1: 

2: 

3: \documentclass[12pt,preprint]{emulateapj}

4: \usepackage{graphicx}

5: %% manuscript produces a one-column, double-spaced document:

6: 

7: %\documentclass[manuscript]{aastex}

8: 

9: %%preprint2 produces a double-column, single-spaced document:

10: %\documentclass[preprint2]{aastex}

11: %\documentclass{aastex}

12: %\usepackage{natbib}

13: %\bibliographystyle{apj}

14: %\usepackage{emulateapj5}

15: \usepackage{reaction}

16: %\documentclass{emulateapj}

17: %\usepackage{apjfonts}

18: 

19: %% If you want to create your own macros, you can do so

20: %% using \newcommand. Your macros should appear before

21: %% the \begin{document} command.

22: %%

23: %% If you are submitting to a journal that translates manuscripts

24: %% into SGML, you need to follow certain guidelines when preparing

25: %% your macros. See the AASTeX v5.x Author Guide

26: %% for information.

27: 

28: \newcommand{\vdag}{(v)^\dagger}

29: \newcommand{\myemail}{suvrath@astro.ufl.edu}

30: 

31: %% You can insert a short comment on the title page using the command below.

32: 

33: % \slugcomment{Not to appear in Nonlearned J., 45.}

34: 

35: %% If you wish, you may supply running head information, although

36: %% this information may be modified by the editorial offices.

37: %% The left head contains a list of authors,

38: %% usually a maximum of three (otherwise use et al.).  The right

39: %% head is a modified title of up to roughly 44 characters.  Running heads

40: %% will not print in the manuscript style.

41: 

42: \shorttitle{Monolithic Michelson for Precision Radial Velocity}

43: \shortauthors{Mahadevan et al.}

44: 

45: %%is the end of the preamble.  Indicate the beginning of the

46: %% paper itself with \begin{document}.

47: 

48: \begin{document}

49: 

50: %% LaTeX will automatically break titles if they run longer than

51: %% one line. However, you may use \\ to force a line break if

52: %% you desire.

53: 

54: \title{An Inexpensive Field-Widened Monolithic Michelson Interferometer for Precision Radial Velocity Measurements}

55: 

56: %% Use \author, \affil, and the \and command to format

57: %% author and affiliation information.

58: %% Note that \email has replaced the old \authoremail command

59: %% from AASTeX v4.0. You can use \email to mark an email address

60: %% anywhere in the paper, not just in the front matter.

61: %% As in the title, you can use \\ to force line breaks.

62: 

63: \author{Suvrath Mahadevan\altaffilmark{1}, Jian Ge, Scott W. Fleming, Xiaoke Wan, Curtis DeWitt, Julian C. van Eyken, \& Dan McDavitt}

64: \affil{Astronomy Department,  University of Florida, 211 Bryant

65: Space Science Center P.O. Box 112055 Gainesville, FL 32611-2055}

66: \email{suvrath@astro.ufl.edu}

67: 

68: 

69: %% Notice that each of these authors has alternate affiliations, which

70: %% are identified by the \altaffilmark after each name.  Specify alternate

71: %% affiliation information with \altaffiltext, with one command per each

72: %% affiliation.

73: 

74: \altaffiltext{1}{Visiting Astronomer, Kitt Peak National

75: Observatory, National Optical Astronomy Observatory. KPNO is

76: operated by AURA, Inc.\ under contract to the National Science

77: Foundation.}

78: %\altaffiltext{2}{Society of Fellows, Harvard University.}

79: %\altaffiltext{3}{present address: Center for Astrophysics,

80: %    60 Garden Street, Cambridge, MA 02138}

81: %\altaffiltext{4}{Visiting Programmer, Space Telescope Science Institute}

82: %\altaffiltext{5}{Patron, Alonso's Bar and Grill}

83: 

84: 

85: %% Mark off your abstract in the ``abstract'' environment. In the manuscript

86: %% style, abstract will output a Received/Accepted line after the

87: %% title and affiliation information. No date will appear since the author

88: %% does not have this information. The dates will be filled in by the

89: %% editorial office after submission.

90: 

91: 

92: \begin{abstract}

93: We have constructed a thermally compensated field-widened monolithic Michelson

94: interferometer that can be used with a medium-resolution

95: spectrograph to measure precise Doppler radial velocities of

96: stars. Our prototype monolithic fixed-delay interferometer is

97: constructed with off-the-shelf components and assembled using a

98: hydrolysis bonding technique. We installed and tested this

99: interferometer in the Exoplanet Tracker (ET) instrument at the

100: Kitt Peak 2.1m telescope, an instrument built to demonstrate the

101: principles of dispersed fixed delay interferometry. An iodine cell

102: allows the interferometer drift to be accurately calibrated,

103: relaxing the stability requirements on the interferometer itself.

104: When using our monolithic interferometer, the ET instrument has no

105: moving parts (except the iodine cell), greatly simplifying its

106: operation. We demonstrate differential radial velocity precision

107: of a few m s$^{-1}$ on well known radial velocity standards and

108: planet bearing stars when using this interferometer. Such

109: monolithic interferometers will make it possible to build

110: relatively inexpensive instruments that are easy to operate and

111: capable of precision radial velocity measurements. A larger

112: multi-object version of the Exoplanet Tracker will be used to

113: conduct a large scale survey for planetary systems as

114: part of the Sloan Digital Sky Survey III (SDSS III). Variants of the techniques and principles

115: discussed in this paper can be directly applied to build large

116: monolithic interferometers for such applications, enabling the

117: construction of instruments capable of efficiently observing many

118: stars simultaneously at high velocity-precision.

119: \end{abstract}

120: 

121: %% Keywords should appear after the \end{abstract} command. The uncommented

122: %% example has been keyed in ApJ style. See the instructions to authors

123: %% for the journal to which you are submitting your paper to determine

124: %% what keyword punctuation is appropriate.

125: 

126: \keywords{ techniques: radial velocities --- techniques:

127: spectroscopic ---  instrumentation: interferometers ---

128: instrumentation: spectrographs --- stars: kinematics --- methods:

129: data analysis}

130: 

131: %% From the front matter, we move on to the body of the paper.

132: %% In the first two sections, notice the use of the natbib \citep

133: %% and \citet commands to identify citations.  The citations are

134: %% tied to the reference list via symbolic KEYs. The KEY corresponds

135: %% to the KEY in the \bibitem in the reference list below. We have

136: %% chosen the first three characters of the first author's name plus

137: %% the last two numeral of the year of publication as our KEY for

138: %% each reference.

139: 

140: \section{INTRODUCTION}

141: Since the discoveries of the first extrasolar planets

142: \citep{Wolszczan92, Mayor95} more than 290 planets have been

143: discovered using radial velocity techniques, transit searches, and

144: microlensing. The majority of the known extrasolar planets have

145: been discovered using precision radial velocity measurements that

146: detect the gravitational influence of the planet on the parent

147: star. The presence of the planet induces a periodic change in the

148: line of sight velocity of the star and this velocity is measured

149: using the small Doppler shift in the spectral lines of the star.

150: Ongoing radial velocity surveys with high resolution echelle

151: spectrographs have achieved velocity precisions of $1-3$ m

152: s$^{-1}$ \citep{Butler96, Rupprecht04}, allowing the detection of $m\sin{i}=

153: 7-20$ Earth mass planets close to their host star \citep{Santos04,

154: Rivera05}. These echelle spectrographs are generally large and

155: expensive instruments which are not easily duplicated, leading to

156: the availability of such instruments becoming a limiting factor in

157: intensifying high-precision radial velocity surveys. High precision radial

158: velocity observations of more stars are necessary in

159: order to discover lower mass planets, explore different regimes of

160: stellar mass and evolution (Johnson et al. 2007), planet formation

161: around young stars (Setiawan et al. 2007), and to discover more

162: planets around low-mass M stars (e.g. Endl et al. 2007).

163: 

164: Over the past few years we have developed the dispersed fixed-delay

165: interferometer (DFDI) technique \citep*{Erskine00, Ge02a, Ge02b}

166: into viable instruments capable of achieving high precision radial velocity

167: measurements on stars and detecting extrasolar planets. Such

168: instruments are substantially less expensive than echelle

169: spectrographs since they do not require high spectral resolutions

170: or large-format gratings. At the heart of these instruments is a

171: stable, field-widened, fixed-delay Michelson interferometer. One

172: such instrument, the Exoplanet Tracker (ET) instrument at the Kitt

173: Peak 2.1m has been used to confirm known exoplanets (van Eyken et

174: al. 2004) and detect a hot Jupiter around HD102195 (Ge et al.

175: 2006). The ability to reliably construct monolithic Michelson

176: interferometers has the potential to make such instruments stable,

177: easy to use, and relatively easy to duplicate. This will enable

178: multiple versions of these instruments to easily be built to

179: complement existing echelle spectrographs in order to meet the

180: need for high precision radial velocity measurements to support upcoming space

181: missions like {\it KEPLER} (Borucki et al. 2003) and {\it SIM}

182: (Shao et al. 2007). An additional advantage of the DFDI technique

183: is the ability to observe a large number of stars simultaneously

184: using a wide field-of-view telescope. This technique will be

185: utilized to conduct MARVELS (Multi-object Apache point observatory Radial Velocity Exoplanet Large-area Survey), a multi-object survey for exoplanets,

186: as part of SDSS III (Ge et al. 2007). MARVELS will be able to

187: simultaneously monitor 120 F,G, and K stars in the 3 degree

188: field of view of the Sloan 2.5m telescope, efficiently searching

189: for Jupiter-mass planets \citep{Ge07}.

190: 

191: The Michelson interferometer is a crucial component of the ET and

192: MARVELS instruments and needs to be stable in order to reliably

193: recover relative velocity drifts of a few m s$^{-1}$. Phase drifts

194: of the interferometer during an exposure will tend to wash out the

195: fringes, reducing the measured fringe visibility and the velocity

196: precision. Large drifts between observing runs make it difficult

197: to achieve coherent velocity links over large timespans. The

198: current Michelson interferometer setup for the ET instrument at

199: Kitt Peak uses a piezoelectric transducer (PZT) to control the path difference. This stabilizes the

200: interferometer fringe and reduces the interferometer fringe phase

201: drifts. The need to run a closed-loop system on the most critical

202: component also makes observations difficult, since the observer

203: has to continually monitor the phase, and restart the PZT after a software or computer failure. In terms of a long duration survey for exoplanets, failure

204: of the PZT can lead to the inability to coherently link radial

205: velocities. The PZT control also becomes increasingly more difficult

206: for multi-object instruments based on ET that use much larger

207: beamsplitters due to the need to accommodate more fibers.

