Instrumental Requirements for Nanoscale LiquidChromatography

J. P. Chervet* and M. Ursem

LC Packings, Baarsjesweg 154, 1057 HM Amsterdam, The Netherlands

J. P. Salzmann

LC Packings Inc., 80 Carolina Street, San Francisco, California 94103

Nanoscale liquid chromatography (nano-LC), with packedcolumns of typically 75 µm i.d. × 15 cm length, packedwith C18, 5 µm of stationary phase, and optimal flow ratesof 180 nL/min, can be considered as a miniaturizedversion of conventional HPLC. Using the down-scalingfactor, which corresponds to the ratio of the columndiameter in square, (dconv/dmicro)2, excellent agreementbetween the theoretically calculated values and the valuesobtained using the down-scaling factor (∼3800) has beenobserved. This factor was applied to all system compo-nents, including flow rate, injection and detection vol-umes, and connecting capillaries. Down-scaling of aconventional HPLC system to a nano-LC system is easyto realize in practice and involves using a microflowprocessor for nanoflow delivery (50-500 nL/min), alongitudinal nanoflow cell (e3 nL), a microinjection valve(e 20 nL), low-dispersion connecting tubing, and zerodead volume connections. Excellent retention time re-producibility was measured with RSD values of (0.1%for isocratic and (0.2% for gradient elution. Platescounts of more than 100 000/m indicate the excellentperformance of the entire nano-LC system. With minimaldetectable amounts of proteins in the low femtomole andsubfemtomole ranges (e.g., 520 amol for bovine serumalbumin), high mass sensitivity was found, making nano-LC attractive for the microcharacterization of valuableand/or minute proteinaceous samples. Coupling nano-LC with concomitant mass spectrometry using nanoscaleion spray or electrospray interfaces looks very promisingand is obviously the next step for future work.

The growing interest in analyzing minute samples in variousfields, such as environmental, clinical, forensic, and pharmaceuticalchemistry and biotechnology, is one of the driving forces for therapid development of microseparation techniques such as micro-LC and capillary LC in combination with mass spectrometry (MS).

Today, micro-LC and capillary LC with column diameters oftypically 300 µm i.d. and flow rates of 4 µL/min have foundacceptance as routine techniques in various laboratories due totheir superiority in situations with limited sample amounts and/or sample concentration.1-12

Recent developments on new electrospray interfaces, such as“nano ion spray”, lead to even lower flow rates, in the range of 20nL/min to 1 µL/min.13,14 Theoretically, such miniaturizationincreases the mass sensitivity and should allow for furtherenhancement of the limit of detection (LOD), in both mass andconcentration sensitivity. The first attempts at using packedcapillary columns with very small inner diameters, ranging from20 to 70 µm, and flow rates on the order of 50-200 nl/min werereported several years ago by Novotny15,16 and Jorgenson.17,18 Thedirect coupling of packed microcapillary columns with ESI/MSwas shown recently by Hunt et al.19,20

In this paper, instrumental requirements such as highlyreproducible delivery of nanoliter flows under isocratic andgradient conditions, highly sensitive UV detection using Z- orU-shaped nanoliter UV flow cells, and the use of zero dead volumeconnections are presented, in order to realize the potential of

(1) Henzel, W. J.; Bourell, J. H.; Stults, J. T. Anal. Biochem. 1990, 187, 228-233.

(2) Griffin, P. R.; Coffman, J. A.; Hood, L. E.; Yates, J. R. Int. J. Mass Spectrom.Ion Processes 1991, 11, 131-149.

(3) Moritz, R. L.; Simpson, R. J. J. Chromatogr. 1992, 599, 119-130.(4) Kassel, D. B.; Luther, M. A.; Willard, D. H.; Fulton, S. P.; Salzmann, J.-P. In

Techniques in Protein Chemistry IV; Angeletti, R. H., Ed.; Academic: SanDiego, CA, 1993; pp 55-64.

(5) Lane, S. J.; Brinded, K. D.; Taylor, N. L. Rapid Commun. Mass Spectrom.1993, 7, 492-495.

(6) Murata, H.; Takao, T.; Anahara, S.; Shimonishi, Y. Anal. Biochem. 1993,210, 206-208.

(7) Kassel, D. B.; Shushan, B.; Sakuma, T.; Salzmann, J. P. Anal. Chem. 1994,66, 236-243.

(8) Lewis, D. A.; Guzzetta, A. W.; Hancock, W. S.; Costello, M. Anal. Chem.1994, 66, 585-595.

(9) Roboz, J.; Yu, Q.; Meng, A.; van Soest, R. Rapid Commun. Mass Spectrom.1994, 8, 621-626.

(10) Battersby, J. E.; Mukku, V. R.; Clark, R. G.; Hancock, W. S. Anal. Chem.1995, 67, 447-455.

(11) Chowdhury, S. K.; Eshraghi, J.; Wolfe, H.; Forde, D.; Hlavac, A. G.; Johnston,D. Anal. Chem. 1995, 67, 390-398.

(12) Simpson, R. C. J. Chromatogr. A 1995, 691, 163-170.(13) Mann, M.; Wilm, M. Intl. J. Mass Spectrom. Ion Phys. 1994, 136, 167ff.(14) Covey, T.; Shushan, B.; Kaudewitz, H. Proceedings of the 1st European

Symposium of the Protein Society; Cambridge University Press: Cambridge,UK, 1995; Poster 456.

(15) McGuffin, V. L.; Novotny, M. J. Chromatogr. 1983, 255, 381-393.(16) Karlsson, K. E.; Novotny, M. Anal. Chem. 1988, 60, 1662 ff.(17) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 61, 1128-1135.(18) Kennedy, R. T.; Jorgenson, J. W. J. Microcolumn Sep. 1990, 2, 120-126.(19) Huczko, E. L.; Bodnar, W. M.; Benjamin, D.; Sakaguchi, K.; Zhu, N. Z.;

Shabanowitz, J.; Henderson, R. A.; Apella, E.; Hunt, D. F.; Engelhard, V. H.J. Immunol. 1993, 151 (5), 2572-2587.

(20) Henderson, R. A.; Cox, A. L.; Sakaguchi, K.; Apella, E.; Shabanowitz, J.;Hunt, D. F.; Engelhard, V. H. Proc. Natl. Acad. Sci. U.S.A. 1993, 90 (21),10275-10279.

Anal. Chem. 1996, 68, 1507-1512

0003-2700/96/0368-1507$12.00/0 © 1996 American Chemical Society Analytical Chemistry, Vol. 68, No. 9, May 1, 1996 1507

higher mass and concentration sensitivity of nanoscale LC (nano-LC).

Further, it is demonstrated how conventional HPLC instru-mentation can be modified for successful use in nano-LC. Pre-injection flow splitting with high split ratios of 1:2000, the use ofhigh-sensitivity nanoliter UV flow cells with volumes as small as3 nL, and different injection techniques for optimal nondispersiveinjection are discussed in detail.

EXPERIMENTAL SECTION(a) Reagents. A reversed-phase test mixture containing uracil

(4.3 µg/mL), naphthalene (57.5 µg/mL), biphenyl (46.0 µg/mL),fluorene (8.21 µg/mL), anthracene (9.85 µg/mL), and fluoran-thene (26.3 µg/mL) was used for efficiency (N/m) measurements.All compounds were purchased from Fluka AG (Buchs, Switzer-land). Acetonitrile (ACN), methanol (MeOH), and water wereall of HPLC grade (LabScan Ltd., Dublin, Ireland).

For the separation of the protein standards, a test mixture wasmade containing 0.2 mg/mL each of ribonuclease, insulin, cyto-chrome c, lysozyme, and bovine serum albumin (Sigma Co., St.Louis, MO). Sample solvent was 0.1% trifluoracetic acid (TFA)in water/ACN (95:5 v/v). To assure high UV transparency at lowwavelengths, supergradient ACN (Catalog No. C2527, LabScan)and TFA from ampules (Catalog No. T-6508, Sigma Co.) were usedfor the preparation of the mobile phases, which consisted of water/ACN/TFA mixtures (see Chromatography, below).

(b) Instrumentation. Figure 1 shows the schematics of thenano-LC instrumentation. The main system components were aconventional HPLC pump (either low- or high-pressure mixinggradient can be used), a conventional UV detector equipped witha U-shaped nanoflow cell, and a microinjection valve. A microflowprocessor (Acurate, Model AC-2000-NAN, LC Packings) placedbetween the pump and the injector was used to split the flow downto the required nanoflow. The microflow processor compensatesfor the viscosity changes during gradient elution and, hence,results in highly constant flow delivery. Further, it streamlinesthe baseline noise generated by the movement of the pump pistonand/or stepper motor. More details about the function andworking of this system have been published previously.21,22

Under isocratic conditions, the split flow was recycled back tothe reservoir. Under gradient conditions, it was guided to waste.

Typical pump settings were 300-400 µL/min to assure properfunctioning of the conventional HPLC pump and accurate pro-

portioning of the gradient. The flow was split by a ratio of ∼1:2000 to achieve the required nanoflow of 150-200 nL/min.

Low-pressure in-line filters (e.g., 0.2 µm disposable filters fromWhatman Inc., Clifton, NJ, placed between the reservoir and thepump) are recommended to avoid clogging of the nano-LC system.Further, the microflow processor contains built-in, replaceable,high-pressure in-line filters to trap particles originating fromnormal pump seal wear and/or the mobile phase.

A commercial microinjection valve equipped with a 20 nLinternal loop (Model C4-1004.02, Valco Europe, Schenkon, Swit-zerland) was used for nanoscale injections. To allow systemevaluation (plate count measurements), injections of 2-3 nL arerequired. Hence, an additional 1:10 split of the sample was appliedby connecting the column to the injector via an additional split T.

For 20 nL large volume injections, the above-mentioned valvecan be used without any sample split, simply by installing theappropriate rotor loop (Model C4-10R4.02, Valco). However,sample focusing (peak compression) conditions must be assured.On the injection side, the column is directly connected to theinjection valve with a 1/16 in. o.d. PEEK sleeve connector, thusguaranteeing minimal dispersion. On the detection side, thecolumn is directly connected to the nanoflow cell with a Teflonconnector (TF-250, LC Packings) that allows for a dead volumefree butt connection.

Detection was performed with a conventional UV detector(model 1050 VWD, Hewlett Packard, Waldbronn, Germany)equipped with a U-shaped nanoflow cell of 8 mm path length and3 nL cell volume (Model UZ-HP-NAN, LC Packings). The flowcell had connecting capillaries (inlet and outlet) of e20 µm i.d.and short length (e15 cm) to assure minimal extracolumndispersion. For further details, see refs 23-25.

(c) Chromatography. For isocratic system evaluation a 75µm i.d. × 30 cm column, packed with C18, 5 µm particles wasused (Fusica, Catalog No. NAN75-30-05-C18, LC Packings). Themobile phase consisted of acetonitrile/water (80:20 v/v) at a flowrate of 180 nL/min. Injection volume was ∼2.0 nL using the 20nL injection valve and a 1:10 sample split (made by a T-piece,Valco, and a fused silica restrictor capillary). Detection was UVat λ ) 254 nm.

A step gradient was measured without a column by directlyconnecting the microflow processor to the UV detector, equipped

(21) Chervet, J. P.; Meyvogel, C. J.; Ursem, M.; Salzmann, J. P. LC-GC 1992,10 (2), 140-148.

(22) Chervet, J. P. (LC Packings). Eur. Pat. Appl. 0597552A1, 1993.

(23) Chervet, J. P.; Ursem, M.; Salzmann, J. P.; Vannoort, R. W. J. High Resolut.Chromatogr. 1989, 12 (5), 278-281.

(24) Chervet, J. P.; Van Soest, R. E. J.; Ursem, M. J. Chromatogr. 1991, 543,439-449.

(25) Chervet, J. P. (LC Packings). Eur. Pat. Appl. 0495255A1, 1991.

Figure 1. Schematics of the nano-LC system. Conventional HPLC system with microflow processor, capillary column, and UV detecor withU-shaped nanoflow cell (modifications in black).

1508 Analytical Chemistry, Vol. 68, No. 9, May 1, 1996

with an ∼100 pL on-column flow cell (20 µm i.d. fused silica) toassure linear response over the entire range and minimal disper-sion.

Gradient separations were performed at ambient temperatureon a 75 µm i.d. × 15 cm column, packed with Vydac, C18, 5 µmdiameter, 300 Å pore stationary phase (Fusica, Catalog No.NAN75-15-05-C18P3, LC Packings).

Gradient was from 25% B to 55% B in 20 min, whereby themobile phase consisted of (A) 0.1% TFA in water/ACN (90:10 v/v)and (B) 0.08 TFA in water/ACN (10:90 v/v). To compensate forthe gradient delay (dwell volume), the 20 nL sample was injected2 min after the gradient start at a flow rate of 180 nL/min.Detection was UV at λ ) 220 nm. The mobile phase was renewedevery second day and used under continuous helium sparging.

RESULTS AND DISCUSSIONNames and Definitions. Table 1 lists the most common

names for the different HPLC techniques. So far, most of thetechniques have been named on the basis of the type or the innerdiameter of the column tubing. With the use of new tubingmaterials (e.g., fused silica capillaries, polymeric tubing, stainlesssteel tubing, micromachined channels on silicon wafers, etc.), thisnomenclature is not sufficiently precise. Therefore, we suggestdefining the chromatographic technique according to the flow raterange used rather than by the inner diameter of the tubing or itsmaterial. Hence, for packed microcolumns with 10-150 µm i.d.and flow rates of 10-1000 nL/min, the technique would be callednano-LC.

Fundamental Factors in Nano-LC. (a) Down-Scale Factor.In nano-LC, as well as in all other microcolumn LC techniques,all volumes must be down-scaled by a factor (f),

where dconv and dmicro are the diameters of the conventional andmicroscale HPLC columns, respectively. This down-scaling isinevitable in order to maintain the performance of the microsepa-ration system and to work under optimal conditions. Down-scalingfrom a 4.6 mm i.d. conventional HPLC column to a 75 µm i.d.packed nano-LC column equals a factor of ∼3800 (3762). Thisfactor applies to all components or parameters of the nano-LCsystem, such as flow rates, injection and detection volumes, andconnecting capillaries.

(b) Required Flow Rates. The volumetric flow rate (F) in achromatographic system is defined as26,27

where u is the linear velocity of the mobile phase, dc the columndiameter, and ε the column porosity.

With a typical linear velocity of u ) 1 mm/s and a columnporosity of ε) 0.7, the volumetric flow rate for a 75 µm i.d. columncan be calculated as 186 nL/min. This value compares favorablywith the value obtained by using the down-scaling factor and aflow rate of 0.8 mL/min for the conventional column.

(c) Maximum Injection Volume. As in conventional HPLC,the maximum injection volume (Vmax) that can be injected ontothe microcolumn can be expressed by the following equation,

where θ is the fractional loss of the column plate number causedby the injection, K is a constant describing the injection profile, Lthe column length, k the retention factor,28 and N the theoreticalplate number.

Allowing a typical value of 5% in volume dispersion (θ ) 0.05)and column porosity ε ) 0.7, assuming that the injection profileis almost an ideal rectangular plug (K ) 4) and that the columnshave good efficiency with reduced plate height h ) 2, andsubstituting N by

where dp is the particle size of the stationary phase, eq 3 thenbecomes

Thus, the maximum injection volume is proportional to the squareof the column diameter (dc

2), the retention factor (k), and thesquare root of the column length (L) and the particle size (dp).

For a typical 75 µm i.d. × 15 cm microcolumn, packed withC18, 5 µm particles, the injection volume for an slightly retainedpeak (k ) 1) should be not larger than 6.1 nL. Similar values areobtained by dividing the typical injection volume of a conventionalcolumn, e.g. 20 µL, by the down-scaling factor.

In practice, however, there are no commercial injection valvesavailable that allow for the injection of volumes less than 20 nL.Therefore, a 1:10 sample split was used for performance measure-ments of the nanocolumns (h/u curve). For large volumeinjections, however, the sample should be dissolved in a weakersolvent than the mobile phase, thus allowing the enrichment(adsorption) of the analytes on top of the column and avoidingpeak-broadening.

(d) Detection Volume. To maintain the high efficiency ofthe microcolumn and to avoid peak dispersion during detection,the volume of the sensing region must be adapted. Using UVdetection, the sensitivity is proportional to the path length of theflow cell (Lambert-Beer), and due to shot noise limitations, theflow cell aperture should be relatively large. Consequently, a

(26) Ishii, D., Ed. Introduction to Microscale HPLC; VCH Publishers, Inc.: NewYork, 1988.

(27) Kucera, P., Ed. Microcolumn HPLC; Journal of Chromatography Library 28;Elsevier Publishers: Amsterdam, 1984.

(28) Ettre, L. S. Pure Appl. Chem. 1993, 65 (4) 819-872.(29) Chervet, J. P.; van Soest, R. E. J.; Salzmann, J. P. LC/GC Int. 1992, 5, 33-

38.(30) Chervet, J. P.; Ursem, M. GIT Spez. Chromatogr. 1992, 12, 38-44.

Table 1. Names and Definitions for HPLC Techniques

column i.d. flow rate name

3.2-4.6 mm 0.5-2.0 mL/min conventional HPLC1.5-3.2 mm 100-500 µL/min mirobore HPLC0.5-1.5 mm 10-100 µL/min micro-LC150-500 µm 1-10 µL/min capillary LC10-150 µm 10-1000 nL/min nano-LC

f ) dconv2 /dmicro

2 (1)

F ) uπdc2ε/4 (2)

Vmax ) θKπεdc2L(k + 1)/xN (3)

N ) L/hdp (4)

Vmax ≈ 0.622dc2(k + 1)xLdp (5)

Analytical Chemistry, Vol. 68, No. 9, May 1, 1996 1509

compromise between maximum sensitivity and minimum peakdispersion must be found. Using Z- or U-shaped capillary flowcells, Chervet et al.21 have demonstrated excellent sensitivitieswith minimal dispersion even under CE conditions. Ideally, themaximum detection volume should be not more than one-tenthof the volume of the peak that is leaving the separation column.For a nonretained peak the dilution over the column is ∼4-fold.From eq 1 or 3, the detection volume for an ideal nanoflow cellcan be calculated as 1 nL. For peaks with k > 1, it has beenshown in conventional HPLC that flow cell volumes are usuallysimilar to the injection volumes. Hence, in our experiments, weused flow cells of 3 nL volume and 8 mm path length.

(e) Maximum Sample Mass (Loadability). Similar to themaximum injection volume as defined in eq 3, the maximumsample mass (Mmax) that can be injected onto a column can beexpressed as26,27

where Cm is the maximum sample concentration of a peak elutedfrom the column. Thus, for two columns of the same length andwith identical packing and performance characteristics, differingonly in the column diameter, the ratio of the maximum injectablemass that can be placed on each column to give the same sampleconcentration (Cm) is equal to the ratio of the column diametersin square,

and equals the down-scaling factor (f) of eq 1.In practice, this means that, for a 75 µm i.d. × 15 cm packed

column, the maximum injected sample mass (loading capacity)for a nonretained compound (k ) 0) should not exceed 50 ng. Inall experiments, the injected sample mass was significant lowerto avoid the risk of any mass overload and to assure optimalseparation conditions.

Flow rate and volume requirement for nano-LC are sum-marized in Table 2.

System Evaluation. To 1evaluate the performance of theentire nano-LC system, chromatographic reproducibility (isocraticand gradient) as well as efficiency and sensitivity measurementshave been conducted.

(a) Isocratic Elution. For reproducibility and efficiencymeasurements, the reversed-phase test mixture was injected underisocratic conditions, using ACN/water (80:20 v/v) for elution. To

avoid peak dispersion, injection volumes were set at ∼2 nL. InFigure 2, a typical chromatogram is shown for the separation ofthe test mixture.

In Figure 3, the h/u curve for fluorene (peak 4) and fluoran-thene (peak 6) is shown. The data for the h/u curve are listed inTable 3.

At a linear velocity (u) of 1.16 mm/s, which corresponds to aflow rate of 175 nL/min, maximum column plate counts of morethan 100 000/m (reduced plate heights h e 2.0) were measured.These high plate counts illustrate the high performance of thenano-LC system. Furthermore, the experimentally determinedflow rate corresponds well with the calculated value of 186 nL/min. With RSD values of (0.1% on retention times (n ) 20),excellent reproducibility under isocratic conditions was observed.

(b) Gradient Elution. Proper gradient elution at nanoliterflow rates is possible only when accurate proportioning, homo-geneous mixing with minimal delay, and reproducible delivery ofthe mobile phase are realized. At the present time, there are nomicropumping systems available that allow for nanoflow delivery

Table 2. Flow Rate and Volume Requirements forNano-LC (50 and 75 µm i.d. Packed Columns)

packed column

50 µm i.d. 75 µm i.d.

flow rate (nL/min)a 80 180injection volume (nL)b e1.5 e3.0flow cell volume (nL) e1.5 e3.0connecting capillaries (µm)c e10 e20

a For maximum plate counts according to the h/u curve. b Fornonretained peak, k ) 0. c Maximum length, 15 cm.

Mmax ) Cmπεdc2Lk/2xN (6)

(Mmax)conv

(Mmax)micro)

dconv2

dmicro2 ) f (7)

Figure 2. Isocratic separation of test mixture for system evalua-tion: uracil (1), naphthalene (2), biphenyl (3), fluorene (4), anthracene(5), and fluoranthene (6); flow rate, 180 nL/min (other conditions, seeExperimental Section).

Figure 3. h/u curve. Column: 75 µm i.d. × 30 cm, packed withC18, 5 µm.

1510 Analytical Chemistry, Vol. 68, No. 9, May 1, 1996

under these conditions. Therefore, the concept of microflowprocessing, using conventional HPLC pumps combined with flowspitting and simultaneous viscosity change compensation, wasapplied.21,22 In comparison to microflow processing in micro-LCand capillary LC, the main difference for nano-LC was the use ofhigher splitting ratios, typically 1:2000. Thus, almost any con-ventional HPLC pumping system can be successfully used fornano-LC (isocratic and gradient). The flow rate of the conven-tional HPLC pumping system is usually set at 400 µL/min, whichresults in a nanoflow of 200 nL/min using a 1:2000 splitting rate.If syringe pumps are used, lower spitting ratios (∼1:500) andsmaller input flows (100 µL/min) can be applied (data not shown).To measure the proportioning accuracy, a step gradient wasprogrammed. Figure 4 shows the step gradient ranging from 0%B to 100% B with steps of 10% B every 5 min at a total flow rateof 180 nL/min (for experimental conditions, see Chromatography,above).

Over the entire gradient range, only minimal deviation fromthe theoretical step profile was observed, thus illustrating homo-geneous mixing and accurate proportioning of the mobile phases.

Further, the system provides a minimal gradient delay of only3 min (without column installed). That corresponds to a dwellvolume of less than 600 nL. Baseline noise over the entiregradient is similar to that of a conventional HPLC step gradient,indicating good mixing and streamlining capabilities of microflowprocessors.

The use of proteinaceous samples rather than small organicmolecules is recommended for sensitive monitoring of gradient

reproducibility.31 Owing to their strong dependency on theretention factor k from the mobile phase composition, they areideally suited as test compounds. Minor changes in the gradientcomposition result in notable retention time variations. Therefore,gradient reproducibility was measured by consecutive injectionsof a test sample containing five standard proteins, as shown inFigure 5.

To circumvent the need for sample splitting during injectionand to avoid the loss of valuable sample, a 20 nL sample wasinjected directly onto the packed 75 µm i.d. nanocolumn, usingsample focusing (peak compression). In the case of our proteinstandards, the proteins were dissolved in acidified water with onlya small amount of organic modifier (0.1% TFA in water/ACN, 95:5, see Experimental Section). This allows for effective samplefocusing without any measurable peak dispersion and the directuse of microinjection valves without any sample loss due to theabsence of the sample split. Optimal handling of minute samplesusing large and extra-large volume injection in nano-LC will be

(31) Dolan, J. W. LC-GC 1989, 7 (1), 18-24.

Table 3. Data for the h/u Curve

N h

flow (nL/min) t0 E0 (porosity) u (mm/s) fluorene fluoranthene fluorene fluoranthene

25 31.48 0.594 0.16 13 150 12 536 4.56 4.7950 13.45 0.508 0.37 24 574 23 549 2.44 2.55

100 7.45 0.562 0.67 28 907 27 561 2.08 2.18150 4.97 0.563 1.01 29 799 28 161 2.01 2.13175 4.30 0.568 1.16 30 127 27 872 1.99 2.15200 3.74 0.565 1.34 28 188 25 982 2.13 2.31225 3.30 0.561 1.52 26 331 23 457 2.28 2.56250 2.90 0.547 1.72 25 483 23 711 2.35 2.53275 2.67 0.554 1.87 23 087 21 842 2.60 2.75300 2.54 0.575 1.97 22 233 21 388 2.70 2.81350 2.24 0.592 2.23 21 712 20 687 2.76 2.90400 1.91 0.577 2.62 19 281 18 332 3.11 3.27450 1.57 0.533 3.18 17 767 17 274 3.38 3.47500 1.48 0.559 3.38 16 657 16 190 3.60 3.71550 1.34 0.556 3.73 16 070 15 261 3.73 3.93600 1.26 0.571 3.97 15 184 14 893 3.95 4.03

Figure 4. Step gradient at 180 nL/min using conventional HPLCpump with microflow processor. Gradient from 0 to 100% B; step,10% every 5 min. (A) MeOH, (B) MeOH + 10% acetone; UV detectionat λ ) 254 nm.

Figure 5. Separation of protein standards: ribonuclease (1), insulin(2), cytochrome c (3), lysozyme (4), and bovine serum albumin (5);flow rate, 180 nL/min (other conditions, see Experimental Section).

Analytical Chemistry, Vol. 68, No. 9, May 1, 1996 1511

discussed in an upcoming paper.32 With the injection of 20 nL ofthe protein standard solution (0.2 mg/mL of each protein), a totalof 4 ng of each protein was injected. These amounts correspondto 292 fmol of ribonuclease, 696 fmol of insulin, 308 fmol ofcytochrome c, 264 fmol of lysozyme, and 60 fmol of bovine serumalbumin. With signal (peak height)-to-noise (S/N) levels rangingfrom 286 for cytochrome c to 710 for insulin, the LODs shown inTable 4 can be calculated.

To confirm the calculated LOD values, the protein standardsolution was diluted 1:100 (using 0.1% TFA in water/ACN, 95:5)and reinjected under the same conditions as described above. With2 times lower LOD, similar values were observed, as reported inTable 4 for the five proteins.

These data demonstrate unambiguously the high mass sen-sitivity of nano-LC systems. In Figure 6, the reproducibility ofseven consecutive injections is shown as a 3D plot. With RSDvalues of (0.2% (n ) 7) for retention times, excellent reproduc-ibility was found, indicating the high performance of the entirenano-LC system.

CONCLUSIONSThe use of nano-LC with packed columns of 75 µm i.d. and

typical flow rates of 180-200 nL/min looks very promising forthe analysis of minute and/or valuable samples.

Conventional HPLC instrumentation can be easily upgradedto nano-LC. Modifications thereof include the use of microflowprocessors for flow splitting, longitudinal capillary flow cells (U-or Z shape) for sensitive UV detection, microinjection valves fornanoliter injections, and connecting capillaries of small innerdiameter (e20 µm) to avoid peak dispersion. Installation is veryeasy, extremely economical due to the use of existing hardware(pumps and detector), and done within a few minutes. Further,the use of conventional HPLC instrumentation in nano-LC allowsthe entire range of HPLC techniques (conventional HPLC, micro-LC, capillary LC, and nano-LC) to be covered with one instrumentversus the need for dedicated hardware. Over the last 2 years,we have been working with such nano-LC systems. We havefound great ease of use and almost no downtime. Both isocraticand gradient elutions are feasible at nanoflows, with excellentretention time reproducibility, similar to that of conventionalHPLC.

The reduced plate height of 2 reflects the overall highperformance of the nano-LC system. Further, large volumeinjections of several nanoliters (g20 nL) can be applied directlyonto the packed capillary column with sample focusing to allowtrace analysis and to enhance the minimal detectable concentra-tion.

With minimal detectable amounts on the order of 35 pg or523 amol for bovine serum albumin, extremely low limits ofdetection (LODs) were realized using simple UV detection. SuchLODs are not attainable by either conventional HPLC or micro-LC. Also noteworthy is the high resolution of the proteinseparation. The increased sample-to-phase ratio in nano-LC mightbe an explanation.

Using nano-LC in conjunction with new MS techniques andnew interfaces, such as the “nano ion spray” or ion trap ESI/MS,one might expect further enhancements of the LOD, required forthe microcharacterization of proteins and peptides.

In an upcoming paper,32 results will be presented for the useof large and extra-large volume injections to further enhance theconcentration sensitivity in nano-LC.

Received for review September 1, 1995. AcceptedJanuary 20, 1996.X

AC9508964

(32) Chervet, J. P.; et al. Manuscript in preparation. X Abstract published in Advance ACS Abstracts, March 1, 1996.

Table 4. Limits of Detection for Proteins UsingNano-LC (UV Detection, λ ) 220 nm)

compound S/N LODa (pg) LODa (fmol)

ribonuclease 330 37 2.7insulin 710 17 2.9cytochrome c 290 42 3.2lysozyme 560 22 1.4bovine serum albumin 340 35 0.5

a S/N ) 3.

Figure 6. Reproducibility nanoscale gradient using protein stan-dards (3D plot). Flow rate, 180 nL/min.

1512 Analytical Chemistry, Vol. 68, No. 9, May 1, 1996