1. Introduction

Two-photon photochemical processes have attracted much of attention [1,2] due to potential in 3D applications, such as microfabrication [3,4], photonic crystals [4,5], memories [6–₁₀], two-photon holography [₁₁,₁₂]. Cationic two-photon induced photo-polymerization has recently been reported [₁₃,₁₄]. The subject has not yet been studied in depth, though two-photon initiation was demonstrated, and several basic parameters of the process were presented. Interest in this process is not only because it is a feasible low shrinkage alternative to its free-radical counterpart in 3D microlithography, but also due to specific unique features of its origination from photoinduced acid generation [₁₅] as a basis for the process itself. In addition to the “negative resist” type of 3D microfabrication, similar to that of free radical photopolymerization, acid generation allows for a “positive resist” type of 3D microlithography based on polymer degradation. Another feature of cationic photopolymerization is a single-molecule type of termination for the chain reaction, which allows for a stronger dependence of the polymerization rate due to light intensity, as compared to the free radical case. In this paper, a more detailed study is presented for two-photon induced photopolymerization of an epoxide sensitized by isopropylthioxanthone and diaryliodonium salt. Targeted issues are dependence of the polymerization rate due to light intensity, and the threshold value dependence on the concentration of the initiator.

2. Experimental techniques and materials

A mode-locked Ti:Sapphire laser (150 fs pulses, 76 MHz repetition rate, average output power 300 mW, wavelength 710 nm) was employed to initiate two-photon photo-polymerization. A long focal length objective with a numerical aperture of 0.35 was used to focus the beam, yielding a focused spot of gaussian intensity distribution with a 1.75 µm half-width [14]. Cycloaliphatic Diepoxide (K-126 from Sartomer) was used as the cationically polymerizable monomer. Diarylidonium hexafluoroantimonate (DAIF₆Sb, here DAI; CD-1012 from Sartomer) was the cationic initiator. Isopropylthioxanthone (ITX, see Fig. 1) was the sensitizer. For comparison, the free radical photo-polymerizable formulation was used, which included as Di-Penta-Erythritol-Penta-Acrylate (DPEPA, SR399 from Sartomer) as a monomer and Benzil-Di-Methyl-Ketal (BDMK) as photo-initiator. An interferometric technique [14] is used for in-situ monitoring of the polymer conversion. Single-photon initiated photo-polymerization was performed with longwave UV source (UVP, model B, Blak-Ray, 115 V, 60 Hz, 2.5 A, filter 365 nm) @ 365 nm wavelength, distance between source and sample being 5″. Infrared radiation from the source was blocked by a water filter with a 2″ path-length to prevent heat-induced contributions to polymerization effect. Polymerizable formulation was placed into the glass cuvette and exposed to 365 nm collimated radiation at various intensities, regulated by optical filters. Exposure time was kept the same for each series of experiments. After exposure the non-polymerized part of the formulation was washed out with acetone:propanol=1:1 solution, then the remaining polymer sample was dried @ 100 0C for 10 min. Thickness of the produced polymer layer was assumed to be the measure of the polymerization rate in single-photon regime [13]. Here single-photon polymerization rate measurements is used to make a comparison of that with the polymerization performance in two-photon initiation case.

 

Fig. 1. Chemical structure of cationic sensitizer Isopropylthioxanthone (ITX; 2&4 isomer mixture; C16H14OS)

Download Full Size | PDF

2.1 Interferometric monitoring of photopolymerization

Similar to free radical polymerization [16,17], we assume for the cationic case the refractive index, n, of the material is linearly proportional to the polymer conversion, N, defined as the fraction of the monomer molecules included into the polymer chains. The reason for this is the incremental change of molecular polarizability of the monomer molecules when connecting to the polymeric chain, and the change of molecular density (shrinkage) that accompanies polymerization. Therefore Eq. (1)–Eq. (12) are equally applicable to both free radical and cationic cases of polymerization:

where R is the polymerization rate, and t is the exposure time. Transformation of the monomer into a polymer is accompanied by the gradual change of n from n₀ of the initial formulation to n_p of the final polymer. As the polymerization progresses, N increases from zero to its saturation level N_sat (for given conditions of temperature, pressure, and chemicals involved), the n changes from n₀ to n_sat=n_p. Since incremental changes in molecular polarizability and molecular density contribute linearly to the change of n, N becomes linearly connected with n:

where n(N) is the refractive index of the polymer at conversion level N, Δn(N)=|n₀-n(N)|, Δn_sat=|n₀-n_sat|. From Eq. (1) and Eq. (2) it follows:

where coefficient k=1/Δn_sat. For a two-photon initiated polymerization process above threshold, polymerization rapidly reaches linear growth and then gradually slows down (saturation). Therefore the value of R, measured at the very beginning of the two-photon polymerization process, represents a characteristic value, R_lin, of the linear polymer growth. Monitoring R(N) via n(N) using Eq. (3) is the basic principle for the suggested interferometric technique [14], shown schematically in Fig. 2.

The two-photon characteristic of the photopolymerization process limits the polymerization zone to the central area of the focused beam, where peak intensity J exceeds the threshold value J_2th. Interference between the wave J₁ passing through the polymerization zone, and the wave J₂ diffracted around it, yields an interference pattern on the screen (Fig. 2) that shows the phase differences between J₁ and J₂. In practice, the quantity measured by the detector in Fig.2 is an average intensity value I produced by the interference of J₁ and J₂, each of which alone produces an average detector response I₁ and I₂ respectively. Thus herein I denotes measured average intensity values, while J denotes peak intensity values in the focal spot.

For simplicity we assume the amplitudes of both waves are equal |J₁|=|J₂|, and therefore the intensity of the interference pattern I is changing from zero (when J₁ and J₂ are completely out-of-phase) up to J_max=4J₁ (when J₁ and J₂ are totally in-phase). The phase shift Δϕ between J₁ and J₂ is determined by the optical path difference within the polymerization region:

where ϕ₁, ϕ₂ are phase changes acquired by J₁ and J₂ when passing through the polymerization zone of length L, λ is the wavelength in a vacuum, and Φ is the phase of the interference pattern. Therefore

The intensity distribution of the interference pattern I, under the assumption |I₁|=|I₂|, is

The temporal change dI/dt is thus

Experimentally choosing the measurement point of dI/dt to be Φ=π/2 (where sinΦ=1) simplifies Eq. (9) to

which together with Eq. (3) and Eq. (6) yields

where coefficient K=2λk/L, and I_max is the maximum average intensity value of the interference pattern at the detector. Eq. (12) allows the retrieval of the polymerization rate from the interferometric data obatined using the setup in Fig. 2.

 

Fig. 2. Schematic for in-situ interferometric monitoring of two-photon photopolymerization. Only the two-photon induced contribution to the local change of the refractive index, Δn_total, is contributing to the signal on the detector. This technique is “blind” to the single photon induced contribution to Δn. The dominant contribution is due to Δn_pol=n(N_pol) - n_mon, i.e. Δn_pol≫Δn_thermal, Δn_bleaching, Δn_{others if any}.

Download Full Size | PDF

3. Experimental results

The following will be studied:

a) Intensity dependence of R_lin

b) J_2th and R_lin dependences on initiator concentration

c) Dynamic range of the material, which is defined as ratio J_damage/J_2th with J_damage denoting the damage threshold. Above J_damage the material experiences unrecoverable loss of transparency.

3.1 Cationic two-photon photo-polymerization

The first observation of two-photon cationic photopolymerization with high dynamic range was reported in earlier work [14]. Here we present a further study of a two-photon cationically photopolymerizable material based on an epoxide (Cycloaliphatic Diepoxide, K-126 from Sartomer). Two-photon sensitivity to 710 nm is achieved by sensitization of diarylidonium salt (DAI) with isopropylthioxanthone (ITX, see Fig. 1) in the formulation K126:ITX:DAI=x:y:z. Testing of the formulation with no sensitizer ITX (i.e. y=0) in the setup of Fig. 2 revealed no measureable signal at maximum achievable intensities J_max=300 GW/cm². This indicates it is ITX which is responsible for two-photon sensitivity at 710 nm.

The process seems to exhibit threshold behavior at a threshold intensity of approximately 1 GW/cm² (Ti:Sapphire laser, 710 nm, 150 fs pulses @ 76 MHz rate). Single-photon sensitization of cationic polymerization of epoxides with thioxanthones (TX) and onium salts is known to be due to an electron transfer from the triplet state of TX to the cation of the onium salt [18]:

TX+2hv→¹TX→³TX (TX*)

TX*+Ph₂I⁺;→Exciplex (TX…Ph₂I⁺)

(TX…Ph₂I⁺)→(TX⁺ +Ph₂I)

Ph₂I→PhI+Ph

The extended conjugated electronic structure of the thioxanthone molecule (see Fig. 1) allows the possibility of low excitation energy via a two-photon absorption process at ⁽¹⁾λmax=375 nm (two-photon equivalent being ⁽²⁾λmax=750 nm; data of λmax are listed for methanol). Ideally, the shape of the experimental plot should be a cos-function of the exposure time, as in the equation (8). As the exposure proceeds, the number of the available reactive double-bonds gets reduced, eventually reaching the saturation level, which corresponds to the saturation of the experimental curve (see Figs.3–5). Deviations from this ideal behavior are possible due to following reasons. In one case (see for instance Fig 3-b of the Ref.[14]), the recorded noise hologram may affect (reduce) the amplitude of the signal’s changes due to contrast reduction of the interference pattern. In other cases (as it is illustrated in Figs.3–5 herewith) the overall shift of the interference pattern may occur due to the beam deflection by the polymerization area itself, and which in turn may be caused by the defects in the lens and/or optical misalignments. In the last case the informative parts of the experimental plots are its initial part and its far-end part. The slope at initial part characterizes the initial polymerization rate R_lin, whereas far-end part indicates saturation of the polymerization by reaching the steady value and thus confirming that the changes in the signal is due to the polymerization reaction, not anything else. The last fact has been used here to verify for each experiment that there are no contributions into the detected phase shift other than due to polymerization reaction. The described overall shift of the interference pattern due to defects in the optical lens or misalignment is strongest at the intermediate levels of polymerization, where product of (Δn)x(dn/dt) is largest, and it has no effect at the initial as well as final stages of polymerization, where either Δn (at the initial stage) or dn/dt (at the final stage) are at its lowest (i.e. close to zero). It is worth noting here, that even though only initial and the very far part of the plots in Fig.3–Fig.5 have been accounted for in the present article, all intermediate points of the interferometric phase shift plots do carry an information on polymer conversion at each and every observation moment. This information is, generally speaking, retrievable by employing technique similar to that in Ref.[16], provided the restriction is met of having no distortions within the observation period. Among presented plots only two would qualify for that, namely, those in Fig.4, a and Fig.5. The rest of the plots in Fig.3–Fig.4 illustrate that even encountering distortions at the intermediate levels of polymerization does not compromise the technique described herewith.

For the validity of the intreferometric registration of polymerization it is important to rule out any possible contributions into Δn which are not due to polymerization itself, i.e. which differ from Δn_pol. Among the candidates for such contributions into Δn most obvious are (see Fig.2): 1) Δn_thermal, caused by local heating in the focal spot; 2) Δn_bleaching, caused by bleaching of the sensitizer. Local heating up by laser radiation in the focal spot can be identified experimentally by its accumulative nature as well as by its dissipation when intensity of the light is dropped down (by optical filter, for example). Contribution from the bleaching can be determined via replacing the photopolymerizable sample by that of fully polymerized polymer film doped with sensitizer. Thus measured, the observed shift of the interference pattern due to bleaching of the sensitizer can be accounted for the case of photopolymerizable material. In our experiments the contribution from the bleaching of the sensitizer is found to be less than sensitivity of the experimental set up (less than half-fringe) and therefore was neglected. Contribution from the heat-induced Δn also is found to be much smaller than that produced by polymerization reaction and therefore can be neglected too. This way it has been verified that for the experiments under consideration Δn_pol≫Δn_thermal, Δn_bleaching and therefore experimentally observed shift of the interference pattern is caused only by polymerization reaction.

In the described experiments it is essential also to provide an ample proof of two-photon nature of the observed polymerization process. Combination of the following features has been considered here as sufficient evidence of the two-photon origin of the process:

1) Absence of the detectable absorption of the sensitizer at the operating laser wavelength (710 nm herewith).

2) Significant absorption at the half-wavelength of the operating laser, i.e. at 355 nm herewith.

3) Absence of the polymerization when exposed to continuous mode operating laser at its maximum output power.

4) Presence of the polymerization when exposed to mode-locked operating laser only.

Features 1 and 2 were established via spectrophotometry, whereas features 3 and 4 were established by comparing the material response at two different modes of Ti:Saphire laser operation, namely, continuous wave and/or mode-locked regime. It is worth to note that while availability of the features 1 and 2 is a necessary prerequisite of establishing the two-photon nature of the polymerization under study, the presence of features 3 and 4 suffice it. This is because under condition 4 the same average energy flux delivered with much higher peak intensity than under condition 3, produces essentially different response, namely, the polymerization occurs only at the operating peak intensity higher than J_2th.

The threshold seems to depend on the concentration of onium salt, and is somewhat inversely proportional to onium salt concentration (see Section 3.2 for more detailed consideration of the issue).. In this regard, the solubility of onium salt limits the achievable threshold value. Results from the interferometric monitoring of cationic two-photon photopolymerization at various intensities and concentrations of DAI are shown in Fig. 3–Fig. 5, where the polymerization rate is proportional to the slope of the curve as indicated by Eq. (12).

 

Fig. 3. Change of two-photon initiated cationic photopolymerization rate (which is proportional to the slope of the curve as indicated by Eq. (12)) at different peak intensities J of the laser beam - (a) J=10 GW/cm²; (b) J=20 GW/cm² - for low concentrations (z=0.5%) of the cationic initiator DAI (formulation K126:ITX:DAI=97:2.5:0.5). Vertical axis is in arbitrary units of I using Eq. (8). Horizontal axis is exposure time in seconds. The polymerization rate R_lin is proportional to the slope of the beginning stages of exposure. Initial zero shift is due to background contribution after the beam is turned on. . Differences in appearance are not essential here, which is commented on in the text of the article.

Download Full Size | PDF

 

Fig. 4. Change of two-photon initiated cationic photopolymerization rate (which is proportional to the slope of the curve as indicated by Eq. (12)) at different peak intensities J of the laser beam - (a) J=10 GW/cm²; (b) J=20 GW/cm² - for high concentrations (z=1.5%) of the cationic initiator DAI (formulation K126:ITX:DAI=96:2.5:1.5). Vertical axis is in arbitrary units of I using Eq. (8). Horizontal axis is exposure time in seconds. The polymerization rate R_lin is proportional to the slope of the beginning stages of exposure. Initial zero shift is due to background contribution after the beam is turned on.

Download Full Size | PDF

 

Fig. 5. Saturation of two-photon initiated cationic photopolymerization rate at near-threshold laser intensity. Formulation is K126:ITX:DAI=96:2.5:1.5. Vertical axis is in arbitrary units of I from using Eq. (8). Horizontal axis is exposure time in seconds. The polymerization rate R_lin is proportional to the slope of the beginning stages of exposure. Initial zero shift is due to background contribution after the beam is turned on.