208: 

209:  To address some of these problems and to make ET-like instruments more

210: stable and substantially easier to use, we have been pursuing a

211: program to design and build an inexpensive monolithic

212: interferometer that satisfies our criteria of field widening and

213: stability (see Mahadevan et al. 2004). Monolithic interferometers

214: have also been discussed by Mosser et al. (2007) for use in

215: asteroseismology from DOME C in Antarctica. Polarizing Michelson

216: interferometers similar to our prototype have been in use for

217: solar oscillation measurements for a number of years in the Global Oscillation Network Group

218: (GONG, Harvey et al. 1995) and Solar and Heliospheric Observatory Michelson Doppler Imager (SOHO MDI)

219: instruments and are the design chosen for the Helioseismic and Magnetic Imager (HMI) instrument

220: (Graham et al. 2003) on the planned Solar Dynamics Observatory.

221: Monolithic interferometer assemblies have also been used to measure the solar resonance fluorescence of OH (Englert

222: et al. 2007). Polarizing interferometers are usually designed to

223: observe a single absorption line, while interferometers built for

224: spatial heterodyne spectroscopy are only effective over a small

225: wavelength region. The monolithic interferometer for the ET

226: instrument needs to be able to function in the wavelength range

227: 5000-6000 \AA, be non-polarizing, and have a slight tilt in one

228: mirror to created the fringes observed with ET.

229: 

230: In this paper we describe the construction of an inexpensive

231: prototype monolithic interferometer and present results

232: demonstrating the ability to acquire precise stellar radial

233: velocities using the monolithic interferometer with the Exoplanet

234: Tracker (ET) instrument.

235: 

236: \section{Fixed Delay Interferometers: History and

237: Background}

238:   By the term fixed-delay interferometer we refer to interferometers

239: that operate in a narrow range of delays (usually a few $\mu$m

240: change in path length) unlike the interferometers used in Fourier

241: transform spectrographs which need to be scanned over large path

242: lengths. Fixed-delay interferometers and interferometers coupled

243: to spectrographs have often been used to aid calibration and

244: enable precise wavelength and velocity measurements. Fabry-Perot interferometers have been

245: used with spectrographs to measure precise radial velocities of

246: stars (McMillan et al. 1994; Connes et al. 1996) by using them to

247: track and calibrate out instrument drifts. Fixed-delay

248: interferometers are also routinely used for high precision

249: velocity measurements of solar oscillations (Gorskii \& Lebedev

250: 1977; Beckers \& Brown 1978; Kozhevatov 1983; Harvey et al 1995

251: and Graham et al. 2003). Their use has also been proposed to

252: detect stellar oscillations and oscillations of Jupiter (Forrest

253: et al. 1978; Mosser et al. 1998; Mosser et al. 2003).

254: 

255:  For an

256: interferometer with an optical path difference $d$:

257: \begin{equation}

258: m \lambda=d

259: \label{eqn:delayfringe}

260: \end{equation}

261: where $m$ is the fringe order and $\lambda$ the monochromatic

262: wavelength of interest. A small change in the wavelength due to a

263: Doppler shift will also change the fringe order since the

264: interferometer delay ($d$) remains the same if the interferometer

265: is stable.

266: \begin{equation}

267: \Delta m=\frac{d}{\lambda^2}\Delta \lambda

268: \label{eqn:delayfringe1}

269: \end{equation}

270: Using the Doppler shift formula for non-relativistic velocities

271: \begin{equation}

272: \frac{\Delta \lambda}{\lambda} = \frac{\Delta v}{c}

273: \end{equation}

274: we can write Equation \ref{eqn:delayfringe1} as

275: \begin{equation} \Delta m = \frac{d}{\lambda} \frac{\Delta

276: v}{c} \end{equation} \begin{equation} \Delta v= \frac{c \lambda}{2

277: \pi  d} \Delta \phi \label{eqn:delayfringe2}

278: \end{equation}

279:  where $\phi=2 \pi m$ is the phase of the interference order.

280:   A change in velocity causes a Doppler shift in the wavelength of

281: the light. This Doppler shift can be measured from the phase shift

282: ($\Delta \phi$) if the interferometer delay ($d$) is known. The

283: phase and phase shift can be measured either by stepping the

284: interferometer through small changes in the optical delay, or by

285: putting a small tilt on one interferometer mirror so that the

286: phase can be measured in a single exposure using a two dimensional

287: image.

288: 

289: Figure \ref{fig:interferometerfringe} illustrates the phase

290: pattern for a monochromatic emission line. The phase pattern shown

291: is for a collimated beam passing through a field-widened Michelson

292: interferometer that has a slight tilt on one mirror. A change in

293: the wavelength of the incident light leads to a shift in phase of the

294: output fringe pattern from the interferometer. The velocity shift

295: can then be estimated using Equation \ref{eqn:delayfringe2} since

296: the shift in phase of the sinusoidal pattern can be easily

297: measured. This ability to measure phase shifts is also applicable

298: for a narrow wavelength band containing a single absorption line

299: (Harvey et al. 1995). Real stellar spectra, however, have

300: thousands of stellar lines and using the largest possible number

301: of lines will yield the best velocity precision. A large

302: wavelength range of light passing through the interferometer will

303: lead to a dramatic reduction of the recorded fringe visibility

304: since different wavelengths will have different phase shifts in

305: the output pattern. Almost no interference fringes will be visible

306: in the output phase pattern, making the estimation of phase shifts

307: and velocity very difficult. Using an interferometer alone one can

308: achieve good velocity precision on bright sources, but the use of

309: only a single absorption line, or a small wavelength region, makes it difficult to obtain the velocity accuracy required to detect planets around stars due to the limited number of photons collected.

310: 

311: 

312:       In 1997 David Erskine (then at Lawrence Livermore National Labs)

313: proposed using a field-widened Michelson interferometer in series

314: with a medium resolution spectrograph for the measurement of

315: precise radial velocities. The spectrograph is used to disperse

316: the light, allowing the measurement of fringe visibilities over a

317: significantly wider wavelength range. The first results from such

318: an instrument are found in Erskine \& Ge (2000) and Ge et al.

319: (2002), and the theoretical basis for the use of such an

320: instrument is outlined in Ge (2002), Erskine et al. (2003) and

321: Mosser et al. (2003). Figure \ref{fig:ETsimulspec} shows a

322: simulation of the spectral format expected from a dispersed fixed-delay

323: interferometer instrument. The interferometer fringes are clearly

324: visible due to dispersion by the medium-resolution

325: spectrograph. The normalized fringe pattern ($I$) in the slit

326: direction for each wavelength channel can be expressed in terms of velocity using

327: Equation \ref{eqn:delayfringe2}

328: \begin{equation}

329: I=1 + \gamma \sin {(2\pi d \frac{\Delta v}{c\lambda} +\phi_0})

330: \label{eqn:a}

331: \end{equation}

332: where $\phi_0$ is the phase of the first pixel and $\gamma$ is the

333: fringe visibility defined using the maximum ($I_{max}$) and

334: minimum ($I_{min}$) intensities of the sinusoidal fringe pattern

335: \begin{equation}

336: \gamma = \frac{I_{max} - I_{min}}{I_{max} + I_{min}} \label{eqn:b}

337: \end{equation}

338: 

339: A Doppler velocity shift of the absorption line manifests itself

340: as a phase shift of this sinusoidal pattern. The accuracy with

341: which the radial velocity can be measured from one fringe is

342: related to the photon noise error and the fringe visibility ($\gamma$) and

343: is discussed in \S 4.2.

344: 

345: \section{The Current Exoplanet Tracker Instrument}

346: 

347: The current Exoplanet Tracker (ET) instrument at Kitt Peak is the

348: end product of many years of experimentation with various design

349: ideas for high throughput DFDI instruments.

350: ET has been built to enable precision radial velocity measurements on

351: slowly rotating main sequence stars, and the exoplanet HD102195b

352: was discovered using radial velocities obtained with this

353: instrument (Ge et al. 2006). Aspects of the instrument design have

354: been mentioned in van Eyken et al (2003, 2004),

355: Ge et al. (2003, 2004, 2006), and Mahadevan et al. (2004, 2008). In this section we

356: briefly review the design aspects of the instrument that are

357: relevant from the point of view of installing and testing a

358: prototype monolithic interferometer. Figure \ref{fig:EToptical}

359: shows the optical layout of the KPNO ET instrument. Relevant

360: components are described below.

361: 

362: \subsection{Optical Fibers and Fiber Feed}

363:   The optical fibers and fiber feed used in ET are inherited from the Fiber Optic Echelle instrument (Ramsey \& Huenemoerder

364: 1986). The efficiency and properties of these optical fibers are

365: described in Barden (1998). Each fiber cable is enclosed in a

366: flexible metal covering and is 22 meters in length. The optical

367: fiber used in the ET instrument is a $200$ $\mu$m high $OH$ fiber.

368: In the f/8 beam of the telescope this fiber subtends an angle of

369: $\sim 2.5$ arc seconds on the sky. The use of the optical fiber

370: enables the ET instrument to be mechanically de-linked from the

371: telescope and in an isolated enclosure with very few moving parts.

372: This allows the instrument environment to be well controlled and

373: eliminates mechanical flexure due to movement, minimizing velocity

374: drifts that need to be calibrated out. The ET instrument can be

375: used with either the 2.1m telescope or the 0.9m Coud\'e

376: telescope simply by switching the 200$\mu$m optical fiber from

377: one telescope to another. The scrambling properties of the optical

378: fiber (Heacox 1986, Hunter \ Ramsey 1991) make the illumination

379: of the Michelson interferometer and spectrograph slit much more

380: uniform.

381: 

382: \subsection{Instrument Location and Enclosure}

383:   The ET instrument is currently installed in a small insulated room within the

384: coud\'e spectrograph room at the base of the 2.1m telescope dome.

385: The spectrograph room is very stable and its temperature only

386: responds slowly to the outside temperature. The instrument

387: itself is built on a vibration isolated optical table. A baseboard heater is installed at one end of the room

388: and a thermostat at the other end senses the ambient temperature

389: of the air and controls the heater. A secondary enclosure made

390: of insulating panels is built on the optical table to further

391: reduce the effect of temperature variations and to minimize the

392: turbulence. Figure \ref{fig:kpnotemperature} shows the temperature

393: of the ET instrument measured at four locations. The temperature

394: of the instrument within the inner instrument enclosure is more

395: stable than the temperature of the room itself.

396: 

397: 

398: \subsection{Iodine Cell}

399:    An iodine cell manufactured by Triad Technologies is used in the ET

400: instrument to provide a simultaneous calibration of the instrument

401: drift. A motor allows the iodine cell to be inserted or removed

402: from the collimated stellar beam path. The cell is operated at a

403: temperature of $60 \pm 0.1^\circ$C. The dense absorption

404: spectra of $I_2$ are superimposed on the stellar absorption lines.

