Mitigation of analyte loss on metal surfaces in liquid chromatography

The adsorptive loss of acidic analytes in liquid chromatography was investigated using metal frits. Repetitive injections of acidic small molecules or an oligonucleotide were made on individual 2.1 or 4.6 mm i.d. column frits.

Journal of Chromatography A 1650 (2021) 462247 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Mitigation of analyte loss on metal surfaces in liquid chromatography Martin Gilar∗, Mathew DeLano, Fabrice Gritti Waters Corporation, 34 Maple Street, Milford, MA 01757, USA a r t i c l e i n f o Article history: Received 16 February 2021 Revised 21 April 2021 Accepted May 2021 Available online 19 May 2021 Keywords: Adsorption Stainless steel Titanium Hybrid surface technology Peak tailing MISER a b s t r a c t The adsorptive loss of acidic analytes in liquid chromatography was investigated using metal frits Repetitive injections of acidic small molecules or an oligonucleotide were made on individual 2.1 or 4.6 mm i.d column frits Losses were observed for adenosine -(α ,β -methylene) diphosphate, 2-pyridinol 1oxide and the 25-mer phosphorothioate oligonucleotide Trecovirsen (GEM91) on stainless steel and titanium frits Analyte adsorption was greatest at acidic pH due to the positive charge on the metal oxide surface Analyte recovery increased when a series of injections was performed; this effect is known as sample conditioning Nearly complete recovery was achieved when the metal adsorptive sites were saturated with the analyte A similar effect was achieved by conditioning the frits with phosphoric, citric or etidronic acids, or their buffered solutions These procedures can be utilized to mitigate analyte loss However, the effect is temporary, as the conditioning agent is gradually removed by the running mobile phase Metal frits modified with hybrid organic/inorganic surface technology were shown to mitigate analyte-to-metal surface interactions and improve recovery of acidic analytes Quantitative recovery of a 15–35 mer oligodeoxythymidine mixture was achieved using column hardware modified with hybrid surface technology, without a need for column conditioning prior to analysis © 2021 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Introduction Liquid chromatography (LC) separations are based on analyte interactions with the sorbent packed in the chromatographic column [1] Typical sorbents are porous particles (or monoliths) with large surface available for the adsorption-desorption interactions While the chromatographic sorbent is the dominant source of analyte retention, it has been reported that additional interactions may participate in the chromatographic process [2–4] Unexpected peak tailing, loss of analyte recovery and shifts in retention time were observed in analysis of peptides [5,6], oligonucleotides [4], proteins [3], glycans [7] and selected small molecules [4,8] due to undesirable interactions Several published reports discuss this phenomenon and its causes The observed behavior was linked to interactions of analytes with the metal surfaces present in LC systems (e.g connecting capillaries and injector) or metallic column hardware (column body, end fittings, and the column frits) [4,9] Metal adsorption is most apparent for acidic molecules, such as those containing phosphate, or multiple carboxylate moieties [4,6,10] ∗ Corresponding author E-mail address: Martin_Gilar@waters.com (M Gilar) Nagayasu et al [10] confirmed that adsorption of carboxylic acids on stainless steel surfaces depends on pH and ionic strength conditions The adsorption strength increased with the number of the charged moieties in the test molecule structure (1–6 carboxyl groups) The authors concluded that the adsorption phenomenon is driven by the charge of the metal oxide layer on the surface of the stainless steel, which can be positively or negatively charged (stainless steel pI~7) It is important to consider that pH also governs the analyte charge, which has a direct impact on the overall adsorption [10] This hypothesis about ionic analyte adsorption on metallic surfaces is supported by the observation that an increase in mobile phase ionic strength leads to decreased adsorption strength of test analytes [10] Sugiyama et al studied ionic strength and pH effects on the adsorption of proteins on stainless steel particles The authors observed that multi-valent phosphate and citrate buffers compete for adsorption sites on the stainless steel surface and effectively minimize the adsorption of acidic proteins [11] Loss of phosphopeptide recovery in LC MS was reported by Fleitz et al and others, most notably for multi-phosphorylated peptides [6,12] LC system treatment with EDTA or high pH mobile phase (pH> 9) presumably reduced the phosphopeptide adsorption on LC instrument and column metal hardware [6,13] https://doi.org/10.1016/j.chroma.2021.462247 0021-9673/© 2021 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 10 x 155.5 ng A - PEEK union AU 0.12 10 x 62.2 ng 0.08 10 x 31.1 ng 0.04 0.00 0.0 5.0 10.0 Minutes 15.0 90 % AU 0.06 B - stainless steel frit 0.04 60 % 0.02 0.00 0.0 5.0 10.0 Minutes 15.0 Fig Results of a MISER experiment consisting of 10 × 31.1 ng injections followed by 10 × 62.2 ng and 10 × 155.5 ng injections of 25-mer phosphorothioate oligonucleotide (A) The injector was connected to the detector via a PEEK union; (B) The PEEK union was replaced with a 4.6 mm stainless steel frit used in HPLC columns The mobile phase was mM ammonium acetate, pH and the peaks were detected by absorbance at 260 nm A B 120 100 100 80 80 % recovery % recovery 120 60 40 20 60 40 20 0 200 400 600 800 1000 1200 1400 200 400 ng injected 600 800 1000 1200 1400 ng injected Fig Cumulative recovery of 25-mer phosphorothioate oligonucleotide injected in MISER experiments on a 2.1 mm i.d stainless steel UPLC frit (A) or a 4.6 mm i.d stainless steel HPLC frit (B) Experimental conditions were the same as in Fig RSD values were estimated from n = experiments for 2.1 mm frits and n = experiments for 4.6 mm frits Table Estimated metal surface of frits, columns, and sample accessible metal surface in selected LC systems Acquity H-class UPLC Arc HPLC a b Frit area (mm2 ) Column areaa (mm2 ) LC system areab (mm2 ) 453 (2.1 mm) 3396 (4.6 mm) 1236 (50 × 2.1 mm) 8237 (100 × 4.6 mm) 513 1453 Calculated as internal column body surface + two frits Estimated from connecting tubing and injector surface exposed to the sample covery and reduce the carryover [3,7,12,16] Despite the published reports, many chromatographers underappreciate the effects of adsorption on metal surfaces on their analytical results This is because the detrimental effects are not apparent for neutral or acidic compound that are not charged at LC conditions The adsorption becomes very obvious for acidic analytes, particularly those containing two or more phosphate groups, when analyzed with low ionic strength LC MS compatible mobile phases (e.g 0.1% formic acid and organic eluent) [4,8] Several strategies were proposed to mitigate metal-surface adsorptive effects Citrates, phosphates, ethylenediaminetetraacetic Nucleotide mono- (NMP), di- (NDP), and triphosphates (NTP) are classes of analytes susceptible to metal adsorption [3,8,14] The peak tailing and sample loss increases in the following order: NMP < NDP < NTP The loss of analytes was conclusively linked to adsorption on metal tubing and metal surfaces, including the MS electrospray needle [4,15] The adsorption of acidic analytes on metal surfaces was also linked to an increased LC carryover [16] The undesirable adsorptive effects in LC prompted the development of chromatographic instruments and columns using nonmetallic materials Polyether ether ketone (PEEK) lined tubing and columns were shown to decrease peak tailing, improve sample re- M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 AU 0.08 0.00 A Ti HST pH 0.04 HST-modified titanium frits We studied the sample loss in the mobile phase pH range between and We also evaluated the efficiency of LC frits passivation with selected solution of acids and chelators Experimental 0.0 6.0 Minutes 12.0 18.0 2.1 Materials and reagents AU 0.08 Ti pH 0.04 0.00 0.0 Ammonium acetate (for molecular biology, purity ≥ 98%), hexylamine (purity ≥ 99%), acetic acid (purity ≥ 99.8%), phosphoric acid (ACS reagent, ≥ 85% w/v solution), etidronic acid (60% w/v solution), ammonium citrate tribasic (purity ≥ 97%), ammonium phosphate dibasic (purity ≥ 99%), citric acid anhydrous (purity ≥ 99.5%), adenosine -(α ,β -methylene) diphosphate (AMPcP, purity ≥ 99.0%), and 2-pyridinol 1-oxide (PNO, purity ≥ 99.0%) were purchased from Sigma-Aldrich (St Louis, USA) Ammonium hydroxide (purity ≥ 28–30%, w/v, J.T Baker) and acetonitrile, HPLC purity, were purchased from Thermo Fisher Scientific (Waltham, MA, USA) HPLC purified 25-mer phosphorothioate oligonucleotide Trecovirsen (GEM91), sequence CTC TCG CAC CCA TCT CTC TCC TTC T, was obtained from Nitto Denko Avecia, Inc (Milford, MA, USA) A Milli-Q water purification system (Millipore, Bedford, MA, USA) was used for preparation of HPLC mobile phases For chemical structures of analytes see Supplementary Fig S1 B 6.0 Minutes 12.0 18.0 AU 0.08 0.00 C ss pH 0.04 0.0 6.0 Minutes 12.0 18.0 AU 0.08 ss pH 0.04 0.00 D 0.0 6.0 Minutes 12.0 2.2 LC instrumentation, columns, and frits All chromatographic measurements were performed using an ACQUITY UPLC H-class Bio system (Waters, Milford, MA, USA) consisting of a quaternary solvent manager (QSM), a column manager (CM) module, a flow through needle (FTN) sample manager, and an ACQUITY photodiode array (PDA) detector equipped with a μL titanium flow cell The flow path of the instrument was modified with HST to minimize the analyte interaction with metal surfaces Empower software was used for data acquisition and analysis Multiple Injection in a Single Experimental Run (MISER) mode was enabled by prototype instrument control software developed inhouse and installed on the SM-FTN We utilized sample manager FTN_V1.65.356, (MISER_HT_V13) firmware The instrument control software enables repetitive sample injections from one or multiple vials while collecting the data in a single chromatogram ˚ 1.7 μm, An ACQUITY UPLC Oligonucleotide BEH C18, 130 A, 2.1 × 50 mm column (stainless steel hardware) and an ACQUITY ˚ 1.7 μm, 2.1 × 50 mm PREMIER Oligonucleotide BEH C18, 130 A, column were compared for the analysis of a 15–35 mer oligodeoxythymidine mixture The denotation PREMIER signifies that the column hardware is modified with HST, as described in Section 2.3 Columns and the MassPREPTM Oligonucleotide Standard were obtained from Waters (Milford, MA, USA) 18.0 Fig Results of MISER experiments consisting of 10 × 10 ng injections followed by 10 × 20 ng and 10 × 50 ng injections of AMPcP on metal frits (A) 4.6 mm i.d titanium frit modified with HST (B) unmodified 4.6 mm i.d titanium frit (C, D) 4.6 mm i.d stainless steel frit The mobile phase was mM ammonium acetate adjusted to pH for experiments A, B, C and pH for experiment D The peaks were detected by absorbance at 260 nm acid (EDTA), acetylacetone [17] or medronic acid have been used as additives to the mobile phase (or sample diluent) to minimize the sample adsorptive effects [3,8,9,18–20] While effective, this approach may have a negative impact on MS sensitivity Some laboratories perform column conditioning with series of injections, until the signal response stabilizes at a desirable level [21,22] Another method is to passivate an entire LC-MS system flow path, sometimes including the column, with phosphoric or citric acid solutions wash [14,19] Recently, methods involving permanent modification of metal surfaces were developed to eliminate the analyte adsorption in LC Lauber et al explored the use of a hybrid organic/inorganic barrier layer applied by vapor deposition The method produces a permanent chemical barrier that minimizes the analyte contact with metal surfaces Vapor deposition readily modifies surfaces of column hardware, connecting tubing in LC system, and porous frits embedded within the chromatographic columns [23] The technology of ethylene bridged siloxane modified metal surface, named hybrid surface technology (HST), was deployed for use in reversedphase (RP) and hydrophilic interaction chromatography (HILIC) applications [24,25] The goal of this study is to evaluate the usefulness of HST metal surface modification for LC analysis of acidic molecules such as oligonucleotides and selected small molecules The metal adsorption of analytes was investigated using stainless-steel, titanium and 2.3 Frit modification with hybrid surface technology The hybrid surface technology, HST, forms an ethylene-bridged siloxane polymer bonded to the metal oxide surface via a vapor deposition process [23] The resulting hybrid organic-inorganic barrier is chemically similar to that of bridged-ethylene hybrid (BEH) chromatographic particles [26] This process was used to modify metal tubing and frit surfaces for the study The vapor deposition technique creates an effective barrier on high aspect ratio substrates, such as tubing with an internal diameter of 100 μm and a length of 368 mm and porous materials such as LC frits This makes it possible to implement the technology for modification of LC instruments and column hardware The vapor deposition process was utilized for improved LC analysis of molecules with strong M Gilar, M DeLano and F Gritti 120 Journal of Chromatography A 1650 (2021) 462247 oligonucleoƟde, stainless steel 100 120 A pH pH pH pH pH 40 60 pH pH pH pH pH 40 200 400 600 800 1000 1200 1400 1600 200 400 600 800 1000 1200 1400 1600 120 D 100 120 AMPcP, Ɵtanium E 40 20 60 pH pH pH pH pH 40 20 800 1000 200 120 400 600 800 1000 G 400 40 20 H 800 1000 800 1000 PNO, HST Ɵtanium I 80 60 pH pH pH pH pH 40 20 600 600 100 % recovery pH pH pH pH pH % recovery 60 400 200 120 PNO, Ɵtanium 80 200 pH pH pH pH pH 40 ng injected 100 80 F 60 ng injected PNO, stainless steel AMPcP, HST Ɵtanium 0 ng injected 100 1000 1200 1400 1600 20 600 800 80 % recovery pH pH pH pH pH % recovery 60 120 600 100 80 400 400 ng injected 100 80 200 200 ng injected AMPcP, stainless steel pH pH pH pH pH 40 0 ng injected 120 60 20 0 C 80 20 oligonucleoƟde, HST Ɵtanium 100 % recovery 60 20 % recovery B 80 % recovery % recovery 80 % recovery 120 oligonucleoƟde, Ɵtanium 100 60 pH pH pH pH pH 40 20 0 200 ng injected 400 600 800 1000 200 400 ng injected 600 800 1000 ng injected Fig Cumulative recovery of 25-mer phosphorothioate oligonucleotide (A, B, C), AMPcP (D, E, F) and PNO (G, H, I) injected on 4.6 mm frits using mobile phase pH values ranging from to All mobile phases were mM ammonium acetate HST modification of the frit minimizes the analyte adsorption on the metal surface (panels C, F, I) 0.06 A 0.04 AU Ti frit, oligonucleotide 0.02 0.00 0.0 5.0 Minutes 10.0 15.0 0.03 B Ti frit, AMPcP AU 0.02 0.01 0.00 0.0 5.0 Minutes 10.0 15.0 Fig Results of MISER experiments consisting of three series of injections separated by 2.5 gaps Totals of 15 injections of 155.5 ng of 25-mer phosphorothioate oligonucleotide (A) or 15 injections of 50 ng of AMPcP (B) were performed on 4.6 mm i.d titanium frits The mobile phase was mM ammonium acetate, pH and the peaks were detected by absorbance at 260 nm The arrows highlight the decline of the signal between the last injection in a series and the initial injection after the time gap M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 A - first load AU 0.06 0.04 0.02 0.00 B - second load AU 0.06 0.04 0.02 0.00 C - third load after basic wash AU 0.06 0.04 0.02 0.00 0.0 5.0 Minutes 10.0 15.0 Fig Results of MISER experiments performed under the same conditions as Fig using a 4.6 mm i.d stainless steel frit and the 25-mer phosphorothioate oligonucleotide sample (A) A repeated experiment performed immediately after the conclusion of the first experiment illustrates the sample conditioning effect resulting in an improved oligonucleotide recovery (B) The third experiment (C) was performed with the same frit after it was washed with 2.8% ammonia solution The observed sample loss was comparable to the result in panel A The basic wash removes the sample conditioning (desorbs the sample from the frit surface) For additional discussion see Supplemental Fig S4 affinity to metal surfaces as described by Lauber, DeLano and colleagues [22] The MISER experiment consisted of one microliter injections in 0.5 intervals, unless noted otherwise The mobile phase was mM ammonium acetate with pH adjusted to desirable value with acetic acid or ammonium hydroxide (for pH see figure captions) Mobile phase flow rate was 0.2 mL/min and the experimental temperature was 25 °C The typical experiment included ten injections of pmole (31.1 ng) of 25-mer GEM91 oligonucleotide sample, followed by ten injections of pmole (62.2 ng) and ten injections of 10 pmol (155.5 ng) of oligonucleotide Due to the absence of a chromatographic column, each sample injection is detected as a single peak Fig 1A illustrates the resulting MISER “chromatogram” of thirty peaks spaced at an 0.5 intervals performed with a PEEK union The experiments with AMPcP or PNO analytes were performed in a similar fashion using 10 × 10 ng injections, followed by 10 × 20 ng and 10 × 50 ng sample injections For all MISER experiments the sample solvent was consistent with the running mobile phase Sample Manager Wash and Sample Manager Purge lines were primed with the solvent matching the running mobile phase (60 s prime for Wash solvent line, 50 cycles prime for Purge solvent line) to ensure that the frits were exposed only to the solvents consistent with the running mobile phase Fig 1B shows that when a stainless steel frit was placed in the sample flow-path, a loss of analyte was detected The amount of analyte loss was estimated by comparing the data to the control 2.4 MISER experimental procedures MISER experimental setup was utilized with the goal to speed up the multiple injection of the sample The samples were injected typically every 0.5 by an FTN sample manager onto a single chromatographic frit placed in a custom-made holder The frits are identical to those used in chromatographic columns The sample signal was recorded using a PDA detector connected to a holder with 30 cm × 75 μm PEEK capillary When evaluating analyte adsorption on the frit metal surface, stainless steel, titanium or HST surface modified titanium frits (2.1 or 4.6 mm i.d.) were placed in the sample flow path A control experiment was performed by replacing the frit with the PEEK union The control experiment (no frit in the flow path) provided the 100% recovery value used for relative quantitation The flow path of an ACQUITY H–Class Bio system was modified using HST to minimize the metal surface area exposed to the sample Therefore, the dominant metal surfaces available for analyte adsorption were the investigated frits The PDA detector flow cell and its internal tubing were built from unmodified titanium The amount of area available for sample adsorption in the PDA flow cell was two orders of magnitude smaller compared to the surface area of a 4.6 mm metal frit M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 experiment in which a PEEK union was used in place of the frit (Fig 1A) Additional MISER experiments were performed with 155.5 ng (oligonucleotide) or 50 ng (small molecules) injections using a series of five injections separated by the time gap The gap was inserted to evaluate the amount of sample desorption from the frit metal surface during the mobile phase wash Another experiment consisted of multiple series of ten injections separated by a 10 gap The design of the MISER experiments can be best understood from the presented chromatograms 3.2 Effects of pH on analyte adsorption on metal surfaces Fig illustrates the results of similar experiments as described in the previous section performed with various types of 4.6 mm i.d frits and a range of mobile phase pH It was reported that the adsorption of acidic molecules is stronger at acidic pH [4,12,15,22] Therefore, we used mM ammonium acetate buffer adjusted to pH with acetic acid for the following experiment Three separate frits were evaluated side-by-side: a titanium frit modified with HSB technology (Fig 3A), an unmodified titanium frit (Fig 3B), and a stainless steel frit (Fig 3C) AMPcP analyte, structurally similar to adenosine diphosphate, but hydrolytically stable in acidic solutions, was selected for the MISER experiment Ten injections of 10 ng AMPcP were followed with ten injections of 20 ng and ten injections of 50 ng of analyte Fig 3A shows that the hybrid surface technology modified titanium frit does not show any adsorptive losses of the AMPcP (compared with a PEEK union experiment, not shown) An unmodified titanium frit exhibited distinct adsorption of the analyte at pH and some peak tailing can be observed (Fig 3B) The stainless steel frit had the most pronounced adsorption and extensive peak tailing (Fig 3C) When the experiment was repeated with a new stainless steel frit and mobile phase adjusted to pH 8, reduced adsorption and tailing were detected (Fig 3D) The observed behavior confirms that analyte adsorption on metal surfaces is detectable for acidic small molecules The degree of adsorption depends on the metal nature and mobile phase pH Because the analyte adsorption is related to the surface charge on the metal [15], and also the charge of the analyte [10], we extended the study to a wide range of mobile phase pH and investigated the adsorption of three analytes Fig summarizes the recovery of oligonucleotide, AMPcP and 2-pyridinol 1-oxide, PNO for mM ammonium acetate mobile phases adjusted to pH 5–9 Oligonucleotide and AMPcP are strongly acidic molecules, expected to be charged in the pH range 5–9 PNO is a weak acid with pKa ~ 6.2; it carries an average charge of −1 at pH >8, −0.87 at pH 7, −0.4 at pH and −0.06 at pH (values calculated with ChemAxon software) Fig illustrates several trends: The adsorption losses for the oligonucleotide and AMPcP are most significant at acidic pH, when the titanium and stainless steel surfaces are predominantly positively charged The recovery of all samples improves at basic pH, presumably due to the alteration of the surface charge on the metal oxide layer from positive to negative This observation is consistent with published reports and our own experience with phosphopeptide analysis using basic mobile phases [18] Another conclusion from the results shown in Fig is that adsorption of the oligonucleotide and AMPcP is most significant for the stainless steel frit, followed by titanium, and HST modified titanium frits Because the HST modification hinders the analyte access to the metal surface, this frit shows limited sample adsorption The minor losses observed at pH and are likely due to sample adsorption on parts of the LC system that were not modified with HSB (detector cell tubing); they are comparable with the magnitude of integration error (~ 1%) Some “dips” in the trends in Fig (most notably in Fig 4H) were caused by significant peak tailing that precluded an accurate peak integration, especially when the sample mass load changed such as in the 11th and 21st injections The peak tailing is related to desorption of analyte trapped on the frit surface; we will discuss this phenomenon in the next section The peak tailing due to metal surfaces is illustrated in Supplemental Fig S5 PNO shows somewhat different behavior than the oligonucleotide and AMPcP The lowest recoveries of PNO are observed at pH for the titanium and stainless steel frit This is due to PNO weaker acidity; the PNO is mostly neutral at pH 5, which leads to improved recovery (Fig 4G, H) and peak shape (data not shown) Results and discussion 3.1 Evaluation of oligonucleotide adsorption on stainless steel using MISER Table illustrates that the surface area of the selected HPLC or UPLC systems accessible to the sample is relatively small compared to the surface area of the chromatographic column Due to the porous nature of metal frits in chromatographic columns, they are the dominant source of the adsorption Therefore, we utilized 2.1 and 4.6 mm LC column frits as surrogates to study the analyte adsorption on metal surfaces Fig outlines the performed MISER experiment Ten injections of 25 mer GEM91 oligonucleotide sample (31.1 ng) were followed with ten injections of 62.2 ng and finally ten injections of 155.5 ng of sample Fig 1A shows the results of this experiment, where the injector was connected to the detector with PEEK tubing and union Little or no sample loss was observed in this experiment; the peak areas in Fig 1A were used for estimation of oligonucleotide recovery in the subsequent experiments with the frits Fig 1B shows the scenario where the PEEK union was substituted with a 4.6 mm i.d stainless steel frit An apparent loss of oligonucleotide peak area was observed in the initial injections followed by a gradual signal buildup in later injections (Fig 1B) Approximately 60% oligonucleotide recovery was observed in the twentieth injection, while the recovery increased to 90% of the expected peak area for the thirtieth injection The experimental results support several conclusions (i) The stainless steel surface adsorbs the oligonucleotide (acidic) sample (ii) When a sufficient mass of the sample is injected, the frit adsorption sites are partially or completely saturated, and the analyte recovery improves These observations are consistent with the experience of chromatographers analyzing samples susceptible to metal adsorption The so-called “sample conditioning” protocol is often utilized prior to analysis to mitigate sample losses, consisting of repetitive injections of the sample of interest or a sacrificial compound of a similar nature We repeated the experiment shown in Fig with a 2.1 mm i.d stainless steel frit Because the 2.1 mm frit surface is 7.5 fold smaller compared to the 4.6 mm frit, proportionally lesser adsorption loss was observed Fig 1B) Nearly 100% signal was observed after a cumulative load of 500 ng of the oligonucleotide (Fig 2A) In contrast, the 4.6 mm i.d frit required a mass load of 1300 ng for the signal to reach approximately 80% recovery (Fig 2B) In the next experiment we investigated whether the sample adsorption on stainless steel frits is repeatable for multiple frits The experiment was repeated with six 4.6 mm and four 2.1 mm i.d new frits that were not previously exposed to the sample The error bars in Fig illustrate that some variability exists between the frits, presumably due to surface area and surface oxide layer variation Raw chromatograms obtained in this experiment are provided as Supplemental material, Fig S2 and S3 The presented chromatograms confirm that the MISER experimental setup is useful for rapid evaluation of sample losses on metal surfaces In the next section we utilize the method to study a wider set of analytes, LC conditions, and types of metal surfaces M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 3.3 Impact of analyte adsorption-desorption on peak tailing and LC analysis 3.4 LC system conditioning using acid wash and chelators Alternative approaches for mitigation of sample adsorption on metal surfaces involve an addition of chelating compounds (citric, medronic or ethylenediaminetetraacetic acid) to the mobile phase or sample [9,18,19,30] Another approach is to passivate the LC system with phosphoric acid solutions [19] We employed the MISER method to investigate the conditioning (passivation) effectiveness for stainless steel frits Unused 4.6 mm i.d frits were sonicated for 15 at 50 °C in an aqueous solution of the selected acids, buffers, or chelating agents After the treatment the frits were placed into the holder, equilibrated with mobile phase, and tested for adsorption using the protocol described in Fig Fig visually represents the oligonucleotide recovery for various frit conditioning treatments The conditioning in phosphoric, citric or etidronic acid solutions was successful; little or no loss of oligonucleotide due to metal adsorption observed The treatments with formic acid and EDTA did not condition the frits successfully, while the nitric acid treatment enhanced the adsorptive loss of the sample; no oligonucleotide signal was observed for thirty injections Phosphate, citrate and etidronate salts solutions were used with the pH between 5.7 and Fig illustrates that the results are comparable with the acid solutions This result leads us to conclude that the conditioning mechanism is the adsorption of multivalent phosphate, citrate, or etidronate ions on the metal surface, rather than acid treatment of the frit Nitrate or formate ions have presumably weaker adsorption to the metal and not act as efficient conditioning agents Birdsall and colleagues observed improved peak shapes for acidic peptides after LC system treatment with phosphoric acid [19] However, the treatment durability was limited; the tailing gradually worsened over several hours of run time This observation supports our hypothesis that competitive adsorption of multivalent ions (phosphate, citrate, etidronate) reduces the interaction of acidic analytes with metallic components of the LC system/column The conditioning with selected agents (Fig 7) is helpful, but not a permanent solution In a separate experiment we investigated the stability of conditioned frits and found that deconditioning is accelerated at elevated mobile phase pH (wash with 2.8% ammonia solution, Supplemental Fig S6) Fig illustrates the temporary nature of stainless steel surface conditioning with ammonium etidronate A new 4.6 mm i.d stainless steel frit was treated with 10 × 10 μL injections of 0.42 M ammonium etidronate, pH The conditioning by repetitive injections of conditioning solution is an alternative to off-line frit sonication in ammonium etidronate buffer; this process is equivalent to the sample conditioning method described in Section 3.1 After the conditioning we immediately proceeded with the MISER experiment presented in Fig Ten series of ten injections of AMPcP were performed spaced by a 10-minute gap after each series Initially nearly quantitative recovery was observed (98%) However, the signal decreased during the experiment and the peak tailing became more pronounced Multiple processes are contributing to the observed loss of conditioning: (i) injected AMPcP competitively displaces a portion of the etidronate ions from the frit, (ii) AMPcP itself is partially desorbed during the 10-minute gaps between the injection series, (iii) running mobile phase buffer is competing with etidronate ions adsorbed on the metal surface and the conditioning agent is gradually displaced Fig illustrates the results of a MISER experiment where three series of five sample injections were separated by gaps of 2.5 The design of the experiment was adjusted in order to investigate the analyte desorption during the time gap when the frit is continuously washed with the running mobile phase The mechanism of tailing in chromatography can be explained by simultaneous participation of weak and strong interactions [27,28], or mass overload [29] In the case of frits the peak shape is a result of flow through (no interaction) and relatively strong analyte interaction with charged metal surface Fig shows the initial sample recovery loss, and a gradual signal build up after titanium surface conditioning with the analyte The oligonucleotide peaks (Fig 5A) show only minor tailing and small losses of recovery (peak height) were observed after the time gaps indicated by the arrows in Fig 5A This result suggests that the oligonucleotide is strongly adsorbed to the metal surface and the desorption kinetics is slow Only a small portion of the oligonucleotide is desorbed from the titanium surface during the 2.5 wash with the mobile phase In contrast, the experiment with AMPcP reveals severe peak tailing and an apparent loss of signal after the time gaps (see arrows in Fig 5B) These results suggest that AMPcP is rapidly desorbed from the titanium surface at pH 6, which means that the effect of sample conditioning is only temporary Fig illustrates that sample conditioning is a viable strategy for LC analysis of the GEM91 oligonucleotide, but not for AMPcP The peak tailing seen for AMPcP in Fig 5B can explain the carryover phenomenon linked to metal adsorption [16] After the sample is injected onto the column (column frits), the adsorbed analyte slowly desorbs from the metal surface as the mobile phase flows through the column This leads to enrichment of the desorbed analyte on the chromatographic sorbent and contributes to the analyte signal in the subsequent injections To investigate the stability of sample conditioning, we performed additional experiments illustrated in Fig First, a new 4.6 mm i.d stainless steel frit was conditioned with the oligonucleotide sample using the experimental conditions described in Fig After frit conditioning (Fig 6A), the frit was washed for with running mobile phase and the experiment was restarted The result for this repeated experiment with an additional thirty injections is shown in Fig 6B Improved oligonucleotide signal was observed in Fig 6B, which is consistent with successful sample conditioning of the frit in the previous run After the second experiment the frit was disconnected from the LC system and 0.2 mL of 2.8% ammonia aqueous solution was pushed through the frit using a syringe, followed with 0.6 mL of mM ammonium acetate, pH The frit in the holder was reconnected to the LC system, equilibrated for with the mobile phase, and an additional thirty injections of sample were performed (Fig 6C) The observed loss of signal was comparable to Fig 6A confirming that the frit conditioning was effectively removed with the ammonia wash This finding is consistent with the results in Fig 4; the basic mobile phase reduces the acidic analyte adsorption on the metal surface by altering the metal oxide layer charge In a separate experiment we pre-loaded a stainless steel frit with the oligonucleotide sample and subsequently executed five 10 μl injections of 2.8% ammonia solution on the frit Recorded UV data confirmed that the injections of the high pH solution effectively desorbed the oligonucleotide from the metal frit (Supplemental Fig S4) M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 0.17 M (NH4)2HPO4 pH 5.7 0.17 M H3PO4 0.52 M (NH4)3 citrate pH 0.52 M citric acid 0.42 M NH4 etidronate pH 0.42 M etidronic acid 2.65 M formic acid mM EDTA 2.24 M nitric acid 0.0 5.0 Minutes 10.0 15.0 0.0 5.0 Minutes 10.0 15.0 Fig Results of MISER experiments executed using the same conditions as Fig Prior to the experiment each individual 4.6 mm i.d stainless steel frit was conditioned for 15 at 50 °C in acid or salt solution Blue chromatograms represent the experiments with successful conditioning with nearly complete oligonucleotide recovery Black chromatograms show loss of oligonucleotide recovery, which indicates an unsuccessful conditioning with formic, nitric, or EDTA acid solutions 98 % 89 % 0.03 61 % AU 0.02 A 0.01 0.00 0.0 30.0 60.0 90.0 120.0 150.0 Minutes Fig Results of a MISER experiment consisting of ten series of ten injections of 50 ng of AMPcP; the series of injections were separated by ten minute gaps inserted to allow for an extended mobile phase wash of the frit Prior to the experiment the 4.6 mm i.d stainless steel frit was conditioned with ten injections of 10 μL of 0.42 M ammonium etidronate, pH The decline in the AMPcP signal is suggestive of a gradual loss of etidronate 3.5 Application of HST modified columns to the analysis of oligonucleotides 10 pmol of each oligonucleotide was injected on column) Adsorption of oligonucleotides on conventional stainless steel column hardware resulted in low peak areas, most notably in the first injection Minor peaks present in the sample eluting between the dominant oligonucleotides are impurities created in the oligonucleotide synthesis The minor impurities were not observed in the first injection and only partially in the second injection on the conventional column Ten injections of the oligonucleotide standard (10 pmol/injection for each dominant oligonucleotide) were required to completely condition the standard column (data not shown) Alternatively, one or two injections of 500 pmol 35 mer oligodeoxythymidine provided sufficient column conditioning However, no conditioning is required when we used column hardware modified with HST 97–100% recovery was observed in the first 10 pmol injection of 15, 20, 25, 30 and 35 mer oligodeoxythymidine mixture (Fig 9B; for recovery see Fig S8) The same batch of chromatographic sorbent was used in both Fig ex- In the previous sections we investigated several strategies for mitigation of analyte loss due to metal surface adsorption in LC While helpful, neither sample conditioning nor washes of LC hardware with conditioning agents provide a permanent solution Some LC manufacturers introduced columns with non-metallic hardware, most commonly constructed from polyether ether ketone (PEEK) or PEEK-lined stainless steel Due to limitations in mechanical strength and chemical compatibility of PEEK, other solutions were explored Columns with metal hardware modified with hybrid surface technology were introduced for the analysis of metal-sensitive analytes, including oligonucleotides, acidic peptides and certain small molecules [22,25,31] Fig shows the effect of column hardware on the recovery of oligonucleotides (15, 20, 25, 30 and 35 mer oligodeoxythymidines, M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 Fig Analyses of 15–35 mer oligodeoxythymidine sample using 50 × 2.1 mm, 1.7 μm Oligonucleotide BEH C18 columns with conventional stainless steel hardware (A) and HST modified column hardware (B) Mobile phase A was 25 mM hexylammonium acetate, pH and mobile phase B was prepared by mixing mobile phase A and acetonitrile in a 1:1 ratio (v:v) A linear gradient was carried out from 50 to 86% B in 12 min, with a flow rate of 0.4 mL/min, and a column temperature of 60 °C 10 pmol of each oligonucleotide was injected on column The peaks were detected by absorbance at 260 nm For both columns three consecutive injections were made (only first two injections are showed), followed by injection of 500 pmol of 35-mer oligodeoxythymidine and the forth injection of 10 pmol 15–35 mer oligodeoxythymidine sample (labelled sample conditioned) (v) Modification of metal column hardware with hybrid surface technology provides nearly quantitative recovery for oligonucleotides and metal-sensitive small molecules Expected signal was observed from the first injection with no column conditioning required prior to analysis Hybrid organic/inorganic surface technology holds promise for LC analysis of analytes susceptible to adsorption on metal surfaces periments The improved recovery obtained with the LC column with HST modified hardware is consistent with the MISER experiment using an HST modified titanium frit (Supplemental Fig S7) No loss of oligonucleotide was detected using a 4.6 mm i.d frit modified with HST Due to the covalent nature of the surface modification produced using HST, the metal surface protection is stable in most LC conditions Accelerated stability studies of HST modified frits under acidic and basic conditions were described in a recent report [22] Declaration of Competing Interest Conclusion The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper In this study we were able to directly observe and quantify the loss of analytes injected on metal frits We suggest several strategies for mitigation of acidic analytes loss in liquid chromatography CRediT authorship contribution statement (i) Loss of acidic analytes observed in LC is predominantly due to ionic adsorption on positively charged metal oxide surfaces (ii) The extent of the loss depends on the nature of the metal and mobile phase pH The observed analyte losses were greater on stainless steel than on titanium frits Titanium frits modified with hybrid surface technology exhibited negligible sample losses (iii) Repetitive exposure of the metal hardware to acidic samples or conditioning with the solutions of multivalent acids (phosphate, citrate, etc.) reduced the analyte adsorptive losses (iv) Although the sample conditioning and multivalent acid washes improve the recovery and peak shape of acidic analytes in LC, the effect is temporary LC hardware exposure to mobile phase, and high pH buffers desorb the conditioning molecules from the metal surface Martin Gilar: Conceptualization, Methodology, Investigation, Writing – original draft, Visualization Mathew DeLano: Methodology, Investigation Fabrice Gritti: Conceptualization, Writing – review & editing Acknowledgement The authors wish to thank Thomas McDonald and Patrick Brophy for their help with setting up the LC system for MISER experiment Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.chroma.2021.462247 M Gilar, M DeLano and F Gritti Journal of Chromatography A 1650 (2021) 462247 References [16] H Sakamaki, T Uchida, L.W Lim, T Takeuchi, Evaluation of column carryover of phosphorylated peptides and fumonisins by duplicated solvent gradient method in liquid chromatography/tandem mass spectrometry, Anal Sci 31 (2015) 91–97 [17] D Siegel, H Permentier, R Bischoff, Controlling detrimental effects of metal cations in the quantification of energy metabolites via ultrahigh pressure-liquid chromatography-electrospray-tandem mass spectrometry by employing acetylacetone as a volatile eluent modifier, J Chromatogr A 1294 (2013) 87–97 [18] Y.Q Yu, I Campuzano, M Gilar, Enhanced nanoLC/MS analysis of phosphopeptides, Waters Application Note (2006) 720001458en [19] R.E Birdsall, J Kellett, Y.Q Yu, W Chen, Application of mobile phase additives to reduce metal-ion mediated adsorption of non-phosphorylated peptides in RPLC/MS-based assays, J Chromatogr B Analyt Technol Biomed Life Sci 1126-1127 (2019) 121773 [20] J.J Hsiao, O.G Potter, T.W Chu, H Yin, Improved LC/MS Methods for the Analysis of Metal-Sensitive Analytes Using Medronic Acid as a Mobile Phase Additive, Anal Chem 90 (2018) 9457–9464 [21] M Gilar, Z Li, J MacLean, Improved Chromatographic Analysis of Oligonucleotides with ACQUITY PREMIER Oligonucleotide BEH C18 Columns, Waters Application Note 720 07012en 2020 [22] M.A Lauber, T.H Walter, M Gilar, D DeLano, C Boissel, K Smith, R.E Birdsall, P Rainville, J Belanger, K Wyndham, Low Adsorption HPLC Columns Based on MaxPeak High Performance Surfaces, White paper, Waters 720 06930EN (2021) [23] A.C Jones, M.L Hithchman, Chemical Vapour Deposition Precursors, Processes and Applications, RSC Publishing, 2008 [24] M.A Lauber, S.M Koza, S.A McCall, B.A Alden, P.C Iraneta, K.J Fountain, High-resolution peptide mapping separations with MS-friendly mobile phases and charge-surface-modified C18, Anal Chem 85 (2013) 6936–6944 [25] M DeLano, T.H Walter, M.A Lauber, M Gilar, M.C Jung, J.M Nguyen, C Boissel, A.V Patel, A Bates-Harrison, K.D Wyndham, Using hybrid organic-inorganic surface technology to mitigate analyte interactions with metal surfaces in UHPLC, Anal Chem 93 (2021) 5773–5781 [26] K.D Wyndham, J.E O’Gara, T.H Walter, K.H Glose, N.L Lawrence, B.A Alden, G.S Izzo, C.J Hudalla, P.C Iraneta, Characterization and evaluation of C18 HPLC stationary phases based on ethyl-bridged hybrid organic/inorganic particles, Anal Chem 75 (2003) 6781–6788 [27] T Fornstedt, G.M Zhong, G Guiochon, Peak tailing and mass transfer kinetics in linear chromatography, J Chromatogr A 741 (1996) 1–12 [28] G Gotmar, T Fornstedt, G Guiochon, Peak tailing and mass transfer kinetics in linear chromatography dependence on the column length and the linear velocity of the mobile phase, J Chromatogr A 831 (1999) 17–35 [29] J Bagge, M Enmark, M Les´ ko, F Limé, T Fornstedt, J Samuelsson, Impact of stationary-phase pore size on chromatographic performance using oligonucleotide separation as a model, J Chromatogr A 1634 (2020) 461653 [30] T Nakamura, K.T Myint, Y Oda, Ethylenediaminetetraacetic acid increases identification rate of phosphoproteomics in real biological samples, J Proteome Res (2010) 1385–1391 [31] M.A Lauber, M.H Delano, S.A McCall, J.L Belanger, T.A Dourdeville, K.M Smith, P.D Rainville, D.D Shah, S.J Shiner, C Doneanu, M Donegan, Waters Technologies Corporation, US 20190086371, 2018 [1] U.D Neue, HPLC columns: theory, technology, and Practice, Wiley-VCH, New York, 1997 [2] J.J Hsiao, T.W Chu, O.G Potter, G.O Staples, D.R Stoll, Troubleshooting LC separations of biomolecules, part ii: passivation and mobile-phase additives, LC GC N Am 38 (2020) 431–434 [3] J.J Hsiao, G.O Staples, D.R Stoll, Troubleshooting LC separations of biomolecules, part i: background, and the meaning of inertness, LC GC N Am 38 (2020) 146–150 [4] R Tuytten, F Lemière, E Witters, W Van Dongen, H Slegers, R.P Newton, H Van Onckelen, E.L Esmans, Stainless steel electrospray probe: a dead end for phosphorylated organic compounds? J Chromatogr A 1104 (2006) 209–221 [5] M Gilar, Y.Q Yu, J Ahn, J Fournier, J.C Gebler, Mixed-mode chromatography for fractionation of peptides, phosphopeptides, and sialylated glycopeptides, J Chromatogr A 1191 (2008) 162–170 [6] A Fleitz, E Nieves, C Madrid-Aliste, S.J Fentress, L.D Sibley, L.M Weiss, R.H Angeletti, F.Y Che, Enhanced detection of multiply phosphorylated peptides and identification of their sites of modification, Anal Chem 85 (2013) 8566–8576 [7] K Sandra, J Vandenbussche, P Sandra, Analyzing phosphorylated N-glycans with full recovery on bio-inert LC systems and PEEK-lined HILIC columns, LC GC Eur 31 (2018) 566–571 [8] J.C Heaton, D.V McCalley, Some factors that can lead to poor peak shape in hydrophilic interaction chromatography, and possibilities for their remediation, J Chromatogr A 1427 (2016) 37–44 [9] D Winter, J Seidler, Y Ziv, Y Shiloh, W.D Lehmann, Citrate boosts the performance of phosphopeptide analysis by UPLC-ESI-MS/MS, J Proteome Res (2009) 418–424 [10] T Nagayasu, C Yoshioka, K Imamura, K Nakanishi, Effects of carboxyl groups on the adsorption behavior of low-molecular-weight substances on a stainless steel surface, J Colloid Interface Sci 279 (2004) 296–306 [11] H Sugiyama, T Hagiwara, H Watanabe, T Sakiyama, Effects of ionic substances on the adsorption of egg white proteins to a stainless steel surface, Biosci Biotechnol Biochem 76 (2012) 467–472 [12] H Sakamaki, T Uchida, L.W Lim, T Takeuchi, Evaluation of column hardware on liquid chromatography-mass spectrometry of phosphorylated compounds, J Chromatogr A 1381 (2015) 125–131 [13] C Song, M Ye, G Han, X Jiang, F Wang, Z Yu, R Chen, H Zou, Reversed-phase-reversed-phase liquid chromatography approach with high orthogonality for multidimensional separation of phosphopeptides, Anal Chem 82 (2010) 53–56 [14] A Wakamatsu, K Morimoto, M Shimizu, S Kudoh, A severe peak tailing of phosphate compounds caused by interaction with stainless steel used for liquid chromatography and electrospray mass spectrometry, J Sep Sci 28 (2005) 1823–1830 [15] M.A Iqubal, R Sharma, Kamaluddin, Surface interaction of ribonucleic acid constituents with spinel ferrite nanoparticles: a prebiotic chemistry experiment, RSC Adv (2016) 68574–68583 10 ... adsorption [3,8,14] The peak tailing and sample loss increases in the following order: NMP < NDP < NTP The loss of analytes was conclusively linked to adsorption on metal tubing and metal surfaces, including... repetitive injections of conditioning solution is an alternative to off-line frit sonication in ammonium etidronate buffer; this process is equivalent to the sample conditioning method described in Section... in the oligonucleotide synthesis The minor impurities were not observed in the first injection and only partially in the second injection on the conventional column Ten injections of the oligonucleotide