0 Bewertungen0% fanden dieses Dokument nützlich (0 Abstimmungen)
296 Ansichten29 Seiten
This is a chapter that I wrote about 20 years ago that describes photoionization and far UV absorbance detectors for capillary GC. This was published by Wiley in 1992 with a title: Detectors for Capillary GC" Edited by H. Hill & D. McMinn
Originaltitel
Far UV Ionization Photo Ionization) and Absorbance Detectors for CapillaryGC
This is a chapter that I wrote about 20 years ago that describes photoionization and far UV absorbance detectors for capillary GC. This was published by Wiley in 1992 with a title: Detectors for Capillary GC" Edited by H. Hill & D. McMinn
Copyright:
Attribution Non-Commercial (BY-NC)
Verfügbare Formate
Als PDF, TXT herunterladen oder online auf Scribd lesen
This is a chapter that I wrote about 20 years ago that describes photoionization and far UV absorbance detectors for capillary GC. This was published by Wiley in 1992 with a title: Detectors for Capillary GC" Edited by H. Hill & D. McMinn
Copyright:
Attribution Non-Commercial (BY-NC)
Verfügbare Formate
Als PDF, TXT herunterladen oder online auf Scribd lesen
HPLC, and SFC, where the lower flow rates ean increase the sensiti,:. such detectors. Background information, analytical applications and sources far-UV radiation are described in detail in the following sections. 1. PHOTOIONIZATION DETECTION 1.1. Background The use of far-UV ionization (photoionization) detection for GC was described in the late 19505 along with flame ionization (FID), electron c:.; (ECD), mass spectrometer, and cross-section detection (l). During the _>, . there was considerable research on the PID but the FID gained por ity so quickly that all commereial gas chromatograph manufacturers 0::: __ the FID along with the thermal conductivity detector (TCD). In the mea!".:. - _ the PID continued to be a sensitive, novel detector but required a ,:1": for operation, was easily fouled by column bleed, was uns- - . and required a skilled operator. By the later 1960s, a decade after its discc : the PID had been virtually abandoned. By the mid-1970s, however, the -: was reborn when several researchers (5,6) found that separating the - discharge from the ion chamber improved the stability of the PID, an": detector was simplified. Driscoll and Spaziani (7) described additional imF . ments to the PID that resulted in an improved range {> 10 7 instead 0:- : improved stability, and lower background. It also minimized the probic column bleed that plagued earlier PIDs. After a slow start in the 1960,. -. PID was finally on its way in the later 19708 to becoming an importan: in the arsenal of the analytical chemist Davenport and Adlard have rey;;: 1989 1976 1960 '811 150 200 Figure 4.1. Comparison of PID geometry before and after 1970. Tmax 1IIIIII,.....t.t.:;............ ..IIIIIIII..1l1.. of :he ::rst ".:o:ure ::'ered -'.lme, .,:uum , ":.lble. . , ,'\cry, .. PID clamp the -'. :orove ::' 10 4 ), of ':'\5. thc . ,,:1t tool , ':',iewed PHOTOIONIZA TION DETECTION 53 the PID for its application with respect to analysis of alkenes and aromatic compounds; they found that it has a high sensitivity and selectivity to these materials (8). Some suggestion for an improved detector for analysis via capillary columns was also included. 1.2. Theory of Operation In 1976, the first commercial PID was described by HNU Systems, Inc., Newton, Massachusetts (9). Schematic drawings of that detector and a recent (third-generation) PID are shown in Figure 4.1. The process of photonioniza tion starts with the absorption of a photon (hv) by a molecule R. If the energy of the photon is more than or equal to the ionization potential of species R and the carrier gas is given by C, then ionization and the other processes shows here occur: Direct R +e Indirect
C* + R---> R+ + e- + C R + hV'---7 R* Quenching R++e----+R e'+C---+C (kd (k2) (k3) (k 4 ) (k s ) (k6) (k7 ) (k s ) The number of ions formed is equal to the summation of the direct and indirect process minus the quenching process: Note that in k s , ions may also result from a photochemical rearrangement process leading to the formation of R!. This process may account for the photoionization signal observed for the 1 0.2-e V lamp with methylene chloride lionization potential = 11.3 eV), Here. the rearrangement and resultant loss of el 2 may form an ethylenic species with a lower ionization potential. The .L 54 FAR-UV IONIZATION AND ABSORBANCE DETECTORS P " . energy of a 10.2-eV lamp, which corresponds to more than 230 kcal m:. greater than the 80 kcal/mol required to break a typical carbon-carbon c: One can calculate the energy (E) in eV or wavelength (nm) from the follo\;. equation: ;;i W(nm) = 1234.5/E(eV)
Thus, a 10.2-eV lamp has an energy of 121 nm. The photoionization yield (Plyield) is the number of ions produced photon absorbed: = 100(Rs:m)/(number of photons absorbed) Pl yicld From this equation, the number of ions produced (Rs:m) is proportion::. .. the product of the absorption coefficient and the intensity of the lamp.