405: Since the starlight passes through the iodine cell, the two

406: intermingled spectra experience exactly the same illumination in

407: the interferometer and spectrograph, and therefore track the same

408: instrument drifts.

409: 

410: \subsection{Input Optics and Michelson Interferometer}

411:   The ET input optics are designed to account for the focal ratio degradation of the 200$\mu$m optical

412: fiber and able to collect most of the photons exiting the  fiber. The output beam from the fiber is collimated to a

413: diameter of $\sim 6.6$ mm and passes through the iodine cell if

414: the cell is inserted into the beam path. A cylindrical lens then

415: collapses the beam in one direction (dispersion, or horizontal

416: direction) and the beam is focused onto the mirrors of the

417: Michelson interferometer. The Michelson interferometer used in ET

418: has a 4 mm BK7 glass etalon in one arm, an air-gap in the other

419: and is designed to have an optical delay of $\sim 7$ mm. The

420: lengths of the two arms are also chosen so that the interferometer

421: optical delay is relatively insensitive to the angle of the

422: converging beam from the cylindrical lens. This `field widening'

423: of the interferometer is described in more detail in \S 4.3. A

424: small tilt in one beam of the interferometer allows the creation

425: of $3-4$ horizontal fringes over the 6.6 mm stellar beam. A slight

426: tilt of the interferometer beam cube allows both the transmitted

427: and reflected beams to be imaged onto the spectrograph slit using

428: relay mirrors.

429: 

430: \subsection{Piezoelectric Transducer for Interferometer Stabilization}

431:  We use a piezoelectric transducer (PZT) to keep the interferometer from

432: drifting during an exposure (drifts lead to decreased fringe visbility) and keep the delay stable. Piezoelectric materials can

433: generate mechanical stress in response to applied electric

434: voltage, enabling the interferometer delay to be maintained even

435: in the presence of vibration and temperature drifts. The PZT is

436: attached to one of the interferometer mirrors and is controlled

437: using a LabView program. A single-mode fiber fed with a stabilized

438: He-Ne laser (632.8 nm) is used to illuminate a small portion of

439: the beam splitter. The horizontal fringes formed are recorded by a

440: video camera and are used as the input to the LabView program. The

441: position of these fringes is then locked by applying the

442: appropriate voltages to the PZT. This method can effectively

443: stabilize the interferometer fringe pattern over the period of an

444: observing run ($7-12$ days). To obtain flats and calibration lamp

445: spectra the PZT voltage is ramped very quickly, leading to the PZT

446: mirror vibrating and washing out the interference fringes. The PZT

447: is turned off after every observing run, re-initialized and dialed

448: back to its previous position at the start of a new observing run.

449: 

450: \subsection{Spectrograph}

451:  The ET spectrograph is a transmissive design with an f/5 collimator, a

452: volume phase holographic transmission grating and a $\sim f/2$

453: camera. The spectrograph optics were designed by Dequing Ren. The

454: stellar beam from the interferometer is imaged on the spectrograph

455: slit and collimated. Unlike normal spectrographs, where the

456: starlight is brought to a focus at the slit, the ET slit is not at

457: the point where the beam is smallest but where its fringe

458: visibility is maximal. The pre-slit optics form an image of one of

459: the interferometer mirrors onto the slit, and change the slow beam

460: to an f/5 beam. The spectrograph then disperses the light, forming

461: an image of the slit (and the interference fringes) on the

462: detector.

463: 

464: \subsection{Spectral Format \& Efficiency}

465:  The ET instrument operates in the grating first order, but its spectral

466: format is two dimensional to allow the measurement of the fringe

467: phase at each wavelength. The $250$ $\mu$m slit is sampled by 6.7

468: pixels on the detector, yielding a spectral resolution of $\sim

469: 6000$. The fringes in the slit direction are sampled by $\sim 58$

470: pixels. Figure \ref{fig:ET_format} shows the spectral format of

471: the instrument. The top and middle spectra are the two outputs

472: from the interferometer. The measured instrument efficiency from

473: the fiber output to the detector plane is $\sim 49 \%$. This

474: efficiency includes losses from the lenses, mirrors,

475: interferometer, slit and the grating, but not the iodine cell or

476: the CCD. The CCD is expected to have a quantum efficiency of $\sim

477: 90\%$ in our wavelength range. Using the measured fiber and

478: telescope transmission in average seeing, and assuming their

479: efficiency to be 35-40$\%$ in average seeing the total instrument

480: efficiency is calculated to be $\sim$ 15.5 -- 17.5 \%. The

481: instrument efficiency in good seeing condition, measured using a

482: flux calibration star, is $\sim 18\%$ and in good agreement with

483: the expected values.

484: 

485: \section{A Monolithic Michelson Interferometer for ET}

486:  A monolithic interferometer for an instrument like ET has a

487:  number of requirements in terms of delay, field widening and

488:  stability. In this section we explore these parameters and the

489:  materials that can be used in the construction of a monolithic

490:  interferometer.

491: \subsection{The Choice of Optical Delay}

492: The spectral format of ET allows one to measure the phase shift of the

493: interference pattern and convert that to velocity. As outlined in

494: Butler et al. (1996) and \citet{Ge02a}, the velocity

495: uncertainty from one pixel of the normalized fringe is:

496: 

497: \begin{equation}

498: \sigma_{RMS}= <\frac{\epsilon_{I}}{dI/dv}>

499: \end{equation}

500: 

501: The uncertainty in the measured intensity, $\epsilon_{I}$, is just

502: the inverse of the signal to noise ratio (S/N$^{-1}$) if the S/N

503: is large enough to ignore the dark current and read noise from the

504: detector, a condition that always needs to be satisfied to achieve

505: high-precision radial velocities with ET. For the sinusoidal interferometer

506: fringes observed with ET:

507: \begin{equation}

508: <\frac{dI}{dv}>= \frac{\sqrt{2}\pi d \gamma}{c \lambda}

509: \end{equation}

510: The radial velocity error from measurements of one fringe is then

511: given by

512: \begin{equation}

513: \sigma_{RMS}=\frac{c\lambda}{\sqrt{2}\pi d \gamma

514: \sqrt{N_{photons}}}

515: \end{equation}

516: 

517: where $N_{photons}$ is the total number of photons recorded for

518: all the pixels in the interferometer fringe. The functional form

519: of this equation means that the highest velocity precision for a

520: given number of photons is obtained when the value of $d \gamma$

521: is maximized. The value of the interferometer delay that yields

522: the largest $d \times \gamma$ for a given wavelength region is, however,

523: a function of the spectral type and the rotational velocity of the

524: star. Since the interferometer effectively heterodynes information

525: to lower spatial frequencies (Erskine et al. 2003), a fast rotator

526: has less high frequency information and requires a smaller delay.

527: 

528: Figure \ref{fig:rotators} shows the $d \gamma$ at different

529: optical delays ($d$) for a K5V star at various rotational velocities.

530: These results are derived from simulations of the ET instrument

531: spectral format using stellar models as an input. The chosen delay

532: for the ET instrument is $\sim 7$ mm since this value provides

533: adequate fringe visibility ($\gamma$) for late F, G and K stars

534: that have $0< v \sin{i} < 12$ km s$^{-1}$. Fast rotators (e.g. the

535: planet bearing star Tau Boo) have very low fringe visibility at

536: our chosen interferometer delay, making precise radial velocities

537: difficult to extract. The optical delay of the interferometer does

538: not need to be exact for our application, It merely needs to be

539: stable. More detailed discussions regarding

540: the calculation of optimal delay for different spectral parameters

541: can be found in Mosser et al. (2003).

542: 

543: \subsection{The Field Compensation Principle}

544: 

545:   The optical delay of a Michelson interferometer is strongly

546: dependant on the incident angle of the light. Deliberately

547: designing an interferometer to decrease this dependence of the

548: optical delay on the incident angle is known as field widening, or

549: field compensation, and is described in more detail in Hillard \&

550: Sheperd (1966), Shepherd et al. (1985), and Thullier \& Sheperd

551: (1985). A Michelson interferometer for ET must have arms of

552: different lengths to create a total optical delay of $\sim 7$ mm,

553: i.e., $\sim 14000$ waves of delay at a wavelength of 5000 \AA.

554: This delay is appropriate for observations of stars with $v

555: \sin{i} < 12$ km s$^{-1}$. For an interferometer with arm lengths

556: $L_1$ and $L_2$, with refractive indices $n_1$ and $n_2$

557: respectively, the interferometer optical delay for an on-axis ray

558: ($d_0$) is given by

559: 

560: \begin{equation}

561:  d_0=2(n_{1} L_{1} - n_{2}L_{2})

562: \label{eqn:delay}

563:  \end{equation}

564: 

565: For a ray entering the Michelson interferometer at an angle $i$

566: the interferometer delay can be written as (Shepherd et al. 1985)

567: 

568: \begin{equation}

569: d = 2[n_{1}L_{1}(1-\frac{\sin^{2}i}{n_{1}^{2}})^{-\frac{1}{2}} -

570: n_{2}L_{2}(1-\frac{\sin^{2}i}{n_{2}^{2}})^{-\frac{1}{2}} ]

571: \end{equation}

572: 

573: Expanding all terms of $\sin{i}$ upto fourth order and ignoring

574: higher order terms we get

575: \begin{equation}

576: d = 2[(n_{1}L_{1} - n_{2}L_{2}) - \frac{\sin^{2} i}{2}

577: (\frac{L_{1}}{n_{1}} - \frac{L_{2}}{n_{2}}) - \frac{\sin^4 i}{8}

578: (\frac{L_1}{{n_1}^3} -\frac{L_2}{{n_2}^3})] \label{eqn:c}

579: \end{equation}

580: 

581: If the same material is used in both arms of the interferometer

582: ($n_1=n_2$) the interferometer delay can be quite sensitive to the

583: angle of the incident light. Such an interferometer is often

584: referred to as an uncompensated interferometer. For example, an

585: interferometer with air gaps in both arms and a 7 mm delay (at

586: 5000 \AA) has 19.1 waves of phase change over a 3 degree

587: half-angle and 4.5 waves over a

588: 1.5 degree half angle. Such large phase changes are unacceptable with the ET

589: instrument since the fringe visibilities will be substantially

590: reduced when the fringes are imaged on the slit.