Download Full Size | PDF

Near-threshold polymerization monitoring results shown in Fig. 5 illustrates a slow down of the polymerization rate at near-threshold conditions of the two-photon cationic initiation. Results show the polymerization rate of two-photon induced cationic photopolymerization increases for higher peak intensities J of the laser radiation (compare (a) and (b) in Fig. 3 and Fig. 4) and for higher concentration levels of the cationic initiator DAI in the formulation (compare Fig. 3 and Fig. 4). The dynamic range of this formulation appears to be very high. Damage thresholds were not observable with the laser we used, which indicates it is >300 GW/cm². This allows for high speed polymerization rates at peak intensities J significantly above the threshold intensity J_2th without damage to the material.

3.2 Processing of interferometric data for cationic photopolymerization

The intensity dependence of the polymerization rate for cationic two-photon photopolymerization differs from that of the free radical case due to a termination mechanism difference. The growth of polymer chains in the free radical case has a bi-molecular type of termination, which has a square root intensity dependence on intensity for single-photon photopolymerization:

where C is a constant coefficient. This dependence becomes linear for the two-photon case:

where J_2th is the threshold intensity, and is introduced empirically to reflect experimental threshold observations.

The cationic case has a mono-molecular type of termination, if at all. Then the respective modifications of Eq. (13) and Eq. (14) give

where for free radical initiation m=1, and for cationic initiation mechanism generally 1<m<2 with m=2 if termination is 100% mono-molecular; m < 2 reflects non-trivial photochemical initiation and termination mechanisms. The capacity of the demonstrated experimental monitoring technique based on the described interferometric approach as determining of m is of importance for interpreting those mechanisms..

Clearly a plot of Eq. (16) on a logarithmic scale should yield a straight line. Therefore the first step in verifying whether the interferometric data for R(J) obeys Eq. (16) is to plot the data on a logarithmic scale to see whether it yields a straight line. If it does, then the exponent m can be derived from the tilt of this experimental line, which in turn is calculated by substituting the data for two experimental points J⁽¹⁾ and J⁽²⁾ into Eq. (16):

where Δ₁₂ denotes the difference in the underlying function’s values for two chosen experimental points J⁽¹⁾ and J⁽²⁾. J_2th needs to be excluded from Eq. (19) by a coordinate shift, or by choosing J≫J_2th. The obtained value of m can be verified by plotting R vs J^m; if the value of m is indeed correct, the plot should yield a straight line. J_2th can be determined by extrapolation.

The above data processing procedure has been applied to the interferometric data obtained from the two-photon induced photopolymerization of the formulation K126:ITX:DAI=x:y:z, where x, y and z are the respective concentrations. Plots of R vs J based on experimental data substituted into Eq. (12) are shown in Fig. 6. The power m=1.7 is deduced from the plots according to Eq. (19), and using this value of m in the m-power scale plot shown in Fig. 6 indeed shows a straight line fit, confirming the correct value for m=1.7. It is worth while noting here that in the described technique the threshold J_2th is determined as a value of J at which the curve R vs J intersects axis J to meet R(J_2th)=0 condition. The attempt to define a threshold relying on less quantitative criteria may lead to large errors due to almost quadratic dependence of R from J. In free radical case the strictness of the criteria is weaker as R linearly depends on J, which allows more qualitative criteria without compromising the accuracy.

 

Fig. 6. Dynamic range of two-photon cationic photopolymerizable formulation K126:ITX:DAI=x:y:z, where x:y:z=96:2.5:1.5 for Series1 and x:y:z=97:2.5:0.5 for Series2. Points A and B are near-threshold conditions for each respective Series; upper plot is on a linear scale (a); lower plot is on a power m=1.7 scale (b), J_max=300 GW/cm².

