Temporal Evolution of an Electron-Free Afterglow ... - ACS Publications

Mar 8, 2008 - Department of Electrical Engineering and Electronics, UniVersity of ... and The School of Pharmacy, UniVersity of Nottingham, UniVersity...
0 downloads 0 Views 448KB Size
3938

J. Phys. Chem. B 2008, 112, 3938-3947

Temporal Evolution of an Electron-Free Afterglow in the Pulsed Plasma Polymerisation of Acrylic Acid Ian Swindells,† Sergey A. Voronin,† Paul M. Bryant,† Morgan R. Alexander,‡ and James W. Bradley*,† Department of Electrical Engineering and Electronics, UniVersity of LiVerpool, Brownlow Hill, LiVerpool, L69 3GJ, United Kingdom, and The School of Pharmacy, UniVersity of Nottingham, UniVersity Park, Nottingham, NG7 2RD, United Kingdom ReceiVed: October 29, 2007; In Final Form: January 10, 2008

By use of time and energy-resolved mass spectrometry, negative ions with masses ranging from m/z ) 1-287 amu have been observed in the afterglow of a low-pressure (10 mTorr) pulsed acrylic acid polymerizing plasma. The most intense peaks, seen at m/z ) 71, 143, 215, and 287, are assigned to the dehydrogenated oligomer of the form [nM-H]- for n ) 1, 2, 3, and 4, respectively. The results strongly suggest that both m/z ) 71 and 143 ions are produced in the on period of the pulse cycle (0.1 ms duration), with higher masses m/z ) 215 and 287 being produced by neutral ion chemistry in the off period (up to 40 ms in duration). The increase in the intensity of the [3M-H]- and [4M-H]- peaks in the off period is accompanied by a rapid fall in the concentration of [M-H]- ions and electrons, the latter decreasing from ∼1015 m-3 to zero within 150 µs. Deep into the afterglow, Langmuir probe measurements show that the charge species only consist of positive and negative ions, present at equal concentrations in excess of ∼1014 m-3 even after 10 ms that is, the plasma is wholly electron free. To describe the growth of large negative ions a number of possible ionneutral chemical pathways have been postulated, and a calculation of the ambipolar diffusion rates to the walls suggests that, in the off period, the positive and negative ion contribution to the deposition rate is small (∼1%) compared to the net total deposition rate. However, the observations do indicate that it may be necessary to update models of film growth in the pulsed plasma polymerization of acrylic acid to account for negative ions.

1. Introduction Plasma polymer deposits are of great scientific interest as a versatile approach to study and control surface chemistry1 and have recently achieved considerable technological importance in application areas as diverse as cellular therapy2 and corrosion and protection of automotive components.3 The highly adherent pinhole-free deposits may be readily formed on surfaces from plasmas struck in a range of volatile organic compounds to achieve many different types of chemistry.4,5 For most organic compounds with a significant vapor pressure under vacuum, it is possible to strike a discharge in the pure monomer, and for low-input powers (a few Watts), the monomer functionality can be retained in the deposit.6 Through important but empirical research, it has been shown that in low-power continuous-wave (CW) radio-frequency (RF) plasmas, reduced monomer fragmentation in the discharge leads to high functional group retention in the films.6-10 In a refinement of the technique, it has been found that for a host of different monomers good control over the surface chemistry can be achieved by pulsing the RF power,11-19 with long off periods toff (milliseconds) and short on periods ton (in the hundreds of microseconds range). The short duration of the on period is sufficient to produce radicals and charged species in the plasma necessary for the polymerization to proceed (primarily in the off period). Pulsing allows the time-averaged power dissipated in the plasma to * To whom correspondence should be addressed. † University of Liverpool. ‡ University of Nottingham.

remain low, comparable to that of CW operation, so attaining high functional retention in the film. It has been shown in a number of studies that the duty cycle, γ ) ton/(ton + toff) is the key factor in determining film chemistry.13,19 In the latter work, it was demonstrated for acrylic acid that the level of COOH functional retention was a unique function of the ratio of the time-averaged power (on-period power P × γ) and the flow rate Φ of fresh monomer into the system. This ratio Emean ) γP/Φ is the energy in the plasma per resident monomer species. Emean is related to the Yasuda parameter via Y ) Emean/M, where M is the molar mass.20 The finding in ref 19 showed explicitly that quoting RF power is misleading, and it is the energy absorbed by each molecule during its residence time in the vessel that is important in relating film chemistry to plasma properties. In plasma polymerization, the relative importance of the neutral and positive plasma species in deposit formation has received considerable attention, much of it speculative, based upon observations of surface chemistry. Yasuda5 has described plasma polymerization as proceeding through a rapid stepgrowth polymerization mechanism (RSGP). More recently, mass spectrometer measurements in CW and pulsed polymerizing plasmas have shown the presence of large oligomer-positive ions, resulting in the proposition of an ion-neutral chain polymerization mechanism in the acrylic acid6 and other systems, e.g., siloxanes.21 In pulsed polymerization, since the deposition rate is greater than would be anticipated by simple scaling of CW deposition

10.1021/jp7104117 CCC: $40.75 © 2008 American Chemical Society Published on Web 03/08/2008

Temporal Evolution of an Electron-Free Afterglow to the on period, it has been assumed that grafting of neutrals onto long-lived surface-free radicals dominates. Such a mechanism has been proposed largely because it is assumed that ionic species have very short lifetimes (i.e., a few microseconds). This assertion is discussed for polystyrene deposition by Friedrich et al.17 However, it has been shown, using a sensitive ion flux probe, that in CW deposition of acrylic acid the positive ions themselves can be responsible for a significant fraction (22% of the mass) of the deposit.10 Interestingly, O’Toole et al.6 and Haddow et al.9 demonstrated using (time-averaged) mass spectrometry that heavy positive ion oligomers of the form [mM+H]+ (plus their dehydrated derivatives) can be detected during CW acrylic plasma operation, and also Fraser et al.12 and Haddow et al.11 found similar results in the pulsed regime. Voronin et al.19 showed by correlating time-resolved mass spectroscopy measurements of the ion energies to the plasma potential that these positive ion oligomers (up to mass m/z ) 217) originated in the plasma and were not an artifact of the detection system. It was also revealed that these heavy ions exist both in the on and off periods of the discharge and that the heavier species, n ) 3 and 4 of the series [nM+H]+, grow as a result of positive ion-neutral gas-phase chemistry. Langmuir probe measurements of the total ionic flux indicate that they contribute up to about 10% of the mass of the deposit19 in general agreement with ref 10. In pulsed plasma polymerization, time-resolved Langmuir probe and mass and energy analysis has been used to yield plasma parameters, (ionic and electron concentrations, plasma potentials, energies, and relative fluxes of species, etc.) in pulsed isopropyl alcohol18 and in acrylic acid19,22-24 plasmas. In ref 19, it was established that the heavy oligomer ions, created primarily in the off period, can persist for up to 1 ms. Here, we present new evidence that both heavy positive and negative ions exist in the plasma for much longer times and in significant concentrations. Since the acrylic acid monomer contains species that readily form negative ions in other plasmas, such as electronegative oxygen,25 it is easy to envisage the existence of negative ions in the discharge. However, in nearly all acrylic acid studies (both chemistry and film-growth studies) the presence of negative ions has been neglected, perhaps because their detection was not achieved. Negative ions have been postulated to be important in other deposition plasmas where they have been detected, particularly Si-based monomers, where very large negative ions are formed in the gas phase leading to cluster formation.15 Recent developments in mass-spectral sampling techniques have enabled us to observe the temporal evolution of the relative fluxes and absolute energies of negative ions in pulsed acrylic acid plasma.26 These investigations have already shown mass spectral negative ion intensities (uncalibrated) at least as large as those of the positive ions. These ions have the form [nMH]- (i.e., n multiples of the monomer unit minus a proton), together with families of associated dehydrated forms. As we shall show here, these are generated by ion-neutral chemistry in the gas phase predominately in the off period. 2. Experimental 2.1. The Pulsed Plasma System. The acrylic acid plasma was sustained inside a 4-way glass vessel by applying RF power at 13.56 MHz on to 4-turn (unterminated) coil electrode wrapped around the vessel. The capacitive coupling was achieved via a matching network. Degassed acrylic acid monomer (CH2dCHC(dO)OH, m/z ) 72, >99% purity, Sigma Aldrich) was introduced to the evacuated reactor chamber through a needle

J. Phys. Chem. B, Vol. 112, No. 13, 2008 3939 valve. The working vessel pressure was set as 10 mTorr (1.3 Pa) before plasma ignition, and this fell to around 6.5 mTorr (0.85 Pa) once the plasma was struck. The exact final pressure depends on the degree of fragmentation of the monomer, which in turn depends on the frequency, power, and gas flow rate. Monomer flow rates of 1.5 and 4.8 sccm were used, calculated, and controlled as described in ref 19. By use of a pulse generator the plasma on period was fixed at 0.1 ms, while the off period was varied from 0 (CW) up to 40 ms using a delay generator. Here, two (on period) powers were used, namely, 15 and 50 W. A diagram of the experimental set up is shown in Figure 1, together with the signal lines for the mass spectrometer and Langmuir probes and also the instrument trigger waveforms. As a time reference, we take time t ) 0 as the start of the on period, with the off period (afterglow) beginning at t ) 100 µs. 2.2. Plasma Mass Spectrometry. The instrument used here was the HIDEN Analytical Ltd EQP 300 energy-resolved mass spectrometer. A “double-gating” technique was used to timeresolve the ion measurements. The technique can deliver a 10µs time-resolution for positive ions,19 and it has already been applied to pulsed acrylic acid24,26 and pure argon27 plasmas. To observe negative ions with this system, the mass spectrometer extractor and beam line potentials are reversed. In this mode the ions are sampled from the plasma bulk by applying a constant positive bias of +65 V to a specially made “puller” electrode, which has a 100-µm orifice at the center. This electrode is situated behind an outermost-earthed electrode end cap with a 4-mm diameter-sampling orifice, see Figure 2. Ions accelerated into the instrument are then allowed to enter the beam line at specific times (for temporal resolution) by a rising pulse on the extractor electrode (raising its potential from -30 to +180 V. The mass spectrometer configuration is however inefficient at sampling negative ions in the driven phase of the discharge. This is a well-known phenomenon in negative ion detection for electronegative plasmas. In the on period the +65 V on the “puller” is only + 5 V above the typical plasma potential (+60 V), and this is insufficient, given the weak penetration of this field into the plasma, to extract many, if any, negative ions. Higher extractor biases cannot be used to solve this problem, as they may perturb the plasma globally and raise the plasma potential. Not only this, in the on period, the low temperature of the negative ions (around 0.1 eV) confines them to the discharge center away from all sheath boundaries. If there are sufficient numbers of negative ions, the plasma becomes structured, with the formation of an electronegative core surrounded by an electropositive halo.28 However, in the off period, when the electric potential across the plasma that confines negative ions to central region collapses (the boundary sheaths collapse) then negative ions are readily accelerated into the mass spectrometer ion-optical system by the + 65 V constant “puller” bias. It should be noted that in quadrupole-based mass spectrometers there is a strong mass-dependent transmission function, with the transmission of ions decreasing strongly with increased mass, m.29 The instrument manufacturer has found empirically that the instrument function lies between m-1 and m-2. In this work, we take a conservative estimate of m-1 and all detected ions signals (including neutrals ionized in the mass spectral ionization unit) have been corrected for this effect. 2.3. The Langmuir Probe System. The Langmuir probe consisted of a 1 cm long tungsten wire of radius 250 µm emanating from a ceramic barrel with a heating coil around it

3940 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Swindells et al.

Figure 1. The experimental system including the Langmuir probe and mass spectrometer together with the triggering voltages. Symbols: Upulse is the signal defining RF on period, Urf is the excitation voltage on the driving coil, Ustrobe defines the extractor potential pulse duration, Ug is the extractor pulse waveform, and Ude is the pulse enable on the SEM detector.