591: 

592: The $\sin^2{i}$ term in Equation \ref{eqn:c} for the delay

593: disappears when

594: 

595: \begin{equation}

596: \frac{L_1}{n_1}=\frac{L_2}{n_2} \label{eqn:compensation}

597: \end{equation}

598: 

599: The delay for a compensated Michelson interferometer (ie. one that satisfies Equation \ref{eqn:compensation}) is then given by:

600: 

601:  \begin{equation}

602: d= d_0[1 +\frac{\sin^4 i}{8{n_1}^2{n_2}^2}]

603: \end{equation}

604: 

605: where $d_0$ is the interferometer delay when $i= 0$, ie. when the

606: ray is along the axis of the interferometer. For small values of

607: $i$ the delay is almost independent of the incident angle. Further

608: increase in field widening (so called super field widening) can be

609: achieved by using two or more glasses with an air space to also

610: reduce the effect of the $\sin^4{i}$ term (eg. Gault, Johnston \&

611: Kendall 1985), but we have not attempted this for our prototype monolithic interferometer.

612: 

613: Any field compensated Michelson interferometer at a specific delay must therefore simultaneously satisfy Equations

614: \ref{eqn:delay} and \ref{eqn:compensation}. The complete

615: disappearance of the $\sin^2{i}$ term in Equation

616: \ref{eqn:c} can only be achieved for a specific design

617: wavelength (5000 \AA\ in this case). The simplest choice for the ET interferometer was to use the same material as the beamspliiter (BK7) in one arm and an air-gap  in the other. To find the values of $L_1$

618: and $L_2$ that satisfy our constraints we calculate the refractive

619: index of BK7 at the design wavelength using the Sellmeier dispersion formula. The

620: dispersion constants used in this equation may vary slightly

621: depending on the glass melt and the values we use are for typical

622: melts of N-BK7. The values of the arm lengths that satisfy the

623: field widening condition are $L_1$(BK7) = 4.04 mm and $L_2$(air) =

624: 2.66 mm. The wavelength dependence of the refractive index and

625: errors in the dimensions of the arms will, in reality, lead to a

626: situation where the compensation condition is not met exactly. To

627: evaluate the performance in such a case we define a term

628: $\epsilon$ which is a wavelength dependant measure of the

629: departure from the field widening condition.

630: 

631: \begin{equation}

632: \epsilon=\frac{L_1}{n_1}-\frac{L_2}{n_2}

633: \end{equation}

634: 

635: Even when $\epsilon=0$ at 5000 \AA\ the variation of refractive

636: index with wavelength results in a non-zero value of $\epsilon$

637: creating a small dependence of the delay on the $\sin^2{i}$ term

638: for wavelengths other than 5000 \AA. The phase difference caused by

639: this effect at the calculated values of $L_1$ and $L_2$ is

640: illustrated in Figure \ref{fig:phaseshift} for a beam with a

641: half-angle of 1.5 degrees. Although the ET instrument only works

642: in the 5000-5600 \AA\ region, such an interferometer can, in

643: principle, be used over wider wavelength regions, and we have

644: calculated the phase difference for 4000-6000 \AA\ which is the

645: spectral range containing most of the radial velocity information

646: for F, G, K stars \citep*{Bouchy01}. For a hypothetical case in

647: which the compensation condition is off by $\epsilon=100$ $\mu m$

648: even at 5000 \AA\ , a phase shift of 0.137 waves occurs for a beam

649: with a half angle of 1.5 degrees. This will lead to a slight

650: decrease in the observed visibility, but is otherwise acceptable

651: as long as the focal ratio or the illumination pattern of the beam

652: does not change significantly. If an iodine cell is used in the

653: beam path, then small changes in the illumination do not matter

654: since the iodine and starlight are both affected in the same way.

655: The scrambling properties of the optical fibers used in ET also

656: ensure that the illumination of the interferometer is stable even

657: in bad seeing conditions.

658: 

659: \subsection{Reducing Sensitivity to Temperature Drifts}

660:      The need for a particular value of the optical delay and field

661: compensation fixes the values of the arm lengths $L_1$ and $L_2$

662: since we have decided to use BK7 glass in one arm and an air-space

663: in the other. However, since our goal is to build a stand alone

664: monolithic interferometer, whose arms are isolated from any

665: external mechanical mounts, the choice of spacer material for the

666: air arm still needs to be made. The delay for a beam going through

667: a compensated Michelson interferometer with a very small angle of

668: inclination is shown in Equation \ref{eqn:delay}. Following

669: \citet*{Title80}, variation of the interferometer phase with

670: temperature can then be written as

671: 

672: \begin{equation}

673: \frac{d (d_0)}{d T}=2(n_1 \frac{d L_1}{dT} + L_1\frac{dn_1}{dT}-n_2 \frac{d L_2}{dT} -L_2\frac{d n_2}{dT})

674: \end{equation}

675: 

676: Using the field widening condition (Equation

677: \ref{eqn:compensation}) the right hand side of this equation can be recast in the form

678: 

679: 

680: \begin{equation}

681: 2\frac{L_1}{n_1}({n_1}^2(\frac{1}{n_1}\frac{dn_1}{dT}+\frac{1}{L_1}\frac{dL_1}{dT})-{n_2}^2(\frac{1}{n_2}\frac{dn_2}{dT}+\frac{1}{L_2}\frac{dL_2}{dT}))

682: \label{eqn:athermal}

683: \end{equation}

684: 

685: which disappears when

686: 

687: 

688: \begin{equation}

689: {n_1}^2(\frac{1}{n_1}\frac{dn_1}{dT}+\frac{1}{L_1}\frac{dL_1}{dT})={n_2}^2(\frac{1}{n_2}\frac{dn_2}{dT}+\frac{1}{L_2}\frac{dL_2}{dT})

690: \end{equation}

691: 

692:     This thermal compensation condition depends only on the refractive

693: indices, their variation with temperature, and the linear thermal

694: expansion coefficients of the materials used, and is independent

695: of the actual arm lengths or path differences. In our case the

696: left-hand-side of the equation is already predetermined by the

697: choice of BK7 as the material in that arm. A wise choice of spacer

698: material for the other arm can then be used to athermalize the

699: interferometer. To evaluate the variation of refractive index with

700: temperature of BK7 we use the equation

701: 

702: \begin{equation}

703: \frac{dn_{abs}}{dT}= \frac{n^2(\lambda, T_0)-1}{2n(\lambda, T_0)}

704: (D_0+2D_1\Delta T+3D_2{\Delta T}^2 +\frac{E_0 +2E_1\Delta

705: T}{\lambda^2-\lambda_{TK^2}})

706: \end{equation}

707: 

708: %\end{equation}

709: 

710: where $\lambda$ is the wavelength of interest, $\Delta T$ is the

711: temperature difference from the reference temperature $T_0$ ($20^\circ$ C) and $n$ is the refractive index for that wavelength

712: calculated at the reference temperature using the Sellmeier

713: dispersion formula. Using this equation, the variation of

714: refractive index of BK7 with temperature can be calculated with

715: constants ($D_0-D_2, E_0, E_1$) provided in the Schott optical glass catalog. Using

716: Equation \ref{eqn:athermal} and assuming vacuum operation, the

717: change in delay and the velocity drift can be estimated for

718: various materials. Using Equation \ref{eqn:delayfringe2}, the

719: shift of one fringe ($2\pi$ in phase) corresponds to a velocity

720: shift of

721: \begin{equation}

722: \Delta v = \frac{c \lambda}{d_0}

723: \end{equation}

724: 

725: For our value of $d_0 \sim 7$ mm, one fringe shift corresponds to a

726: velocity shift of $\sim 21.5$ km s$^{-1}$. Table

727: \ref{tbl:velshift} shows the velocity shift for a one degree

728: change in temperature using BK7, copper, CaF$_2$, or aluminium

729: as a spacer material for an interferometer operating in vacuum.

730: CaF$_2$ proves to be the best spacer but is difficult to work with

731: since it is fragile. For assembling our prototype interferometer

732: we chose copper as a spacer material since it provides a

733: significant improvement from a BK7 spacer, and is easy to work

734: with. We have chosen to design the interferometer for vacuum

735: operation since this significantly increases the stability by

736: removing the effect of the refractive index of air changing due to

737: pressure and humidity variations (J. Sudol, private

738: communication).\footnote{http://gong.nso.edu/instrument/instrument\_performance/vres/vres.html}

739: While designed for operation in vacuum, this interferometer can

740: also be operated in air, although the effect of a one degree

741: temperature increase will rise from $\sim 250-500$ m s$^{-1}$ to

742: $\sim1000$ m s$^{-1}$. This still provides a significant

743: improvement from using a BK7 spacer in air or vacuum. The ET

744: instrument is temperature stabilized and the temperature of the room at the location

745: of the interferometer rarely changes beyond $0.3^\circ$ C and is often stable to less than $0.2^\circ$ C over

746: the course of an observing run (see Figure

747: \ref{fig:kpnotemperature}). An interferometer built with a copper

748: spacer would therefore be expected to be stable at a level of

749: $\sim$200 m s$^{-1}$ in air over an observing run if the observed

750: velocity drift is purely due to the interferometer delay changing

751: as a function of temperature.

752: 

753: \section{Assembling a Monolithic Fixed-Delay Interferometer}

754: 

755: \subsection{Components}

756:  We used commercially available optical components

757: to assemble the monolithic interferometer prototype since the goal

758: was to determine if a relatively inexpensive interferometer can be

759: used in a DFDI instrument to

760: derive precise radial velocities. The components used in the

761: assembly were a Newport 1 inch broad-band non-polarizing beam

762: splitter, an uncoated BK7 window of diameter

763: 1 inch and thickness 4 mm from Edmund Optics to

764: use as the optical delay in one arm, and 1 inch mirrors from

765: Thorlabs. The Newport beam splitter cube had a proprietary

766: broad-band anti-reflection coating on it made of alternate layers

767: of TiO and SiO. %, with the outermost layer being SiO

768: A copper ring of thickness 2.66 mm was also machined for use as a

769: spacer material in the air arm. Small holes were drilled around

770: the surface of the copper ring to enable the evacuation of air so

771: that the final assembled interferometer could also be operated or

772: tested in a vacuum vessel. Figure \ref{fig:exploded} shows each component in an exploded view of the final

773: interferometer assembly.

