1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

3 downloads 0 Views 3MB Size Report
County, New York (4.46% TOC; Lower Oatka Creek Mbr, Marcellus Shale); ..... Wastewater time series data from Greene County, PA reported by Rowan et al.
 

  1  

Rapid  desorption  of  radium  isotopes  from  black  shale  during  hydraulic  fracturing.  2.  A  model  

2  

reconciling  radium  extraction  with  Marcellus  wastewater  production.  

3  

 

4  

Joshua  D.  Landis,  Mukul  Sharma*,  and  Devon  Renock    

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  

Radium   in   hydraulic   fracturing   wastewaters   derives   from   two   isotopically   distinct   end-­‐ members   in   the   shale,   labile   228Ra   phase   hosted   by   mineral   surfaces   (226Ra/228Ra   atom   ratio   ~250)  and  exchangeable   226Ra  hosted  by  organic  surfaces  (226Ra/228Ra  ~10,000).    Here  we  use   mass   balance   and   isotope   mixing   models   to   reconcile   extraction   of   Ra   from   these   phases   with   mechanisms   of   Marcellus   wastewater   production.     Radium   isotopic   mass   balance   requires   that  the  characteristic  water-­‐rock  ratio  between  wastewater  and  shale  is  exceedingly  low,  on   the  order  of  0.04,  and  that  this  ratio  decreases  with  time  during  wastewater  production.    An   evolving   water-­‐rock   interaction   drives   increasing   Ra   concentration   (=[Ra])   and   226Ra/228Ra   ratios  during  wastewater  production,  mediated  by  increasing  [Ca2+]  that  favors  desorption  of   226 Ra   from   organics.     Our   observations   and   models   of   Ra   isotope   geochemistry   are   best   reconciled  with  observations  of  water  and  salinity  mass  balance,  δ 18O,  Na-­‐Br-­‐Cl,  and  87Sr/86Sr   if   wastewater   is   produced   by   mixing   of   injected   fluids   with   a   limited   volume   of   pore   brine   (on   the   order   of   13%   by   volume),   accompanied   by   contemporaneous   extraction   of   excess   alkaline   earth   elements   by   water-­‐rock   exchange.   Validated   using   Ra   isotope   data,   this   model   attributes   the   extreme   salinity   and   [Ra]   in   wastewaters   to   the   progressive,   hydrologic   enrichment  of  injected  fluids  during  hydraulic  fracturing.      

26  

Keywords:     radium,   isotope,  

27  

sorption,  exchange,  water,  rock    

 

226

Ra,  

228

Ra,   shale,   Marcellus,   wastewater,   produced   water,    

1  

hydraulic fracturing wastewater

organic surface, 1-15Å

nic pore orga

Ra

226

mineral p

ore

Ra

reaction progress

ture

228

frac

mineral surface

water-rock ratio [Ca2+]/Na+] [Ra2+] 226 Ra/228Ra

water-rock ratio [Ca2+]/Na+] [Ra2+] 226 Ra/228Ra

 

  28  

1.  Introduction  

29  

Hypersaline   wastewaters *  generated   during   hydraulic   fracturing   of   Marcellus   Shale   have  

30  

extraordinary   Ra   concentrations   and   require   handling,   regulation   and   disposal   as   hazardous  

31  

waste.     Understanding   wastewater   is   necessary   for   managing   its   future   production,   but   faces  

32  

two  outstanding  puzzles–  the  deficit  of  water  and  excess  of  salt  in  Marcellus  wastewaters,  the  

33  

‘salinity   dilemma’   (Engelder   et   al.   2014).     Hydraulic   fracturing   requires   on   the   order   of   6x106  

34  

liters  of  water  to  be  injected  into  the  well,  but  just  10-­‐40%  of  this  volume  returns  to  the  well-­‐

35  

head   during   the   course   of   gas   production.     Conversely,   total   salinity   and   concentrations   of  

36  

alkaline  elements,  including  Ra,  increase  dramatically  and  continuously  in  wastewater  recovered  

37  

over  the  lifetime  of  production  (Hayes  2009,  Rowan  et  al.  2015).    Due  to  a  distinctive  226Ra/228Ra  

38  

isotopic  composition  shared  between  Marcellus  wastewaters  and  Marcellus  Shale,  and  half-­‐life  

39  

restrictions   on   both  

40  

wastewaters  reflects  sources  within  the  shale  and  processes  that  regulate  their  release  during  

41  

fracturing.     We   sought   to   reconcile   Marcellus   Shale   Ra   source   and   isotope   geochemistry  

42  

described   in   our   companion   work   (Landis   et   al.   2018)   with   mechanisms   of   wastewater  

43  

production.  

44  

 

45  

Three  mechanisms  have  been  proposed  for  the  generation  of  hydraulic  fracturing  wastewater,  

46  

based   on   existing   geochemical   and   geophysical   data.     These   are   not   mutually   exclusive,   but  

47  

might  contribute  to  differing  degrees  for  different  elements,  

226

Ra   (half-­‐life   1600   years)   and  

228

Ra   (half-­‐life   5.8   years),   the   Ra   in  

48  

(I)   replacement   of   injected   water   by   large-­‐volume   brines,   i.e.   basin   brines,   sequestered   in  

49  

fracture   networks   or   isolated   facies   exogenous   to   the   shale   matrix   (Engle   and   Rowan   2014,  

50  

Rowan   et   al.   2015,   Stewart   et   al.   2015,   Kondash   et   al.   2017).     This   mechanism   is   consistent   with  

51  

a   common   genetic   Na-­‐Br-­‐Cl   signature   among   Marcellus   wastewaters   and   other   Appalachian  

52  

Brines   (Engle   and   Rowan   2014,   Rowan   et   al.   2015),   with   temporal   trends   in   δ   18O   data   that  

53  

suggest   the   mixing   of   different   water   sources   (Rowan   et   al.   2015),   and   with   differing   major  

54  

cation   ratios   between   experimental   shale   leachates   and   wastewaters   that   could   indicate  

55  

production  of  wastewaters  outside  the  shale  (Stewart  et  al.  2015).    

                                                                                                                *Hydraulic  fracturing  industry  describes  these  waters  as  flowback  when  recovered  before  gas  

production  commences,  and  produced  water  after  production;  collectively  we  term  them   wastewaters  as  they  cannot  be  discharged  to  the  environment  or  waste  facilities  without  substantial   treatment.  

 

2  

 

  56  

(II)   diffusion   of   salts   from   ubiquitous,   endogenous   formation   water   within   the   Marcellus  

57  

Shale  nanopore  network  into  the  injected  fluid  volume  (Balashov  et  al.  2015).    This  mechanism  

58  

is   consistent   with   the   lack   of   free   water   in   the   Marcellus   Shale   that   could   contribute   to  

59  

wastewater   volume   (Engelder   2012,   Engelder   et   al.   2014),   but   also   accommodates   a   brine  

60  

geochemical  signature  in  wastewaters  by  invoking  the  presence  of  a  labile,  capillary-­‐bound  ‘pore  

61  

brine’.     This   brine   would   share   Na-­‐Br-­‐Cl   signature   common   to   other   Appalachian   Brines   due   to   a  

62  

shared   marine   origin   and   basinal   geologic   history,   but   would   also   have   acquired   distinctive  

63  

Marcellus   87Sr/86Sr  (Chapman  et  al.  2012,  Capo  et  al.  2014,  Warner  et  al.  2012),  Li  (Phan  et  al.  

