Visible Spectrometer

Experiment VIS

University of Florida — Department of Physics

PHY4803L — Advanced Physics Laboratory

Objective mass of the electron-proton me -mp system,

The Balmer spectral lines emitted from a hy- me mp

µ= (2)
drogen discharge lamp are observed with a me + mp
transmission grating spectrometer and ana- A photon is emitted when the electron
lyzed to obtain the Rydberg constant. Wave- makes a transition from a higher energy level
length calibration is achieved by measuring to a lower level with the photon carrying away
diffraction angles for spectral lines of “known” the excess energy ∆E = Eni − Enf . Because
wavelengths from a helium discharge tube and the energy of a photon and its wavelength λ
fitting this data to the grating equation. The in vacuum are related by λ = hc/∆E, Eq. 1
wavelengths of the hydrogen lines are then de- predicts the relation
termined and fit to the Rydberg formula. Ba- ( )
sic spectroscopic and statistical analysis tech- 1 1 1
niques are covered. = RH − 2 (3)
λ n2f ni

where
Theory µe4
RH = (4)
8ϵ20 h3 c
The Hydrogen Spectrum
is called the reduced mass Rydberg. If the nu-
The Bohr model assumes the electron moves cleus were infinitely heavy, the reduced mass
around the proton in circular orbits with only µ becomes me , the mass of the electron. The
quantized angular momentum L = n~ al- combination of physical constants
lowed, where n = 1, 2, ..., is a positive integer me e 4
and ~ = h/2π where h is Planck’s constant. R∞ = 2 3 (5)
8ϵ0 h c
These assumptions lead to discrete energy lev-
els that can be expressed is called the infinite mass Rydberg or simply
the Rydberg. Although RH and R∞ differ by
µe4 less than 0.1%, with care, the measurements
En = − 2 2 2 (1)
8ϵ0 h n you make should be just accurate enough to
distinguish the difference between them.
where e is the electron charge, ϵ0 is the per- Sets of wavelengths (series) are categorized
mittivity of free space, and µ is the reduced by the quantum number nf of the lower level of

VIS 1
VIS 2 Advanced Physics Laboratory

transmission
collimator
grating tral lines that are finely spaced about the Bohr
central prediction—too finely spaced to see them as
beam
θi θr separate with our apparatus.
slit diffracted
grating beam
normal
telescope Exercise 1 (a) Use this Bohr model to derive
Eq. 1. Give the n = 3 to n = 2 energy level
difference in eV. (b) Learn about the hydrogen
Figure 1: Top view of a transmission grating fine structure and hyperfine structure and de-
spectrometer. Note that the incident angle θi scribe their origins. The largest of these effects
and the diffraction angle θr are relative to the is the fine structure splitting resulting in two
grating normal. n = 2 energy levels. Look up this splitting and
give it in eV and as a fraction of n = 3 to
the transition. The Lyman series is obtained n = 2 transition energy.
for nf = 1, ni = 2, 3, ... and is in the ultra-
violet part of the spectrum; the Balmer series The Diffraction Grating
corresponds to nf = 2, ni = 3, 4, ... and is in Wavelength measurements in this experiment
the visible; the Paschen series, for nf = 3, is are based on the interference of a large number
in the infrared; etc. of waves scattered from the grooves of a trans-
The Bohr model is overly simplistic and not mission grating illuminated by incident plane
in agreement with a more complete quantum waves of various wavelengths. The geometry
mechanical treatment. For example, whereas is shown in Fig. 1.
the Bohr theory predicts the lowest energy or
ground state (for n = 1) has an angular mo- Exercise 2 Draw a figure showing adjacent
mentum ~, quantum mechanics says it has no grating grooves (spaced d apart) and show that
angular momentum at all. constructive interference occurs for
Besides the principal quantum number n,
quantum mechanics describes states of hydro- mλ = d(sin θr − sin θi ) (6)
gen by additional quantum numbers. For ex- where the incident angle θ and the diffraction
i
ample, the orbital angular momentum quan- angle θ are measured relative to the grating
r
tum number ℓ can range from ℓ = 0 to ℓ = normal and have the same sign when they lie
n − 1. Thus for n = 1, only ℓ = 0 is allowed on opposite sides of the grating normal (as in
(the 1s state). For n = 2, only ℓ = 0 (the 2s Fig. 1. The diffraction order m is a positive
state) and ℓ = 1 (the 2p state) are possible. integer (θ > θ ) negative integer (θ < θ ) or
r i r i
For higher n, higher ℓ states are possible. zero (θr = θi ).
The Bohr model also does not take into ac-
count the intrinsic angular momentum (spin)
Dispersion and Resolution
of the electron or of the nucleus. Where
the Bohr model says the internal energy de- Two important spectrometer properties are
pends only on n, the real hydrogen atom has dispersion and resolution. Dispersion is a mea-
more than one energy level for each value of sure of the rate of change of spectral line posi-
n that depend on ℓ and other quantum num- tion with λ. With higher dispersion, the spec-
bers associated with the electron and nuclear tral lines are more separated from each other
spins. These extra energy levels lead to spec- or more spread out. With our spectrometer,