774: 

775: \subsection{Bonding Technique}

776:  To assemble these components into a single interferometer we used

777:  a hydroxide catalyzed bonding procedure developed by Jason Gwo at Stanford University to meet the high stability requirements of the

778: Gravity Probe B mission (Gwo 1998a,b). Such optical bonding

779: techniques are also currently being explored for use in the LISA

780: space mission. This bonding technique can link two surfaces by

781: forming a silicate-like network through hydroxide catalyzed

782: hydration and dehydration. The bond that is formed is transparent

783: to visible wavelengths, and expected to be very stable. The

784: hydroxide catalyzed bonding can work at room temperature, and

785: interface thickness are expected to be less than $200 \mu$m. The

786: technique can be applied to any  surface capable of forming a

787: silicate-like network. For such a surface (eg. fused silica, BK7,

788: alumina) the equations for the hydration and dehydration in the

789: presence of a hydroxide catalyst is \citep{Gwoa, Gwob}

790: \begin{equation}

791: Si-O-Si + H_20 \reactionrevarrow{KOH}{} Si-OH + Si-OH

792: \end{equation}

793: 

794: The bonding process is illustrated for fused silica in Figure

795: \ref{fig:optobond}. The presence of the hydroxide ion catalyzes

796: the formation of hydrated fused silica, which is then able to form

797: a bond, linking the two disks together. As the two disks are

798: allowed  to cure at room temperature the loss of water from the

799: interface leads to a stronger bond. The two components can be

800: manipulated during the bond cure time to achieve the desired

801: optical alignment. This bonding technique is applicable to

802: materials that can both form silicate-like networks, and also if

803: one of the materials can generate a silicate-like network and the

804: other is capable of forming a layer of hydroxyl groups (-OH). This

805: allows a number of metals (aluminium, brass, copper etc.) and

806: metal oxides to be bonded to glasses capable of forming

807: silicate-like networks (eg. BK7, fused silica).

808: 

809: \subsection{Assembly}

810: The optical components were cleaned by immersing them in acetone

811: and agitating in an ultrasonic bath for five minutes. To eliminate

812: any residue after the acetone cleaning the components were also

813: cleaned by immersing them in iso-propanol and de-ionized water

814: (DI) and agitated again in the ultrasonic bath. The copper

815: ring was immersed in acetone for 24 hours to remove grease from

816: the machining process and then cleaned using the same procedure as

817: that followed for the optical components. The entire assembly of

818: the interferometer was performed in a clean room. To assemble the

819: BK7 arm of the interferometer we mixed a solution of KOH of 1:500

820: molar ratio. To prevent contamination by particulate matter the

821: KOH solution was applied to the optical surfaces of the mirror

822: using a syringe with a 0.2 $\mu$m particulate filter attached to

823: its tip. The 4 mm BK7 glass window was then placed on the mirror,

824: and more KOH applied to the exposed surface. Finally the

825: beamsplitter was placed on the assembly. One of the advantages of

826: this bonding procedure is that the bonding is not instantaneous,

827: and small alignments are possible. The three pieces were then

828: aligned together, with the weight of the beamsplitter keeping the

829: three components in contact till they were bonded. An optically

830: transparent bond was achieved in $48-72$ hours and the pieces could

831: not be separated by hand.

832: 

833: Bonding the copper ring to the mirror to create the air arm of the

834: interferometer was more challenging. One side of the

835: copper ring was polished with various grades of polishing paper

836: until interference fringes of the required density and tilt were

837: observed when the copper ring and mirror were placed on the

838: beamsplitter assembly and observed with a 543 nm He-Ne laser. Copper cannot form a silicate-like

839: network, and the machined copper ring was not optically flat. To

840: bond the copper to the mirror we used a sodium silicate solution.

841: The silicate in this solution allows for easier bonding and can

842: fill in small gaps between the pieces. Optical transparency of the

843: bond was not an issue here, since the copper ring is only a spacer

844: for the air arm. A small amount of sodium silicate solution was

845: placed on the copper ring using the tip of a syringe and the

846: mirror was gently lowered onto the ring. Unlike KOH, the sodium

847: silicate solution left substantial residue, and we were careful

848: not to let it get on the exposed mirror surface. The copper spacer, now

849: bonded to the mirror, was then tested again with the beamsplitter

850: to ensure that the fringes were present. A small amount of further

851: polishing and alignment of the copper spacer was required to ensure that the correct

852: fringe pattern was still observed.

853: 

854: We were apprehensive about

855: using sodium silicate solution to bond the copper ring to the

856: beamsplitter, and so we used UV curing optical glue (NOA 73). The

857: glue was applied to the copper ring, and a UV lamp was used to

858: flash cure the glue when the correct optical alignment was

859: achieved. The entire assembly was then allowed to settle for 48

860: hours before further testing.

861:   Three prototype interferometers were assembled using the procedures

862: described above. One of the interferometers was installed in the

863: Kitt Peak ET instrument, and its performance is described in more

864: detail in the following section. Another interferometer was placed

865: in a custom designed vacuum enclosure on a vibration isolated

866: table, and its fringe pattern was monitored for more than 50 hours

867: using a stabilized He-Ne laser (632.8 nm). Since the optical delay

868: of the interferometer is known, the phase shift of the

869: interferometer pattern can be converted to a measured velocity

870: shift using Equation \ref{eqn:delayfringe2}. The results of this

871: test are shown in Figure \ref{fig:labtest} and demonstrate that

872: our prototype monolithic interferometer is capable of achieving a

873: velocity stability of $\sim 500$ m s$^{-1}$ over a few days. The

874: interferometer vacuum enclosure itself was not temperature

875: controlled and we suspect that the quasi-periodic variations in

876: velocity are caused by diurnal variations in temperature. Such

877: variations are quite acceptable, since the fringe shifts due to instrument drifts will be very small over a typical stellar exposure using ET (15-20 minutes), and will not impact the observed visibilities.

878: 

879: \section{On-Sky Results with the Monolithic Interferometer}

880: One of the monolithic interferometers was transported to Kitt Peak

881: National Observatory and installed in the ET instrument in May

882: 2005. The monolithic interferometer replaced the existing actively-controlled interferometer (with its PZT mirrors and mounts) for

883: the duration of this observing run. Figure \ref{fig:passivesetup}

884: shows the passive interferometer set-up with the ET instrument.

885: The input optics, iodine cell and the slit assembly are also

886: clearly visible in the figure.

887: 

888: \subsection{Observations}

889:  Once the monolithic interferometer was installed and aligned with ET we were

890: able to observe a number of stars using the 0.9m Coud\'e

891: telescope. During the observing run (29th May 2005 through 6th

892: June 2005) we obtained multiple data points on stars known to be

893: stable over the short term, such as $\eta$ Cas (V=3.45 mag) and 36 UMa

894: (V=4.83) as well as stars with known planetary companions (55 Cnc,

895: $\rho$ Crb). The standard observing procedure with this instrument

896: is described in Ge et al. (2006) and Mahadevan et al. (2008), and

897: we mention it only briefly here. Individual star and iodine

898: templates were obtained for each star, and each star was then

899: observed with the iodine cell inserted in the path of the stellar

900: beam. The iodine spectrum acts as a fiducial for determining and

901: subtracting off the instrument drift. A number of additional spectra of the

902: iodine cell (illuminated by a quartz lamp) were also observed over

903: the duration of the run. This iodine cell data allows one to

904: characterize the instrument drift during the run. To verify the

905: performance of the passive interferometer a single mode fiber, fed

906: with a stabilized He-Ne laser (632.8 nm), was used to illuminate a

907: small portion of the passive interferometer and this phase pattern

908: was continuously recorded by a video camera. This phase pattern

909: allowed us to continuously monitor the total drift of the passive

910: interferometer over the entire observing run.

911: 

912: \subsection{Data Analysis}

913: The data acquired with the monolithic interferometer could not be

914: processed automatically with the standard ET data pipeline because

915: the standard calibration sequence of non-fringing tungsten lamp flats and

916: non-fringing Thorium-Argon (Th-Ar) spectra were not acquired.

917: These are usually acquired with the ET instrument by

918: jittering the PZT very fast to wash out the interferometer

919: fringes. Since the monolithic interferometer was deliberately

920: designed to have no moving parts such flats cannot be obtained. The tungsten

921: lamp flats are used to correct pixel-to-pixel variations and

922: partially correct for illumination variations, and the

923: non-fringing Th-Ar emission lamp spectra are used to determine the

924: slant of the spectra in the slit direction. Not correcting the

925: pixel-to-pixel variations may lead to excess radial velocity

926: noise, and correcting the slant is essential to determine

927: accurate velocity shifts.

928: 

929: We were unable to reliably measure the

930: slant from fringing Th-Ar spectra and we decided to use the slant

931: correction parameters from the preceding observing run. The

932: observed slant is primarily caused by distortions introduced by

933: the spectrograph optics, and by the positioning of the slit. The

934: use of the slant solution from a previous run (May 27 2005) is

935: justified since we were careful to not move the spectrograph, and

936: to maintain the same slit optics and beam paths while replacing

937: the actively controlled interferometer with the monolithic

938: interferometer. Using the slant solutions from the previous run

939: enabled radial velocities to be extracted using the standard ET

940: data processing pipeline.

941: 

942: The pipeline processing steps are described in more detail in van

943: Eyken et al. (2004), Ge et al. (2006), and Mahadevan et al. (2008).

944: The data produced by ET were processed using standard IRAF

945: procedures, as well as software written in Research System Inc.'s

946: IDL software. The images were corrected for biases, dark current,

947: and scattered light and then trimmed, illumination corrected,

948: slant corrected and low-pass filtered. The visibilities ($V$, same

949: as $\gamma$) and the phases ($\theta$) of the fringes were

950: determined for each channel by fitting a sine wave to each column

951: of pixels in the slit direction (see Figures 2, 3). To determine

952: differential velocity shifts the star+iodine data can be

953: considered as a summation of the complex visibilities (${\bf

954: V}=Ve^{i\theta}$) of the relevant star ($V_Se^{i\theta_{S_0}}$)

955: and iodine ($V_Ie^{i\theta_{I_0}}$) templates (van Eyken et al.

956: 2004; Erskine 2003, van Eyken et al. 2007). For small velocity

957: shifts the complex visibility of the data, for each wavelength channel, can be written as

958: 

959: \begin{equation}

960: V_D e^{i\theta_D} = V_S

961: e^{i\theta_{S_0}}e^{i\theta_S-i\theta_{S_0}}

962: +V_Ie^{i\theta_{I_0}}e^{i\theta_I-i\theta_{I_0}}

963: \end{equation}

964: 

965: where $V_D, V_S,$ and $V_I$ are the fringe visibilities for a

966: given wavelength in the star+iodine data, star template and iodine

967: template respectively, and $\theta_D, \theta_{S_0}$, and

968: $\theta_{I_0}$ the corresponding measured phases.

969:  In the presence of real velocity shifts of the star and instrument

970:  drift, the complex visibilities of the star and iodine template best match the

971:  data with phase shifts of $\theta_S - \theta_{S_0}$ and $\theta_I -

972:  \theta_{I_0}$ respectively. The iodine is a stable reference and the iodine phase shift tracks the instrument drift. The

973: difference between star and iodine shifts is the real phase shift

974: of the star, $\Delta \phi$, corrected for any instrumental drifts

975: 

976: \begin{equation}

977: \Delta \phi = (\theta_S - \theta_{S_0}) - (\theta_I -

978: \theta_{I_0})

979: \end{equation}

980: 

981: This phase shift can be converted to a velocity shift, $\Delta v$,

982: using Equation \ref{eqn:delayfringe2}. The ET data analysis

983: pipeline finds the shift in phase of the star and iodine templates

984: that are a best match for the data, and uses these phase shifts to

985: calculate the velocity shift of the star relative to the stellar

986: template.