Electrodes High l UVW;odow Temperature Seal Collector L Electrode perture Accelerating A ;Ii Figure 4.2. Exploded view of ion chamber for PID. PHOTOIONIZA TION DETECTION 55 : "caljmol, j, - -.ypicai PID has two functional parts: an excitation source and an - -bon bond .:.:ion chamber. An expanded view of the lamp and ionization chamber - following - . xn in Figure 4.2. A potential of 100-200 V is applied to the accelerating _ - :de to push the ions formed by UV ionization (above) to the collection _ :de where the current (proportional to concentration) is measured. :hat the collection electrode is shielded to reduce the background (9) ;; configuration of the ion chamber is axial. It can be shown that the most .<0 design of an ionization chamber is one with a coaxial configuration - - _ pc _'" the field is given by E = VJ2.3rlog(a/b) t - is the applied voltage between the collector of radius a and the . -:ional . electrode of radius b, while E is the electric field at any point in c::ce r from the center of the accelerating electrode. Thus, the field increases - as r -4 b. If the collection electrode is in the UV beam, the background by one or two orders of magnitude. On the other hand, placing electrode in the UV beam does not increase the background HNU ion chamber is novel because the collection electrode is shielded - :he UV beam. This design resulted in the lowest background possible -.:lC widest dynamic range. These two features provided a PID with a low ,;;ound current, high sensitivity, and an improvement of 2-3 orders of in linear range (7,9) with a dynamic range of 10 7 1.3. Excitation Sources 1.3.1. Discharge ::xcitation source may be a discharge lamp excited by direct current :,;V), a radio frequency (75-125 kHz), a microwave (2450 MHz), or a laser. :he sealed discharge lamp, the lamp is filled with a nonreactive gas at low . ,,'ure and produces an emission line or lines in the far- UV region. A typical "ion spectrum for a d.c. excited discharge is shown in Figure 4.3. Note . _- :he 11.7 eV has two emission lines at 104.4 and 106.6 nm, the two argon :"_ance lines. or alkaline earth fluorides are used most commonly for the lamp - .20WS since fcw other materials transmit in this region. The short-wave cutoffs for these windows arc lithium fluoride, 105 nm; magnesium 112 nm; calcium fluoride, 122 nm; strontium fluoride, 128 nm. The >::V lamp has a high temperature seal that will withstand temperatures :0 T for short periods of time and tempcratures of 275 "C for proionged -. l1li
56 FAR-CV IONIZATION AND ABSORBANCE DETECTORS 106,7
iii :z UJ ;; 104,8 ........ .. ..'-". '-"-"'- WAVELENGTH (nm) Figure 4.3. Emission spectrum of an 1 L7-eV lamp. periods. This lamp has a lifetime of > 5000 h. The lifetimes of the 9.5- 2::.. 8.3-eV lamps are similar. The 11.7-eV lamp, however, has a lithium fluor:::: window that has a problem with solarization (color center formation) tr.. limits its useful lifetime to a few hundred hours. A recently improved laIC shown in Figure 4.4, has a nickel cathode to bleach the color centers by absorption of the short-wavelength photons and trapping of electrons below the conduction band of lithium fluoride) and extends the lifetime 1000 h. The trapped electrons are released by irradiation with UV (abc'. 7 eV) from the Ni lines in the lamp. Another problem with this lamp is :':-.: shift in short-wavelength cutoff to a longer wa velength as the temperature the window increases. Since the argon emission lines (104.4 and 106.6 nm) close to the cutoff at ambient temperatures, heating the window above 125 reduces the transmission of lithium fluoride and therefore the photon flux.
'I, 1 PHOTorONIZATION DETECTION 57 of the 9.5- an..:: . ;!hium fluoride :'ormation) tha ::1proved lam!" . -::enters (forme..:: electrons the lifetime L ,ith UV (aboL :his lamp is the temperature - :::-.d 106.6nm) aI: - _:.\\ above 125 C :Jhoton flux. A .c igure 4.4. Construction of an improved 1 L 7-10V lamp: #18, metal cathode; #28. lithium fluoride ::dow; #14 and #26, discharge electrodes. 2 FID 11.7 10.2 9.5 8.3 Increasing Selectivity ... . 4.5. Comparison oflhe relative response of the FID with a PID containing various lamps , ;ies). . - same time, the solarization of the window increases as a result of the .. :eased absorption of photons by lithium fluoride. The most popular PID lamp is the 1O.2-eV model, which has the highest :on flux and therefore the greatest sensitivity. This lamp uses a magnesium . -ide window. Figure 4.5 illustrates the variation in sensitivity relative to .;:' energy. In comparing the two chromatograms in Figure 4.6, the differ , -c in sensitivity (attenuation) of benzene should be noted (10). The 10.2-eV J 58 FAR-UV IONIZATION AND ABSORBANCE DETECTORS PID PID 11.7 eV 10.2 eV 1 x 8 10 x 16 ." ." E c 0 u c c ru 0 c '" c " c " '" " " ru '" ru '" e .g 1? " 9 u S u .25 0 2 c " Q 2
1: '" <:: u " 0 jg 1i3 .f' E ro !l1 2'" Figure 4.6. Comparison of the response of a PID with 10.2- or 11.7-cV lamps for a mixlurt aromatic and chlorinated hydrocarbons. lamp is more selective than the 11.7-eV lamp, as it does not respond to low-molecular-weight chloroalkanes. The peak heights are approximately t:-: same, but the attenuation differs by a factor of 20. There are two reasons the difference in intensity of these lamps: (lJ the transmission of lithic: fluoride (in the 11.7-eV lamp) is only that of magnesium fluoride (in f': 1O.2-eV lamp) at the appropriate resonance line output, and (2) the absoh::: energy output of the 1O.2-eV lamp is much higher than that of the 11.7-,,-. lamp. J.3.1. Lasers The UV discharge lamps just described differ from lasers in several respec' They are oflow intensity (about 10 12 photons/s) and are noncoherent. Sin(: --'-_........- - .......-. _IL to ':-:imately . : "easons . lithi .. (in :. - absoL: ::':el1.-:' respee> . Sic-. .u PHOTOIO?>;IZATION DETECTION 59 excited state Figure 4.7. Morse curve of potential energy ground state vs interatomic distance for the ground state (RoJ and the first electronically excited state Atomic Distance (Rd I' I , I 1(: sensitivity of the PID is proportional to the intensity of the light source, .::.sers will have an inherently higher sensitivity. The development of high and high-repetition-rate tunable lJV lasers has made multi Joton ionization (MPI) feasible (II). In this process, the first photon excites molecule from the ground state to the first excited electronic state and the cJsorption of the second photon produces ionization. Typical Morse curves . r photoionization and two-photon ionization are shown in Figure 4.7. In two-photon ionization process, the sum of the energy of the two photons .. :lst be greater than the ionization potential of the molecule and the two can have either identical or different frequencies. Since many large ":".0Iecules have ionization potentials from 7 to 11 eV (176 to 112 nm), two Joton ionization can be initiated with near-UV laser sources. Typical laser ; :,urccs can be excimer pumped frequency doubled dye lasers. Two-photon ionization can produce ionization yields of several percent . )mpared with 10- for conventional photoionization processes (above). :>,ovided the repetition rate (duty cycle) is sufficiently high, the "potential" ; ::nsitivity is higher than that obtained with a typical discharge lamp. Klimetand Wessel (12) showed that laser ph6toionization is ':lotentially" more sensitive than ionization with a discharge lamp, their ':3 ta is not conclusive since the low-picogram detection limits reported can be .::hieved using a conventional UV discharge lamp in the PID. Ogawa et al. : 3) also report detection limits of 6 pg for pyrene using laser two-photon
Two major problems with a laser replacing a discharge lamp in a PID are cost ($50,000 vs. $300) and the size (several cubic meters vs several cubic Should these problems be overcome, the possibility of selectivity .::ctermining isomers or classes of eompounds by two-photon ionization will -;:come a reality . Itt t 111\, III SO, CO 2 COS CS, N,O NO, II (,' Iii. IX 1234 13.79 1US 10.08 12.90 9.78 I I" ' I I" 'U \( 'I' :,( 'Ilt ;\'lwIIPll lUl) (,!('ill 'I, ,( 'I CF ,CeI, (Freon 113) Alipbalic Alcobols,
Ii II II \ \,\}<) 11.78 \ill'hull' \UIHII", :u,,' Amides Methyl amine Ethyl amine ,,_Propyl amine IP 8.97 8,86 8.78 I cll ;lhydl' Illtl,lll Th'\ophcnc Aromatic Compounds 'J,t',
IP 0\ II,', N \I J Paraffills and Cycioparaffins Methane Ethane Propane II-Butane JI-Pentane n-Hexane (I-Heptalle Cyclohc,alle Alkyl Halides Methyl chloride D1chloromcthanc Trichloromethane -retrachloromcthane In HI 10.1 IP 12,98 1165 11.07 1063 10.35 10.18 1008 988 IP 11.28 11.35 11.42 11.47 I hiub. and Sulfides Methyl alcohol Ethyl alcohol n-Propyl alcohol Dimethyl ether Methanethiol Ethancthiol 1- Propanethiol Dimethyl sulfidc Ethyl methyl sullide Diethyl sulfide AIi,.hatic Aldehydes and Ketoues Formaldehyde Acetaldehyde Propionaldehyde IP 10.85 IOAS 10,20 10.00 9,440 9.285 9.195 g,6X5 8,55 8.430 II' 10.87 10.21 9.98
Acetamide Nitromethanc Nitroethane Other Aliphatic Molecules with a N Atom HCN Acetonitrile Propionitrile Olefins, Cydo-olefins, Acetylenes Ethylene Propylene I-Butene tl'afl.)-2,-Butenc I-Hexene 1,3-Butadiene Acetylene l-Butync 10,25 9,77 1l,08 11188 IP \3,91 12,22 1\.84 IP 10.515 9.73, 9.58 9.J.1 9,46 907 I J AI 10.18 BenZeJ1C Toluene Ethylbellzene Biphenyl Phenol N a ph thalenc Styrene a-Xylene p-Xylene Mesltyienc Aniline Fluorobenzcne Chlorobenzenc Bromobcnzcne Todobenzene Miscellaneous Molecules Ethylene oxide Propylene oxide
Melhyl disulfide Ethyl disulfide Phosgene 9,245 8.82 8,76 8.27 8,50 8.12 8,47 8.56 8,445 840 7.70 9.195 9.07 8.98 8.73 IP lO.565 10.22 9.13 8.46 8.27 11.77 I 62 FAR-tiV IONIZATION AND ABSORBANCE DETECTORS 1.4. Gas Chromatography Applications The 11. 7-eV lamp will respond to many low-molecular-weight comp('_. that have more tightly bound electrons and hence higher ionization poter .. Typical applications include the detection of low- or sub-ppm levels of formaldehyde (14), which cannot be measured directly by any.: detector at these levels. The 11.7-eV lamp is also useful for the low-molecular-weight compounds ( < 150 amu) such as CC1 4 , CHCI 3 , and C Z H 2 that have IPs> 10.5 eV. A comparison of the response for typical hazardous waste components using a 10.2- and a 11.7-eV shown in Figure 4.6. Chloroalkanes are not detected with the 1O.2-eV L:- The 11.7-eV lamp PID is similar in response to the FID for many carbons (except for methane, which does not respond), and it also resp: . to inorganic compounds such as C1 2 , PH 3 , and 1 2 , The FID only resp.:-_ to carbon-containing compounds. A list of ionization potentials (IPs) is l_! in Table 4.1. I The major difference between the 10.2- and 9.5-eV lamps is the intensity of the lines. There are certain applications where the 9.5-e V lar: preferable to the I 0.2-e V lamp. These applications include aromatics I:. _ aliphatic matrix such as pentane, mercaptans in the presence of H 2 S. c. _ amines in the presence of ammonia 05). With the 8.3-eV lamp there considerable increase in selectivity compared to the 10.2-eV lamp. A ber.:: or toluene solvent prod uces no response on a PID (8.3 eV). The 8.3-e V l.: was selected for the determination of polyaromatic hydrocarbons (16) bee::_ of the increased selectivity. Typical detection limits were low- to gram levels. Chromatograms ofcomplex mixtures using these three lamp:'.;. . the FID are shown in Figures 4.8 and 4.9. The pattern for the 10.2-eV PID i,,: is similar to that of the FID, as noted previously (17), whereas the chrorr . .:. grams are simpler for the 8.3-eV lamp. In a previous publication (101. reported a lower "apparent" sensitivity for biphenyl (IP 8.27 eV) use: _ a normalization compound. In later work, we found that if (lP = 7.5 eV) is used for normalization, a 10-fold higher response is obta:: for the 8.3-e V lamp, which is attributable to the increased efficienc:. anthracene. II. Langhorst (18) determined the sensitivities for nearly two hundred 'f' ";" pounds for a PID with a 10.2 e V lamp. She found that the PID was a car: counter (on a molar basis); that the sensitivity for alkanes < alkenes matics; that sensitivities for cyclic> monocyclic and branched> r:: branched; and that for substituted benzenes, ring activators increased: sensitivity whereas ring deactivators decreased the sensitivity. Figure..:. depicts a typical curve for a homologous series of hydrocarbons comparee benzene. A comparison of the sensitivity (normalized to benzene) for a var:: - l compounG potentiaL Inanograr. any detection ( _HCI 3 , C 2 K - . ::se for so IT. --cV lamp . . 1.2-e V lam :11any hydL .:.150 respon..: respoDe , ,IPs) is gi\c , the absolc lamp .. : -matics in : - :' of H 2 S, a:-.. .:.:::p there i5 -.;J. A benze: - 8.3-e V lar: ;; I. 16) becm.: . :0 sub-nar.. lamps a:. =eVPIDlal: .'1c chroma: :ion (10), :- cV) used anthrace: : . :-.S:: is obtaiL .:: cfficienc) cc:' =Ixas a cart . .::lkenes < - . .:.::ched > n( : :1creased . Figure ,.l. ;; .::ompared for a varic U.--_ w ,''''._lllll111i11'-------- 63 : ;:ure 4.8. Chromatogram of leaded gasoline by FID compared with Chr(lmatograms by PID :. 9.5. and 8.3 eV). 1 64 FAR-UV IOKIZATION AND ABSORBANCE DETECTORS 2,4,4 TRIMETHYLPENTANE OCTENE trans 1,4 DIMETHYLCYCLOHEXANE 4 DIMETHYLCYCLOHEXANE PID 9,5 eV :r ~ I ~ " ~ , 1 HEXENE 2 3 1 4 5 6 7 TOLUENE 8 m - XYLENE 9 p. XYLENE 10 C U ~ E N E 11 0 - XYLENE 12 MESITYLENE Figure 4.9. Chromatogram of components of gasoline by FlD compared with chromatogri.:: by PID (10.2, 9,5, and 8.3eV). PHOTOIONIZAnON DETECTION 65 ;,a-SltiVlty (MR) ReI. to Benzene ----=-j
Carbon Number Aliphatic HC Olefinic HC Aromatic HC " 10. Comparison of relative molar response (MR) VB. carbon numberfor the PID (10.2 eV). -':ituted benzenes is shown in Figure 4.11. The GC-PID response for :.: ted benzenes changes slightly with electronegative substituents, yet -_ <'onse for the same compounds using LC-PID changes dramatically 1.5, below). - : -iSS (19), a review of the first decade following the commercial introduc .: :he PID showed that only of the applications involved capillary :J;lring the past few years, this has changed significantly--with 30% of using capillary columns. The primary reasons for this change " wider use of capillary columns and the development of an improved '.J 20). A review of capillary column applications with the PID was done _ -:scoll (21). Several papers on the analysis of aromatics in water were ::':.lblished (22,23). - -e PID described in 1986 was a third-generation detector with many - - .ed features (21). We investigated two different cell volumes (175 and _' with similar geometries and found that the latter cell operated opti at low flow rates ( < 1-15 mLjmin) whereas the larger cell provided .. results at the higher flow rates required for packed columns (10). The \olume cell displayed a threefold lower sensitivity at 30 mL/min com -, : :0 the 175-.uL cell. The data are shown in Figure 4.12 . ..: _\\ do the PIDs differ? As a result of the axial ion chamber design _-jed earlier, the only way to reduce the dead volume is to reduce the 66 fAR-UV rONrZATION AND ABSORBANCE DETECTORS POOH. I I I PhGOOH Ph"NO, "0 " '" Ph"Br "" E "I ""el u" ! Ie Phil " I I I I_ I j I I I : I
___ 4-' Figure 4.11. Comparison of the molar response for 'j- ! 10 2') L", j" 50 1,1 GC-PlD with LC-PrD. Respor.se :E 0> 'iii .<::: .:.: ll! a. Figure 4.12. Comparison of peak height (sensitivity) vs. flow rate for the 40- and 175'/lL PID volumes. aperture size. This also reduces the light source flux and therefore the sensi: ity of the PID. If we can reduce or eliminate the makeup flow needed, we ," regain some of the lost sensitivity. We find that if we use a total flow 4 2-3 mL/min for the carrier gas on a capillary column (no makeup), the dead volume detector (40 ItL) will be about three to five times more sensi: i I Makeup flow rate (mLlmin)
PHOTOIONIZA TION DETECTION 67 3.7 nglcmpd. NO MAKEUP Attn: 10X32 ., .. " <II c: .. " " <II e c " J:) " " c D '" 0
r. " "" :c '0 " c:: '5 t1. e
.. " " "'. " u
.. .tl e <> .9 .:: '5
4.13. Effect of eliminating makeup gas for the low dead volume PID, :he 175-,uL cell as a result of the sharper peaks. Peak height vs. flow -,,5 for the 175- and 40-,uL detectors are shown in Figure 4.12, as well .s:tivity vs. flow rate for each detector. It is obvious from the curves .' . ";'.12 that the 175-,uL cell is more suitable when packed and capillary to be used, since makeup gas offsets the dead volume at low flow -." -i-O-,uL cell is most useful for capillary (0.05-0.75 mm) applications. of two chromatograms with and without makeup gas using the ::- : D is shown in Figure 4.13. The addition of 1.5 mL/min of makeup - .. "Hed the tailing and achieved baseline resolution for all the compo ;:h the standard 175-,uL cell, it took 20-30mL/min of makeup gas _ ':: the same result. Other capillary chromatograms obtained for the . 'hown in Figures 4.14 and 4.15 and compared with the FID using P:: - :apillary columns. Note the large difference in sensitivity for the -'::tween the two detectors. '. _ :'ment of "fast GC" using a PID has been underway at the - ::' of Michigan (24) for some time. By using a cryogenic inlet system, '::'0 heated rapidly to inject the sample onto the capillary column .- - . chromatogram in Figure 4.16 was obtained. This shows fast . (baseline) separation of six components within 4. Some add i on a fast PID for GC at the University of Eindhoven, Netherlands, FAR-lJV IONIZATION AND ABSORBANCE DETECTORS 68 C 10 L "'9 ca PIO cscs C 10.2 eV _-+ ~ JL\ 7 . _ L _ - , L _ - - ~ ' - ' - " - - FlO c9 C 10 c 8 C 7 C s c 6 Figure 4.14. Comparison of capillary chromatograms for the FID and PID for alkanes (Nordi bond SE54, 25 M x 0.32mm; temp. prog. 50-180'C at 10C/min). Courtesy ofHNU-Nordion. used a similar cryogenic trap approach but with smaller diameter (O.OS-mm) capillary columns and longer retention times (minutes vs. seconds) as a result of the molecular weight of the components selected for analysis. Results for a peppermint oil sample obtained using a O.OS-mm capillary column are shown in Figure 4.17. Van Es and Rijks (25) used makeup gas for the O.OS-mm columns with no adverse effect on the performance of the detector. They found a minimum detectable quantity of 0.2 pg for anthracene. This is an order of magnitude lower than packed column detection limits and about two to three times better than detection with the tow dead volume cell and no makeup gas (10). Although the reduction of cell volume and detector pressure are equivalent alternatives for other concentration detectors such as the TCD, van Es and Rijks (2S) found that, for the PID, the reduction of dead volume was more critical than reducing the pressure. They also found that the effect of addition of makeup gas and reduced pressure on detection limits was equivalent. ;'or alkanes (Nc' , ofHNU-Norc, ,::neter (O.OS-n' as are, _'.sis. Results L ' :.'lumn are she :'Of the 0.05 :;;.;:tor. They fc ' is an ." ,:]out two to t .-.0. no makeu[ : ,:re are TCD, van E, 'ulume was :: effect of adZ: equivalent PID "-::1.2 eV >. .f: " " c: >1 '" " 0 >. c: 1: " (!! ::' E " ;; N c: " .0 .s:: ., .. mparison of capillary chromatograms for the FID and PID for aromatic hydro : ::ibond SE54, 25 M x 0.32 mm; temp. 40 C/3 min, 40-120 DC at 5CjminJ, , :';:--;CNordion. PHOTOIO]\,lZATION DETECTION 69 '" " c: " '" .ll : " " '" " " " o 2 4 TIME (secends) c::romatogram with high-speed PID. Courtesy of S. Levine. University of 1 70 FAR-UV IONIZATION AND ABSORBANCE DETECTORS rr FrD II I I ! II I I \III
Fro II II!I I : I I '\ /1 i 11 __ \LJ o---I---r------a---4----s-=:::niln-- 7 -- Figure 4.17. Fast chromatogram of peppermint oil using PID and FID in series (50-JlM column). Courtesy of J. Rijks, University of Eindhoven. 1.5. Liquid Chromatography Many papers have been published over the past 15 years on interfacing a PID to a liquid chromatograph (26), yet there are no commercial PIDs available for liquid chromatography (Le). This detector remains one of the more interesting detectors for HPLC and awaits development, as noted several years ago by the author (to). One interesting feature of the PID is that with
_______._._._. PHOTOIONIZATION DETECTION 71 TabJe 4.2. IOllizatioll Potelltials of Some Common So}yents fOF HPLC Ionization Potential Solvent (eV) \Vater 12.35 \fethanol 10.85 Chloroform 11.42 Carbon tetrachloride 11.47 Acetonitrile 12.22 Pentane 10.35 ..::TiP, most of the common solvents used for reversed phase LC . , 'nized (see Table 4.2). -.:..:' for LC have involved flowing the liquid directly into the cell or - ; ::-:'c liquid before passing it through the cell (27). Both these ap '. i: drawbacks, and neither solves the problem of quenching, which . ::.; D to the detection of quantities above the low-microgram (by the . .:ch) or nanogram level (by the vaporization approach), in LC. . 