October 14, 2022

Visible Spectrometer VIS 3

spectral line positions are measured as an an-

gle θr and dispersion is best represented by the
value of dθr /dλ.

Exercise 3 (a) Obtain an expression for

dθr /dλ from the grating equation (Eq. 6) in
terms of d, m, and θr . Show that it increases
for larger order number m, smaller d, and
larger θr . (b) For a given λ, m, and d, will
the dispersion increase, decrease, or remain
the same as θi increases? Hint for part b: As
θi increases, θr must change as well. Figure
out how θr must change. Then from part (a)
you will know how the dispersion will change. Figure 2: Top: Many-slit diffraction pattern
The end result is that the answer to part b will for a single wavelength. Bottom: The thick
depend on the sign of m, i.e., to which side line is the sum of two diffraction patterns
of the grating normal you are measuring. The for two different wavelengths separated by the
dispersion will increase on one side and de- Rayleigh criterion.
crease on the other. You should derive and
explain this dependence.
slit width and focusing quality, higher disper-
◦
Exercise 4 If θi were 25 and the grating had sion generally implies higher resolution. But
600 lines/mm, find all angles θr where a red the narrowest linewidths are ultimately lim-
line at 650 nm and a blue line at 450 nm ited by the number of grating grooves illumi-
should appear in first order (m = ±1) or sec- nated. As the number of grooves contributing
ond order (m = ±2). Hint: You should get to the diffraction increases, the angular width
only seven angles not eight. Which one isn’t of the diffraction pattern (and hence the mini-
possible? Assume you made measurements at mum linewidth) decreases. (Large gratings are
these seven angles—each with an uncertainty used when high resolution is needed.)
in θr of 4 minutes of arc. (One minute of The Rayleigh criterion gives a generally ac-
arc is 1/60 of a degree.) Provide an analytic cepted minimum “just resolvable” ∆λ. The
expression for the propagated uncertainty in λ criterion is that the peak of the diffraction
and determine its value for each measurement. pattern of one line is at the first zero of the
diffraction pattern of the other. See Fig. 2.
Resolution describes the ability of the spec- Diffraction theory can be used to show this
trometer to separate two nearby spectral lines. condition occurs for two lines of wavelengths
Imagine the wavelength separation between λ ± ∆λ/2 when
the two spectral lines becoming smaller. The
λ
observed lines will begin to overlap each other = mN (7)
and at some small separation ∆λ the ability to ∆λ
discern the lines as separate will be lost. Res- where N is the number of grooves illuminated,
olution is a measure of the smallest discern- and m is the order of diffraction in which they
able ∆λ. When the spectral linewidths are are observed. λ/∆λ is called the resolving
dominated by spectrometer settings, such as power.

October 14, 2022

VIS 4 Advanced Physics Laboratory

Figure 4: The angular reading is 133◦ 9′ . The

0 mark is just after the 133◦ line on the main
′
Figure 3: The spectrometer for this experi- scale. The 9 mark is directly aligned with a
ment. The various components are described mark on the main scale.
in the text.
ciated with these elements. The outer circle
Spectrometer is rigidly attached to the telescope and has an
angular scale from 0 to 360◦ in 0.5◦ increments.
Fig. 3 shows the location of the various spec- The adjacent inner circle is rigidly attached to
trometer adjustment mechanisms. Familiarize the table base and has two 0 to 30′ (minutes
yourself with them. Components in italics be- of arc) vernier scales on opposite sides.
low will be referred to throughout the write- When measuring angles it is important to
up. take readings at both vernier scales. Because
Note the two sets of rotation adjustments of manufacturing tolerances, the two readings
◦
(J and K) near the legs of the spectrometer. will not always be exactly 180 apart. By us-
The upper one affects rotations of the telescope ing them both, more accurate angular mea-
(D), and the lower one affects rotations of the surements are obtained. See Fig. 4 for an ex-
table base (N) on which the actual table (C) ample of a reading at one vernier scale.
(holding the grating, not shown) is inserted. The table is mounted on a post that is in-
Also not shown is a mounting post on the table serted into the table base and locked into place
for clamping the grating to the table. Both ro- with the table locking screw (I). Three table
tational motions are about the main spectrom- leveling screws (O) under the table allow the
eter axis (Q) which is vertical and through pitch and yaw of the table to be varied.
the center of the instrument. Each rotation A cross hair inside the telescope is brought
adjustment has a locking screw and a tan- into focus by sliding the eyepiece (L) in or out.
gent screw. When the locking screw is loose, The telescope is focused by rotating the tele-
the corresponding element (telescope or table scope focusing ring (E). The collimator (B)
base) can be rotated freely by hand. When (on which the entrance slit (A) is located) is
tightened, the corresponding element can be adjusted for focus by sliding the inner colli-
rotated small amounts with fine control using mator tube in or out. An index ring (F) on
the tangent screw. There are two precision the collimator tube allows the focusing posi-
machined circles (called divided circles) asso- tion to be maintained once found. The ring

October 14, 2022

Visible Spectrometer VIS 5

has a V-shaped protuberance that fits into one in or out and on the distant object us-
of the V-shaped cutouts spaced 90◦ apart on ing the telescope focus. Both the tele-
the outer collimator tube. Making registra- scope and cross hair images should now
tion with one of the V’s once the proper focus- be focused at infinity and there should
ing and slit orientation is obtained, the ring be no parallax. As you move your eye
is then tightened into place. This permits the slightly from side to side, the cross hair
slit orientation to be changed 90◦ without los- should not move relative to the image of
ing the focus by sliding the tube out slightly, the bulletin board. If it does move, the
rotating it 90◦ , and pushing it back into the image is not at the cross hair and the
other V-shaped cutout. telescope focus still needs adjustment. It
The telescope and collimator also have lev- may help to de-focus the telescope and
eling screws (P) that allow their optical axes then readjust the eyepiece to focus only
to tilt up or down. These adjustments are on the cross hair before trying to refocus
used in a special procedure to orient both optic the telescope. Once both the cross hair
axes in a common plane perpendicular to the and bulletin board are focused and show
main spectrometer axis. To use them, loosen no parallax, bring the spectrometer back
the upper (locking) screw and rotate the lower to the lab bench.
(adjustment) screw to tilt the optic axis up or
down. When properly adjusted, finger tighten 2. Loosen the index ring, move it back
both screws (mostly the locking screw) while against the entrance slit assembly and re-
maintaining the axis orientation. tighten it. You should now be able grab
A grating mounted on a glass plate with a the index ring to move the collimator tube
grating spacing d ≈ 1/600 mm will be used for smoothly in and out and rotate it. Line
the measurements. The grating should only up the telescope to look into the colli-
be handled by the edges of the glass plate. Do mator and place a incandescent bulb be-
not damage the grating by touching it. hind the entrance slit. Looking through
the telescope, move the collimator tube
Alignment Procedure in or out until the open slit is in sharp
focus. Do not touch the telescope fo-
This spectrometer is a semi-precision instru- cus ring. It must be left where it was
ment that can be damaged with improper use. from the previous step. Adjust the en-
Parts should not be removed for any reason trance slit for a narrow width and orient
without first checking with an instructor. it horizontally. Check the slit orientation
by rotating the telescope slightly and ver-
1. Carefully take the spectrometer out into ifying that the cross hair moves parallel
the hallway and rest it on a portable table to the slit, but don’t be concerned if it
or lab stool. Look through the telescope is slightly high or low relative to the slit
at the bulletin board at the end of the cor- image. Carefully—without moving the
ridor. This is far enough to be an effec- collimator tube—loosen the index ring
tively infinite object distance. With both completely. Then gently move it into one
eyes open and the unaided eye focused on of the V-grooves and retighten it—again,
the distant object, simultaneously focus without moving the collimator tube.
on the cross hair by sliding the eyepiece At the end of this step you should have