987: 

988: % by a known phase-to-velocity

989: %scaling factor ($\frac{c \lambda}{2 \pi  d}$) that is a function

990: %of the optical delay in the Michelson interferometer ($d$), the

991: %wavelength ($\lambda$) and the speed of light ($c$). These

992: %parameters are connected by the relation

993: %\begin{equation} \Delta v= \frac{c \lambda}{2 \pi  d}

994: %\Delta \phi \label{eqn:delayfringe2}

995: %\end{equation}

996: 

997: \subsection{Interferometer Drift \& Environmental Stability}

998: Our experimental setup allowed us to measure and assess the

999: performance of the interferometer and its stability in two

1000: different ways:

1001: 

1002: \begin{itemize}

1003: \item{By monitoring the phase drift of a fringe pattern generated

1004: using a using a stabilized He-Ne laser (632.8 nm). This fringe

1005: pattern is generated by shining the laser light into the

1006: interferometer via a single-mode optical fiber on a separate path. This fringe

1007: pattern is continuously recorded by a video camera, and the phase

1008: variation can be converted into velocity shifts.}

1009: \item{By tracking the velocity drifts of the iodine templates

1010: obtained throughout the observing run. The iodine in the

1011: temperature-stabilized cell is very stable by design, and any measured

1012: velocity shifts have to be a result of drifts in the

1013: interferometer or spectrograph. }

1014: 

1015: \end{itemize}

1016: 

1017: The results of these measurement techniques are expected to be

1018: highly correlated, but not identical since they monitor different

1019: parts of the instrument, use different wavelength regions and

1020: different data analysis techniques are used. The interferometer

1021: drift measured from the phase patterns recorded by the video

1022: camera is shown in Figure \ref{fig:kpnodrift} as a solid line, and

1023: the measured iodine velocity drift is plotted as filled circles.

1024: These two velocities are forced to the same zero point at the

1025: beginning of the observing run. The dotted line shows the

1026: temperature measured at the location of the interferometer. The

1027: passive interferometer drifted by $\sim 7000$ m s$^{-1}$ during

1028: the observing run. The sharp dip in the temperature and the

1029: velocities seen towards the end of the observing run was caused

1030: when the oxygen alarm was triggered for the coud\'e spectrograph

1031: room. Safety considerations required the door of the coud\'e

1032: spectrograph room to be opened to the cold night air to ensure the

1033: safety of the observer. Although the ET instrument room was not

1034: opened, the cold air rushing into the telescope basement caused a

1035: temperature dip in the ET instrument room. The prototype monolithic interferometer tested with the Kitt Peak ET exhibits

1036: substantially more drift than the one tested in the vaccum

1037: chamber. While some of this drift can be attributed to the

1038: temperature and pressure shifts in the instrument enclosure,

1039: Figure \ref{fig:kpnodrift} demonstrates that the temperature does

1040: not shift enough to cause such a large drift. We suspect that a

1041: significant part of this drift is caused by a variation in the

1042: tilt due to slow warps of the copper ring, but the exact cause of

1043: the drift is not trivial to determine. We surmise that

1044: this drift was a consequence of internal stresses in the copper

1045: ring being released as the interferometer attempted to achieve

1046: equilibrium with the thermal environment.

1047: 

1048: \subsection{Radial Velocity Results}

1049: 

1050:   The radial velocities obtained for our velocity reference and planet

1051: stars are shown in Figure \ref{fig:velocities}. The mean photon noise error for $\eta$ Cas is $\sim \sigma=5.5$ m s$^{-1}$,

1052: and the data points exhibit a very low scatter only because of

1053: chance alignment. $\eta$ Cas is a known long period binary whose

1054: radial velocity is expected to be quite stable in the short term.

1055: For 55 Cnc and $\rho$ Crb the expected radial velocity curves of

1056: the stars due to the planets orbiting them (using orbital

1057: parameters from \cite{Butler06}) are plotted as a solid line. The monolithic

1058: interferometer drift of $\sim 7$ km s$^{-1}$ over the 12 day

1059: observing run yields a mean long term drift of $\sim 24$ m

1060: s$^{-1}$ hour$^{-1}$, which is acceptable since such a small drift

1061: does not significantly affect fringe visibilities for even the

1062: longest exposures which are typically ~20 minutes. The use of the

1063: iodine cell for simultaneous calibration allows us to effectively

1064: determine and subtract out the instrumental drift, yielding precise radial velocities.

1065: 

1066: \section{Discussion}

1067: Using off-the-shelf components and a novel bonding technique we

1068: have built a monolithic fixed-delay Michelson interferometer for

1069: the ET instrument. With this inexpensive prototype we have

1070: effectively demonstrated the ability of such an interferometers to

1071: recover precise radial velocities. The results for the stable

1072: and planet bearing stars demonstrate that the prototype

1073: monolithic interferometer, developed using only off-the-shelf

1074: components, is capable of allowing us to achieve a velocity

1075: precision of $\sim 10$ m s$^{-1}$ or better over observing runs as

1076: long as two weeks. Monolithic interferometers are significantly

1077: more insensitive to vibrations, and do not need active locking

1078: using a PZT device. The temperature-compensated, field widened

1079: interferometer also makes the ET instrument significantly easier

1080: to use since the observer no longer has to constantly monitor the

1081: phase locking software.

1082:   The performance of one of our prototype interferometers in the lab is

1083:  substantially better than the one tested at KPNO. We believe that

1084:  the large drifts seen at Kitt-Peak were caused by the warping on the copper ring. Nevertheless, we have demonstrated the ability of such a monolithic

1085:  interferometer to aid in the precise measurement of radial

1086:  velocities using a suitable reference to track the instrument and

1087:  interferometer drifts.  The hydrolysis catalyzed bonding technique we have

1088:  employed is transparent over a wide wavelength region, and stable

1089:  over a large range in temperatures.

1090:   The next goal in our development efforts is a larger and very stable

1091:  field-widened and temperature-compensated fixed-delay

1092:  interferometer that can be used in a multi-object instrument to

1093:  observe many stars simultaneously. We expect that the issues related to the warping of the ring can be avoided if a second glass material were used instead of an airgap. Better designs of such

1094:  interferometers, coupled with better environmental control, may

1095:  make it possible to build interferometers that are intrinsically stable enough to not

1096:  require simultaneous iodine calibration in order to measure to radial velocities to a high precision. With high enough stability and good thermal control it may become possible to achieve acceptable velocity precision by bracketing the stellar observations with a velocity reference, or using another reference fiber to track the star fiber. With minor variations, such interferometers can

1097:  potentially be employed in a variety of dispersed fixed delay

1098:  interferometer instruments including large multi-object

1099:  instruments planned for upcoming surveys like MARVELS.

1100: 

1101: \acknowledgments  We thank Dequing Ren and Bo Zhao for useful

1102: discussions and their contributions to the ET project. Jerry

1103: Friedman shared his expertise in mechanical design for tests of an

1104: earlier version of this interferometer. We are grateful to Richard

1105: Green, Skip Andree, Daryl Wilmarth and the KPNO staff for their

1106: generous support. This work is supported by National Science

1107: Foundation grant AST 02-4309, JPL, the Pennsylvania State

1108: University, and the University of Florida. JCvE., SM. and SWF. acknowledge travel support from KPNO. JCvE. and SM.

1109: acknowledge the Michelson Fellowship. SWF was supported by the Florida Space Grant Fellowship. This research has made use

1110: of the SIMBAD and Vizier databases, operated at ADC, Strasbourg,

1111: France. This work was performed in part under contract with the

1112: Jet Propulsion Laboratory (JPL) funded by NASA through the

1113: Michelson Fellowship Program. JPL is managed for NASA by the

1114: California Institute of Technology. This work is based on data

1115: obtained at the Kitt Peak 2.1 m telescope.

1116: 

1117: %\bibliography{references}

1118: \begin{thebibliography}{}

1119: 

1120: 

1121: \bibitem[\protect\citeauthoryear{{Barden}}{{Barden}}{1998}]{Barden98}

1122: {Barden}, S.~C. 1998, in ASP Conf. Ser. 152: Fiber Optics in

1123: Astronomy III, ed.

1124:   S.~{Arribas}, E.~{Mediavilla}, \& F.~{Watson}, 14

1125: 

1126: \bibitem[Beckers\& Brown(1979)]{1979MmArc.106..189B} Beckers, J.~M., \& Brown, T.~M.\ 1979, Oss.~Mem.~Oss.~Astrofis.~Arcetri, Fasc.~106, p.~189 -

1127: 203, 106, 189

1128: 

1129: \bibitem[Borucki et al.(2003)]{2003SPIE.4854..129B} Borucki, W.~J., et al.\

1130: 2003, \procspie, 4854, 129

1131: 

1132: \bibitem[Bouchy et

1133: al.(2001)]{Bouchy01} Bouchy, F., Pepe, F., \& Queloz, D.\ 2001,

1134: \aap, 374, 733

1135: 

1136: \bibitem[\protect\citeauthoryear{{Butler} et~al.}{{Butler}

1137:   et~al.}{1996}]{Butler96}

1138: {Butler}, R.~P., {Marcy}, G.~W., {Williams}, E., {McCarthy}, C.,

1139: {Dosanjh}, P.,

1140:    \& {Vogt}, S.~S. 1996, \pasp, 108, 500

1141: 

1142: \bibitem[\protect\citeauthoryear{{Butler} et~al.}{{Butler}

1143:   et~al.}{2006}]{Butler06}

1144: {Butler}, R.~P., {Wright}, J.~T., {Marcy}, G.~W., {Fischer},

1145: D.~A., {Vogt},

1146:   S.~S., {Tinney}, C.~G., {Jones}, H.~R.~A., {Carter}, B.~D., {Johnson}, J.~A.,

1147:   {McCarthy}, C.,  \& {Penny}, A.~J. 2006, \apj, 646, 505

1148: 

1149: \bibitem[\protect\citeauthoryear{{Connes}, {Martic}, \& {Schmitt}}{{Connes}

1150:   et~al.}{1996}]{Connes96}

1151: {Connes}, P., {Martic}, M.,  \& {Schmitt}, J. 1996, \apss, 241, 61

1152: 

1153: \bibitem[Endl et al.(2007)]{2007arXiv0709.0944E} Endl, M., Cochran, W.~D.,

1154: Wittenmyer, R.~A., \& Boss, A.~P.\ 2007, ArXiv e-prints, 709,

1155: arXiv:0709.0944

1156: 

