Soc 141 (10), 1994, 2834–2841.

[5] H. Morinaga, T. Futatsuki, T. Ohmi, E. Fuchita, M. Oda and C. Hayashi, Behavior of ultrafine metallic particles on silicon wafer surface, Journal of the Electrochemical Society Electrochem Soc 142 (3), 1995, 966–970.

[6] H. Horibe, M. Yamamoto, E. Kusano, T. Ichikawa and S. Tagawa, Resist removal by using atomic hydrogen, J. Photopolym. Sci. Technol. Photopolym Sci Technol 21 (2), 2008, 293.

[7] M. Yamamoto, H. Horibe, H. Umemoto, K. Takao, E. Kusano, M. Kase, et al., S.TagawaPhotoresist removal using atomic hydrogen generated by hot-wire and effects on Si-wafer surface, Jpn. J. Appl. Phys. J Appl Phys 48 (2), 2009, 026503-1-026503-7.

[8] M. Yamamoto, T. Maruoka, A. Kono, H. Horibe and H. Umemoto, Substrate temperature dependence of the photoresist removal rate using atomic hydrogen generated by a hot-wire tungsten catalyst, Jpn. J. Appl. Phys. J Appl Phys 49, 2010, 16701–16706.

[9] M. Yamamoto, T. Maruoka, Y. Goto, A. Kono, H. Horibe, M. Sakamoto, et al., E.KusanoH.SekiS.TagawaRemoval of ion-implanted photoresists using atomic hydrogen, J. Electrochem. Soc. Electrochem Soc 157 (3), 2010, H361.

[10] M. Yamamoto, T. Maruoka, A. Kono, H. Horibe and H. Umemoto, Order of reaction between photoresist and atomic hydrogen generated by a tungsten hot-wire catalyst, Appl. Phys. Phys Express 3 (2), 2010, 026501–0265013.

[11] David R. Lide, CRC Handbook of Chemistry and Physics88th Ehandbook of chemistry and physics, 88th edition, 2007, by CRC Press, 97808493048809780849304880 - CAT# 488, June 25.

[12] B.J. Garrison and R. Srinivasan, Laser Ablation of Organic Polymers: Microscopic Models for Photochemical and Thermal ProcessesJournal of Applied Physics.ablation of organic polymers: microscopic models for photochemical and thermal processes, J Appl Phys
57, 1985, 2909–2914.

[13] R. Srinivasan, B. Braren and R.W. Dreyfus, Ultraviolet Laser Ablation of Polyimide FilmsJ. Appl.Phys.laser ablation of polyimide films, J Appl Phys 61 (1), 1987, 372–376.

[14] D. Bäuerle, ’Chapter 2 Thermal, Photophysical, and Photochemical Processes’Laser Processing and CChapter 2 thermal, photophysical, and photochemical processes, Laser processing and chemistry, 4th ed., 2011, Springer-Verlag; Berlin Heidelberg, 13–38,
https://doi.org/10.1007/978-3-642-17613-5_2.

[15] B.V. Spitzyn, L.L. Bouilov and B.V. Derjaguin, J. Cryst. Cryst Growth 52, 1981, 219.

[16] A.P. Dementiev and M.N. Petukhou, Diamond RelatMater.Diamond Relat Mater 6, 1997, 486.

[17] G. Geuskens and C David, Photo oxidation of polymers a comparison with low molecular weight compounds, Pure & Applied ChemistryAppl Chem 51, 1979, 233–240.

[18] G Geuskens and C David, Polymer photo degradation: mechanisms and experimental methods, Pure and Applied ChemistryAppl Chem 51, 1979, 233–240.

[19] M. Takata, K. Ogushi, Y. Yuba, Y. Akasaka, K. Tomioka, E. Soda, et al., N.KobayashiPhotoresist removal process by hydrogen radicals generated by W catalyst, ’Thin Solid Films 516, 2008, 847–849.

[20] Kouhei Hashimoto, Atsushi Masuda, Hideki Matsumura, Tomoatsu Ishibashi and Kazuhisa Takao, Systematic study on photoresist removal using hydrogen atoms generated on heated catalyzer, ’Thin Solid Films 501, 2006, 326–328.

[21] Akira Izumi and Hideki Matsumura, ’Photoresist Removal using Atomic Hydrogen Generated by Heated CatalyzerJapanese Journal of Applied PhysicPhotoresist removal using atomic hydrogen generated by heated catalyzer, Jpn J Appl Phys 41, 2002, 4639–4641.

[22] H. Horibe, M. Yamamoto, T. Maruoka, Y. Goto, A. Kono, I. Nishiyama, et al., S.TagawaIonimplanted Resist Removal Using Atomic HIonimplanted resist removal using atomic hydrogen, Thin Solid Films 519, 2011, 4578–4581.

[23] Akihiko Kono, Yu Arai, Takeshi Maruoka, Masashi Yamamoto, Yousuke Goto, Seiji Takahashi, Takashi Nishiyama and Hideo Horibe, High removal rate of cross-linked SU-8 resist using hydrogen radicals generated by tungsten hot-wire catalyzer, Thin Solid
 Films 562, 2014, 632–637.

[24] T. Maruoka, Y. Goto, M. Yamamoto, H. Horibe, E. Kusano, K. Takao, et al., S.TagawaRelationship between the thermal hardening of ion-implanted resist and the resist removal using atomic hydrogen, J. Photopolym. Sci. Technol. Photopolym Sci Technol 22 (3),
2009, 325.

Highlights

• Photoresist removal process by 248 nm KrF Excimer laser is studied in gaseous media viz., hydrogen, helium, argon, nitrogen and air.

• Despite its higher thermal conductivity, hydrogen gas showed remarkable increase in photoresist removal rate compared to vacuum and other gas media studied.

• As far the main effects are concerned, laser energy and hydrogen gas flow rate showed increasing trend on photoresist removal rate, while pulse repetition rate showed decreasing trend.

• From a set of experiments, two models viz., Response Surface Methodology and Artificial Neural Network were tested to predict the photo resist removal rate satisfactorily.

• A mechanism has been proposed to explain the observed phenomenon of higher photoresist removal rate in hydrogen environment.

Queries and Answers

Query: Your article is registered as a regular item and is being processed for inclusion in a regular issue of the journal. If this is NOT correct and your article belongs to a Special Issue/Collection please contact s.jayan@elsevier.com
immediately prior to returning your corrections.
Answer:

Query: The author names have been tagged as given names and surnames (surnames are highlighted in teal color). Please confirm if they have been identified correctly.
Answer:

Query: Please check that the affiliations link the authors with their correct departments, institutions, and locations, and correct if necessary.
Answer:

Query: Highlights must be provided as 3-5 bullet points, each bullet point having a maximum of 125 characters. Please rephrase the following highlights or provide new highlights.
Answer: