1
Rapid desorption of radium isotopes from black shale during hydraulic fracturing. 1.
2
Source phases that control the release of Ra from Marcellus shale.
3
4
Joshua D. Landis, Mukul Sharma*, Devon Renock, and Danielle Niu
5
Department of Earth Sciences, Dartmouth College, Hanover NH 03755
6
7
*correspondence to
[email protected]
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Hydraulic fracturing of the Marcellus Shale produces wastewaters that are hypersaline and highly enriched in isotopes of radium. Radium is understood to derive from the Marcellus Shale itself, but its source phases and their contributions to wastewater production have not been described. Using sequential extractions and experimental leachates, we characterize two distinct end-‐members that could contribute Ra to wastewaters, (1) a labile 228Ra mineral phase with 226Ra/228Ra atom ratios ~250, and (2) an exchangeable 226Ra organic phase with 226Ra/228Ra ~10,000. In leaching experiments we observed rapid extraction of Ra from these phases, with high ionic strength solutions leaching up to 14% of Ra from the shale in just hours. Radium concentrations and 226Ra/228Ra ratios increase with [Ca2+] of the leaching solution, and solutions approaching 1M Ca2+ produce 226Ra/228Ra ratios compatible with Marcellus wastewaters. In contrast, pure water removes 450 C. Additional descriptions of these rocks are given in Renock et al. (2016) and Niu et al. (2016). Hand samples were split, selected for clean faces and powdered in a ring mill to a median particle size of 2-‐3 µm with measured BET surface area of 20-‐30 m2 g-‐1 (Renock et al. 2016). We note that the rock powder particles are considerably larger than typical shale pores, which have median sizes of a few nanometers. We anticipate that the exchange properties and presence of pore brine are maintained through crushing (see Balashov et al. 2015). 3.2. Sequential extractions under oxidizing conditions To provide insight into sources of Ra within the Marcellus shale we performed sequential extractions that target specific, operationally-‐defined phases of the shale (Tessier et al. 1979). Major phases of the shale including calcite, organic matter and clay minerals may each be expected to host U or Th and their decay daughters including 226Ra and 228Ra. Operationally, these phases would be dissolved by acetic acid, acidic peroxide (H2O2) and hydrofluoric (HF) acid, respectively (see also Galindo et al. 2007, Stewart et al. 2015, Phan et al. 2015). Major elements and U, Th were measured for all fractions. Ra isotopes were measured on leachate fractions for all three shale samples, and on refractory phases for the Chenango Co. sample only. These extractions were performed under ambient (oxic) conditions with 5-‐gram aliquots of shale powder and 30 mL of leachate. The following extractions were adapted from Eagle et al. (2003): (f1) deionized water to dissolve soluble salts and labile components; (f2) 1M CaCl2 to remove exchangeable cations; we chose Ca2+ as an exchanger to best replicate exchange conditions in wastewaters; (f3) 4N acetic acid to dissolve carbonates; (f4a) repeated extractions using 30% H2O2 at pH ~1.5 with HNO3 to target oxidizable organic matter, continued until the shale residual was a uniform tan-‐gray color, visually indicating loss of organic matter; pyrite will also be dissolved; (f4b) 1M NH4-‐acetate rinse to desorb species released during oxidation but resorbed to residual solids; (f5) 0.2M hydroxylamine hydrochloride to reduce oxides; (f6a) repeated additions of HF in increasing strengths to destroy silicates (primarily clays and quartz) while minimizing precipitation of alkaline earth fluorides; (f6b) saturated AlCl3 to dissolve insoluble fluorides formed during HF digestion; (f7) 0.5 N Na2CO3 carbonate replacement to dissolve residual barite (Curie and Debierne 1904), either a primary component of the shale or produced secondarily during preceding chemical treatments; (f8) strong acid digestion to attack refractory minerals. Note that for step f4a we used acidified H2O2 as oxidant as opposed to commonly used bleach (NaOCl) to prevent trace metal precipitation as oxides at high pH of hypochlorite, as a compromise between selective decomposition of target phases and successful extraction of target elements. Scanning electron microscope characterization of residue (90%. For these samples we added 1M Na2SO4 to provide SO42-‐ equimolar to Ba, with Ra yields of >95% as measured by γ-‐spectrometry and gravimetrically by weighing the precipitate and assuming BaSO4 stoichiometry. Prior to further processing the precipitates were rinsed repeatedly in deionized water to a pH of ~5 that signaled the removal of excess H2SO4 from the precipitates. The wastewaters are enriched in Ba that is sufficient to produce several mg of Ba(Ra)SO4 precipitate. Sulfate precipitates were dissolved by a carbonate replacement technique developed by Marie Curie (Curie and Debierne 1904, Cohen and O’Nions 1991): approximately 10 mg of Sr(Ra)SO4 or Ba(Ra)SO4 were transferred to Teflon vials in 10 mL 0.5N Na2CO3 and were oven-‐heated at 105°C for about 12 hours. This permits a complete replacement of the sulfate precipitate by acid-‐soluble carbonate precipitate. After cooling the samples were centrifuged and rinsed in deionized water until rinses showed a pH of ~ 5. The resulting Ba (or Sr) carbonates were dissolved in 2 mL of 1N HCl. These solutions were then evaporated, re-‐dissolved in 3N HNO3, and passed through columns packed with Eichrom Sr-‐spec resin to separate Ra from Sr or Ba (Chabaux et al. 1994; Fig. 3b). Samples were passed twice on 2 mL columns, then twice on 0.1 mL micro-‐columns. In the final pass the samples were eluted in 0.03N HNO3 to minimize elution of hydrolyzed resin organic material. To further reduce organic interference, the final Ra fraction was combined with aqua-‐regia and evaporated, then re-‐dissolved in 3N HNO3 and irradiated with ultra-‐violet light. Samples were finally evaporated to dryness and loaded onto filaments for thermal ionization mass spectrometry. 3.4.5. Thermal ionization mass spectrometry (TIMS)
8
332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
Radium isotope analysis was performed on the Dartmouth Triton thermal ionization mass spectrometer. Purified Ra fractions were loaded on tungsten (W) filaments in 2.5N HCl between a ‘sandwich’ load of tantalum chloride solution. Ra+ ions were measured using a Mascom secondary ion multiplier (SEM) operated in ion counting mode. To preserve our detector fromaccruing dark noise due to radiation damage, we adopted a static analytical protocol (see Yokoyama and Nakamura 2004). Samples were sequentially baked at 1200°C and 1240°C to remove final traces of organics. The filament temperature was then raised to 1320°C for analysis. We measured 30-‐100 cycles as necessary to achieve an internal precision of ~3%. Effective ionization efficiencies were in the range of 1-‐10%. SEM dark noise was monitored throughout our analysis period and was maintained at less than 0.05 counts per second. Ion counts were corrected for the dark noise contribution. The final 226Ra/228Ra atom ratio of our 228Ra tracer measured using TIMS is 6.60±0.04 (n =11), within error of certified OKA2 ore value of 6.38±0.14. Standards with different sized Ra loads and running temperatures do not show fractionation beyond analytical uncertainty. Final concentration of the 228Ra tracer solution was determined by gamma spectrometry (Landis et al. 2012) with an estimated expanded uncertainty (2σ) of 3.4%. We used this primary 228Ra spike solution to determine sample Ra concentration by isotope dilution using TIMS. We also created a mixed spike from our 226Ra and 228Ra primary tracer solutions to provide cross-‐calibration of the two solutions. Based on γ-‐ calibration of tracer solutions the expected 226Ra/228Ra ratio of the mixed spike was 1065±72 (mean±2σ). TIMS analyses yield a 226Ra/228Ra ratio of 1043±14 (n=9). All tracer 228Ra concentrations and 226Ra/228Ra ratios are decay-‐corrected to a common reference time of 1/1/2015. Total procedural Ra blanks were too low to be distinguished from 0.1 fg loads of our 228Ra spike during measurement by isotope dilution, and are less than 0.01 fg at 99% confidence interval.
356
4. Results and Discussion
357
Quantitative XRD reveals that the black shale samples contain clay minerals (illite and
358
chlorite), quartz, calcite, organic matter, and pyrite ± barite (Table 1). Evaluation by
359
scanning electron microscopy reveals that calcite is present as intergranular cement, and
360
organic matter and clay minerals form intercalated organo-‐clay composites at scales that
361
vary from sub-‐mm to nanometer (Figure 4). Ra concentrations (= [Ra]) and 226Ra/228Ra
362
ratios of three black shale samples are given in Table 1, as are activities of the Ra isotopes
363
and their U and Th parents. All rocks appear to be in 226Ra-‐238U and 228Ra-‐232Th radioactive
364
equilibrium. Due to the hosting of U in organic matter, bulk rock [Ra] and 226Ra/228Ra ratios
365
increase with total-‐organic-‐carbon (TOC; Figure 2a) -‐-‐ this drives a strong Ra isotope mixing
366
relationship between organic and mineral solids of Marcellus Shale (Figure 2b).
367
368
4.1 Sequential extractions under oxidizing conditions, atmospheric pressure
369
Table 2 gives the sequential extraction yields of U, Th, Na, Ca as well as the ionic strength of
370
the solutions produced under oxidizing conditions at room temperature and atmospheric
371
pressure. Table 3 gives the sequential extraction yields of Ra, Ba and S and their
372
corresponding 226Ra/228Ra ratios. Prior to describing these results, we note that elements
373
can and do fractionate from each other during leaching procedures, which are
9
374
“operationally defined” according to both the general properties of the extracting reagents
375
and the nature of the target phase (Tessier et al. 1975, Bacon and Davidson 2008). Different
376
elements in decay couples such as U-‐226Ra or Th-‐228Ra may also be differentially susceptible
377
to extraction based on elemental behavior rather than properties of their source. In fact,
378
decay couples provide the best means for observing elemental effects in extractions since
379
these elements are known to originate from the same source; when released in equilibrium,
380
as evidenced by the same activity and mass percent of bulk rock, no elemental fractionation
381
is implicated. But where disequilibrium is apparent, elemental chemistry has dictated
382
release over mineralogy, due to either selective extraction of one radionuclide, or
383
precipitation or resorption of another (e.g., Lucey et al. 2007). Repartitioning of
384
radionuclides to solids can be expected to be favored in the order Th >> Ra > U (Ivanovich
385
and Harman 1992).
386
387
4.2.1. Reservoirs of 238U and 226Ra in Marcellus Shale
388
Among shale samples the amount of U removed in the oxidizable (f4) fraction increases
389
approximately exponentially with TOC. This is in agreement with the control of organic
390
matter abundance on shale U content (Leventhal 1981; Figure 2a). However, consistent
391
with the findings of Phan et al. (2015), the fraction of total U removed in the oxidizable (f4)
392
fraction constitutes a minority (