1157: \bibitem[Englert et al.(2007)]{2007AGUFMSA13B..01E} Englert, C.~R.,

1158: Stevens, M.~H., Siskind, D.~E., Harlander, J.~M., \& Roesler,

1159: F.~L.\ 2007, AGU Fall Meeting Abstracts, 1

1160: 

1161: \bibitem[\protect\citeauthoryear{{Erskine}}{{Erskine}}{2003}]{Erskine03} {Erskine}, D.~J. 2003, \pasp, 115, 255

1162: 

1163: \bibitem[\protect\citeauthoryear{{Erskine} \& {Ge}}{{Erskine} \&

1164:   {Ge}}{2000}]{Erskine00}

1165: {Erskine}, D.~J.,  \& {Ge}, J. 2000, in ASP Conf. Ser. 195:

1166: Imaging the

1167:   Universe in Three Dimensions, ed. W.~{van Breugel} \& J.~{Bland-Hawthorn},

1168:   501

1169: 

1170: \bibitem[Forrest \& Ring(1978)]{Forrest78} Forrest, A.~K., \&

1171: Ring, J.\ 1978, High resolution spectrometry, 462

1172: 

1173: \bibitem[Gault et al.(1985)]{1985ApOpt..24.1604G} Gault, W.~A., Johnston,

1174: S.~F., \& Kendall, D.~J.~W.\ 1985, \ao, 24, 1604

1175: 

1176: \bibitem[Ge(2002)]{Ge02a} Ge, J.\ 2002, \apjl, 571, L165

1177: 

1178: \bibitem[\protect\citeauthoryear{{Ge}, {Erskine}, \& {Rushford}}{{Ge} et~al.}{2002}]{Ge02b} {Ge}, J., {Erskine}, D.~J.,  \& {Rushford},

1179: M. 2002, \pasp, 114, 1016

1180: 

1181: \bibitem[Ge et al.(2003)]{2003SPIE.4838..503G} Ge, J., van Eyken, J.~C.,

1182: Mahadevan, S., DeWitt, C., Ramsey, L.~W., Shaklan, S.~B.,

1183: \& Pan, X.\ 2003, \procspie, 4838, 503

1184: 

1185: \bibitem[Ge et al.(2004)]{2004SPIE.5492..711G} Ge, J., Mahadevan, S., van

1186: Eyken, J.~C., DeWitt, C., Friedman, J.,

1187: \& Ren, D.\ 2004, \procspie, 5492, 711

1188: 

1189: \bibitem[Ge et al.(2006)]{} Ge, J., et al.\ 2006, \apj,648, 683

1190: 

1191: \bibitem[Ge et al.(2007)]{Ge07} Ge, J., et al.\ 2007,

1192: American Astronomical Society Meeting Abstracts, 211, \#132.09

1193: 

1194: \bibitem[Gwo(1998)]{1998SPIE.3435..136G} Gwo, D.-H.\ 1998, \procspie, 3435,

1195: 136

1196: 

1197: \bibitem[Gwo(1998)]{1998SPIE.3356..892G} Gwo, D.-H.\ 1998, \procspie, 3356,

1198: 892

1199: 

1200: \bibitem[\protect\citeauthoryear{{Gwo}}{{Gwo}}{2001}]{Gwoa}

1201: {Gwo}, D. 2001, United States Patent 6284085

1202: 

1203: \bibitem[\protect\citeauthoryear{{Gwo}}{{Gwo}}{2003}]{Gwob}

1204: 

1205: {Gwo}, D. 2003, United States Patent 6284085

1206: 

1207: \bibitem[\protect\citeauthoryear{{Harvey} \& {The GONG Instrument

1208:   Team}}{{Harvey} \& {The GONG Instrument Team}}{1995}]{Harvey95}

1209: {Harvey}, J.,  \& {The GONG Instrument Team}. 1995, in ASP Conf.

1210: Ser. 76: GONG

1211:   1994. Helio- and Astro-Seismology from the Earth and Space, ed. R.~K.

1212:   {Ulrich}, E.~J. {Rhodes}, Jr., \& W.~{Dappen}, 432

1213: 

1214: \bibitem[\protect\citeauthoryear{{Heacox}}{{Heacox}}{1986}]{Heacox86}

1215: {Heacox}, W.~D. 1986, \aj, 92, 219

1216: 

1217: \bibitem[\protect\citeauthoryear{{Hilliard} \& {Shepherd}}{{Hilliard} \&

1218:   {Shepherd}}{1966}]{Hillard66}

1219: {Hilliard}, R.~L.,  \& {Shepherd}, G.~G. 1966, Journal of the

1220: Optical Society

1221:   of America (1917-1983), 56, 362

1222: 

1223: \bibitem[\protect\citeauthoryear{{Hunter} \& {Ramsey}}{{Hunter} \&

1224:   {Ramsey}}{1992}]{Hunter91}

1225: {Hunter}, T.~R.,  \& {Ramsey}, L.~W. 1992, \pasp, 104, 1244

1226: 

1227: \bibitem[Johnson et al.(2007)]{2007ApJ...665..785J} Johnson, J.~A., et al.\

1228: 2007, \apj, 665, 785

1229: 

1230: \bibitem[\protect\citeauthoryear{{Kogelnik}}{{Kogelnik}}{1969}]{Kogelnik69}

1231: {Kogelnik}, H. 1969, The Bell System Technical Journal, Vol.~48,

1232: no.~9,

1233:   November 1969, pp.~2909-2947, 48, 2909

1234: 

1235: \bibitem[\protect\citeauthoryear{{Kozhevatov}}{{Kozhevatov}}{1983}]{Kozhevatov%

1236: 83} {Kozhevatov}, I.~E. 1983, Issledovaniia Geomagnetizmu

1237: Aeronomii i Fizike

1238:   Solntsa, 64, 42

1239: 

1240: \bibitem[Mahadevan et al.(2008)]{2008ApJ...678.1505M} Mahadevan, S., van

1241: Eyken, J., Ge, J., DeWitt, C., Fleming, S.~W., Cohen, R., Crepp, J.,

1242: \& Vanden Heuvel, A.\ 2008, \apj, 678, 1505

1243: 

1244: \bibitem[\protect\citeauthoryear{{Mayor} \& {Queloz}}{{Mayor} \&

1245:   {Queloz}}{1995}]{Mayor95}

1246: {Mayor}, M.,  \& {Queloz}, D. 1995, \nat, 378, 355

1247: 

1248: \bibitem[McMillan et

1249: al.(1994)]{1994Ap&SS.212..271M} McMillan, R.~S., Moore, T.~L.,

1250: Perry, M.~L., \& Smith, P.~H.\ 1994, \apss, 212, 271

1251: 

1252: \bibitem[\protect\citeauthoryear{{Mosser} et~al.}{{Mosser}

1253:   et~al.}{1998}]{Mosser98}

1254: {Mosser}, B., {Maillard}, J.~P., {Mekarnia}, D.,  \& {Gay}, J.

1255: 1998, \aap, 340, 457

1256: 

1257: \bibitem[\protect\citeauthoryear{{Mosser}, {Maillard}, \& {Bouchy}}{{Mosser}

1258:   et~al.}{2003}]{Mosser03}

1259: {Mosser}, B., {Maillard}, J.-P.,  \& {Bouchy}, F. 2003, \pasp,

1260: 115, 990

1261: 

1262: \bibitem[Mosser

1263: \& The Siamois Team(2007)]{2007EAS....25..239M} Mosser, B., \& The

1264: Siamois Team 2007, EAS Publications Series, 25, 239

1265: 

1266: 

1267: \bibitem[Ramsey

1268: \& Huenemoerder(1986)]{1986SPIE..627..282R} Ramsey, L.~W., \&

1269: Huenemoerder, D.~P.\ 1986, \procspie, 627, 282

1270: 

1271: \bibitem[\protect\citeauthoryear{{Rallison}, {Rallison}, \&

1272:   {Dickson}}{{Rallison} et~al.}{2003}]{Rallison03}

1273: {Rallison}, R.~D., {Rallison}, R.~W.,  \& {Dickson}, L.~D. 2003,

1274: in Specialized

1275:   Optical Developments in Astronomy. Edited by Atad-Ettedgui, Eli; D'Odorico,

1276:   Sandro. Proceedings of the SPIE, Volume 4842, pp. 10-21 (2003)., ed.

1277:   E.~{Atad-Ettedgui} \& S.~{D'Odorico}, 10

1278: 

1279: \bibitem[\protect\citeauthoryear{{Rivera} et~al.}{{Rivera}

1280:   et~al.}{2005}]{Rivera05}

1281: {Rivera}, E.~J., {Lissauer}, J.~J., {Butler}, R.~P., {Marcy},

1282: G.~W., {Vogt},

1283:   S.~S., {Fischer}, D.~A., {Brown}, T.~M., {Laughlin}, G.,  \& {Henry}, G.~W.

1284:   2005, \apj, 634, 625

1285: 

1286: \bibitem[\protect\citeauthoryear{{Rupprecht} et~al.}{{Rupprecht}

1287:   et~al.}{2004}]{Rupprecht04}

1288: {Rupprecht}, G., {Pepe}, F., {Mayor}, M., {Queloz}, D., {Bouchy},

1289: F., {Avila},

1290:   G., {Benz}, W., {Bertaux}, J.-L., {Bonfils}, X., {Dall}, T., {Delabre}, B.,

1291:   {Dekker}, H., {Eckert}, W., {Fleury}, M., {Gilliotte}, A., {Gojak}, D.,

1292:   {Guzman}, J.~C., {Kohler}, D., {Lizon}, J.-L., {Lo Curto}, G., {Longinotti},

1293:   A., {Lovis}, C., {Megevand}, D., {Pasquini}, L., {Reyes}, J., {Sivan}, J.-P.,

1294:   {Sosnowska}, D., {Soto}, R., {Udry}, S., {Van Kesteren}, A., {Weber}, L.,  \&

1295:   {Weilenmann}, U. 2004, in Ground-based Instrumentation for Astronomy. Edited

1296:   by Alan F. M. Moorwood and Iye Masanori. Proceedings of the SPIE, Volume

1297:   5492, pp. 148-159 (2004)., ed. A.~F.~M. {Moorwood} \& M.~{Iye}, 148

1298: 

1299: \bibitem[\protect\citeauthoryear{{Santos} et~al.}{{Santos}

1300:   et~al.}{2004}]{Santos04}

1301: {Santos}, N.~C., {Bouchy}, F., {Mayor}, M., {Pepe}, F., {Queloz},

1302: D., {Udry},

1303:   S., {Lovis}, C., {Bazot}, M., {Benz}, W., {Bertaux}, J.-L., {Lo Curto}, G.,

1304:   {Delfosse}, X., {Mordasini}, C., {Naef}, D., {Sivan}, J.-P.,  \& {Vauclair},

1305:   S. 2004, \aap, 426, L19

1306: 

1307: \bibitem[Setiawan et al.(2007)]{2007ApJ...660L.145S} Setiawan, J., Weise,

1308: P., Henning, T., Launhardt, R., M{\"u}ller, A., \& Rodmann, J.\

1309: 2007, \apjl, 660, L145

1310: 

1311: \bibitem[Shao et al.(2007)]{2007arXiv0704.0952S} Shao, M., et al.\ 2007,

1312: ArXiv e-prints, 704, arXiv:0704.0952

1313: 

1314: \bibitem[\protect\citeauthoryear{{Shepherd} et~al.}{{Shepherd}

1315:   et~al.}{1985}]{Shepherd85}

1316: {Shepherd}, G.~G., {Gault}, W.~A., {Miller}, D.~W., {Pasturczyk},

1317: Z.,

1318:   {Johnston}, S.~F., {Kosteniuk}, P.~R., {Haslett}, J.~W., {Kendall}, D.~J.~W.,

1319:    \& {Wimperis}, J.~R. 1985, \ao, 24, 1571

1320: 

1321:    \bibitem[\protect\citeauthoryear{{Thuillier} \& {Shepherd}}{{Thuillier} \&

1322:   {Shepherd}}{1985}]{Thullier85}

1323: {Thuillier}, G.,  \& {Shepherd}, G.~G. 1985, \ao, 24, 1599

1324: 

1325: \bibitem[\protect\citeauthoryear{{Title} \& {Ramsey}}{{Title} \&

1326:   {Ramsey}}{1980}]{Title80}

1327: {Title}, A.~M.,  \& {Ramsey}, H.~E. 1980, \ao, 19, 2046

1328: 

1329: \bibitem[van Eyken et al.(2003)]{2003SPIE.5170..250V} van Eyken, J.~C., Ge,

1330: J.~C., Mahadevan, S., DeWitt, C., \& Ren, D.\ 2003, \procspie, 5170, 250

1331: 

1332: \bibitem[van Eyken et al.(2004)]{2004ApJ...600L..79V} van Eyken, J.~C., Ge,

1333: J., Mahadevan, S., \& DeWitt, C.\ 2004, \apjl, 600, L79

1334: 

1335: \bibitem[\protect\citeauthoryear{{Wolszczan} \& {Frail}}{{Wolszczan} \&

1336:   {Frail}}{1992}]{Wolszczan92}

1337: {Wolszczan}, A.,  \& {Frail}, D.~A. 1992, \nat, 355, 145

1338: 

1339: \end{thebibliography}

1340: 

1341: \clearpage

1342: 

1343: \begin{table}[htbp]

1344:   \begin{center}

1345:     \caption[Velocity shift of the interferometer]{Velocity shift of the interferometer, for a one degree Celsius drift in temperature, when different materials are used as spacers. The numbers shown here are for an interferometer operating in vacuum. }

1346:      \label{tbl:velshift}

1347:     \begin{tabular}{lllllll}

1348:       \hline

1349: SPACER MATERIAL  &  VELOCITY SHIFT (m s$^{-1}$ degree$^{-1}$)\\

1350: \hline

1351:   BK7       &             $\sim$ 2700\\

1352:   Copper      &          $\sim$ 250-500\\

1353:   CaF$_2$     &           $\sim$ 50\\

1354:   Aluminum    &          $\sim$ -1000\\

1355: \hline

1356:  \end{tabular}

1357:   \end{center}

1358: \end{table}

1359: 

1360: \clearpage

1361: 

1362: 

1363: \begin{figure}[htbp] \begin{center}

1364: \includegraphics*[width=3in,angle=0]{f1.eps}

1365: \end{center} \caption[Phase shift of the interferometer fringes]

1366: {This pattern shows the expected output when a field compensated Michelson interferometer with a small tilt in one arm is illuminated with monochromatic light. The cartoon on the right shows the phase shift of the interferometer fringes when the wavelength of

1367: the light experiences a Doppler shift.}

1368: \label{fig:interferometerfringe}

1369: \end{figure}

1370: 

1371: \begin{figure}[htbp] \begin{center}

1372: \includegraphics*[width=3in,angle=0]{f2.eps}

1373: \end{center} \caption[Two dimensional spectral format]{Simulated two dimensional spectral format for a dispersed fixed-delay interferometer instrument. Fringes are clearly visible at the location of the stellar absorption lines.} \label{fig:ETsimulspec} \end{figure}

1374: 

1375: \begin{figure}[htbp]

1376: \begin{center}

1377: \includegraphics*[width=6in,angle=0]{f3.eps}

1378: \end{center}

1379: \caption[Optical layout of the Kitt Peak ET instrument]{Optical

1380: layout of the Kitt Peak ET instrument, showing the Michelson

1381: interferometer, spectrograph camera and collimator, VPH grating,

1382: and the iodine cell that can be inserted into the path of the

1383: collimated beam.} \label{fig:EToptical}

1384: \end{figure}

1385: 

1386: \begin{figure}[htbp]

1387: \begin{center}

1388: \includegraphics*[width=4in,angle=0]{f4.eps}

1389: \end{center}

1390: \caption[The temperature measured at four points in the ET

1391: instrument]{The temperature measured at four points in the ET

1392: instrument during a typical observing run. The instrument

1393: enclosure helps stabilize the temperature and dampens the effect

1394: of the temperature variations outside the enclosure. The iodine

1395: cell and the electronics inside the detector head are the dominant

1396: source of heat inside the enclosure.} \label{fig:kpnotemperature}

1397: \end{figure}

1398: 

1399: \begin{figure}[htbp]

1400: \begin{center}

1401: \includegraphics*[width=5in, height=5in, angle=90]{f5.eps}

1402: \end{center}

1403: \caption[The spectra format of the ET instrument]{The 4k $\times$ 4k pixles image obtained with ET instrument. Both outputs from the interferometer

1404: are imaged onto the detector. The two-dimensional spectral format

1405: allows the measurement of the fringe phase at each wavelength

1406: without having to physically change the interferometer delay.}

1407: \label{fig:ET_format}

1408: \end{figure}

1409: 

1410: \begin{figure}[htbp] \begin{center}

1411: \includegraphics*[width=5in,angle=0]{f6.eps}

1412: \end{center} \caption[Optimal delay for a K5V star]{Optimal delay for a K5V star at different rotational velocities. The optimal value of  the interferometer delay maximizes $\gamma d$. This maximum value of $\gamma d$ has been normalized to one for all the plots.}

1413: \label{fig:rotators} \end{figure}

1414: 

1415: 

1416: \begin{figure}[htbp]

1417: \begin{center}

1418: \includegraphics*[width=4in,angle=0]{f7.eps}

1419: \end{center}

1420: \caption[Phase shift for a compensated Michelson

1421: interferometer]{Phase shift for a compensated Michelson

1422: interferometer that has $L_1$(BK7) $=4.04$ mm, $L_2$(air) $=2.66$

1423: mm. The wavelength dependent phase shift show here is for a

1424: converging beam with a half angle of 1.5 degrees.}

1425: \label{fig:phaseshift}

1426: \end{figure}

1427: 

1428: 

1429: 

1430: \begin{figure}[htbp]

1431: \begin{center}

1432: \includegraphics*[width=3in,angle=0]{f8.eps}

1433: \end{center}

1434: \caption[Phase shift for a compensated Michelson

1435: interferometer]{Exploded view of the monolithic Michelson

1436: interferometer showing each component of the final assembly.}

1437: \label{fig:exploded}

1438: \end{figure}

1439: 

1440: \begin{figure}[htbp]

1441: \begin{center}

1442: \includegraphics*[width=4.6in,angle=0]{f9.eps}

1443: \end{center}

1444: \caption[The formation of a bond between two fused silica

1445: disks]{The formation of a bond between two fused silica disks in

1446: the presence of water and the hydroxide catalyst.  A similar

1447: mechanism occurs for any material capable of forming a

1448: silicate-like network. In this figure a) shows the fused silica

1449: disks, b) demonstrates the formation of hydrated fused silica, and

1450: c) shows the formations of a bond linking the two disks. This

1451: figure is adapted from \citet{Gwob}.} \label{fig:optobond}

1452: \end{figure}

1453: 

1454: \begin{figure}[htbp]

1455: \begin{center}

1456: \includegraphics*[width=3.6in,angle=90]{f10.eps}

1457: \end{center}

1458: \caption[The velocity drift of a passive interferometer during lab

1459: testing]{The velocity drift of a passive interferometer during lab

1460: testing. The interferometer was placed in a vacuum chamber and the

1461: fringes were monitored using a stabilized He-Ne laser and a

1462: LabView program.} \label{fig:labtest}

1463: \end{figure}

1464: 

1465: 

1466: \begin{figure}[htbp]

1467: \begin{center}

1468: \includegraphics*[width=3.6in,angle=0]{f11.eps}

1469: \end{center}

1470: \caption[The assembled monolithic interferometer aligned in the

1471: Kitt Peak ET instrument]{The assembled monolithic interferometer

1472: aligned in the Kitt Peak ET instrument. The copper spacer and the

1473: BK7 arm of the interferometer are clearly visible. Also seen are

1474: the iodine cell, the input optics and the slit assembly.}

1475: \label{fig:passivesetup}

1476: \end{figure}

1477: 

1478: \begin{figure}[htbp]

1479: \begin{center}

1480: \includegraphics*[width=3.6in,angle=90]{f12.eps}

1481: \end{center}

1482: \caption[The velocity drift of the monolithic interferometer

1483: during an observing run at Kitt Peak]{The velocity drift of the

1484: monolithic interferometer during an observing run at Kitt Peak, as

1485: measured with a stabilized He-Ne laser, is shown as solid black line.

1486: Filled circles are the measured drift of the iodine calibration

1487: spectra obtained during the run. The grey line shows the

1488: measured interferometer temperature.} \label{fig:kpnodrift}

1489: \end{figure}

1490: 

1491: \begin{figure}[htbp]

1492: \begin{center}

1493: \includegraphics*[width=3.6in,angle=90]{f13.eps}

1494: \end{center}

1495: \caption[Radial velocity results from the ET instrument and the

1496: passive interferometer]{Radial velocity obtained with the

1497: monolithic interferometer. The stars selected are the known radial

1498: velocity stable stars $\eta$ Cas  and 36 UMa , and the known

1499: planet bearing stars 55 Cnc and $\rho$ Crb. The expected velocity

1500: amplitude caused by the known planetary systems is plotted as a

1501: solid line.} \label{fig:velocities}

1502: \end{figure}

1503: 

1504: \end{document}

1505: