Appl. Phys. A 83, 147–151 (2006)
Applied Physics A
DOI: 10.1007/s00339-005-3478-8
Materials Science & Processing
d.m. bubb1,2,u s.l. johnson3 r. belmont3 k.e. schriver3 r.f. haglund jr.3 c. antonacci4 l.-s. yeung4
Mode-specific effects in resonant infrared ablation and deposition of polystyrene 1 Department
of Physics, Rutgers University–Camden, Business and Science Building, 227 Penn Street, Rutgers University, Camden, NJ 08102, USA 2 Naval Research Laboratory, 4555 Overlook Ave. SW, Washington, DC 20375, USA 3 Department of Physics and Astronomy and W.M. Keck Foundation Free-Electron Laser Center, Vanderbilt University, Nashville, TN 37235, USA 4 Department of Chemistry and Biochemistry, Seton Hall University, 400 South Orange Ave., South Orange, NJ 07079, USA
Received: 8 August 2005/Accepted: 12 December 2005 Published online: 20 January 2006 • © Springer-Verlag 2005 ABSTRACT Low molecular weight polystyrene (∼ 10 kDa) was ablated with a free-electron laser at 3.31 and 3.43 µm and deposited as thin films on Si(100) substrates. The vibrational bands at 3.31 and 3.43 µm correspond to phenyl-ring CH and backbone CH2 modes, respectively. Even though the absorption coefficients of these two modes are nearly the same, the ablation yield was approximately 50% higher for the ring-mode excitation compared with the backbone mode. Based on spectral line width, the ring-mode lifetime is approximately triple that of the backbone mode, leading to a higher spatiotemporal density of vibrational excitation that more effectively disrupts the relatively weak Van der Waals bonds between neighboring polymer chains and consequently to higher ablation efficiency of the ring mode. Molecular weight assays of the deposited films showed that relatively little bond scission occurred and that the average molecular weight of the films was similar to that of the starting material. PACS 61.41.+e;
1
78.30.-j; 81.05.Lg; 81.15.Fg
Introduction
The ablation of polymers and other organic materials by infrared (IR) lasers that are resonant with a vibrational mode in the target has been the subject of several recent papers [1–3]. The mechanism of the resonant IR laser interaction in particular is substantially different than that observed when using ultraviolet (UV) or visible lasers [4, 5], where absorption occurs by exciting electrons out of the ground state. In UV and visible ablation of polymers, it is generally believed that single- or multiphoton excitations lead to transitions to the conduction band; relaxation processes leading to photochemical interactions and bond breaking that determine both the photofragmentation and the kinetic-energy distributions of the ejecta [6, 7]. In contrast, with resonant infrared ablation, localized anharmonic vibrational modes are directly excited, generating the energy density and nuclear motion required for vaporization by mechanisms that are still poorly understood. In a previous study of resonant infrared ablation of polyethylene glycol [1], we observed that the ablation yield u Fax: 1-856-225-6624, E-mail:
[email protected]
depends strongly on the particular vibrational band that is excited. From quite general considerations, one expects that the ablation yield should depend on the energy density per unit volume as follows: Yield ∝
E = Fα(ω, I ) , V
(1)
where F is the laser fluence (J/cm2 ) and α(ω, I ) is the sum of the linear and (possibly) intensity-dependent absorption coefficients (cm−1 ). If nonlinear processes substantially affect the absorption coefficient, then the volumetric energy density can become significantly larger than one would expect from linear processes alone, provided the intensity is high enough [8]. However, in [1], we found that the deposition rate was not correlated with the linear absorption coefficient during resonant infrared ablation of polyethylene glycol. The deposition rate when exciting the end-group −OH mode was significantly larger than for the CO backbone vibrational mode, even though the absorption coefficient is approximately 30 times greater for the CO mode. Therefore, (1) predicts that at similar fluences, the ablation yield should have been 30 times greater for excitation of the CO bond than the −OH bond. These results were interpreted to mean that exciting the terminal −OH bonds produced especially efficient desorption, because this mode is involved in the hydrogen bonding that is the dominant intermolecular interaction in polyethylene glycol. Here we have undertaken a study with relatively low molecular weight polystyrene (PS, ∼ 10 kDa) in order to compare the ablation yield of PS when the resonant excitation targets two bonds of similar absorption coefficient in different parts of the polymer chain. In particular, we compared the ablation yields resulting from resonant excitation of the aromatic stretching modes in the phenyl rings (3.31 µm) and of the CH2 backbone modes (3.43 µm). Ablation via the ring-mode excitation is found to have approximately 50% greater yield than that produced by backbone-mode excitation over a range of fluences, suggesting a quite different ablation mechanism in the two cases. 2
