Evaluation of helium–argon mixed gas plasmas for bulk and depthresolved analyses by radiofrequency glow discharge atomic emission spectroscopy Matthew L. Hartenstein, Steven J. Christopher and R. Kenneth Marcus* Department of Chemistry, Howard L . Hunter Chemical Laboratories, Clemson University, Clemson, SC 29634-1905, USA Received 24th February 1999, Accepted 27th April 1999 Studies were performed to determine the practical benefits of mixed discharge gases (Ar and He) in the bulk and depth-resolved analysis of solids using a radiofrequency glow discharge atomic emission source.This study examined the characteristics of analyte emission intensity and yield, sample sputter rate and crater shape as a function of added He gas, for both conductive and non-conductive sample matrices. In comparison with pure Ar plasmas, the addition of He does not ultimately improve the limits of detection in the bulk analysis of conductive (metallic) solid samples. However, non-conductive sample matrices such as glasses, which are greatly aVected by discharge gas pressure to maintain sputtering conditions (crater shape, sputter rate), may benefit from the addition of He to the plasma gas as a means of enhancing the excitation conditions.The shape of the sputtered craters was minimally aVected by the addition of He, indicating that Ar partial pressure is the parameter that most critically aVects sputtering characteristics. Overall, the addition of He to the discharge plasma has been found to enhance analyte emission intensity without significantly influencing sputtering characteristics.In terms of depth-resolved analyses, optimized discharge conditions (specifically pressure) derived for the pure Ar case, which are sometimes accompanied by a loss of analytical sensitivity, can be augmented by the addition of He to the discharge to yield improved analytical responses while retaining the desired sputtering characteristics.for MIPs employed as gas chromatographic detectors.7 Argon Introduction has been the dominant plasma gas in analytical GD devices Advances in materials development have illustrated distinct for several reasons. Argon ions are eYcient sputtering agents, advantages which may be gained from the use of coatings the plasmas generate high electron temperatures and densities (paint, PTFE) and multi-layered (optics, semiconductors) sys- and the metastable energy levels of 11.5 and 11.7 eV are tems.The radiofrequency glow discharge atomic emission suYcient to ionize a majority of the elements in the Periodic (rf-GD-AES) source approaches the ideal for characterizing Table. Recently, as applications have pushed analytical requirethese types of materials as sample composition and thickness ments beyond traditional capabilities, there has been renewed information may be generated quickly and reliably in a single interest in alternative plasma gases. For example, helium oVers run.The duality of the GD plasma as both an atomization the potential to satisfy many of the needs as it yields a higher (sputtering) and excitation source makes depth-resolved analy- energy plasma than does argon. This advantage is especially sis possible. However, these processes within the plasma are significant to the analysis of hard to excite elements such as intertwined and cannot be independently controlled by the the non-metals. parameters aVecting them, specifically the applied power and The possible advantages of the use of alternative discharge discharge gas pressure.Because the ionization potential, the gases in GD plasmas have been illustrated by earlier studies; presence of metastable energy levels and atomic (molecular) however, the potential benefits to depth profile analysis, in mass of the discharge gas are determining factors in discharge particular, have not been examined extensively. Wagatsuma atomization and excitation characteristics, it is reasonable to and co-workers have conducted several studies involving the expect that there may be instances where diVerent gases, or use of alternative gases in direct current (dc) GD-AES,8–10 mixtures thereof, may be employed to advantage. illustrating that the emission spectrum of a plasma is strongly While argon has been the dominant discharge gas employed dependent on the discharge gas employed.For example, helium across the range of spectrochemical plasmas [i.e., arcs, sparks, and neon were found to increase significantly the intensities inductively coupled plasmas (ICPs), direct current plasmas of analyte lines, both ionic and atomic, regardless of sputter (DCPs), microwave induced plasmas (MIPs) and glow dis- rate.8,9 The eVects, both positive and negative, are strongly charges], gases other than argon have been of interest for element and transition dependent.Enhancements in ionic some time.For example, helium-based ICP studies were con- emission intensities were attributed to charge transfer between ducted by Reed nearly 40 years ago.1 In the area of ICP plasma gas (Ar, He, Ne) ions and analyte atom species, spectrometry, alternative gases have been employed as means varying with the respective energetics of the noble gas and of removing spectral interferences in ICP-MS, providing analyte ion states. The advantageous use of Ne as a means of enhanced matrix volatilization, enhancing excitation con- enhancing F (I ) emission illustrates the possible analytical ditions and improving the capacity to tolerate organic solu- utility of diVerent discharge gases.10 In the field of glow tions.2–5 Nitrogen has found acceptance as the plasma gas for discharge mass spectrometry (GD-MS), Woo et al.found that the addition of even small amounts (4%) of He to the discharge high power (>500 W) MIPs,6 while helium is the standard J. Anal. At. Spectrom., 1999, 14, 1039–1048 1039plasma increased the ionization eYciency by as much as 25 conditions yield the most desirable crater shapes: for the conductive (steel ) sample, an Ar base pressure of 5 Torr, and times.11 Teo and Hirokawa12 compared depth profiles of metal coatings obtained using pure argon and an argon–helium for the non-conductive (glass) sample, an Ar base pressure of 3 Torr.Sputter rate and spectral data were collected at these mixture in a Grimm-type discharge.The presence of He in the discharge gas was found to enhance the emission intensity for pressures and each subsequent pressure of added He above this pressure (3 Torr intervals equaling total pressures of 8, elements having a simple energy level structure (Cu and Zn in this case), but not for Ni, which has a much higher density of 11 Torr, etc. for the metal, and 6, 9 Torr, etc. for the glass). In each case the range of added He was 3–15 Torr. Optical and states.The former eVect was attributed to the higher excitation and ionization potentials of He than Ar. Helium was also sputtering data were also collected at the same total pressures using pure Ar. Another set of studies involved a variation of found to reduce sputtering rates by a factor of two, while also producing broader interface transitions. A recent study pub- the He5Ar ratio at a fixed total pressure. In this case, an absolute total gas pressure of 10 Torr was maintained and the lished by this laboratory touched on the possible practical advantages for He–Ar mixed gas plasmas in rf-GD-AES.13 He5Ar ratio was varied from 100% Ar to 100% He.While involving only conductive samples, the study reported Determination of emission and sputtering characteristics increases in analyte emission intensity of up to 300% over a 0–15 Torr range of increasing He pressure. Consistent with all The following procedure was employed for the collection of of the above mentioned studies, a decrease in sputter rate of optical emission intensities and the subsequent determination up to 50% was observed over the same pressure range.of sputtering rates. After a pressure equilibration period of In this work, we set out to compare the sputtering and approximately 1 min and a 3 min pre-sputter period, the emission yield characteristics of the rf glow discharge atomic discharge was ignited and emission profiles were recorded over emission source for He–Ar mixtures, from pure Ar to pure a ~0.05 nm window about each monitored transition. While He plasmas.Since plasma characteristics may vary greatly stabilization usually required less than 1 min, the time delay from sample to sample, especially between conductive and ensured a stable emission signal prior to data collection. The non-conductive matrices, both sample types were employed in plasma was extinguished after a total burn time of 5 min, a the evaluation. Trends for several He5Ar ratios were examined suYcient sputtering period to facilitate meaningful crater at fixed and increasing total discharge gas pressure situations. (sputtered area) profile measurements.The depth and shape Benefits in depth profiling applications are demonstrated in of each crater were measured using a Tencor (Mountain View, the analysis of layered samples at selected He5Ar pressure CA, USA) Model P-10 surface profilometer. Relevant ratios. In general, the data indicate that the addition of He to operating parameters of the profilometer include a 15 mg force an Ar plasma enhances the excitation process without signifi- on a 12.5 mm diameter, diamond-tipped stylus, a horizontal cantly aVecting the sputtering characteristics with respect to scan length of 10 mm across the breadth of each crater (2 mm the pure Ar case.resolution) and a vertical (depth) resolution of 25 A° . The average crater depth was determined for each burn from the central 2000 points of the total of 5000 in the lateral scan.Experimental The corresponding sputter rate is obtained by dividing the Optical emission studies average depth by the total sputtering time of 5 min for each sample. The data reported here represent the average of the The spectrometric equipment employed in this study has been triplicate runs at each of the chosen set of conditions. The described previously,14 so only a brief description is given precision of the emission intensities, over the three runs for here.All rf-GD-AES experiments were performed on a Model each condition, was typically better than 5% RSD for all 5000 RF polychromator system (Jobin-Yvon, Longjumeau, elements monitored (extreme pressures yielded poorer France). The 0.5 m Paschen–Runge polychromator permits precision values, although <15%). the simultaneous monitoring of optical emission from the desired analyte(s) and the discharge gas elements. The source Samples emission is focused with MgF2 optics on to a 2400 grooves/mm-1 ion-etched, holographic grating. The optical Since conductive and non-conductive matrices are known to respond optimally at diVerent plasma conditions, both types path of the spectrometer is nitrogen purged and operates over the wavelength range 110–620 nm, with the practical spectral of samples were examined in this study.NIST SRM 1250 High Temperature Alloy was chosen for the study of conduc- resolution for this instrument being ~0.01 nm.The discharge gas pressure in the GD source was monitored with a silicon tive samples. The following components’ emission lines were monitored (% m/m composition): Al (I ) 396.15 nm (0.99%), diaphragm pressure transducer (Model PX811-005) and a matching controller/meter (Model DP41-S, Omega, Stamford, Co (I ) 345.35 nm (16.1%), Cr (I ) 425.43 nm (0.077%), Cu (I ) 327.39 nm (0.022%), Ni (I ) 341.47 nm (37.78%) and Ar (I ) CT, USA). The pressure transducer allowed the absolute pressures of both He and Ar gases (individually and in total ) 404.44 nm.This sample was prepared by sanding the surface with 600 grit sandpaper to an even finish and subsequently to be measured, without bias, at a resolution of ~0.2 Torr and an experimental reproducibility of better than 5% RSD wiping with a lint-free cloth. Common, 1 mm thick microscope slides (Fisher, Pittsburgh, (at 3 Torr). The source body (anode) is constructed of stainless steel PA, USA, catalog No. 12-544-1) were chosen for the study of non-conductive samples. This product is an ideal sample choice with the dimensions of 8×6.5×6.5 cm. The flat sample is sealed against a Teflon O-ring (located on the demountable for several reasons: (1) the standard deviation of thickness, from slide to slide, is very low (0.6% RSD for 10 samples), limiting orifice plate) by means of a pneumatically controlled piston. A ceramic disk around the O-ring ensures a constant (2) the sample oVers a large surface area and (3) low cost and ease of preparation.Each glass slide required only rinsing with anode-to-cathode spacing (0.5 mm). In all cases, a 4 mm id limiting (anode) orifice was employed and 30W of power were methanol before use. To ensure consistency, 10 separate glass slide samples were analyzed (4 Torr, 20 W), yielding a pre- applied using a modified Dressler (Berlin, Germany) generator. The specific pressures chosen for analysis were based on data cision of <5% RSD for background-corrected signals of Al (I ) 396.15 nm, Mg (II ) 280.27 nm and Si (I ) 288.15 nm lines.from previous studies13,15 and unpublished trials conducted in preparation for this study. Based on the criterion of producing Two very diVerent types of layered materials were employed to evaluate the possible diVerences in operating characteristics crater bottoms parallel to the sample surface, the following 1040 J. Anal. At. Spectrom., 1999, 14, 1039–1048of mixed gas plasmas.The first, a ceramic-coated steel, was undertaken here, Belkin et al. characterized He–Ar working provided by a steel manufacturer who was interested in gas systems in this laboratory by Langmuir probe and emission evaluating rf-GD-AES as a means of characterizing layer intensity studies.18 Those studies indicated that whereas elecformation processes. The non-conductive coating was made tron and ion number densities decrease with helium addition, up predominately of Si, Al and Mg oxides, with a thickness the measured average electron energy e and electron temof ~1 mm on the 1 mm thick steel substrate.This layered perature (Te) values increase. The net result of these diVerences system can be generally characterized as having a broad, was a general increase in analyte atomic emission intensities. diVusional interface rather than a very clean, finite interface. The precise degree of the enhancement in excitation eYciency The second layered system was a Cu/Cr–Ni superlattice mate- cannot be assessed without a corresponding knowledge of rial provided by IFW (Dresden, Germany).This sample changes in sputtering rate. Hence the assessment of the relative consists of 10 alternating layers of Cu and Cr–Ni, each having emission yields for diVerent gas mixtures is of both analytical a thickness of ~100 nm, on a 0.5 mm thick Si substrate. The and fundamental relevance. interfaces in this system are well structured.As listed in Table 1, an increase in emission yield is observed for all of the monitored analytes in NIST SRM 1250 High Temperature Alloy as a function of the total source pressure, Results and discussion for both pure Ar and for the mixed gas (He–Ar) conditions. Comparison of emission yield from pure Ar and mixed gas The data set for the pure Ar discharge reveals an increase in plasmas REY of up to ~25-fold over the pressure range 5–20 Torr.The increase in REY for the mixed gas plasma over the same Analyte emission intensity is a product of two factors: the pressure range is also significant, yet not nearly as great for number of analyte atoms in the plasma (atomization rate) and most elements. These eVects surely reflect enhanced excitation the emission yield (excitation eYciency). Since the discharge eYciencies due to increases in e and Te.18 The Cu 327.96 nm gas type and pressure are known to influence both atomization line exhibited approximately the same overall increase in REY and excitation processes, intensity is not a valid indication of for both cases (~400%); however, for the Cr 425.43 nm line, excitation conditions.Relative emission yield is an eVective there was a nearly 10-fold greater increase in REY as the means of assessing the eVects of plasma conditions (power, pressure was increased for pure Ar in comparison with the pressure, etc.) on analyte excitation because intensity is normixed gas condition (2600% for Ar versus 300% for He–Ar).malized by both the sample sputter rate and analyte concen- This diVerence in REY between the two analytical lines is tration. In all following discussions, the relative emission yield surely due to diVerences in the excitation energies and path- (REY) is expressed as ways for each of the respective excited states. REYx=Ix/(SDsam[X ]) (1) Comparison of the REY data for the NIST SRM 1250 alloy as a function of added discharge gas (Table 1) indicates that for element X, where I is the background-corrected analyte the pure Ar plasma excites the atomic species in the discharge emission intensity in volts, SD is the sputtered depth (mm, per plasma more eYciently than does helium.The same trend is 5 min run) for the respective sample and [X] is the concenobserved on comparing the Ar and Ar–He plasmas at a fixed tration of the element in the sample. The subject of which pressure, as Fig. 1 depicts the trend with a plot of REY versus discharge parameter(s) (current, voltage, or pressure) most He5Ar ratio at a fixed total pressure of 10 Torr. (Data for the strongly aVects the emission yield of analytes in GD plasmas pure He condition were not included with this set as a stable has been debated for some time.16 Payling et al. have published plasma could not be maintained at this pressure.) The emission data that support a theory that discharge gas pressure is yield for Cr and Cu declined sharply on addition of any He, primarily responsible for influencing this phenomenon.17 While by ~80 and ~70%, respectively, over the range of gas several studies have focused on the eVects of power (mostly mixtures.The eVect of added He on the REY of other dc) and pressure on emission yield, few have been conducted monitored emission lines was not as dramatic, but the overall to compare the relative emission yields produced by diVerent decrease was within the same range (decreases of ~70, ~80 discharge gases across a range of diVerent pressures.Since and ~50% for Ni, Co and Al, respectively). Hence it is clear each discharge gas has unique excited state (metastable) and that, on a same-pressure basis, the pure Ar plasma is more ionization energies, each can be expected to influence the emission yield to diVerent extents. As a prelude to the studies eYcient at exciting these particular transitions. Very diVerent Table 1 Relative emission yields (REY) as a function of discharge gas pressure and composition for NIST SRM 1250 High Temperature Alloy (rf power=30 W) REY/V mm-1 %-1 Ar pressure/ He pressure/ Total pressure/ Cu (I ) Co (I ) Cr (I ) Al (I ) Torr Torr Torr 327.39 nm 345.35 nm 425.43 nm 396.15 nm 5 — 5 0.126 0.006 0.022 0.014 8 — 8 0.202 0.012 0.074 0.034 11 — 11 0.278 0.021 0.160 0.075 14 — 14 0.377 0.033 0.277 0.134 17 — 17 0.448 0.042 0.352 0.185 20 — 20 0.637 0.065 0.573 0.306 Increase (%): 500 1150 2600 2150 5 — 5 0.126 0.006 0.022 0.014 5 3 8 0.096 0.010 0.022 0.031 5 6 11 0.180 0.015 0.034 0.056 5 9 14 0.268 0.019 0.041 0.081 5 12 17 0.407 0.024 0.049 0.111 5 15 20 0.675 0.036 0.070 0.177 Increase (%): 530 630 310 1240 J.Anal. At. Spectrom., 1999, 14, 1039–1048 1041at 280.27 nm are diVerent from any other line in the study as the