October 14, 2022

VIS 6 Advanced Physics Laboratory

tilt adjustment other two tilt adjustment screws as shown

screw holes grating
surface
in Fig. 5. Tighten it under the clamp.
post & Install the grating table into the table
arm
base and adjust the table height so the
center of the grating is on the optic axis
of the telescope/collimator. Tighten the
grating table table locking screw. Rotate the telescope
until it makes an 80-90◦ angle with the
collimator. Adjust the table base rota-
tion angle as necessary to see (through the
Figure 5: Schematic top-view of the grat-
telescope) the reflection of the collimator
ing placement relative to the three table tilt
slit (still oriented horizontally and illumi-
screws.
nated with the light bulb) off the front
surface of the grating. Tighten the table
a horizontal and sharp image of the en- base locking screw. Adjust the telescope
trance slit and a sharp image of the cross rotation angle and either of the two table
hairs. leveling screws to center the reflected im-
age of the slit on the cross hairs. This step
3. Adjust the telescope leveling screws to get
guarantees the grating normal is perpen-
the cross hair centered on the narrow en-
dicular to the main rotation axis. Explain
trance slit. Rotate the telescope along
why! (Hint: it requires that the telescope
the slit and verify the cross hair stays
and collimator are aligned with one an-
centered on the slit. Tighten the level-
other before adjusting the table tilt to line
ing screws while maintaining the vertical
up the reflection.)
alignment. This ensures the telescope and
collimator optical axes are aligned to one
5. Rotate the telescope half-way to the col-
another but not necessarily perpendicular
limator so that the telescope is then ap-
to the main rotation axis.
proximately perpendicular to the grating
surface. Shine a small flashlight into the
side hole near the telescope eyepiece and
4. Locate the transmission grating plate, further adjust the telescope rotation un-
which should already be installed in a til you can see light reflected from the
black mounting bracket. The grating is a grating surface. As you get close to per-
decal glued on one side of a glass plate. pendicular, a circle filled with reflected
This side should already be facing the light and having about the same diam-
front side of the bracket. (The grat- eter as the telescope field of view should
ing is installed from the back side of the appear. Some adjustment of the position-
bracket.) ing of the light source going into the side
Mount the grating assembly at the cen- port may be needed to see this reflected
ter of the table so that the front surface field of view, which will probably not be
lies along the diameter directly in line centered vertically. Adjust the telescope
with the mounting post and the opposite leveling screws to center the reflected field
tilt adjustment screw, thus bisecting the of view. If the telescope focusing is cor-

October 14, 2022

Visible Spectrometer VIS 7

rect, you should be able to see a reflected The spectrometer should now be in proper
cross hair in the reflected field of view. If alignment. The telescope and collimator focus
you can not see the reflected cross hair, it should be at infinity, Both the cross hair and
means the telescope was not focused at in- slit should be in sharp focus with no parallax
finity. Adjust the telescope focus by very between them and they should be vertically
small amounts to find and focus on the re- aligned. The telescope and collimator optical
flected cross hair. Adjusting the telescope axes should be perpendicular to the spectrom-
focus so that both the direct and reflected eter rotation axis.
cross hairs are in focus (with no parallax
between them) focuses the telescope at in- 8. Put the table/grating back in the table
finity and is an alternative to focusing it base. Make sure that the front surface
on a distant object. Continue adjusting of the grating is still centered along the
the telescope leveling screw, rotation an- table diameter.
gle and focus to align (overlap) the two fo-
cused cross hairs with no parallax. Over- 9. Loosen the table base locking screw, ad-
lapping the cross hair and its reflection just the vernier scales so they are roughly
is called autocollimation, and will also be 90◦ to the collimator (and thus easy to
performed in other steps. In this step, view) and re-tighten the table base lock-
only the vertical alignment is important; ing screws.
it ensures the telescope axis is parallel to
the grating normal and thus perpendicu- 10. Rotate the table to get an angle of inci-
lar to the main rotation axis. Tighten the dence θi around 5-10◦ . The open end of
telescope leveling screws while maintain- the grating mount should face the colli-
ing its vertical alignment. mator. Otherwise, the mount will block
the grating at large angles. Adjust the ta-
ble height so that the grating is centered
6. Loosen the table locking screw and care-
on the collimator/telescope optic axis and
fully remove the table and set it aside
tighten the table locking screw.
without upsetting the position of the
grating. Align the telescope and colli- 11. Again, shine a light into the telescope side
mator to directly view the entrance slit. hole and adjust the telescope rotation and
Adjust the telescope focus, if necessary, the table leveling screws to achieve auto-
to remove parallax. The cross hair and collimation. This step ensures the grating
entrance slit must be in focus simulta- normal is perpendicular to the main rota-
neously. Adjust the collimator leveling tion axis.
screws so that the telescope cross hair is
in vertical alignment with the entrance 12. With the slit still horizontal and illumi-
slit, and then re-tighten them. This step nated, rotate the telescope to one side and
makes the collimator axis parallel to the the other and find the “rainbows” on each
telescope axis and thus perpendicular to side. Adjust the grating dispersion plane
the main rotation axis. using the inline table leveling screw (op-
posite the post in Fig. 5) so that the rain-
7. Go back to Step 4 and refine your align- bows on each side remain centered on the
ment. cross hairs.

October 14, 2022

VIS 8 Advanced Physics Laboratory

13. Repeat from Step 11 until no further ad- where the autocollimation reflection oc-
justments are necessary. curs and the direct and reflected cross
hairs are aligned both vertically and hor-
The positioning of the grating on the table, izontally. The telescope angular readings
the table itself, and the table base rotation are then a measure of the grating normal
angle determine the grating incidence angle θi direction.
and must be kept fixed throughout the exper-
iment. Thus the grating, the table, and the 15. Turn the entrance slit vertical and make
table base must not be moved again. In par- it reasonably narrow (a few tenths of a
ticular, it is easy to accidentally turn the table millimeter). You will have to make some
base tangent screw when you mean to turn the trade-offs between the slit width and the
telescope tangent screw. ability to see weak spectral lines. The
At this point the spectrometer is ready for narrower the slit, the sharper the lines,
measurements. but they also get harder to see.

16. Place the helium discharge tube just be-

Measurements hind and nearly touching the entrance
The measurements consists of adjusting the slit. Adjust its position for maxi-
telescope rotation angle to center the cross mum brightness while viewing the zero
hair on a colored slit image, a zero-order order (straight-through, all-wavelengths)
slit image, or an autocollimation reflection slit image. Measure and record the angu-
and recording the angular readings from both lar reading Ai and Bi for the zero order
verniers. Do not make any subtractions or av- image. This reading is a measure of the
erages before recording. Label column head- incidence angle θi . Note the “ghost” lines
ings A for one vernier and B for the other. from imperfections in the grating. These
A common spectroscopic technique for mea- ghost lines may also be seen on stronger
suring unknown wavelengths from one source spectral lines. Ignore them.
is to first measure known wavelengths from
some other sources. These initial measure- 17. Leave about ten rows clear at the top of
ments are used to calibrate the spectrome- the spreadsheet (for intermediary and fi-
ter — in this case determine constants such nal results) and make a spreadsheet ta-
as the grating groove spacing d and the an- ble with columns for the color of the
gle of incidence θi . While it is perhaps a line and both vernier readings Ar and Br
bit artificial, in this experiment the hydrogen which would then be a measure of the
wavelengths will be considered unknown and diffraction angle θr . Record the readings
those from any other sources will be consid- for the brighter lines of the calibration
ered known calibration wavelengths. It is rec- source in all orders attainable on both
ommended that you start with a helium dis- sides of the zero order image. For he-
charge tube as a calibration source, but feel lium, Table 1 gives the colors and air
free to use other sources as you see fit. The wavelengths1 of only the brightest line
following measurement and analysis steps out- 1
Air wavelengths are measured at roughly standard
line how this is done. temperature and pressure and thus they are affected
by the index of refraction of air. Comprehension Ques-
14. Record the telescope angles An and Bn tion 5 discusses the issue of vacuum vs. air wavelengths

October 14, 2022

Visible Spectrometer VIS 9

for each color. These four wavelengths should be able to see the Balmer lines
measured in four orders should be suf- corresponding to ni = 3, 4, 5, and 6 in
ficient for a good calibration, but you several orders. In order starting at and
can use the reference tables in the aux- away from the zero-order image on either
iliary material or data on the NIST web side, they appear as a violet (weak, some-
site: www.nist.gov/pml/data/asd.cfm times extremely weak), blue violet, blue
for other calibration sources or for addi- green, and red. Use the video camera to
tional wavelengths of other helium lines. see the weaker lines of hydrogen. Be sure
Include a column for the order, ±1, ±2, the camera aperture is fully opened and
(and maybe ±3 if you see a violet line in simply position it where you put your eye
third order.) Use positive orders for lines and focus it. The lines for ni = 6 and 7
observed on the side of the zero order im- (and perhaps higher) should be measur-
age where sin θr > sin θi and negative or- able in first (and perhaps higher) order.
ders for those observed on the other side. If any of the lines are difficult to measure,
There should be about 16 measured lines. feel free to assign them different angular
uncertainties σr .
18. Add a column for σr — the uncertainty in
the measured Ar or Br . It can be given in
minutes of arc and can be the same for all CHECKPOINT: The procedure should
lines. It can be adjusted in later analysis be complete through the prior step, and
steps. analysis should be complete through
Step 5.
19. Replace the helium discharge tube with a
hydrogen tube. Be careful as the tubes 21. Use the sodium lamp and the nar-
can get quite hot. The hydrogen dis- rowest possible entrance slit. Observe
charge should have a red section in the the sodium doublet lines at 589.0 and
middle of the tube growing pinkish to- 589.6 nm in first order (m = 1). They
ward the electrodes. If there are less than should be easily resolved — appearing as
a couple of centimeters of red in the mid- two separate yellow lines. Now place the
dle, it is time to change the tube. Ad- auxiliary slit over the end of the collima-
just the source/spectrometer heights to tor and orient it vertically. Slowly de-
get the red section just about touching crease its width. Since the light leav-
the entrance slit and the zero order im- ing the collimator is a parallel beam, as
age is as bright as possible. the auxiliary slit is narrowed, less and
less grating grooves will be illuminated.
20. A row or two below the data table for Narrow the slit until the doublet is no
the calibration lines, and with the same longer resolved and appears as a single
columns, make a similar table of color and line. Measure the auxiliary slit width
the Ar and Br readings for the observ- at the point where the sodium lines are
able lines of the hydrogen spectrum and no longer resolved. Determine how many
the order number they are observed at us- grating grooves are illuminated. Do the
ing the same scheme for plus and minus lines then become resolvable when viewed
orders as for the calibration lines. You in second order? Discuss the significance
for this experiment. of this mini-experiment. Be quantitative.

October 14, 2022

VIS 10 Advanced Physics Laboratory

color wavelength of arc. It can be the same for all spectral

(nm) lines.
violet 447.15
blue-green 501.57 3. In a separate area (above the data table)
orange 587.57 enter the incidence angle readings Ai and
red 667.82 Bi and convert to an Hi . Similarly record
the grating normal readings An and Bn
and convert to an Hn . Give a an estimate
Table 1: The stronger visible lines from a he- of their uncertainties σi and σn .
lium discharge.
Recall that the incidence angle θi and the
What is the resolving power of the spec- diffraction angles θr are relative to the grating
trometer in first order assuming diffrac- normal and are thus given by
tion limited performance when the auxil-
θi = Hi − Hn (8)
iary slit is removed?
and
Data Analysis θr = Hr − Hn (9)
Calibration Equation 6 in terms of H-readings becomes:

The first step is to reduce each pair of angular mλ = d [sin(Hr − Hn ) − sin(Hi − Hn )] (10)
readings A and B to a single value H by aver-
aging the readings for A and B ±180◦ . Choose The values d, Hi , and Hn are constants in the
the sign in B ± 180◦ such that this term is fit and all three can be determined from a lin-
near that of the A reading. For example, with ear regression.
A = 32◦ 33′ and B = 212◦ 30′ , use the − sign;
but for A = 325◦ 15′ and B = 145◦ 17′ , use the C.Q. 1 (a) Use the trigonometric identity
+ sign. If the set of H values ends up with sin(a ± b) = sin a cos b ± cos a sin b, but only
values in both the first (0◦ to 90◦ ) and fourth for the term sin(Hr − Hn ), to show that Eq. 10
(270◦ to 360◦ ) quadrants, subtract 360◦ from can be written:
those in the fourth quadrant. That leaves the
mλ = Ds sin Hr + Dc cos Hr + D0 (11)
angle in the fourth quadrant, but with values
from 0◦ to −90◦ . This makes the angles cover where
a continuous range so plots with H values on
the horizontal axis will come out without an Ds = d cos Hn (12)
odd-looking gap. Dc = −d sin Hn
1. Add a column to your data table convert- D0 = −d sin(Hi − Hn )
ing the Ar and Br for each spectral line
to an Hr . Equation 11 is in the form of a linear
regression of mλ (the y-variable) on the
2. The uncertainty σr is yours to decide terms sin Hr , cos Hr and a constant (the x-
based on your estimate of the precision variables). The numerical values for the three
in measuring the angles, but should cor- coefficients: Ds , Dc , and D0 obtained from
respond to no more than a few minutes the regression can then be used with Equation