Figure 2. Schematic of the orifice and end-cap arrangement used for the detection of negative ions. Isolation of the orifice by insulating spacers (gray) allowed a constant +65 V to be applied to the orifice, while the end cap and spectrometer barrel remained at ground potential. The extractor is gated (-30 to + 180 V) to allow time-resolved measurements of the negative ions.

made of platinum wire. The tip was positioned at the center of the vessel, 3 cm in front of the mass spectrometer orifice (see Figure 1a), and could be retracted out of the plasma to reduce deposition build up on its tip between measurements. Since this is a depositing plasma, the probe tip was heated between measurements to remove any insulating layer. The plasma parameters (ionic and electron densities, electron temperature, plasma and floating potentials) were obtained with a 10-µs time resolution, using a time-resolved acquisition system.27 Since we now know the plasma to be electronegative, an appropriate cylindrical probe theory must be used to obtain the electron positive and negative ion densities from the measured

probe voltage-current characteristic. In the on period the plasma is not expected to be highly electronegative so that the plasma potential, floating potential, electron density, and temperature can be obtained by the usual methods (see Supporting Material). The plasma electron temperature (kTe/e) was obtained from the inverse slope of the logarithm of the probe current in the retardation region (between plasma and floating potentials); the plasma potential (Vp) was found from the zero crossing of the second derivative of the total probe current. Finally, the electron density was obtained from the electron thermal current at Vp. In the plasma on period, due to significant RF excitation of the plasma potential Vrf (with Vrf > kTe/e)22 the probe was actively compensated, as described further in ref 19. It was not necessary to compensate the probe in the off period since Vrf drops to zero in a few microseconds after the end of the on time. To measure the negative ion parameters the often-used modified Druyvesteyn method30 was found not to be applicable in our case because of the unacceptable level of noise prevalent in the second derivative of the probe characteristic. An alternative method based on the theory of Allen, Boyd, and Reynolds (ABR31) can be used to obtain the negative ion density during the on period. This is done by fitting the (collisionless) cold ion radial motion approximation developed by Amemiya et al.32 to the measured ion current. This method has been adopted here. However, the theoretical curves consistently overestimated the measured ion current despite varying the fitting parameters over a wide range. This may be due to ion-neutral collisions in the probe sheath, which reduces the collected ion current to below that obtained for a collisionless sheath.25 Even at these low pressures the collision cross section for these large molecular complexes would be expected to be large in comparison to atomic ions. During the on period it was not possible to obtain

Temporal Evolution of an Electron-Free Afterglow

Figure 3. Typical time-averaged neutral (a), positive (b), and negative (c) ion mass spectra of an RF acrylic acid plasma with a 0.1-ms onperiod duration and 10-ms off-period duration.

the negative ion densities, and only the electron density and temperature could be determined. During the plasma off period, once the electrons have diffused out of the plasma, the negative and positive ion densities could be obtained from the probe characteristic. Because of the lack of experimental data for ion-neutral collision cross sections, it is not possible to determine the degree of collisionality of the sheaths adjacent to the probe and the applicability of ion collection theories. However, it is well-known that the collisionless cold ABR theory31 gives better agreement than the alternative orbital motion theory for cylindrical probes even in sheaths that are transitional between the collisionless (kinetic) and fully collisional (mobility limited) regimes (see ref 33 and references therein). Therefore it has been used here. Because of the absence of electrons, the negative ions were taken to be the retarded species when fitting to the measured positive ion current. To fit the measured ion current to a theoretical curve, the ionic masses and positive (Tp) and negative ion (Tn) temperatures are required. In this approach the accelerating species are assumed to be essentially cold with kTp/e ≈ kTn/e ≈ 0.15 eV, which is consistent with other studies of electronegative plasmas.30,34 The weighted average positive ion mass mp was taken from Figure 9 of ref 19 to be 211 amu. The average negative ion masses mn for different operating conditions are calculated in section 3 (Figure 5). It should be noted that the derived ionic densities in the off period are expected to be underestimated because of the use of the collisionless ABR theory. 3. Results 3.1. Time-Averaged Mass Spectroscopic Measurements. Figure 3 shows an example of the time-averaged neutral, positive, and negative ion mass spectra taken from ref 26 for an off period of 10 ms. The applied power in the on period (0.1 ms) was 50 W, and the monomer flow rate was 4.8 sccm. Measurements were made with the extraction orifice 3 cm from the geometric center of the vessel. The principal peak in the neutral spectrum (m/z ) 72) is that of the acrylic acid molecule after undergoing ionization in the electron bombardment chamber of the mass spectrometer (M + e- f M+ + 2e-). Assignments and discussion of the positive and neutral species

J. Phys. Chem. B, Vol. 112, No. 13, 2008 3941 have been published before6,11 and are annotated in Figure 3. In the positive ion spectra, the series [nM+H]+ was apparent at m/z 73 [M+H]+, 145 [2M+H]+, and 217 [3M+H]+ with associated peaks at m/z 55, 127, and 199 representing the loss of water. Negative ions up to m/z 287 are detected (mass limit of the spectrometer is 300 amu), with intense peaks at m/z 143, 215, and 287, which correspond to the series [nM-H]- with n ) 2-4, i.e., the subsequent additions of the monomer unit to the monomer after proton loss. Figure 4 shows in more detail the time-averaged spectra for negative ion families for a variation in applied power (15 and 50 W) and off-period durations (from 0.5 to 20 ms) at 1.5 sccm flow rate. Here the n ) 1 peak in the system [nM-H]- can be seen. It is clear that their time-averaged relative intensities are strongly influenced by the duration of plasma off period. For 20-ms off periods (Figure 4f), we see mainly high mass negative ions centered on m/z ) 215. At shorter off periods, more low mass peaks are observed. The intensities of the negative ion signals are comparable to those measured for the neutral and positive ions, see Figure 3, indicating similar quantities of negative ions. However, these are uncalibrated fluxes, and one must take care: these numbers should not be compared directly due to different extraction and transmission probabilities through the orifice and mass spectrometer system. The m/z ) 71 peak can be assigned to the monomer molecule after undergoing heterolytic cleavage of a proton. In agreement with refs 19 and 26, we see no neutral species with masses greater than m/z ) 72; thus formation of the larger species through a radical polymerization mechanism followed by electron attachment is thought to be unlikely. Successive reaction of the negative [M-H]- ion with neutral monomer may be postulated to produce the products [2M-H]-, [3M-H]-, and [4M-H]-. Comparison with conventional anionic solution polymerization35 supports such a mechanism. A likely formation route is shown in Scheme 1. Depending on the initiation molecule, that is, the product of 1A or 1B, different structures may be postulated for m/z 143 and 215, which cannot be differentiated on the basis of the mass spectra alone. Additional peaks were observed at 26 m/z units below the [nM-H]- peaks at m/z 45, 117, 189, and 261 respectively, that may be assigned using the following structures: [nM-H-C2H2]-. These may arise from rearrangement or polymerization products through loss of C2H2 or reactions with smaller fragments, for example, the formate ion (Scheme 1, 1-4). A structure may be drawn to account for the other significant peak, at m/z 163, which may arise from small fragment combination as for the m/z 117 species (Scheme 1, 5). Further work using different monomer structures or isotopic labeling will be necessary to provide firmer assignments to reach definitive conclusions. From the mass spectroscopic data in Figure 4, the average negative ion mass has been calculated as a function of power P, off period, and flow rate Φ, and this is shown in Figure 5. The average mass (taken to be in amu units with z ) 1) increases with flow rate and with increasing off time. As the energy per monomer (Emean ) γ P/Φ) is decreased the average mass increases to over 200 amu. The dependence on power is weaker, with the average mass decreasing with power as the monomer units are fragmented and is only significant with short off times (5 ms, after which their concentrations grow and dominate the negative ion content of the plasma. These timeaveraged results indicate strongly that mass m/z ) 71 and 143 ions are created in the on period and decrease in the off period through ion-neutral reactions to produce heavier species, namely, m/z ) 215 and 287. It appears that off periods in excess 2 ms are necessary for this to happen. To investigate this phenomena further, time-resolved mass spectroscopic and Langmuir probe measurements have been made as detailed below. 3.2. Time-Resolved Mass Spectroscopic Measurements. The temporal evolution of the negative ion fluxes (un-normalized) for m/z ) 71, 143, 215, and 287 at 3 different off periods (2, 10, and 40 ms) are shown in parts a-c of Figure 8. No data can be collected from the on period due to the inefficiency of negative ion extraction. The flow rate and power were 1.5 sccm and 50 W, respectively. The form of the curves in parts a-c of Figure 8 are similar, with the most striking feature that species m/z ) 215 and 287 persist for long times into the afterglow, in fact, up to at least t ) 30 ms as shown in Figure 8c. We did not attempt to measure species at m/z ) 71 or 143 in Figure 8c as their signals are diminishingly small over most of the off time. From parts a and b of Figure 8, it is clear that, at the start of the off period, both m/z ) 71 and m/z ) 143 are remnants from the plasma on time. The latter falls slowly, whereas the former falls rapidly to undetectably small levels in only 150 µs. This is the same time scale for the electron loss from the system. Both m/z ) 215 and 287 ions have undetectably small initial values at the start of the off period, but their fluxes grow (in several hundreds of microseconds) to significant values, which subsequently decay slowly with characteristic decay times of about 5-10 ms. 3.3. Langmuir Probe Results. Langmuir probe characteristics have been obtained both in the on time and afterglow for a fixed off period of 40 ms and same operating conditions as in Figure 8. A number of these, with accompanying explanation, are shown in parts a-c of Supporting Figure S1. During the on period (Figure S1a), the electron current dominates the characteristic, with its value at the visible knee of the characteristic being about 2000 times larger than the ion current (∼1 µA). This asymmetry of the Langmuir probe characteristic indicates that the plasma in the on time may not be strongly electronegative. At the beginning of the off period the electron current decreases rapidly (by 2 orders of magnitude) within 150 µs, to a point where the probe current-voltage traces become approximately symmetrical, see Figure 9. In highly electronegative plasmas this usually indicates the dominance of heavy negative ions and the absence of electrons.30 The time scale for loss of electrons is similar to that of m/z ) 71 negative ions as described in section 3.2. After about 15 ms into the afterglow the positive and negative ion currents have decreased to the level of the signal noise background. By contrast, the more sensitive mass spectrometer measurements shows negative ions persisting up to 30 ms. At t ) 1 ms, the symmetric characteristic in Figure 9 yields a positive ion density of np ) 9 × 1014 m-3 (assuming mp ) 211 amu) and a negative ion density of nn ) 1.2 × 1015 m-3 (assuming mn ) 287 amu). A least-squares cubic spine-fitting algorithm36 has been used to smooth the data allowing ABR theory to be implemented. From the characteristics, the ionic densities in the afterglow have been calculated, and these are