Experimental apparatus and procedure
The laser used in these experiments was the rflinac-driven free-electron laser (FEL) at the W.M. Keck Foundation Free-Electron Laser Center at Vanderbilt Uni-
148
Applied Physics A – Materials Science & Processing
versity [9]. The accelerator in the FEL is powered by an Sband klystron at 2.865 GHz, and is continuously tunable over the wavelength range 2 – 10 µm. The klystron produces 4-µs macropulses at a repetition rate of 30 Hz; each macropulse comprises some 104 micropulses of approximately 1-ps duration. The optical bandwidth of the pulses is approximately 1% of the center frequency (FWHM) and the energy of each micropulse is typically several µJ, yielding a peak unfocused irradiance of order 107 W/cm2 . The frequency chirp of approximately 1% from beginning to end of the macropulse is inconsequential for our experiments. Because of its unusual pulse structure, the intensity of the focused FEL beam is determined by the micropulse energy, while the integrated fluence is controlled by the macropulse duration. The FEL beam was introduced into the vacuum chamber by way of a BaF2 entrance window and focused onto the target by a 50-cm focal length BaF2 lens. Using a pyroelectric joulemeter, we measured the transmission through the lens and entrance window to be approximately 0.8. The wavelengths used to deposit polystyrene thin films were 3.31 µm (resonant with aromatic ring CH stretching modes) and 3.43 µm (resonant with CH2 asymmetric stretching modes). The spot size was measured using heat-sensitive paper, and was 0.006 cm2 for both the 3.31- and 3.43-µm depositions. The macropulse fluence, monitored before and after each deposition experiment, varied between 2.8 and 3.4 J/cm2 due to secular fluctuations in the FEL beam. Ablation at a wavelength of 6.25 µm (1600 cm−1 ) was also attempted at a similar fluence as above, but no film was deposited even after 1000 shots. The base pressure of our deposition chamber was 10−5 Torr or less. During the depositions, the pressure in the chamber typically rose to about 10−3 Torr, a testament to the efficiency of FEL ablation of polystyrene. The beam was rastered across the surface of the target at a frequency of approximately 15 Hz and the target was slowly rotated as well (∼ 2 Hz) to maintain a relatively even ablation profile across the surface of the drop-cast target. Broad-spectrum visible emission was observed from the plume during the depositions as reported previously for infrared-laser ablation of polystyrene [10]. Similar emission has been observed for a number of other polymers, including polyaniline [11], poly(tetrafluoroethylene), and poly(pyrrhole) targets. The source of this emission is unknown at this time because the spectrum was not resolved, but is probably due to electronic processes associated with bond breaking in the laser ablation plume. Such processes are not unique to laser ablation, and have been shown to be associated with all kinds of bond breaking, including simple mechanical bond breaking [12, 13]. The chemical structure and infrared spectrum of polystyrene are shown in Fig. 1. The bands near 3000 cm−1 divide neatly into two categories; above 3000 cm−1 (3.31 µm) the absorption band results from aromatic CH stretching modes in the phenyl rings, while below 3000 cm−1 (3.43 µm) the absorption is due to CH2 single bonds in the backbone. The spectral weight for the peak at 3.31 µm, computed by integrating the absorption spectrum, is 0.45 times that of the 3.43-µm peak, as indicated in Fig. 1. The ablation targets were prepared by melt casting a narrow molecular weight distribution poly-
The infrared spectrum of polystyrene near 3000 cm−1 . The chemical structure is shown along with the ratio of the spectral weights of the 3.31- and 3.43-µm peaks FIGURE 1
styrene powder obtained from Aldrich in 2.5-cm-diameter stainless steel dies on a hot plate. No other processing or special preparation of the sample material was undertaken. The deposited films were analyzed by size-exclusion chromatography (SEC) and matrix-assisted laser desorption– ionization (MALDI) mass spectrometry. For the SEC analyses, a Shimadzu LC-10AS pump was employed with a Phenomenex Prodigy ODS-2 column and a Spectroflow 783 UV detector; data were acquired using the Perkin-Elmer TurboChrom 4.0 software package. The mobile phase was tetrahydrofuran (THF); the injection volume was 0.5 µL and the flow rate was 0.5 mL/min. Five polystyrene molecular weight standards in the range 2330 – 29 300 g/mol were used in order to calibrate the column. MALDI mass spectra were acquired using a Voyager/DE STR time-of-flight reflectron mass spectrometer. Samples of polystyrene were eluted from the Si substrates on which they were deposited, using THF. The polymer was mixed with sil-
FIGURE 2 Yield (deposition rate) vs. number of laser shots. The average deposition rate for both wavelengths is indicated
BUBB et al.
Mode-specific effects in resonant infrared ablation and deposition of polystyrene
ver trifluoroacetate (Ag-TFA) and a dithranol matrix in a ratio of 1 : 10 : 360 (polymer : Ag-TFA : dithranol). The Ag serves to provide a source of cations for the subsequent desorption and ionization step. The solution was prepared in THF, then 2 µL of the sample was spotted onto a gold sample plate; evaporation of the solvent leaves behind a solid matrix of microcrystallites. The MALDI sample was then ablated and ionized using a nitrogen laser (337 nm) in the mass spectrometer; mass spectra shown in Fig. 3 were typically averaged over 10 laser shots. 3
Results and discussion
The normalized ablation yield per shot was measured by weighing sections of silicon wafers before and after deposition of polystyrene; the mass gain was divided by the substrate area and number of shots. In Fig. 2, the yield is shown for both 3.31- and 3.43-µm excitation wavelengths. Similarly to the results of [10], even though the absorbance at 3.31 µm is less than at 3.43 µm, the average yield at 3.31 µm is 1.5 times higher (26 vs. 17 ng/cm2 pulse). In addition to measuring the polymer deposition rate at each wavelength, we also measured the molecular weight distribution of our deposited films by MALDI and SEC. In Fig. 3, the MALDI results are shown. A small piece of the melted target was used as a standard and the average molecular weight is about 10 500 g/mol. MALDI spectra from targets deposited at both wavelengths clearly show a significant number of ions with smaller mass-to-charge ratios than the starting material; however, the relative yield of ions at mass-to-charge ratios around 5000 is about 50% higher (normalized to the central maximum of the ion distribution at m/z ∼ 10 500) for the 3.43-µm film. SEC was carried out in order to determine the molecular weight of the films as a function of deposition wavelength. These results are shown in Fig. 4. Samples A and B were
149
deposited at 3.31 µm and samples C and D were deposited at 3.43 µm. While all the deposited samples all have Mw ∼ 7000 g/mol with polydispersities around 1.15, the two films deposited using 3.31-µm irradiation have Mw values that are higher by 400– 500 g/mol. Sample D, deposited at 3.43 µm, shows significantly more low-mass counts than sample C. The results of this experiment can be summarized as follows: (1) PS molecules are ablated reasonably intact from a bulk PS target in both ring and backbone modes; (2) the lifetimes of the two modes differ by a factor of about three based on their respective absorption spectra, with τring ≈ 500 fs and τbackbone ≈ 150 fs; (3) the efficiency of the ablation as measured by relative yield is about 50% higher for excitation of the ring mode than for the backbone mode; (4) the FEL macropulse is thermally confined, but not pressure confined, implying that a laser-driven bipolar stress wave may participate in the ablation mechanism; and (5) thermal effects are present, but are not severe enough to damage the PS molecules or cause substantial cross linking or other denaturation effects. To understand these facts, we need to understand something of the composition of the PS target and its thermodynamic properties. For a polymer melt, with the number of monomer units N ∼ 102 , Flory theory predicts that the average volume per molecule is roughly that to be expected from the simple random-walk model, or ν ∼ 1.3 × 103 nm3 . Assuming a penetration depth at the indicated wavelengths of order 10−5 m, the total number of polymers in the IRabsorbing volume is of order 1012 . With respect to the laser excitation, two time constants are important. One is the thermal confinement time, τthermal = L 2p /Dthermal , and the other is the pressure or stress confinement time, τstress = L p /Cs , where L p is the optical penetration depth and Dthermal and Cs are respectively the thermal diffusivity and speed of sound. For polystyrene, these properties are given in Table 1. From the data, we find that the macropulse
FIGURE 3 MALDI mass spectra of the target and two films
150
Applied Physics A – Materials Science & Processing FIGURE 4 Size-exclusion chromatograms of four films. Samples A and B were deposited with 3.31-µm laser light and samples C and D were deposited at a wavelength of 3.43 µm
is not stress confined, but that it is thermally confined and that the micropulses are both stress and almost thermally confined. Given the poor thermal conductivity of PS, it is no surprise that both micropulses and the FEL macropulse are thermally confined; hence, the entire absorbed laser energy is converted into heat in the focal volume before the end of the macropulse. Also, given the stress-confinement times shown in Table 1, each polymer in the focal volume absorbs about 10 micropulses before the photomechanically induced stress is communicated to the ablation target outside the focal volume. The mass-spectrometer data show that the deposited polymers are preserved surprisingly intact during ablative deposition. Therefore, in conceptualizing the process mechanistically we have to account for how the polymers might be ejected intact into the gas phase. There are at least two distinct processes at work. In thermal equilibrium, due to the long macropulse, the temperature is raised by some tens of degrees. Since the viscosity satisfies the Volger–Fulcher law η B = exp , η0 T − T0
3.31 µm
Property Absorption coefficient (cm−1 ) Penetration depth (µm) Thermal diffusivity (cm2 /s) Speed of sound (m/s) τthermal = L 2p /Dthermal (ms) τstress = L p /Cs (ns) TABLE 1
(2)
3.43 µm
1290 1450 7.8 6.9 1.06 × 10−4 (at 410 K) 2200 (at 400 K) 5.7 4.5 3.5
Physical and optical properties of polystyrene
3.1
with B ∼ 710 and T0 ∼ 50 ◦ C, the rise of about 100 ◦ C during the macropulse means that the viscosity decreases by a factor of more than 100. At the same time, the polymer is expanding due to the rise in temperature in the focal volume, and PS molecules near the surface of the melt zone can reduce their free energy by uncoiling and getting into the gas phase due to competition between the entropic configuration energy and the gain in energy due to the loss of free volume in the coiled polymer [14]. If there is a stress wave, with propagation times of order nanoseconds in the drop-cast target, this could abet the ejection of surface molecules, as indeed shadowgraphs of the ablation plume seem to indicate. Given the relative lifetimes of the ring and backbone modes, and their vibrational frequencies (ν ≈ 1014 Hz), every time a photon is absorbed by a PS molecule, the localized anharmonic mode vibrates approximately one hundred times (for the backbone mode) or three hundred times (for the ring mode) before decaying thermally. The photon energy of 0.3 eV – and hence the energy of these anharmonic vibrations – is comparable to typical hydrogen-bond energies in polymers. Hence, the greater ablation efficiency of the ring-mode ablation can be viewed simply as a matter of greater integrated bond-breaking probability in this case; from the standpoint of statistical mechanics, the number of attempts made at breaking a hydrogen bond is three times as great for this mode. This line of argument is also consistent with the evidence from mass spectrometry that there is somewhat less fragmentation of the polymer when exciting the ring mode. It is appropriate to compare the experimental results presented here with previous mechanistic studies of wavelengthselective infrared laser induced photodesorption. In these experiments, some carried out with the free-electron laser FELIX and others with conventional table-top lasers, monolayers of small molecules such as NH3 and ND3 [15], N2 O [16], benzene [17], and CD3 F [16] were desorbed by
BUBB et al.
Mode-specific effects in resonant infrared ablation and deposition of polystyrene
resonant infrared radiation under ultra-high-vacuum and lowtemperature conditions. Typical fluences used in the desorption studies were on the order of 5 mJ/cm2 , a factor of 1000 less than used in our experiment. The present experiments report the ablation (rather than the desorption) and deposition of much larger molecules at room temperature and high vacuum. Moreover, the dynamics of tangled polymer chains being ejected from a bulk sample are much different to those involved in the monolayer desorption of small molecules. Thus, the only commonality linking these earlier experiments to ours is the demonstration that molecular bonds can be broken by resonant vibrational excitation, leading to ejection of the molecules into the gas phase. However, the inherent differences that exist between the two are such that a direct comparison of them is not particularly instructive. 4
Conclusions
The use of a tunable, picosecond free-electron laser to ablate polystyrene via each of two distinct vibrational excitations with virtually identical linear absorption coefficients offers a unique opportunity to unravel the effects of ablation by coupling to specific modes from ablation due to thermal vaporization. The former arise from the localized anharmonic excitations of the polymer chain, while the latter originate from non-specific transfer of absorbed photon energy to the phonon bath states. Since the absorption coefficients of the two modes are virtually identical, the equilibrium thermodynamic response of the polymer target must be roughly the same regardless of which mode is excited. Hence, we propose that the observed 50% difference in ablation efficiency in favor of the ring-mode excitation arises from its relatively longer lifetime. That longer lifetime leads to a higher number of bond-breaking trials per absorbed micropulse, translating in turn into a higher probability of breaking local hydrogen bonds and weakening the cohesive energy of the polymer melt. Given the micropulse intensity at which these experiments are carried out, and the bandwidth of the FEL micropulse, it is entirely possible that nonlinear or multiphoton effects also play a role. The photon flux in a micropulse (about 1013 photons deposited in a volume of order 10−6 cm3 ) is such that each −CH bond has a probability of about P ∼ 0.05 of being excited in a micropulse. This means that any given polymer has a significant probability of absorbing multiple
151
quanta during a single micropulse. It is tempting to suggest that such nonlinear processes are in fact occurring here and also influence the higher ablation efficiency. However, the drawing of firmer conclusions on this point must await other experiments that measure intensity-dependent ablation yields. ACKNOWLEDGEMENTS One of the authors (DMB) thanks J.M. Joseph and J. Hanson for useful discussions, and B. Ward for assistance with the manuscript. Research at Rutgers University–Camden is supported by the National Science Foundation under Grant No. DMI-0323621 and a Cottrell College award from the Research Corporation. Research on freeelectron laser ablation at Vanderbilt University is supported by the Medical Free Electron Laser Program administered by the Air Force Office of Scientific Research (F49620-01-1-0429) and by the Naval Research Laboratory (N00173-05-P-0059). We are grateful to the staff of the W.M. Keck Foundation Free-Electron Laser Center for expert operation of the light source.
REFERENCES 1 D.M. Bubb, J.H. Callahan, J.S. Horwitz, R.A. McGill, E.J. Houser, D.B. Chrisey, R.H. Haglund, M.R. Papantonakis, M. Galicia, A. Vertes, J. Vac. Sci. Technol. A 19, 2698 (2001) 2 D.M. Bubb, M.R. Papantonakis, B. Toftmann, J.S. Horwitz, R.A. McGill, D.B. Chrisey, R.F. Haglund, Jr., J. Appl. Phys. 91, 9809 (2002) 3 M.R. Papantonakis, R.F. Haglund, Jr., Appl. Phys. A 79, 1617 (2004) 4 Y. Tsuboi, M. Goto, A. Itaya, J. Appl. Phys. 89, 7917 (2001) 5 See G.B. Blanchet, J. Appl. Phys. 80, 4082 (1996) and references therein 6 R. Srinivasan, B. Braren, Chem. Rev. 89, 1303 (1989) 7 Y.G. Yingling, B.J. Garrison, Chem. Phys. Lett. 364, 237 (2002) 8 H.K. Park, R.F. Haglund, Jr., Appl. Phys. A 64, 431 (1997) 9 G.S. Edwards, D. Evertson, W. Gabella, R. Grant, T.L. King, J. Kozub, M. Mendenhall, J. Shen, R. Shores, S. Storms, R.H. Traeger, IEEE J. Sel. Top. Quantum Electron. 2, 810 (1996) 10 D.M. Bubb, M.R. Papantonakis, J.S. Horwitz, R.F. Haglund, Jr., B. Toftmann, R.A. McGill, D.B. Chrisey, Chem. Phys. Lett. 352, 135 (2002) 11 D.M. Bubb, S.M. O’Malley, C. Antonacci, R. Belmont, R.A. McGill, C. Crimi, Appl. Phys. A 81, 119 (2005) 12 J.T. Dickinson, Fracto-Emission, in Encyclopedia of Materials: Science and Technology, ed. by K.H.J. Buschow, R.W. Cahn, M.C. Flemings, B. Ilschner, E.J. Kramer, S. Mahajan (Elsevier, London, 2001) pp. 3254– 3256 13 J.T. Dickinson, S.C. Langford, S. Nakahara, L. Scudiero, K.W. Hipps, M.-W. Kim, N.-S. Park, Spatial and temporal probes of fracture, wear, and deformation, in Fractography of Glasses and Ceramics III, ed. by J.R. Varner, V.D. Fr´echette, G.D. Quinn (Ceram. Trans. 64) (American Ceramic Society, Westerville, OH 1996) pp. 193–256 14 R.A.L. Jones, Soft Condensed Matter (Oxford, New York 2002) 15 T.J. Chuang, I. Hussla, Phys. Rev. Lett. 52, 2045 (1984) 16 B. Redlich, L. van der Meer, H. Zacharias, G. Meijer, G. von Helden, Nucl. Instrum. Methods Phys. Res. Sect. A – Accel. Spectrom. Detect. Assoc. Equip. 507, 556 (2003) 17 R. Braun, P. Hess, J. Chem. Phys. 99, 8330 (1993)