64  

2017)   and   226Ra/228Ra   isotope   ratios   (Rowan   et   al.   2015)   acquired   through   diagenesis   over  

65  

geologic  time  scales  as  the  brine  equilibrates  with  adjacent  mineral  (clay)  phases.  

66  

(III)   contemporaneous   water-­‐rock   interactions   between   injected   fluids   and   the   fractured  

67  

shale,  which  may  include  dissolution  of  soluble  salts  (Blauch  et  al.  2009)  or  cation  exchange  with  

68  

clay  minerals  (Renock  et  al.  2016);  clay  surfaces  might  be  enriched  in  divalent  alkaline  elements  

69  

by   membrane   filtration   (Engelder   et   al.   2014).     Evidence   for   these   interactions   comes   from  

70  

water  and  salt  mass  balance,  indicating  that  pore  brine  alone  cannot  accommodate  the  total  salt  

71  

extracted   from   the   shale   (Engelder   et   al.   2014);   from   excess   divalent   cations   extracted   during  

72  

Marcellus   flowback   (Barbot   et   al.   2013);   from   the   rapid   rate   (hours   to   days)   at   which   both  

73  

wastewaters  and  experimental  shale  leachates  increase  in  salinity  (Renock  et  al.  2016);  and  the  

74  

control   of   isovalent   exchange   reactions   on   cations   that   distinguish   Marcellus   wastewater  

75  

(Renock  et  al.  2016).      

76  

 

77  

Experimental   226Ra/228Ra   data   indicate   that   contemporary   water-­‐rock   reactions   play   a   critical  

78  

role   in   the   regulation   of   wastewater   Ra   chemistry.     We   proposed   in   a   companion   work   that   two  

79  

distinct  end-­‐members  within  the  Marcellus  Shale  can  contribute  Ra  to  wastewaters  (Landis  et  al.  

80  

2018).    These  are  (1)   labile  228Ra  in  a  mineral  phase  with  low  226Ra/228Ra  (~250)  that  is  accessible  

81  

to   low   ionic   strength   solutions   such   as   surface   waters.   The   226Ra/228Ra   ratio   of   this   phase   is  

82  

comparable  to  that  of  Appalachian  brines,  and  like  the  pore  brine  described  by  Balashov  et  al.  

83  

(2015)  is  accessible  to  pure  water  leachates;  for  this  reason  we  describe  both  as  labile  phases.    

84  

But  while  extraction  of  labile   228Ra  is  enhanced  at  high  pressure,  that  of  Na,  Ca  and  Ba  are  not,  

85  

and  228Ra  thus  appears  to  be  distinct  from  any  brine  per  se.    Importantly,  neither  labile  228Ra  nor  

86  

any   labile   pore   brine   can   reproduce   high  

87  

wastewaters.  

 

226

Ra/228Ra   ratios   that   distinguish   Marcellus  

3  

 

  88  

 

89  

The   second   Ra   source   within   the   Marcellus   Shale   is   (2)   exchangeable   226Ra   from   an   organic  

90  

phase  with  very  high   226Ra/228Ra  (~104)  and  that  is  only  accessible  in  very  high  [Ca2+]  solutions.      

91  

The  exchangeable   226Ra  is  physically  isolated  from  both  the  mineral   228Ra  source  and  any  labile  

92  

pore   brine   present   within   the   shale;   this   is   consistent   with   hydrophobicity   and   poor   water-­‐

93  

accessibility   of   organic   pore   networks   (Gu   et   al.   2015,   Zholfaghari   et   al.   2017).     While   226Ra   is  

94  

isolated  from  both   228Ra  and  pore  brine,  and  is  not  accessible  to  low-­‐ionic  strength  leachates,  it  

95  

is  rapidly  released  upon  external  additions  of  high  [Ca2+].    In  solutions  of  1M  Ca2+  over  10%  of  

96  

total  Ra  can  be  leached  from  the  shale  in  just  hours.    These  experimental  results  are  consistent  

97  

with  the  control  of  [Ca2+]/[Na+]  on  [Ra]  and  226Ra/228Ra  ratios  observed  in  Marcellus  wastewaters  

98  

(Landis  et  al.  2018).  Taken  together  these  observations  indicate  that  contemporary  water-­‐rock  

99  

interactions  and  ion  exchange  processes  control  Marcellus  wastewater  Ra  chemistry.    

100  

 

101  

Radium  in  Marcellus  wastewaters  derives  from  within  the  shale  itself,  and   226Ra/228Ra  ratios  in  

102  

shale  leachates  distinguish  Ra  in  wastewaters  derived  from  brines  and  solid  sources  within  the  

103  

shale.    But  the  timing  and  mechanisms  that  generate  extremely  high  [Ra]  and   226Ra/228Ra  ratios  

104  

in   Marcellus   wastewaters   have   not   been   described.     Here   we   use   isotope   mass   balance   and  

105  

mixing   models   to   reconcile   Marcellus   Ra   geochemistry   with   the   production   of   Marcellus  

106  

wastewater.     We   begin   with   a   compilation   of   Ra   data   in   Marcellus   Shale   and   wastewaters   in  

107  

order  to  provide  a  context  for  experimental  data  (Sections  4.1  and  4.2).    We  then  apply  a  simple  

108  

mass   balance   model   to   describe   the   water-­‐rock   interaction   between   shale   and   wastewater  

109  

(Section   5),   and   an   isotope   mixing   model   to   reproduce   wastewater   [Ra]   versus   226Ra/228Ra  

110  

relationship  (Section  6).    We  finish  with  a  discussion  of  mechanisms  contributing  to  wastewater  

111  

production  in  the  context  of  Ra  observations,  models  and  water-­‐rock  interactions  (Section  7).  

112  

 

113  

2.  Background  

114  

The   abundance   of   radium   isotopes   226Ra   (half-­‐life   1600   years)   and   228Ra   (half-­‐life   5.8   years)  

115  

reflects   secular   equilibrium   with   their   respective   U   and   Th   parents,   and   provides   insight   into  

116  

processes  operating  over  up  to  ca.  10  half-­‐lives  of  each  Ra  daughter  isotope  (about  15,000  years  

117  

for   226Ra  and  60  years  for   228Ra).    Radium  thus  does  not  record  a  genetic  or  diagenetic  origin  of  

118  

any   brine,   as   might   stable   or   long-­‐lived   isotope   systems   such   as   7Li/6Li   or   87Sr/86Sr.     Instead   226Ra  

119  

and   228Ra   concentrations   in   formation   waters   or   brines   reflect   contemporary   equilibrium   of  

 

4  

 

  120  

decay,  recoil  and  desorption  with  adjacent  U  and  Th-­‐bearing  solid  phases  (Krishnaswami  et  al.  

121  

1982,  Tricca  et  al.  2001).    

122  

 

123  

Wastewater   production   and   extraction   of   Ra   during   hydraulic   fracturing   is   rapid   relative   to   half-­‐

124  

lives   of   both   226Ra   and   228Ra,   consequently   we   ignore   recoil   and   decay   processes.     Similarly,  

125  

decay   of   the   long-­‐lived   Ra   isotopes   is   very   slow   relative   to   desorption   and   this   promotes   a  

126  

correlation   between   [Ra]   with   total-­‐dissolved-­‐solids   in   saline   aquifers   (Sturchio   2001).     Ra  

127  

extraction  during  the  timeframe  of  hydraulic  fracturing  thus  simplifies  to  a  problem  of  dilution  

128  

and   leaching   (desorption   plus   dissolution),   and   the   Ra   isotopic   composition   of   shale   leachates  

129  

can  be  used  to  infer  properties  of  host  phases  that  regulate  Ra  release  (Landis  et  al.  2018).    

130  

 

131  

3.  Experimental  approach  and  methods.  

132  

Our   experimental   data   are   described   in   a   companion   paper   (Landis   et   al.   2018).     For   these  

133  

experiments   we   selected   three   samples   of   Marcellus   Shale   that   span   a   range   of   total-­‐organic-­‐

134  

carbon  content  (TOC),  Ra  concentration  and  226Ra/228Ra  isotopic  composition.  These  samples  are  

135  

from   Indiana   County,   Pennsylvania   (2.0%   TOC;   undifferentiated   Marcellus   Shale);   Chenango  

136  

County,   New   York   (4.46%   TOC;   Lower   Oatka   Creek   Mbr,   Marcellus   Shale);   and   Yates   County,  

137  

New  York  (9.9%  TOC;  Upper  Oatka  Creek  Mbr,  Marcellus  Shale).    Thermal  maturity  of  all  rocks  is  

138  

overmature  and  in  the  dry  gas  window.    More  detailed  descriptions  of  these  rocks  are  given  in  

139  

Landis  et  al.  (2018),  Renock  et  al.  (2016)  and  Niu  et  al.  (2016).    Experimental  leachates  of  these  

140  

rocks   include   a   pure   water   fraction   (f1),   0.2M   Ca2+   produced   in   situ   by   dissolution   of   carbonates  

141  

fraction   (fX),   and   1M   CaCl2   (f2).     Significant   amounts   of   Na,   Ca,   Sr,   Ba   and   S   are   released   in   pure  

142  

water,   providing   [Ca2+]   around   0.01M   (Renock   et   al.   2016).     Analytical   details   are   given   in   Landis  

143  

et   al.   (2018),   with   analysis   of   major   elements   by   inductively-­‐coupled   plasma   optical-­‐emission  

144  

spectroscopy   (ICPOES)   and   Ra   isotopes   by   isotope-­‐dilution   thermal   ionization   mass  

145  

spectrometry  (TIMS).    A  summary  of  descriptive  data  for  these  rocks  is  given  in  Table  1.      

146  

 

147  

To   provide   a   context   for   comparing   experimental   leachates   to   Marcellus   wastewaters,   we  

148  

compiled   wastewater   Ra   data   from   the   following   sources,  Rowan   et   al.   (2011),   Haluszczak   et   al.  

149  

(2012),  Rowan   et   al.   (2015),  PADEP   (2015).    We  report   226Ra/228Ra   atom   ratios,  and  all  literature  

150  

activity  data  were  converted  using  relation:  N226/N228  =  (A226/A228)·∙(λ228/λ   226),  and  λ228/λ226  ≈278,  

151  

where  N  specifies  atom  abundance,  A  radioactivity,  and   λ  decay  constant.      We  also  compiled  

 

5  

 

  152  

bulk   shale   Ra   data   from   reported   U   and   Th   concentrations   (w/w)   and   calculated   [226Ra]   and  

153  

[228Ra]   assuming   secular   equilibrium   with   [226Ra]   =   3.40×10-­‐7·∙[U]   and   [228Ra]   =   4.01×10-­‐10·∙[Th],  

154  

and  the  atom  ratio  226Ra/228Ra  =  840·∙(U/Th);  half-­‐lives  of  226Ra  =  1600  y,  238U  =  4.47x109  y,  228Ra  =  

155  

5.75   y,   and   232Th   =   14.05x109   y.     Data   sources   include   Leventhal   et   al.   (1981),   Lash   and   Blood  

156  

(2012),  and  Niu  et  al.  (2016).  

157  

 

158  

4.  Results  

159  

4.1  Comparing  226Ra/228Ra  in  wastewater  and  Marcellus  Shale  

160  

Compiled  data  are  presented  in  Figure  1.  The  Ra  isotope  ratios  of  Marcellus  wastewaters  (log-­‐

161  

normal   distribution,   with   median   226Ra/228Ra   =   1848;   n   =   79)   are   similar   to   those   of   compiled  

162  

Marcellus   Shale   measurements   (226Ra/228Ra   =   1507;   n   =   331).     To   estimate   Ra   composition   of  

163  

Marcellus   Shale   we   also   used   the   correlation   of   organic   matter   content   with   226Ra/228Ra   ratios  

164  

(see  Fig.  2  in  Landis  et  al.  2018),  as  this  approach  avoids  possible  sampling  bias  towards  organic-­‐

165  

rich  facies  (Wang  and  Carr  2013).    Analysis  of  18  wells  in  southwestern  PA  and  northern  VA  by  

166  

Wang  and  Carr  (2013)  shows  that  Marcellus  Shale  subjected  to  hydraulic  fracturing  is  typically  

167  

ca.  5%  total-­‐organic-­‐carbon  (TOC),  which  corresponds  to  a  bulk  rock  226Ra/228Ra  atom  ratio  of  ca.  

168  

1050.    Facies  with  TOC  >  6.5%  are  considered  organic  rich,  with  corresponding   226Ra/228Ra  atom  

169  

ratios  >1470.    

170  

 

171  

In  contrast  to  Marcellus  Shale  and  its  wastewaters,  the  226Ra/228Ra  composition  of  non-­‐Marcellus  

172  

brines   of   the   Appalachian   basin   is   low   (median   =   250)   and   similar   to   that   of   the   upper  

173  

continental   crust   (∼200;   Wedepohl   1995;   Figure   1).     With   some   overlap   in   their   distributions,  

174  

Marcellus   wastewater   Ra   concentrations   (median   =   4550   pg   L-­‐1)   are   an   order   of   magnitude  

175  

higher   than   those   of   Appalachian   Basin   brines   (median   =   450   pg   L-­‐1;   Figure   1).   The   distinctive   Ra  

176  

composition   of   Marcellus   wastewater   and   its   similarity   to   Marcellus   Shale   implies   that  

177  

wastewater   Ra   is   derived   from   solid   phases   of   the   shale   with   limited   contributions   from   any  

178  

exogenous  basin  brines.  

179  

 

180  

Direct   comparison   between   distributions   of   Marcellus   Shale   and   wastewaters   is   sufficient   to  

181  

demonstrate  that  wastewaters  likely  derive  from  within  the  shale  (Rowan  et  al.  2015),  and  that  

182  

typical   wastewater   appears   to   have   somewhat   higher   226Ra/228Ra   compositions   than   typical  

183  

shale.     But   more   detailed   inference   is   not   possible   and   a   modeling   approach   is   required   to  

 

6  

 

  184  

understand   the   relationship   between   Marcellus   Shale   and   wastewaters   due   to   (1)   the   relative  

185  

scales   of   rock   and   water   samples,   where   individual   rock   measurements   reflect   tens   to   hundreds  

186  

of   grams   of   shale,   and   individual   wastewater   measurements   are   likely   to   integrate   many  

187  

thousands  of  liters  of  wastewater  extracted  from  millions  of  cubic  meters  of  shale  formation;  (2)  

188  

incongruent   weathering,   i.e.   different   susceptibilities   of   multiple   host   phases   to   Ra   extraction  

189  

from  the  shale,  and  (3)  the  roles  that  evolving  wastewater  ionic  strength,  [Ca2+]  and  Ca/Na  ratios  

190  

play  in  regulating  release  from  these  phases  (Landis  et  al.  2018).      

191  

 

192  

4.1.2  Signature  of  Ba(Ra)SO4  scaling  in  Marcellus  wastes  

193  

In   estimating   median   [Ra]   of   wastewaters   we   omitted   samples   that   fall   outside   of   the   typical  

194  

field   shown   in   Fig.   1   due   to   their   likely   modification   by   precipitation   of   radiobarite   or   co-­‐

195  

precipitation   with   calcite.     We   did   not   exclude   them   from   estimating   226Ra/228Ra.     To   justify   this,  

196  

we  note  that  thermodynamic  modeling  of  produced  water  chemistries  reported  in  Haluszczak  et  

197  

al.   (2012)   indicates   that   all   produced   water   samples   are   saturated   with   respect   to   barite.    

198  

Whereas   bulk   Marcellus   shale   consists   of   several   weight-­‐percent   sulfur,   and   laboratory   leaching  

199  

of   the   shale   demonstrates   extraction   of   sulfur   even   in   anoxic   conditions   (Renock   et   al.   2016,  

200  

Landis   et   al.   2018),   wastewaters   are   depleted   in   sulfur.     The   reasonable   conclusion   is   that  

201  

measured  wastewater  chemistries  typically  reflect  some  degree  of  sulfate  precipitation  as  CaSO4  

202  

and/or  Ba(Ra)SO4.      

203  

 

204  

Radium   isotopes,   and   radium   versus   barium,   should   not   be   fractionated   from   one   another  

205  

(Porcelli   et   al.   2001)   once   solubilized   in   hydraulic   fracturing   wastewater.     However,   scaling   of  

206  

(Ba,Ra)SO4   will   reduce   the   concentration   of   both   Ra   and   Ba   in   liquids   versus   unaltered  

207  

wastewaters,   and   necessarily   increase   their   concentrations   in   waste   solids   versus   bulk   shale.     Of  

208  

reported  wastewaters  compiled  from   Haluszczak  et  al.  (2012),  Rowan  et  al.  (2011),  Rowan  et  al.  

209  

(2015),   PADEP   (2015)   and   this   study,   17%   fall   outside   of   this   characteristic   Ra   field.   In  

210  

comparison,  56%  of  recycled  or  treated  wastewaters  fall  outside  of  the  characteristic  field  (Fig.  

211  

2a).     Considering   solid   waste   materials,   78%   of   filter   cake   solids   from   wastewater   treatment  

212  

facilities  (PADEP  2015)  fall  outside  the  characteristic  bulk  shale  field  (Fig.  2b).  

213  

 

214  

5.  Radium  (isotope)  mass  balance  during  water-­‐rock  interaction  

215  

5.1.  Estimating  a  characteristic  water-­‐rock  ratio  during  hydraulic  fracturing  

 

7  

 

  216  

Shale   leaching   experiments   can   provide   insight   into   the   Ra   chemistry   of   Marcellus   wastewaters,  

217  

provided   an   understanding   of   the   water-­‐rock   (W-­‐R)   ratio   for   the   alteration   of   shale   by   fluid  

218  

interaction  during  hydraulic  fracturing  (Renock  et  al.  2016),  

219  

 

220  

!

221  

 

222  

Here  Crock  is  the  Ra  concentration  in  the  bulk  shale  [pg  kg-­‐1],  Cwater  is  the  Ra  concentration  in  the  

223  

wastewater   [pg   L-­‐1],   γ   is   the   extraction   efficiency   of   Ra   from   the   rock,   and   W/R   is   the   water-­‐rock  

224  

ratio   of   the   interaction   [L   kg-­‐1].     For   these   long-­‐lived   Ra   isotopes   we   ignore   radioactive   decay  

225  

(Krishnaswami   et   al.   1982,   1991),   and   Eq.   1   is   thus   equivalent   to   the   mass   balance   approach  

226  

used  for  water-­‐rock  interactions  in  hydrothermal  systems  (Alt-­‐Epping  and  Smith  2001)  or  for  Ra  

227  

exchange   in   aquifer   systems   (Sturchio   et   al.   2001).     In   the   latter   case   the   behavior   of   Ra   is  

228  

modeled  according  to  instantaneous  first-­‐order  exchange  reactions,  and  it  can  be  shown  that   γ  

229  

is  equivalent  to  the  inverse  of  the  effective  dimensionless  partition  coefficient  K.      

230  

 

231  

5.2.  Water-­‐Rock  ratio  and  mixing  of  Ra  source  phases  

232  

The   calculation   of   Eq.   1   provides   an   upper   bounding   limit   on   W-­‐R   ratio   of   fracturing,   ignoring  

233  

radiobarite  as  a  Ra  sink.    With  median  rock  and  wastewater  concentrations   as  CR/  CW  =  7141  pg  

234  

kg-­‐1/4550  pg  L-­‐1,  the  maximum  water-­‐rock  ratio  of  fracturing  is  ca.  1.6  L  kg-­‐1  if  100%  of  Ra  were  

235  

extracted   from   the   rock.     The   effect   of   scaling   would   be   to   bias   the   W-­‐R   estimate   too   high.    

236  

Experimental   Ra   extraction   efficiencies   from   the   shale   are   on   the   order   of   10%   (Landis   et   al.  

237  

2018),   and   the   W-­‐R   ratio   typical   of   hydraulic   fracturing   must   thus   be   an   order   of   magnitude  

238  

lower,  or  less  than  ca.  0.10  L  kg-­‐1.    

239  

 

240  

We   stress   that,   because   wastewater   Ra   must   be   derived   from   the   Marcellus   Shale,   the  

241  

extraordinarily  high  [Ra]  that  characterizes  Marcellus  wastewater  can  only  be  attained  through  

242  

very   low   W-­‐R   values   during   the   water-­‐rock   interaction.   To   obtain   late-­‐stage   Marcellus  

243  

wastewaters   with   high   [Ra]   and   226Ra/228Ra   compositions   congruent   with   that   of   Marcellus  

244  

Shale,  W-­‐R  ratios  spanning  ca.  0.025  to  0.15  are  required  as  shown  in  Fig.  1b.    Our  calculation  

245  

using   Ra   indicates   W-­‐R   ratios   similar   to   those   using   Ba,   Na   or   Sr   (Balashov   et   al.   2015,   Renock   et  

246  

al.   2016),   but   ours   is   absolute   because   Ra   is   demonstrably   derived   from   the   Marcellus   Shale  

!

 

=𝛾∙

!!"#$ !!"#

%$    

 

 

 

 

 

 

 

 

Eq.  1  

8  

 

  247  

itself   whereas   the   provenance   of   other   elements   cannot   be   guaranteed   to   exclude   exogenous  

248  

contributions.  

249  

 

250  

Understanding   the   W-­‐R   ratio   of   hydraulic   fracturing   provides   a   critical   basis   for   interpreting  

251  

mechanisms  of  wastewater  production  using  experimental  leachate  data.    Very  low  W-­‐R  ratios  

252  

are   not   achievable   in   the   laboratory   in   static   batch   experiments.     However,   using   Eq.   1   we   solve  

253  

Cwater   for   all   of   our   leachates   to   predict   wastewater   Ra   composition   from   a   given   shale   facies,  

254  

using   measured   rock   Ra   concentrations,   and   leachate   Ra   extraction   yields   and   226Ra/228Ra  

255  

compositions.     Experimental   leachates   can   then   be   compared   directly   to   wastewater   data,   as  

256  

shown   in   Figure   3.     Leachates   appear   as   mixing   trajectories   where   [Ra]   vs.   226Ra/228Ra   ratio  

257  

follows  increasing  [Ca2+].  Importantly,  trajectories  defined  by  leachates  from  the  Chenango  Co.  

258  

(with   TOC   and   Ra   composition   typical   of   Marcellus   Shale)   approximate   [Ra]   vs.   226Ra/228Ra  

259  

compositions  in  basin-­‐wide  wastewater,  provided  a  W-­‐R  ratio  of  0.04  to  0.10  (Fig.  3a).  Aggregate  

260  

trajectories  from  various  leached  shale  domains,  as  represented  by  our  test  rocks  with  a  range  

261  

of  organic  content,  define  the  characteristic  Ra  signature  of  Marcellus  wastewaters.    These  are  

262  

shown  in  Figure  3b.      

263  

 

264  

Comparable  trajectories  in  226Ra/228Ra  composition  among  leachates,  rocks  and  wastewaters  are  

265  

attributable   to   mixing   within   each   between   the   same   two   principal   Ra   end-­‐members,   mineral  

266  

and   organic,   and   to   the   control   of   organic   carbon   on   total   Ra   abundance   (Landis   et   al.   2018).    

267  

Isotopic  disequilibrium  between  leachates  and  bulk  Marcellus  Shale  is  thus  governed  by  (1)  the  

268  

stronger  effective  partitioning  of  Ra  in  organic  surface  sites  relative  to  that  derived  from  mineral  

269  

sites,   (2)   increasing   susceptibility   of   the   226Ra   organic   phase   to   exchange   at   higher   leachate  

270  

[Ca2+],  (3)  the  much  larger  reservoir  of  Ra  associated  with  shale  organic  carbon.      

271  

 

272  

5.3.  Constraining  the  physical  scale  of  reactive  rock  surfaces  

273  

Low  W-­‐R  ratios  are  necessary  to  explain  Ra  mass  balance  and  isotope  mixing,  but  we  require  a  

274  

physical  model  that  can  accommodate  W-­‐R  ratios  on  the  order  of  0.04.    Such  low  ratios  require  

275  

an   extraordinary   rock   surface   area   that   is   ionically   accessible   to   injected   fluids.     This   is   likely  

276  

achieved  within  the  shale  pore  network  where  pores  average  ca.  3  nm  diameter,  or  smaller  for  

277  

pores  within  organic  matter  (Gu  et  al.  2015).    If  a  pore  is  considered  to  be  a  fluid-­‐filled  sphere,  

278  

i.e.  a  single  pore  model,  the  rock  enclosing  the  pore  can  be  considered  to  have  a  reactive  layer  

 

 

9  

 

  279  

subject  to  leaching.    The  thickness  of  this  layer  defines  the  effective  water-­‐rock  (W-­‐R)  ratio  for  

280  

the  pore.    A  pore  of  diameter  3  nm  has  a  volume  of  14  nm3  calculated  as  4/3πr3.    A  reactive  layer  

281  

of   thickness   t   has   a   volume   of   4/3π(r’3-­‐   r3),   where   r’   =   r   +   t.     This   reactive   rock   volume   is  

282  

converted  to  mass  assuming  a  bulk  density  of  2.6  g  cm3.    A  pore  volume  to  reactive  rock  mass  

283  

ratio  (W-­‐R)  of  0.04  requires  a  thickness  of  ca.  15  Å.    Smaller  pores  require  thinner  reactive  layers  

284  

to  accommodate  the  W-­‐R  relationship.  

285  

 

286  

In  a  dynamic  diffusion  model,  Balashov  et  al.  (2015)  estimated  that  a  volume  of  injected  water  

287  

interacts  with  ca.  30x  its  volume  in  pore  space  to  produce  wastewater  salinity.    Balashov  et  al.  

288  

presume   that   the   pore   volume   is   filled   with   formation   brine.     Their   estimated   ratio   (α)   of  

289  

injectate   to   pore   volume   is   0.03,   which   is   comparable   to   our   volume-­‐mass   (W-­‐R)   estimates  

290  

based   on   Ra   isotopes.     Pore   volume   can   be   converted   to   a   reactive   rock   volume   using   the  

291  

geometric   relationship   as   above,   where   the   volume   ratio   of   pore   to   rock   =   r3   /(r’3-­‐   r3).     Rock  

292  

volume   is   again   converted   to   mass   using   a   density   of   2.6   g   cm-­‐3.     W-­‐R   can   thus   be   calculated  

293  

from  α  as  follows,  

294  

 

295  

!

296  

 

297  

From  this  relation  the  reactive  layer  thickness  (t=r’-­‐r)  can  be  estimated.    Thickness  of  just  1.5  Å  

298  

provides  a  W-­‐R  ratio  of  0.04.      

299  

 

300  

In  this  estimate  we  have  omitted  any  contribution  to  water  volume  from  formation  brines  that  

301  

may  be  present  in  the  pores.    While  in  their  model  Balashov  et  al.  (2015)  presume  that  soluble  

302  

elements   in   the   shale   reflect   pore   volume   that   is   filled   with   formation   brine,   our   Ra   leaching  

303  

data  demonstrate  that  the  bulk  of  wastewater  Ra  is  not  derived  from  a  labile  brine  but  instead  

304  

from   organic   surfaces;   in   this   case   our   omission   of   the   pore   brine   volume   is   valid   if   either  

305  

formation  salts  are  not  present  as  brine  as  assumed,  or  if  Ra  is  extracted  from  additional  pore  

306  

volume   which   is   not   water   saturated.     If   we   follow   Balashov   et   al.   in   assuming   that   2%   of   the  

307  

shale   is   water-­‐saturated   pore   volume,   and   further   assume   that   this   water   fully   equilibrates   with  

308  

injected  fluids  (which  seems  unlikely),  the  total  water  volume  in  the  W-­‐R  interaction  increases  

309  

by   a   factor   of   30.     This   case   provides   an   upper   bounding   limit   on   reactive   rock   thickness  

!

 

=  𝛼 ∙

!! !!! !! !

÷ 𝜌  

 

 

Eq.  2  

10  

 

  310  

necessary  to  provide  a  W-­‐R  ratio  of  0.04  and  converges  to  that  of  the  single  pore  model,  which  

311  

is  15Å.      

312  

 

313  

A  reactive   rock  layer  thickness  between  1.5  and  15  Å  approaches  the  order  of  a  chemical  bond  

314  

length   and   thickness   of   the   Double-­‐Electrical   Layer   at   shale   clay   surfaces   (Kwon   et   al.   2004),  

315  

suggesting   that   the   physical   constraints   for   W-­‐R   ratios   on   the   order   of   0.04   can   be   satisfied  

316  

within   the   ionic   exchange   environment   of   the   shale.     Future   work   might   focus   on   reconciling  

317  

geophysical  constraints  with  the  chemical  W-­‐R  estimated  here  using  Ra.  

318  

 

319  

6.  Isotope  mixing  models  predict  wastewater  Ra  composition  

320  

Based   on   an   understanding   of   the   physical   scale   of   the   shale-­‐wastewater   W-­‐R   interaction,  

321  

mixing  between  mineral  228Ra  and  organic  226Ra  end-­‐members,  and  the  role  of  [Ca2+]  in  releasing  

322  

organic   226Ra,  we  constructed  isotope-­‐mixing  models  to  see  if  we  could  reproduce  the  [Ra]  vs.  

323  

226

324  

predicted   by   the   model   is   derived   from   standard   isotope   mixing   as   described   in   Appendix   1.    

325  

Briefly,  Ra  isotopic  composition  of  a  leachate  is  predicted  by  the  following:  

326  

 

327  

𝑄!"# = !

328  

 

329  

Here   Qmix   is   the   226Ra/228Ra   isotopic   ratio   (Quotient)   of   the   leachate,   γm   is   the   extraction  

330  

efficiency   of   total   rock   226Ra   that   is   attributable   to   the   m   phase   (mineral   Ra   atoms   extracted   per  

331  

kilogram  rock),   γo  that  of  the  organic  phase,  CR226  is  the  total  concentration  of   226Ra  in  the  rock  

332  

[atoms  kg-­‐1];  the   mass   of   rock   subjected   to   leaching,   R,   is   omitted   as   written   for   a   unit   mass   of   1  

333  

kg.    

334  

 

335  

Total  [Ra]  is  predicted  as  a  function  of  [Ca2+]  as  follows:  

336  

 

337  

[𝐶𝑎 !! ] =

338  

 

339  

𝑦! = 𝑓 𝐶𝑎 !!      

340  

 

Ra/228Ra  isotopic  composition  that  characterizes  Marcellus  wastewaters.    Radium  composition  

!! ∙!!!!" ∙!!! ∙!!!!" !!" !! ∙!!!!" ! ∙!! !! ! !!

 

!!!" ∙ !!" ! !

 

= !!

!! !!!   !! !! ! !!

 

 

 

 

 

Eq.  3  

 

 

 

 

 

 

 

 

Eq.  4  

 

 

 

 

 

 

 

 

Eq.  5  

11  

 

  341  

Coupling   Eqs.   4   and   5,   our   model   thus   assumes   that   226Ra   yield   of   the   organic   phase   (γo)  

342  

increases  with  decreasing  W/R,  but  scaling  between  the  two  is  a  fitted  parameter  mediated  by  

343  

increasing  concentrations  of  Ca2+.    We  adjusted  the  mixing  model  to  reproduce  wastewater  data  

344  

by  assuming  a  range  of  [Ca2+]  vs.  γo  relationships.  Organic  Ra  yield  (γo)  is  constrained  at  ca.  50%  

345  

in  1M  [Ca2+]  as  derived  by  Landis  et  al.  (2018).    Total  Ca2+  and  Ra  mineral  yields  (γm)  might  also  

346  

be  adjusted  as  fitted  parameters  but  here  are  assumed  to  be  constant  for  all  W/R.    Ca2+  yield  is  

347  

constrained   at   5-­‐10%,   similar   to   observed   yields   in   pure   water   and   exchangeable   fractions  

348  

(Renock  et  al.  2016,  Stewart  et  al.  2015).    Mineral  Ra  yield  (γm)  is  constrained  at  0.5-­‐1%,  similar  

349  

to  those  observed  in  pure  water  and  deduced  from  CaCl2  leachates  (similar  amounts  of  mineral  

350  

Ra  are  extracted  both  in  pure  water  at  high  pressure  and  in  1M  CaCl2  at  atmospheric  pressure).  

351  

 

352  

Within  these  constraints  we  considered  a  range  of  mixing  scenarios  (described  in  Table  A1  and  

353  

Figure   A1)   instead   of   attempting   a   robust   fit   of   the   wastewater   data   since   the   relationship  

354  

between   organic   yield   (γo)   and   [Ca2+]   is   not   well   known   at   the   relevant   scales   and   conditions.    

355  

Nonetheless,  all  scenarios  reproduce  wastewater  Ra  data  as  shown  in  Fig.  4.    In  all  scenarios  the  

356  

critical   parameters   [Ra],   226Ra/228Ra,   [Ca2+]   and   ionic   strength   are   comparable   to   wastewater  

357  

measurements,  without  requiring  external  sources  of  Ra  or  Ca  as  from  basin  brine.    That  is,  all  

358  

wastewater  data  can  be  explained  by  the  leaching  of  Ra  from  the  fine  shale  fabric  as  evident  in  

359  

the   hand   sample,   and   does   not   require   invocation   of   large-­‐scale   features   such   as   isolated   facies,  

360  

fracture  networks  or  basin  brines.    Isotope  mixing  model  results  suggest  that  the  shale  mineral  

361  

228

362  

90%  derived  from  organics  by  leaching  (Table  A1).    

363  

 

364  

Wastewater  time  series  data  from  Greene  County,  PA  reported  by  Rowan  et  al.  (2011)  provide  a  

365  

test  case  for  assessing  rapid  water-­‐rock  interactions  for  a  single  fracture  network.    In  this  case  

366  

saline,   Ra-­‐enriched   wastewater   from   a   previous   well   was   recycled   as   the   fracturing   fluid.    

367  

Salinity,  [Ra]  and  226Ra/228Ra  ratio  of  wastewater  increase  with  time  beyond  that  of  the  injectate,  

368  

proceeding  rapidly  over  days  following  initiation  of  flowback  and  continuing  to  increase  for  over  

369  

one  year  (Figure  4b;  Rowan  et  al.  2011,  Rowan  et  al.  2015).    A  three-­‐member  mixing  scenario  

370  

between   recycled   injected   fluid,   extracted   mineral   and   organic   phases   of   the   shale,   and   a  

371  

decreasing  W-­‐R  ratio  over  this  time  period,  can  explain  the  observed  evolution  in  wastewater  Ra  

372  

composition   as   illustrated   in   Fig.   4b.     In   this   scenario,   the   fraction   of   Ra   derived   from   the  

 

Ra  phase  likely  contributes  ca.  10%  of  total  wastewater  Ra  (by  mass)  with  the  balance  of  ca.  

12  

 

  373  

injected  fluid  is  ca.  47%  on  the  first  day  of  wastewater  production,  dropping  to  ca.  11%  by  the  

374  

end   of   one   year.     Radium   contributions   from   mineral   and   organic   phases   of   the   shale   evolve  

375  

from  6%  and  48%  on  the  first  day,  to  5%  and  84%  by  day  438,  respectively.    

376  

 

377  

7.  Reconciling  Ra  isotope  geochemistry  with  Marcellus  wastewater  production    

378  

Three   mechanisms   are   proposed   for   wastewater   production,   (I)   mixing   with   exogenous   basin  

379  

brine,   (II)   diffusion   from   endogenous   pore   brine,   or   (III)   contemporaneous   water-­‐rock   reactions.    

380  

These   are   not   mutually   exclusive.     However,   because   wastewater   Ra   must   be   derived   from   solid  

381  

phases  within  the  Marcellus  Shale  itself,  it  provides  an  absolute  tracer  of  physic-­‐chemical  water-­‐

382  

rock   interactions   that   govern   wastewater   production   and   may   thus   help   to   discriminate  

383  

contributions  from  different  mechanisms.      

384  

 

385  

Radium  isotope   mass   balance   requires  that  the  characteristic  W-­‐R  ratio   of   hydraulic   fracturing  

386  

must  be  on  the  order  of  0.04.    With  ratios  of  this  order,  the  extraction  of  endogeneous  elements  

387  

in   the   shale   matrix   including   Na,   Ca,   Ba   and   Sr,   can   reproduce   wastewater   salinity   without  

388  

contributions  from  exogenous  large-­‐volume  brines  (I).    This  is  an  important  observation  because  

389  

Ra  or  Sr  isotopes  alone  cannot  rule  out  the  role  that  fracture  networks  could  play  in  transporting  

390  

exogenous   fluids   through   the   fracture   network   (e.g.,   Warner   et   al.   2012).     The   rapid   rate   at  

391  

which  Sr  and  Ra  can  be  leached  from  experimental  shale  samples  suggests  that  any  brine  routed  

392  

through   the   Marcellus   Shale   would   rapidly   acquire   a   Marcellus   signature   irrespective   of   its  

393  

source.    However,  a  low  W-­‐R  ratio  for  the  fluid-­‐rock  interaction  and  mass  balance  considerations  

394  

must   still   apply.     Following   are   additional   insights   and   considerations   from   Ra   isotopes   that  

395  

discriminate  mechanisms  of  wastewater  production.  

396  

 

397  

7.1.  Salt  and  water  mass  balance  in  Marcellus  wastewater  

398  

The  limited  amount  of  water  available  within  the  Marcellus  Shale  (Engelder  2012,  Engelder  et  al.  

399  

2014)   and   the   small   wastewater   volume   recovered   at   the   well   head,   relative   to   the   amount  

400  

injected,  argue  further  against  the  role  of  exogenous  brines  (I)  in  wastewater  production  –  the  

401  

system  is  water  limited.    Analysis  of  Marcellus  well  logs  from  New  York  shows  that  on  average  

402  

the  porosity  of  the  shale  ranges  from  4.7%  to  7.9%  with  a  permeability  of  about  one  hundred  

403  

nano-­‐Darcy.  We  determined  averages  of  total   porosity  (=𝜙)  and  pore  volume  occupied  by  water  

404  

(=water  saturation,  Sw)  in  the  Marcellus  Shale  using  density,  resistivity  and  gamma-­‐ray  logs  for  

 

13  

 

  405  

three   wells   in   New   York   (Advanced   Resources   International   2011)   and   used   these   data   to  

406  

calculate  the  mass  of  water  present  per  unit  mass  of  the  shale  as  follows:  

407  

 

408  

  =

409  

 

410  

where   𝜌   denotes   density   and   subscripts   ‘B’   and   ‘R’   denote   formation   water   and   rock,  

411  

respectively.  Assuming  that  the  densities  of  the  brine  and  shale  are  1.1  g  cm-­‐3  and  2.73  g  cm-­‐3,  

412  

respectively,  we  find  that    ratio  varies  from  2  mg  g-­‐1  to  12  mg  g-­‐1  with  an  average  of  5  mg  g-­‐1  or  

413  

0.5%  w/w.      

414  

 

415  

With  an  estimated  0.5%  of  shale  mass  comprised  of  water  and  a  hydraulic  fracturing  W-­‐R  ratio  

416  

of   0.04,   the   contribution   of   pore   brine   or   formation   water   to   wastewater   volume   is   less   than  

417  

13%  (calculated  as  1/(0.04÷0.005)).    This  assumes  no  additions  of  brine  from  isolated  fractures  

418  

or  facies.  

419  

 

420  

Following   our   assumptions,   the   balance   and   majority   of   wastewater   volume   is   comprised   of   the  

421  

original  injected  fluid  rather  than  formation  water.    Given  this,  late  stage  produced  water  cannot  

422  

represent  pure  brine  as  proposed  by   Kondash  et  al.  (2017).  With  TDS  exceeding  200,000  mg  L-­‐1,  

423  

the   dilution   of   this   wastewater   by   a   factor   of   ca.   10   implies   a   brine   with   impossibly   high  

424  

concentrations   exceeding   200%   by   weight.     Engelder   et   al.   2014   described   this   as   the   ‘salinity  

425  

dilemma’.     Water   and   salinity   mass   balance   in   wastewaters   are   not   satisfied   strictly   by  

426  

extraction  of  brines,  and  additional  mechanisms  must  be  invoked.    

427  

 

428  

7.2.  Geochemical  tracers  of  labile  brine  versus  water-­‐rock  interactions  

429  

7.2.1  Na-­‐Br-­‐Cl  and  δ18O  

430  

The   contribution   of   endogenous   pore   brines   (II)   to   wastewater   salinity   is   necessary   to   explain  

431  

some   geochemical   tracers   and   their   evolution   during   wastewater   recovery,   but   insufficient   to  

432  

explain  others.    Contributions  of  brine  are  necessary  to  reconcile  wastewater  Na-­‐Br-­‐Cl  signature,  

433  

which   is   derived   from   paleo   seawater   and   requires   contributions   from   a   liquid   brine   rather  

434  

dissolution  of  soluble  salts  (Engle  and  Rowan  2014,  Rowan  et  al.  2015).  Similarly,  δ18O  trends  in  

435  

wastewaters  reported  by  Rowan  et  al.  (2015)  suggest  that  mixing  of  liquid  end-­‐members  within  

!

!×!! ×!!

!

!!

     

 

 

 

 

 

 

 

 

Eq.  3  

! !

 

14  

 

  436  

the   shale   is   dominated   by   a   brine   component.     However,   in   the   Rowan   et   al.   data   recycled  

437  

wastewaters   were   reinjected   for   fracturing   and   the   wastewater   signature   is   approximately  

438  

uniformly  depleted  in  δ18O,  about  -­‐3  to  -­‐1  (Rowan  et  al.  2015);  the  apparent  degree  of  mixing  

439  

between   injected   fluids   and   endogenous   brine   is   limited,   as   the   endmember   brine   δ18O  

440  

signature  is  unknown.    Conventional  brine  δ18O  values  measured  by  Dresel  and  Rose  (2010)  span  

441  

-­‐6  to  +2,  but  are  highly  sensitive  to  alteration  by  depleted  groundwaters  with  δ18O  compositions  

442  

lying  on  the  meteoric  water  line.    Rowan  et  al.  (2015)  conclude  that  salinity  and  not  δ18O  is  the  

443  

strongest  argument  for  mixing  between  distinct  fluids  in  the  generation  of  wastewaters.      

444  

 

445  

7.2.2.  alkaline  earth  elements  

446  

Consistent   with   the   extraction   of   divalent   cations   from   an   exchangeable   phase,   Barbot   et   al.  

447  

(2013)   demonstrate   non-­‐conservative   behavior   of   divalent   cations   versus   Br   and   Cl   during  

448  

flowback.     Unlike   conventional   brines,   low   salinity   wastewaters   are   depleted   in   alkaline   earth  

449  

elements   relative   to   Br   and   Cl.     Over   time   the   cations   Mg,   Ca,   Sr,   and   Ba   become   more   enriched  

450  

in   wastewaters   relative   to   Br   or   Cl.     As   the   divalent   elements   increase   but   both   SO4   and   CO3  

451  

decrease   with   time   (Barbot   et   al   2013),   this   trend   might   be   interpreted   as   the   continued  

452  

extraction  of  alkaline  cations  from  the  rock  in  late  wastewaters  rather  than  their  loss  to  scaling  

453  

in  early  water  as  Barbot  et  al.  imply.      

454  

 

455  

In  the  data  of  Rowan  et  al.  (2015)  we  also  see  that  while  Na-­‐Br-­‐Cl  may  increase  by  a  factor  of  ~2-­‐

456  

3  during  wastewater  production,  Mg,  Ca,  and  Sr  increase  by  a  greater  amount,  and  Ba  and  Ra  

457  

may  increase  by  factors  up  to  10  over  the  same  time  period.    Ba  and  Ra  are  principally  derived  

458  

from   the   shale   by   ion   exchange   reactions,   from   clays   (Renock   et   al.   2016)   and   organics,  

459  

respectively   (Landis   et   al.   2018),   and   thus   provide   sensitive   indicators   of   water-­‐rock   interaction.    

460  

Both  [Ba]  and  [Ra]  increase  most  rapidly  in  wastewaters  among  all  cations.    Critically,  226Ra/228Ra  

461  

also  increases  while  Ba/Ra  remains  about  constant,  suggesting  that  the  increase  in  [Ba]  and  [Ra]  

462  

is   attributable   to   their   enhanced   extraction   as   wastewater   salinity   increases,   rather   than   to  

463  

exhaustion   of   sulfate   and   cessation   of   radiobarite   precipitation;   the   latter   processes   would  

464  

affect  both  Ra  isotopes  identically.  

465  

 

 

15  

 

  466  

Both  the  occurrence  of  scaling  and  isovalent  exchange  with  other  alkaline  earth  elements  likely  

467  

contribute   to   incongruence   of   major   element   ratios,   e.g.,   Ca/Na,   between   experimental  

468  

leachates  and  wastewaters  as  observed  by  Stewart  et  al.  (2015).  

469  

 

470  

7.2.3.  lithium  and  strontium  isotopes  

471  

Both   Li   and   Sr   isotopes   in   wastewaters   bear   an   isotopic   signature   attributed   to   Marcellus   clay  

472  

mineral   diagenesis   (Phan   et   al.   2016,   Chapman   et   al.   2012),   acquired   during   brine   emplacement  

473  

and   residence   in   the   shale,   and   distinct   from   other   Appalachian   brines   (Warner   et   al.   2012).    

474  

Water  soluble,  NH4-­‐exchangeable  and  clay  interlayer  (HCl-­‐extractable)  fractions  extracted  from  

475  

shale  all  show  Sr  isotope  ratios  (0.709-­‐0.711)  consistent  with  (or  slightly  lower  than)  wastewater  

476  

87

477  

0.75;  Stewart  et  al.  2015;  Balashov  et  la.  2015).  Thus,  while  experimental  shale  extractions  are  

478  

compatible   with   wastewater   composition,   Sr   isotopes   do   not   discriminate   between  

479  

contributions  from  any  labile  brine  and  Sr  extracted  from  clays  by  water-­‐rock  reactions,  and  thus  

480  

do  not  offer  insights  into  the  mechanisms  or  timescale  of  extraction.      

481  

 

482  

7.2.4.  radium  isotopes  

483  

Radium   isotopes   distinguish   leachable   phases   in   the   shale.     The   226Ra/228Ra   ratios   of   water  

484  

leachates  are  low  (~250)  at  both  low  and  high  pressure,  demonstrating  that  labile  brine  cannot  

485  

generate   the   distinctive   226Ra/228Ra   signature   of   Marcellus   wastewaters   (Landis   et   al.   2018).    

486  

Additional   processes   must   be   invoked   to   explain   wastewater   226Ra/228Ra.     Moreover,   if   water-­‐

487  

extractable  salts  reported  here  and  elsewhere  (Stewart  et  al.  2015,  Balashov  et  al.  2015,  Renock  

488  

et   al.   2016)   are   attributed   to   pore   brine   within   the   shale   matrix   as   Balashov   et   al.   have   done,  

489  

corresponding   concentrations   of   Na   and   Ca   in   the   brine   would   exceed   1M   and   would   be  

490  

sufficient  to  extract  organic-­‐derived  Ra  with  high  226Ra/228Ra  composition,  if  this  solution  were  in  

491  

contact   with   226Ra-­‐bearing   phases.     This   does   not   appear   to   be   the   case.     The   high   226Ra/228Ra  

492  

ratios   and   high   [Ra]   that   distinguish   Marcellus   wastewaters   can   only   be   extracted   from   the  

493  

Marcellus   Shale   by   physical   modification   of   the   shale,   as   we   have   done   experimentally   with  

494  

additions  of  high  [Ca2+]  (Landis  et  al.  2018)  and  as  must  occur  during  hydraulic  fracturing.    

495  

 

496  

7.3.  A  mechanistic  model  reconciling  geochemical  and  geophysical  data  

 

Sr/86Sr   ratios   (0.710-­‐0.712).     The   remaining   refractory   fractions   are   highly   radiogenic   (0.73-­‐

16  

 

  497  

We  have  shown  that  the   226Ra/228Ra  ratios  of  Marcellus  wastewater  can  be  attributed  to  mixing  

498  

of   distinct   end-­‐members   within   the   shale,   regulated   by   water-­‐rock   reactions   and   proceeding  

499  

over   a   timescale   of   days,   consistent   with   the   duration   of   flowback   recovery.     The   evolution   in  

500  

226

501  

mineral   and   exchangeable,   organic   phases   within   the   shale   matrix.     The   source   of   Ca2+   in  

502  

wastewater   is   likely   a   mixture   of   contributions   from   salts,   carbonate   dissolution,   clay   surface  

503  

exchange,   and   endogenous   pore   brine.     Around   5%   of   total   Ca   in   the   shale   is   water-­‐soluble  

504  

(Renock  et  al.  2016,  Stewart  et  al.  2015)  and  consistent  with  a  labile  pore  brine  component,  with  

505  

another   10-­‐20%   exchangeable   from   clays   (Stewart   et   al.   2015).     Carbonate   dissolution   likely  

506  

constitutes   only   a   minor   source   of   Ca2+   to   wastewaters  -­‐-­‐   injected   fluids   with   pH   ~2   have   only  

507  

enough   free   acidity   to   dissolve