'land, low-picogram quantities are detected in Gc. If these GC . c. ,,]s were approached more closely, this detector would become _ .. LC. , ..::-.d Jorgensen (28) have described a PID that uses open tubular ., :C'Jd of conventional LC columns. The sample is vaporized before . ::le PID. Berthold et aL (29) have used laser two-photon photo - .: detect 28 polyaromatic hydrocarbons (PAHs) in an LC effluent. limits were 50-70pg for the PAHs compared with 6pg . ':- :rted by Ogawa et al. (13) for GC-PID. Note that the quenching _:<:-:bt:d above also occurs with the laser source. A typical HPLC-PID , '.aporizes the solvent and sample before detection is shown in , Selected detection limits obtained (26) by LC-PID are shown in sensitivity of this detector, and therefore the detection limits, :;;rably from those obtained by Langhorst (18) for the GC-PID ..: . :\e. Why? The mechanism appears to be different. The simple - ::escribed above for GC-PID [R +hv -> R + +e- (kdJ is per ::-.e presence of high concentrations of electronegative species .. where the solvent (S) is volatilized along with low levels of ::. '. the mechanism is quite different. The reaction (kl) occurs to a . the majority of the UV radiation is absorbed by the - - ionization potential of these common LC solvents is higher ::.:-:cp energy; therefore the following indirect photoionization ---- ---- --- 1 72 FAR-liV IONIZATION AND ABSORBANCE DETECTORS r----------------- INJECTOR I I I VAPOR COIL COLUMN I I I ! ! I I I L__ Figure 4.18. Schematic or LC-PfD, ........PID --------1 , I I -"OVEN Table 4.3. Selected Detection Limits Obtained by LC-PID Detection Limits Compound (ng) Bromobenzene 4 Chlorobenzene 4 Nitrobenzene 2 Phenol 3 Benzoic acid 2 Dioctyl phthalate 16 Benzene 15 Anthracene 75 mechanism will undoubtedly predominate: 5 + hv --.5* (kg) 5* R ----> R+ + e - 5 (k 10 ) The quenching will be considerably greater in the LC mode than the GC mode because of the higher concentrations of electronegative species in the condensed solvents compared to the inert gases in the GC mode. This leads to the following quenching reactions: S+e----+S* (kill 5* R--.R -'-e-+S (k12l .......... I 74 FAR-CV IONIZATION AND ABSORBANCE DETECTORS - - ~ - ~ - ~ - Dichlorobiphenylsulfone in Dichloromethane , - - - - - - - ~ . ---- ~ - - - - - - - - , -1.7 pg Injection Triazopan in Acetronitrile Figure 4.19. SFCPID of Jow-picogram levels of triazopan. Reprinted with permission from l Hayhurst and B. Magill, Int. Labmate 14. (Aug., 1988). Copyright [988, International Labmate. results attainable by Gc. The results of Hayhurst and Magill (34) show sub nanogram levels of anthracene detected by PIO using similar conditions. This suggests some quenching or loss due to the CO 2 carrier gas. Hayhurst and Magill (34) used a 6 m x 50 11m capillary column system and detected low- or sub-picogram quantities of dichlorobiphenyl sulfone and triazopan. A typical chromatogram of a low-picogram sample oftriazopan is shown in Figure 4.19. Advantages of SFCPIO include reduced analysis time by eliminating solvent tailing, use of a wider range of solvents that produce "no response" (such as methanol, methylene chloride, acetonitrile, and Freons), detection of a wide range of compounds that do not absorb in the UV, and increased sensitivity over the FlO.