enhancement in REY was found to be greater for the mixed gas condition (~600× vs. 1500×) over the pressure range. Clearly, this disparity in response with respect to all of the other transitions points to diVerences in excitation mechanisms.Since the energies associated with Arm* and Mg+* are 11.7 and 12.07 eV, respectively, a plausible excitation pathway in this situation is the Penning-type reaction Arm*+Mg0�Mg+*+Ar0 (2) The diVerence of ~0.4 eV (endothermic) between the two energy states can be supplied by the thermal energy of the plasma. Of course, for helium to influence Mg(II ) significantly, the number of argon metastables would also need to increase as a function of helium content.Owing to the small energy diVerence between helium metastable and argon ion excited states, collisional energy exchange between the two states influences Ar+ populations in mixed gas plasmas.9 The forma- Fig. 1 EVect of He5Ar discharge gas ratio on relative emission yields tion of Ar metastables from the neutralization of Ar ions for NIST SRM 1250 High Temperature Alloy (total pressure= completes the pathway 10 Torr, rf power=30W).Hem+Ar0�He0+Ar+ (3) from the response observed for transitions of the sputtered Ar++e�Arm* (4) atoms, Fig. 1 also depicts the raw intensity of the Ar (I ) Arm*+Mg0�Mg+*+Ar0 (2) 404.44 nm transition. In this case, the ability of the He plasma to excite this high-lying transition is dramatic as the Ar (I ) For the glass sample matrix, diVerent He5Ar pressure ratios intensity is essentially unchanged while the actual concen- (at constant absolute pressure) yielded results slightly diVerent tration is decreased by 90%.The underlying reasons for this from those for the NIST SRM 1250 sample. As can be seen diVerence in response are discussed below. in Fig. 2, Si (I ) and Mg (II ) lines yielded the greatest REY In general, rf-GD-AES analyses of glasses have shown that values at He5Ar ratios of ~555 (456 for Si). The Al (I ) line the most favorable analytical responses occur at discharge conditions of lower gas pressures (2–6 Torr Ar) than for metallic samples.14 Similar to trends observed for the alloy sample Table 2 illustrates that both the pure Ar and mixed gas plasmas exhibit a steady increase in REY over the 3–18 Torr range.The highest REY values were observed for pure Ar plasma operating at 15 Torr (an REY increase of nearly two orders of magnitude for Si 288.15 nm). The highest Ar pressure, 18 Torr, produced an unstable plasma, as indicated by REY values that do not follow the trend of the five preceding data points.A very interesting (and important) selfconsistency is borne out on inspection of the response of Al, the lone common element in the metallic and glass matrices, in Tables 1 and 2. In both data sets, the percentage increase from the lowest to highest pressures for the pure Ar and mixed gas plasmas diVers by twofold. Therefore, to a first approximation, the influence of He on the excitation eYciency is the same for conductive and non-conductive samples.Fig. 2 EVect of He5Ar discharge gas ratio on relative emission yields for glass slide samples (total pressure=10 Torr, rf power=30 W). The results observed for the monitored Mg (II ) transition Table 2 Relative emission yields (REY) as a function of discharge gas pressure and composition for glass slide samples (rf power=30 W) REY/V mm-1 %-1 Ar pressure/ He pressure/ Total pressure/ Si (I ) Mg (II ) Al (I ) Torr Torr Torr 288.15 nm 280.27 nm 396.15 nm 3 — 3 0.0004 0.0059 0.0105 6 — 6 0.0008 0.0067 0.0205 9 — 9 0.0025 0.0114 0.0404 12 — 12 0.0190 0.0260 0.0918 15 — 15 0.0295 0.0355 0.1281 18 — 18 0.0070 0.0221 0.0279 Increase (%): 7250 600 1200 3 — 3 0.0004 0.0059 0.0105 3 3 6 0.0007 0.0137 0.0156 3 6 9 0.0012 0.0304 0.0245 3 9 12 0.0021 0.0484 0.0364 3 12 15 0.0036 0.0749 0.0579 3 15 18 0.0046 0.0892 0.0737 Increase (%): 1130 1500 700 1042 J.Anal. At. Spectrom., 1999, 14, 1039–1048at 396.15 nm displayed a drastically lower REY upon He addition.As discussed previously, the addition of He has been shown to increase the average electron energy and electron temperature (e and Te) in Ar plasmas.13 At constant pressure, it reasonably follows that transitions with lower energy requirements [Al (I ) 396.15 nm, 3.15 eV ] will be favored at lower concentrations of He than those with higher energy requirements [Si (I ) 288.15 and Mg (II ) 280.27 nm, 5.09 and 12.07 eV, respectively].The response of the Ar (I ) 404.44 nm signal depicted in Fig. 2, increasing steadily with increasing He5Ar ratio (decreasing Ar partial pressure), is surprising. This trend was observed in earlier He–Ar mixed gas studies, 13,18 and supports the cascade model depicted in eqns. (3) and (4) as that transition is an iportion of the pathway from the ionic to the metastable state. In addition, the high energy requirements (14.7 eV) of the Ar (I ) 404.44 nm line Fig. 3 EVect of discharge gas pressure on sputter rates for NIST SRM may exhibit enhancements by virtue of the higher electron 1250 High Temperature Alloy. %, Pure Ar discharge; 1, mixed gas energies which may contribute to more eVective direct electron discharge, He added to 5 Torr Ar base pressure (rf power=30W). impact population of the excited state. Direct ionization is supported through previous Langmuir probe studies which show that electron energies tend to be much higher in the case of oxide (glass) sample analyses versus metals.19 Addition of He appears to accentuate this diVerence.Influence of He fraction on sample sputtering rates The mass of a sputtering ion logically influences sputter rate, and several theories on sputtering have been published.20–23 In particular, argon is a much more eYcient sputtering agent (at a given accelerating potential ) than helium owing to its greater mass. In fact, the sputtering yield for Ar at a copper target for a nominal ion energy of 400 eV is about eight times higher than that for He (~1.65 vs. 0.21).22,23 The particular argon pressure that yields maximum sputter rates is often specific to the matrix being analyzed, and above this pressure the sputter rate typically declines. These diVerences are particu- Fig. 4 EVect of discharge gas pressure on sputter rates for glass slide larly pronounced when comparing conductive and nonsamples. %, Pure Ar discharge; 1, mixed gas discharge, He added to conductive sample matrices.15 There are several factors that 3 Torr Ar base pressure (rf power=30W).explain this influence.24 At low pressures (mTorr range), the mean free path of discharge gas ions impinging the cathode surface is long, translating into high ion energies. With increas- to the initial base Ar pressure reveals optimum pressure conditions with regard to sputtering rates, whereas addition ing Ar pressure, collisional processes become significant as Ar+–Ar0 collisions result in either elastic scattering or charge of He tends to depress the sputtering rates only slightly relative to the initial (pure Ar) values.In fact, Fig. 4 demonstrates transfer reactions. Both processes reduce the average Ar ion energies, which are manifest in rf-GDs as dc-bias values that beyond the Ar optimum, the sputter rates for the mixed gas plasmas are considerably greater than those for the pure steadily decrease.25,26 To a first approximation, this would be expected to result in lower sputtering yields/rates.The diVusion Ar plasma at the same pressure (i.e., the depression in sputtering rate is greater for the pure Ar discharge). In the case of rate of sputtered species away from the sample surface is also reduced at higher pressures as mean free paths are shorter, the glass sample, it may be that the higher ionization potential (and thus bias potentials) of He helps to maintain the sputter- contributing to higher redeposition rates.When discharge pressure is increased through He addition, however, these ing rates at levels above those of the argon plasma. In general, it appears that addition of He to an Ar plasma tends to have interactions are much less significant. Ar+–He0 charge transfer reactions are not possible and, owing to its low mass, helium only minimal eVects on sputtering rates beyond those established for that pressure of Ar. This characteristic, which is does not strongly influence the trajectory of the impinging ions or the sputtered analytes.The latter point might suggest consistent throughout the studies described here, points clearly to a situation wherein Ar+ is the major positive charge carrier that smaller amounts of redeposition may be realized at the same total pressures for the mixed gas plasma versus the pure in the plasma and thus the predominant sputtering agent. This is a reasonable assumption given the much lower ionization Ar discharge.Thus helium can be expected to aVect sputtering characteristics only mildly when added to an argon plasma. potential of Ar than He. At some point, He would be expected to become ionized to a larger extent as the Ar concentration As can be seen in Figs. 3 and 4, the sputter rates for pure Ar and Ar–He plasmas yield diVerent responses in the cases is reduced, so as to maintain the overall ionization level of the plasma. of the metallic and glass sample matrices. The pressure ranges chosen here begin at Ar values below the optimum values for Fig. 5 is a plot of the sputter rates for both sample types at diVerent He5Ar ratios and fixed absolute pressure. As was pure argon discharges, hence addition of either gas allows passage through maxima in sputtering rates. For both sample observed for increasing pressures with added He, the sputter rates for NIST SRM 1250 decreased steadily with increasing types, a comparison of sputter rate data shows that the mixed gas conditions produce a much more uniform response than He5Ar ratio.As would be expected from Fig. 3, the highest sputtering rate is obtained for the pure Ar (10 Torr) plasma. does pure Ar, collected over the same absolute pressure ranges. Both Figs. 3 and 4 illustrate the fact that addition of Ar gas Note also that sputter rates do not decrease proportionally J. Anal. At. Spectrom., 1999, 14, 1039–1048 1043Fig. 5 EVect of He5Ar discharge gas ratio on sputtering rates.%, NIST SRM 1250 High Temperature Alloy; 1, glass slide samples (total pressure=10 Torr, rf power=30 W). with He dilution. For example, 5 Torr He and 5 Torr Ar yielded a sputtered depth of 6.4 mm whereas pure Ar at 5 Torr yielded a sputtered depth of 8.1 mm. A 50% dilution of Ar yielded only a 20% decrease in sputter rate. Again, this example illustrates that, with respect to Ar, He has little influence over sputtering conditions in the GD plasma. Sputter rate results for the glass analyses at constant pressure closely follow trends set by the pure Ar data in Fig. 4. The data indicate that the increased pressure from added He Fig. 6 EVect of discharge gas pressure on resultant sputtered crater (decreasing the Ar partial pressure) increases the observed shapes for NIST SRM 1250 High Temperature Alloy: (a) pure Ar sputtering rates, reaching maximum values under the con- discharge and (b) He added to 5 Torr base Ar pressure (rf power= ditions where the Ar pressure is in the range 3–4 Torr.While 30 W). added He has been shown generally to decrease sputter rates for Ar-based plasmas, these data indicate that optimum sputprobably the most desirable results for this sample, in depth tering conditions (i.e., Ar partial pressure) may require tailorprofiling terms. Note that as the discharge gas pressure is ing with respect to added He and sample matrix type. This is increased, the crater bottom becomes more curved and the consistent with the pure Ar plasmas also.As expected, there crater walls more sloped. The higher pressure conditions yield is limited correlation between sputtering trends and REYs. a crater profile that is somewhat W-shaped (convex), owing to Since REYs have been shown to increase with increases in a complex gradient of sputtering and redeposition rates across discharge gas pressure (Ar or He), the addition of He to an the analysis area. Work by Parker et al.15 and Angeli et al.27 Ar plasma can be expected to increase the analyte intensity has suggested numerical expressions to quantify better the while minimally influencing the sample sputtering conditions.overall flatness of a crater. However, it is believed that the For example, one could envision a situation wherein sputtering results of this study are best expressed by a comparison of characteristics (rates) can be optimized with respect to Ar crater profiles. Fig. 6(b) illustrates the eVect of mixed gas discharge gas pressure and then He added to improve analytical conditions on crater shapes for the sputtering of NIST SRM emission responses. 1250 alloy. Over a broad range of added He from 0 to 12 Torr (5 to 18 Torr total pressure), the craters maintain their rec- Influence of He fraction on resultant crater shapes tangular shape. Only at the highest pressure (15 Torr added He, 20 Torr total ) does the crater shape become unsuitable for DiVerent from the case of bulk analyses, the eVect of plasma conditions on the shape of sputtered craters is a crucial aspect depth profile analysis.Hence, absolute discharge pressure may be significantly increased by the addition of He, without a of the quality of depth profiling analyses. The most desirable crater is one in which the sides (walls) are perpendicular to, significant aVect on crater shape. These results are consistent with those of the sputter rate studies. and the base parallel to, the original sample’s surface.Such a sputtering profile should yield intensity information most For the glass matrix sample, discharge pressures of 3 and 6 Torr pure Ar yield crater shapes which suggest favorable representative of the sample strata. Previous studies by Parker et al. evaluated the respective roles of the discharge conditions conditions for depth profiling [Fig. 7(a)]; however all craters produced at higher pressures are undesirable for depth profiling of applied rf power and discharge gas pressure.15 In general, rf power was found simply to aVect the rate of cathodic purposes.Interestingly, the 18 Torr condition yielded a crater not nearly as deformed as those at 9, 12 and 15 Torr. The sputtering, whereas Ar discharge gas pressure tended to aVect both the rate and the shape. As with other GD sources, high uniform shape of the 18 Torr crater is probably due to the very low sputter rate at this pressure. Fig. 7(b) is an overlay pressures tend to form concave-shaped craters, whereas low pressures produce more convex craters.As would be expected, of crater profiles from the mixed gas sputtering of the glass sample, using a 3 Torr base Ar pressure. The craters obtained optimum crater shapes for non-conductive samples were achieved at lower pressures than for metallic samples (~3 Torr from this set of analyses remain more uniform as the discharge gas pressure is increased, as compared with pure Ar. The crater vs. 6 Torr Ar). Fig. 6(a) displays an overlay of crater profiles from the GD profiles produced by mixed gas conditions, up to absolute pressures of 9 Torr, appear to remain acceptable for depth sputtering of NIST SRM 1250 at increasing pure Ar pressures from 5 to 20 Torr. The 5 Torr condition yields the flattest and profile analysis. However, above 9 Torr the convexity of the 1044 J. Anal. At. Spectrom., 1999, 14, 1039–1048optimal discharge conditions for depth resolution and intensity can be mutually exclusive. For example, increasing discharge power or pressure to enhance the intensity of analyte optical emission typically increases the sputter rate and in some instances alters the sputtered crater shape, thus decreasing the depth resolution. Depth profiling applications can be envisioned to be enhanced by a process wherein optimum sputtered crater shapes are achieved by adjustment of the Ar discharge gas pressure and then He gas is added to achieve enhanced analyte emission responses.The following examples illustrate the eVect of He gas addition on the depth profile analyses of two distinctly diVerent sample types. The first sample type is a ceramic-coated steel and the second a semiconductive (Si) substrate coated with multiple metal layers (e.g., superlattice). Signal intensity data from the analyte elements were used to calculate and compare depth resolution at diVerent discharge conditions for the well structured superlattice material.Depth resolution was determined from the intensity–time profile of each sample, based on changes in analytical response as sputtering progressed from one layer to the next as described by HoVmann and co-workers.28 Table 3 presents depth resolution (Dz) data as defined by Dz=q Imax (Dt/DI) (5) where q is the sputter rate, Imax is the maximum intensity for the respective analyte line and Dt/DI is the slope of the intensity–time profile. Slope values were calculated using points at 16 and 84% of Imax for each element, in accordance with Fig. 7 EVect of discharge gas pressure on resultant sputtered crater the definition employed by HoVmann and co-workers. For the shapes for glass slide samples: (a) pure Ar discharge and (b) He added multi-layered sample, depth resolution was determined using to 3 Torr base Ar pressure (rf power=30W). data from both the topmost and deepest layers to compare resolution as a function of depth. The sputtering rate, signal crater bottom becomes severe. As indicated previously, sputter- intensity and depth resolution values for the two sample types ing characteristics for the glass sample are more sensitive to are summarized in Table 3.It should be noted that the samples discharge gas conditions than metallic samples. Although not and discharge conditions employed here were chosen to illuspursued in this study, it is believed that flat craters may be trate the eVects of He gas addition to Ar plasmas, in terms of obtained at even higher total pressures, with the proper depth profiling, rather than to define the analytical limits of optimization of operating conditions. rf-GD-AES.Since it is known that dc-bias and ion mean free path Depth profiles of the ceramic-coated, metallic sample are conditions change with increased pressure,24–26 a logical expla- pictured in Fig. 8. This sample consists of a steel substrate nation for the consistency of crater shape (as He is added) with a ~1 mm ceramic (primarily SiO2 and MgO) coating on may be based on the consistency of Ar density within the the surface.Note that this sample is not expected to yield a discharge plasma. Remember that added He does not increase step-type response of the analytes in the interface region as the frequency of charge exchange reactions as does Ar; however, this is probably a spray coated layer that is further annealed a small eVect on redeposition can be expected. It follows that at high temperature.As such, a great deal of interdiVusion of at suYciently high total pressures, the overall density of gas alloy and oxide materials is expected, as is revealed in the species will ultimately lead to deterioration of sputtering depth profiles. The sample was originally analyzed at 10 Torr characteristics as helium gas is added. Because the capacitance pure Ar and 20W power [Fig. 8(a)]. Based on previous steel manometer reflects the total gas pressure in the cell, regardless specimen sputtering data,15 a lower Ar operating pressure of the gas mixture, the gas-phase density is constant in these (e.g., ~5 Torr) would have produced a more desirable crater studies between the pure Ar and mixed gas cases.On the other profile. However, an Ar pressure of 10 Torr was chosen as a hand, there will most certainly be diVerences in total gas flow compromise value for comparing a range of Ar, He and mixed as the ratios vary.The fact that the introduction of the gases gas pressures. Analysis of the same sample at 5 Torr Ar, 5 Torr is far removed from the cathode surface suggests that any He (10 Torr total ) and the same power produced analyte convective perturbations would be small. The influence of gas signals slightly higher for Mg (II ) and three times higher for make-up is especially evident in the changes observed for the Si (I ) [Fig. 8(b)]. Increasing the He pressure to 10 Torr glass sample.Since Ar pressure critically influences sputtering (15 Torr total ) yielded greater returns in raw analyte signal conditions for the glass sample, excessive He partial pressures [Fig. 8(c)]. The net Si signal was, however, largely unchanged may require adjustments of the Ar partial pressure. Further owing to a concurrent increase in background at the Si (I ) investigation of this phenomenon is warranted in future studies, 288.15 nm line. Note that the Fe (I ) signal decreases signifi- but in general results consistent with pure Ar pressures can be cantly from the 10 Torr Ar condition to the 5 Torr He, 5 Torr expected for partial pressures of Ar in He–Ar mixed gas Ar condition, but nearly regains its original intensity under systems. the final sputtering conditions (Table 3).The decrease in Fe (I ) response is consistent with those of low-lying excited states, Assessment of depth resolution: analysis of layered samples as illustrated in previous sections of this paper, where the further increase in He pressure probably provides a gas phase The preceding discussion suggests that depth profile analysis can be enhanced by the addition of He gas to rf-GD methods atmosphere more like that of a 10 Torr Ar discharge.Very qualitative computation of the depth resolution for the analy- previously employing only pure Ar. As mentioned previously, J. Anal. At. Spectrom., 1999, 14, 1039–1048 1045Table 3 Sputtering rates, analyte emission intensities and depth resolution as a function of discharge gas composition for layered materials Imax/V Sample/ Rf power/ Ar pressure/ He pressure/ Sputtering rate/ type Fig.W Torr Torr mm s-1 Si (I ) Mg (II ) Fe (I ) Ceramic-coated steel 8(a) 20 10 — 0.017 0.45 3.95 5.26 8(b) 20 5 5 0.013 1.36 5.81 2.66 8(c) 20 5 10 0.013 1.99 9.05 4.72 Dz/mm Cu (I ) Cr (I ) Cu (I ) Cr (I ) Superlattice — Top layer 9(a) 15 3 — 0.011 1.79 0.61 0.074 0.061 9(b) 15 2 9 0.010 5.32 1.07 0.037 0.041 10 1.5 10 0.006 3.25 0.70 0.020 0.039 10 1.25 11 0.005 2.32 0.518 0.016 0.048 Bottom layer 9(a) 15 3 — 0.010 1.41 0.59 0.071 0.107 9(b) 15 2 9 0.009 4.55 1.01 0.056 0.093 10 1.5 10 0.006 2.96 0.59 0.046 0.103 10 1.25 11 0.004 2.05 0.37 0.053 0.103 ses remained largely unchanged based on Mg and Fe intensities, although, as seen in the temporal profiles in Fig. 8(b) and (c), the definition was greatly enhanced based on the Si signal following He addition.The reduction in sample sputtering rate does not account for this improvement, although improvements in crater shape would be predicted. The improvement here is simply due to the overall improvement in the Si (I ) S/N upon He addition, providing better definition in terms of DI. While the coated steel sample by nature does not contain highly defined layers, it serves as a good example of modern coated systems as the profiles clearly point to very diVerent spatial distribution of the Mg and Si in the ceramic coating. Overall, the addition of He to the plasma increased analyte emission signals while maintaining or enhancing depth resolution.The depth profiles illustrated in Fig. 9 obtained for the superlattice material reflect a sample with highly defined layers. This sample consists of 10 layers, each approximately 100 nm in thickness, deposited on a semiconductor grade silicon wafer. The ‘superlattice’ includes five layers of Cu alternating with five layers of Cr and Ni.The analysis conditions of 3 Torr pure Ar and 15 W power were chosen to best illustrate the highly defined layers of the superlattice, with the resulting depth profile shown in Fig. 9(a). As can be seen, the individual layers are well defined, with the calculated resolution values (shown in Table 3) for the topmost layers being of the order of 70 nm and an average of ~90 nm in the deepest layer, depending on the chosen element. A comparison to the resolution for the upper layer is inferior to that obtained by HoVmann and co-workers, who calculated a value of 25 nm,28 while the lowest levels in their study produced a value of ~60 nm.The addition of 9 Torr He to the base pressure of 3 Torr Ar (profile not shown) increased the sputter rate and changed the sputtering profile (crater shape) to the extent that the last few individual layers of the sample were no longer well defined in the analysis. As suggested by the crater shape study, lowering the Ar partial pressure from 3 to 2 Torr was expected to restore the sharp, well defined profile that was collected for the pure Ar condition [Fig. 9(b)]. These conditions yielded increases in intensity from 70 to 200% (Cr and Cu, respectively). In this case, the depth resolving powers were also enhanced, with the values in the uppermost layers being ~40 nm, and an average of ~75 nm. It must be stressed that this improvement represents the realization of better crater characteristics, as the actual sputtering rates are unchanged. Fig. 8 EVect of discharge gas composition on depth (temporal ) profiles While optimized depth resolution capabilities were not the of ceramic-coated (~1 mm) steel specimens: (a) 10 Torr Ar, (b) 5 Torr He–5 Torr Ar and (c) 10 Torr He–5 Torr Ar (rf power=20 W). ultimate aim of this study, the values obtained here are not 1046 J. Anal. At. Spectrom., 1999, 14, 1039–1048increased significantly with an increasing ratio of He5Ar pressure.Since the monitored Mg (II ) and Ar (I ) lines require relatively high energies for excitation, compared with the other lines monitored, He might be expected to enhance the sensitivity for diYcult to excite species and emission lines, such as non-metals. Practical application of He5Ar systems for rf-GD analysis requires further study and selection of appropriate analytical lines for all analytes of interest. If only atomic lines are considered, this study clearly indicates that Ar is the discharge gas of choice for bulk analysis, owing to the eYciency of Ar plasmas in exciting atomic transitions.Further investigation into the use of ion lines for bulk analysis using He–Ar mixed gas plasmas may well reveal that the limits of detection currently available using Ar based plasmas can be surpassed. The sputtering characteristics of mixed gas plasmas are surprising in comparison with previous published work which reported that the use of He severely compromises sputtering rates.In simple terms, He–Ar mixed gas plasmas exhibit nearly the same sputtering rates as those obtained given Ar pressure alone. Slight depressions in sputtering rate are observed with He addition, but not nearly in proportion to what might be expected based on the stoichiometry of the gas mixture. By the same token, the resultant crater shapes obtained for the case of mixed gas sputtering are analogous to those which would be obtained for a pure Ar plasma.Therefore, a scenario can be envisioned wherein an optimum Ar pressure may be established for the desired sputtering characteristics for a given sample and then He gas added beyond that point to optimize the optical emission response further. This phenomenon can be used to particular advantage in the case of depth profiling Fig. 9 EVect of discharge gas composition on depth (temporal ) profiles of Cu/Cr–Ni superlattice materials (100 nm each layer on silicon): applications.Optimum Ar pressure conditions for depth profi- (a) 3 Torr Ar; and (b) 9 Torr He–2 Torr Ar (rf power=15W). ling tend to be very specific and can restrict analyte sensitivity. However, sensitivity enhancement via added helium is not detrimental to depth resolution as He does not strongly far from those reported by others where this was the intended goal. Analyses at even higher partial pressures of helium (not influence the resultant crater shapes from GD sputtering.In summary, helium may be used as an additional parameter, shown because they are not visually diVerent) were conducted to probe the extent of signal enhancement. For each subsequent along with power and Ar pressure, to tune the glow discharge plasma for specific applications. addition of He, sputter rates decreased and, as a consequence, the analyte intensities also decreased, although not to the extent of the sputtering rates. Interestingly, depth resolution Acknowledgements determinations (Table 3) based on the uppermost layer of the sample improved with increasing He content, although the Financial support from the National Science Foundation under calculations based on the final (deepest) layers of the sample grant No.DMR-9727667 and from Jobin-Yvon, Division of remained substantially constant over the range of conditions Instruments SA, is gratefully acknowledged. (<5% RSD for Cr). Overall, depth resolution appears to be aVected only by sputtering profile (crater shape) and not by References average sputter rate.With this in mind, the 9 Torr He–2 Torr Ar condition produced the highest gains in analyte sensitivity 1 T. B. Reed, J. Appl. Phys., 1961, 32, 821. with no degradation to depth resolution. 2 J. W. H. Iam and G. Horlick, Spectrochim. Acta, Part B, 1990, 45, 1313. 3 F. G. Smith, D. R. Wiederin and R. S. Houk, Anal. Chem., 1991, Conclusions 63, 1458. 4 S. Nam, R. Masamba and A. Montaser, Anal.Chem., 1993, 65, The atomization and excitation characteristics of He–Ar mixed 2784. gas plasmas were compared with the traditionally used pure 5 A. Montaser, K. D. Ohls and D. W. Golightly, in Inductively argon discharge gas for rf-GD-AES analyses. Specifically, Coupled Plasmas in Analytical Atomic Spectrometry, ed. emission yield, sputter rate and crater shape were studied for A. Montaser and D. W. Golightly, VCH, New York, 2nd edn., 1992. both conductive and non-conductive samples as a function of 6 K.Oishi, T. Okumoto, T. Lino, M. Koga, T. Shirasaki and He and Ar partial pressures. The results from this comparison N. Furuta, Spectrochim. Acta, Part B, 1994, 49, 901. were applied to the depth profile analysis of coated and multi- 7 HP G2350A Atomic Emission Detector for Gas Chromatography, layered samples. Hewlett-Packard, Palo Alto, CA. The REY values were found generally to be lower for He 8 K. Wagatsuma and K. Hirokawa, Spectrochim. Acta, Part B, based plasmas than Ar plasmas at the same pressures, for the 1987, 42, 523. 9 K. Wagatsuma, J. Anal. At. Spectrom., 1996, 11, 957. monitored emission lines. This is surprising, as helium has 10 K. Wagatsuma, K. Hirokawa and N. Yamashita, Anal. Chim. been reported generally to enhance emission intensities. Only Acta, 1996, 324, 147. one ion line was monitored [Mg (II ) 280.15 nm], and this line 11 J. Woo, D. Moon, T. Tanaka, M. Matsuno and H. Kawaguchi, was found to have greater REYs for the mixed gas condition. Anal. Sci., 1996, 12, 459. The results agree with other studies that indicate helium aids 12 W. B. Teo and K. Hirokawa, Surf. Interface Anal., 1989, 14, 143. in more eYciently exciting ion lines than does argon. At a 13 S. J. Christopher, M. L. Hartenstein, R. K. Marcus, M. Belkin and J. A. Caruso, Spectrochim. Acta, Part B, 1998, 53, 1181. fixed total pressure, the intensity of the Ar (I ) 404.44 nm line J. Anal. At. Spectrom., 1999, 14, 1039–1048 104714 M. A. Parker, M. L. Hartenstein and R. K. Marcus, Spectrochim. 22 N. Laegreid and G. K. Wehner, J. Appl. Phys., 1961, 32, 365. 23 D. Rosenberg and G. K. Wehner, J. Appl. Phys., 1962, 33, 1842. Acta, Part B, 1997, 52, 567. 24 B. Chapman, Glow Discharge Processes,Wiley, New York, 1980. 15 M. A. Parker, M. L. Hartenstein and R. K. Marcus, Anal. Chem., 25 C. Lazik and R. K. Marcus, Spectrochim. Acta, Part B, 1992, 1996, 68, 4213. 47, 1309. 16 A. Bengtson, Spectrochim. Acta, Part B, 1994, 49, 411. 26 M. Parker and R. K. Marcus, Spectrochim. Acta, Part B, 1995, 17 R. Payling, D. G. Jones and S. A. Gower, Surf. Interface Anal., 50, 617. 1993, 20, 959. 27 J. Angeli, K. Haselgrubler, E. M. Achammer and H. Burger, 18 M. Belkin, J. A. Caruso, S. J. Christopher and R. K. Marcus, Fresenius’ J. Anal. Chem., 1993, 346, 138. Spectrochim. Acta, Part B, 1998, 53, 1197. 28 F. Prassler, F. HoVmann, J. Schumann and K. Wetzig, J. Anal. 19 Y. Ye and R. K. Marcus, Spectrochim. Acta, Part B, 1996, 51, 509. At. Spectrom., 1995, 10, 677. 20 P. Sigmund, Phys. Rev. A, 1969, 20, 855. 21 G. Carter and J. S. Colligon, Ion Bombardment of Solids, Heinemann Educational Books, London, 1968, ch. 7. Paper 9/01517E 1048 J. Anal. At. Spectrom., 1999, 14, 1039–1048