3944 J. Phys. Chem. B, Vol. 112, No. 13, 2008

Swindells et al.

Figure 6. Time-averaged negative ion energy distribution functions of mass-to-charge (m/z) peaks at 71, 143, 215, and 287. The off periods vary from 0.5-10 ms at a flow rate of 4.8 sccm and an on-period power of 50 W. There is clearly some breakthrough of m/z 71 ions at +60 V (and lower). These may be from the on-period plasma.

Figure 7. Time-averaged integrated counts I ) ∫ f(E) dE (normalized to the maximum count rate for m/z ) 71) at 50 W, 0.5 ms off period for m/z ratios of 71, 143, 215, and 287 as a function of off-period for different time-averaged powers.

shown in Figure 10. Note that these values are expected to be underestimated due to ion-neutral collisions in the probe sheath (see section 2.3). It was not possible to obtain negative ion densities for times t < 150 µs due to the presence of a significant electron signal in the characteristics. However, the derived on period electron densities (between 1 and 1.7 × 1015 m-3) and temperature (6-8 eV) are shown in Figure S2.

Although the negative ion density falls in the afterglow to values of 2 × 1014 m-3 (after 10 ms), it still represents a significant fraction of the plasma density during the on period. For times t > 0.5 ms, the negative and positive ion densities merge to the same value reflecting the quasineutral nature of the plasma in an absence of electrons. In the early afterglow, t < 0.5 ms, the negative ion density exceeds that of the positive

Temporal Evolution of an Electron-Free Afterglow

J. Phys. Chem. B, Vol. 112, No. 13, 2008 3945

Figure 9. Typical Langmuir probe current-voltage characteristic measured at 1000 µs. The dotted line is the least-squares cubic spine fit and the solid line is the theoretical ion current giving densities of 1.1 × 1015 m-3 (for negative ions of weighted mass 287 amu) and 0.9 × 1015 m-3 (positive ions of weighted mass 212 amu).

Figure 10. The negative and postive ion densities against time during the afterglow derived from fitting the theoretical ion current to the smoothed time-resolved Langmuir probe characterstics.

Figure 8. Time-resolved integrated counts I ) ∫f(E) dE for m/z ) 71, 143, 215, and 287 for an on time of 0.1 ms and off periods of (a) 2, (b) 10, and (c) 40 ms. In the last case only 215 and 287 were measured due to the small signal of 71 and 143 at long times. The flow rate is 1.5 sccm and 50 W power.

ion density, which rapidly decreases, and this is likely to be due to a very small remnant of electrons in the probe characteristic distorting the results. At t ) 0.3 ms, there is a peak in the positive ion density, which agrees with our assertion in ref 19 that the heavy positive ions are created in the afterglow. From a plot of the natural logarithm of the negative ion density data (figure S3) we find a 2-fold decay rate: initially a 3-ms time constant, slowing to 5.6 ms in the deep afterglow. This is in general agreement with the mass spectrometry results; see parts a-c of Figure 8 where the long-term decay constants of mass selected negative ions are 5-10 ms. Figure 11 shows the temporal evolution of the plasma potential Vp and floating potential Vf. During the on period Vp gradually decreases from 66 V at t ) 0 µs to 52 V at t ) 90 µs. At the beginning of the afterglow (t ) 100 µs) both Vp and Vf

Figure 11. The plasma and floating potential as a function of time in the afterglow derived directly from the Langmuir probe data. Insert: note that the floating potential during later times are negative.

decrease rapidly with Vp falling to 0.93 V after 120 µs and Vf falling to ∼-0.4 V after 150 µs. It was no longer possible to determine Vp after 120 µs since the probe characteristic became highly symmetrical. However, it is likely that Vp ≈ Vf due to the nearly equal positive and negative ion fluxes, and this is assumed here. The difference Vp - Vf can be estimated assuming an ideal planar probe in collisionless conditions (see supporting material). Because of the slightly higher average negative ion mass, as observed here, the difference Vp - Vf is about 0.4 V (with kTn/e ) 0.15 eV) in contrast to an electropositive plasma in argon where Vf would be about -15 V below Vp for kTe/e ) 3 eV. Although the ion energies in Figure 6 never fall below 6

3946 J. Phys. Chem. B, Vol. 112, No. 13, 2008 V (section 3.1), the data here illustrates that Vp does actually collapse, and the ion energies in Figure 6 are artificially high. 4. Discussion The mass spectral and Langmuir probe results provide evidence that not only are negative ion concentrations significant in the afterglow but heavy ionic species are being created through gas-phase ion-neutral reactions in the absence of electrons. This idea is highlighted by the observation that the m/z ) 215 and 287 fluxes increase at the expense of m/z ) 71 and 143 as time into the afterglow increases. The data in Figure 7 would suggest that the necessary reaction time for such reactions is at least 2 ms, and such reactions cannot proceed for off periods with reaction times less than this. The data in Figures 3-5 suggests that lowering the time-averaged energy per molecule, Emean, leads to less fragmentation of m/z ) 71 and 143 ions in the on period and subsequently more heavy ions in the afterglow. By use of simple assumptions it is possible to develop a stepwise phenomenological model of negative ion-neutral chemistry and then test it against the mass spectral results. This will be done in a follow-up paper. It is useful however, to postulate here on the important polymerizing reactions that may be occurring as part of our general discussison of the results in section 3. For instance, we can say that, in the on period, light negative ions are formed through dissociative electron attachment of the monomer unit M (m/z ) 72). These ions then may react with the monomer to form larger ions. It appears [M-H]and [2M-H]- are stable against fragmentation in the relatively warm plasma on period (electron temperatures measured here of ∼6 and 3 eV in ref 19). Dissociative electron attachment has previously been reported for carboxylic acid groups for electron energies above 1.25 eV;37 however, energetic electrons may fragment larger ions, and these are not detected at the start of the off period. There will also be electron detatchment leading to an equilibrium concentration of m/z ) 71 and 145. Therefore, the production mechanisms in the on period are

M + e- f [M-H]- + H [M-H]- + [M] f [2M-H]In the afterglow (t > 100 µs) electrons are rapidly lost from the plasma. This may be through two mechanisms: firstly the loss of fast electrons through the collapsing sheaths to the walls and secondly through their attachment to fresh monomer molecules to form [M-H]-. Because the electron density rapidly decreases and also kTe/e falls to values below 0.2 eV within about 150 µs in the off period,19 electron-induced impact processes leading to fragmentation may be greatly reduced compared to that in the on time. By assumption of stepwise ion-neutral reactions, the most important production reactions in the off period are

M + e- f [M-H]- + H [nM-H]- + M f [(n + 1)M-H]- n ) 1, 2, 3, etc. In our experiments, we were only able to detect species up to n ) 4 due to the limitations of the mass spectrometer; however, heavier negative ions may exist in the decaying plasma. 4.1. The Ionic Fluxes to the Substrate. For diffusiondominated transport, the flux Γ of postive and negative ions (density n) to the walls (separation 2L) is given by the ambipolar flux equation, Γ ) -Damb ∇n ≈ Damb (n/L).38 In the afterglow, given the absence of electrons, the ambipolar diffusion coef-

Swindells et al. ficient Damb can be written as Damb ) (µnDp + µpDn)/(µn + µp), where Dp, Dn, µp, and µn represent the postive and negative diffusion and mobility terms, respectively. In general, for ionic mass mj and temperature Tj we have Dj ) kTj/(mjνj), where νj is the ion-neutral collision frequency. Since it is reasonable to assume that Tn ≈ Tp and µp ≈ µn and given the average negative ion mass exceeds that of the postive ions, then Damb ≈ Dn ≈ Dp. It should be noted that, as far as the authors are aware, there is very little data on the collisional processes of the chemical species present in these polymerizing plasmas. Consequently, the collision frequencies and ionic mobilities are unknown. Despite this, if we lump all the ions together to determine the average effect, we can write Γ ≈ nVD where VD ) Damb/L is the average diffusion velocity. Given an average diffusion time scale of τ ) L2/Damb, the relation Γτ ≈ nL is obtained. From the data in Figure 8 the diffusion time scale must be at least τ ) 10-2 s. Therefore, using n ) 2 × 1014 m-3 and L ) 2.5 × 10-2 m, we find a flux Γ ≈ 5 × 1014 m-2 s-1 to the walls and substrate. Since the positive and negative ions arrive at the walls at the same rate, this implies a total mass depostion by all ions in the off period of 0.32 µg m-2 s-1, which is small (∼1%) compared to the average net depostion rate over the cycle of 10-50 µg m-2 s-1.19 This analysis yields an average diffusion coefficient of ∼6 × 10-2 m2 s-1, giving an effective collision frequency of 1.2 × 106 s-1, which is typical for low-pressure discharges.38 Since the mass spectroscopic measurements suggest that the weighted-average negative ion mass exceeds that of the postive ions (mn > mp), with Tn ≈ Tp, the ambipolar electric field Eamb across the plasma in the afterglow is approximately Eamb ≈ (kBTp - kBTn)/eL ∼> 0 being very weak.38 This is consistant with the measured Langmuir probe floating potential, which is approximately zero, actually being about 0.4 V below ground potential (see insert Figure 11). The slightly negative probe floating potential suggests that the negative ion flux is slightly higher than the positive ion flux. The weak ambipolar electric field (with no requirment for a conventional positive ion sheath to form) then promotes negative ion transport to the walls so that the net current there is approximately zero. The foregoing analysis also suggests that the plasma potential is also likely to be near ground with the potential drop across the plasma approximately given by V ≈ (kBTp - kBTn)/e ≈ 0, being slightly positive (