Download Full Size | PDF

Fig. 6 shows that polymerization rate R_lin increases linearly with concentration z of the initiator DAI. Additional evidence of two-photon nature of the observed polymerization @ 710 nm is provided by comparative data on single-photon initiated polymerization rate versus intensity i @ 365 nm shown in Fig.7,a for the same formulation as that of Fig.6. Data in Fig.7,a confirm that polymerization rate R for cationic single-photon initiated reaction is in about linear proportion to the light intensity i, which meets expectation. Accuracy of +15% achieved with the technique in Fig.7,a allows to conclude that the observed difference between intensity dependence of R for single- and two-photon cases is as expected – the exponent is about twice stronger in two-photon case. Some deviation of the measured difference from being exactly twice may be either due to the accuracy limits of single-photon technique used, or due to photochemical specifics of the two-photon polymerization process, which interferometric technique is capable of detecting. For comparison, Fig.7,b illustrates how single-photon performance of free radical formulation depends on the light intensity i, exposure conditions being similar to that of Fig.7,a.

It is also seen in Fig. 6 that threshold intensity J_2th depends on z, and experiences significant reduction when z increases. This trend is presented in more details in Fig.8. It is the finding of the present article that the threshold of two-photon cationic polymerization is inversely proportional to the m-th power of initiator’s concentration z, which is evidenced by the data in Fig.8 for the studied concentration range of 0.5%<z<1.5%. By way of contrast, Fig.9 illustrates that similar trend for free-radical case exhibits threshold reduction as z-^m/2, m being re-defined=1 as in Eqs. (13)–(16) for bi-molecular termination case.

 

Fig. 7. Polymerization rate increase with the increase of the light intensity i for single-photon initiated polymerization reactions: (a) cationic photo-polymerization of the formulation K126:ITX:DAI=96:2.5:1.5 ; (b) free radical photo-polymerization of the formulation DPEPA:BDMK=95:5. UV light of 365 nm wavelength is from UVP lamp (Blak-Ray). Exposure time intervals are 2 min for (a) and 30 sec for (b).

Download Full Size | PDF

 

Fig. 8. Increase of initiator’s concentration z allows the reduction of the threshold peak intensity J_2th of cationically induced two-photon photopolymerization of the formulation K126:ITX:DAI=x:2.5:z, where x+2.5+z=100%, 0.5%<z<1.5%.

Download Full Size | PDF

 

Fig. 9. Threshold peak intensity J_2th of two-photon free radical polymerization reduces as the concentration z of the initiator increases in the formulation DPEPA:BDMK=x:z, where x+z=100%, 1%<z<5%.

Download Full Size | PDF

3.3 Resolution

The highly localized photochemical response of free radical processes allows high-resolution 3D microfabrication. The smallest features are obtained by keeping the intensity slightly above the threshold condition. In this case only the central portion of the gaussian focal spot will cause polymerization, and outside this central spot no polymerization occurs. Increasing of the light intensity extends of the polymerization area. With an objective lens of NA=0.35, the smallest features obtained were approximately 1 micron at near-threshold conditions. Increasing the light intensity to 10 times the threshold value gave polymerization areas that were approximately 5 times larger.

Although cationic two-photon processes are technically more responsive to localized intensity, the response of the acid generation may be followed by a diffusion of the acid outside the focal spot. This leads to lower resolution. Other conditions being equal, the smallest features obtained in cationic formulations are 3 to 5 times larger than the free radical results. Imposing a limit on acid diffusion offers the potential for resolution improvement in cationic two-photon initiated processes.

4. Conclusions

Cationic two-photon photo-polymerization with a low threshold J_2th (~1 GW/cm²) and a high dynamic range (>100) is demonstrated using an isopropylthioxanthone/diarylidonium salt initiating system for the cationic polymerization of di- epoxide. The material can be fully polymerized at intensities >100 times the threshold level without damage. The polymerization rate R is found to be R=[C (J-J_2th)]^m=[C (J-J_2th)]^1.7, i.e. proportional to the m=1.7 power of the intensity, which implies a significantly stronger intensity-induced localization of the photochemical response than that of free radical photoinitiators. Concentration of the initiator (iodonium salt) is found to affect both characteristic values - the polymerization rate R_lin, and threshold intensity J_2th. R_lin increases linearly with the initiator’s concentration z, whereas J_2th reduces in proportion to z^-m (m=1.7). This is a first detailed study of two-photon cationic photopolymer materials, and is performed with an original interferometric in-situ monitoring technique.