October 14, 2022

Visible Spectrometer VIS 11

set 12 to determine d, Hn and Hi . (b) Show cos Hr columns in the Jacobian can be
that d and Hn can be determined from reused for this column, but also note that
√ the derivative depends on the fitting pa-
d = Ds2 + Dc2 (13) rameters, which have not yet been cal-
( )
Hn = − tan−1
Dc culated. Thus, reserve a 3 × 1 column
D vector of cells above the data table for
( s ) ( )
D Dc these values and simply start with Ds =
−1 0 −1
Hi = − sin √ − tan Dc = D0 = 1. Use these cell addresses
Ds2 + Dc2 Ds
when making the column of σmλ values.
Keep in mind that Excel’s trig functions You will be overriding the parameter val-
need arguments in radians and the inverse trig ues with correct values in Steps 10 and 11
functions return angles in radians. The con- before accepting results as final.
version factor is π/180 (radians/degree) and Since the fit depends on the uncertain-
Excel has a PI() function for the value of π. ties σmλ and the uncertainties depend on the
Excel can also translate between degrees and derivatives and the derivatives depend on the
radians with the functions RADIANS(angle in fit, if we are not careful we will create a cir-
degrees) or DEGREES(angle in radians). cular reference. We can take care of that
by telling Excel to allow circular references
4. Add a column for the Table 1 reference but restrict how many iterations around the
wavelengths λ of each line and another circle are allowed. This would work for the
for mλ — the product of the order number analysis here, but not if the Excel Solver pro-
and the wavelength. Leave these columns gram must be used. Since we might want to
in units of nanometers. use Solver as a check on the analysis, rather
5. In three adjacent columns, but in the than describe the details of using circular ref-
same rows with the data, create the Ja- erences, it is recommended that you use a
cobian [Jbmλ ]ij = ∂(mλi )/∂bj where the Copy|Paste Special|Values technique (described
bj are the three fitting parameters Ds , shortly) which works in all cases where uncer-
Dc and D0 . That is, make side-by-side tainties depend on the best fit.
columns for sin Hr and cos Hr (for the 7. You already have the calculated column
Ds and Dc , respectively) and a column for σmλ , but it is based on incorrect pa-
of ones (for the D0 term). Note that a rameter values: Ds = Dc = D0 = 1.
subscript or superscript b in Jacobian and So, one column to the right of the cal-
covariance matrices will always represent culated column, create a temporary col-
this set of fitting parameters. umn of ones for the constant copy. Us-
ing this column of ones in the linear
6. Make a column for σmλ , the uncertainty in
regression analysis will make it equally
mλ due to uncertainty in Hr . From prop-
weighted. After this equally-weighted fit
agation of error, this quantity is given by
is complete, you will use the Copy|Paste
σmλ = |∂(mλ)/∂Hr |σr where σr is your
Special|Values method to achieve a more
uncertainty in Hr but it must be con-
properly weighted linear regression.
verted to radians in this equation.
Determine the formula for the derivative 8. Use the weighted linear regression equa-
from Eq. 11 to see that the sin Hr and tions and Excel array formulas to first cre-

October 14, 2022

VIS 12 Advanced Physics Laboratory

2 −1
ate the N × N weighting matrix [σmλ ] . Special|Values from the (now almost cor-
There is no reason to display the weight- rect) calculated values of σmλ to the con-
ing matrix explicitly on the spreadsheet stant column from which the weighting
so name it and create it in the Name matrix is made and the new set of fitting
Manager (located in the Formula toolbar). parameters will immediately update to a
Recall the Excel array formula for the more properly weighted fit. It won’t be
weighting matrix is =MUNIT(N)/array^2 totally correct until you do the Copy|Paste
where N is the number of calibration Special|Values a few more times and the fit
wavelengths to be fitted and array is the stops changing.
N × 1 column vector containing the con-
stant copy of σmλ values. As described 12. Fill in the blank row next to the fitting
above, it is OK that we start with a col- parameters with their uncertainties — the
umn of ones, which will produce a first square root of the diagonal elements of
set of parameters based on an equally- the covariance matrix, [σb2 ].
weighted fit. 13. In the data table, make a new column for
the fitted values of mλ (based on Eq. 11).
9. Next to, and in the same rows as the Make plots of mλ vs. Hr . For the mea-
fitting parameters reserved for them in surements use symbols no line. Error bars
Step 6, leave one column blank (for the should be too small to show on this plot.
parameter uncertainties) and then to the Overlay a plot for the fit using a smoothed
right of the blank column and in the same line, no symbols. You may have to sort
rows as the fitting parameters make [σb2 ], the rows (sort all rows in order of Hr ) for
the 3 × 3 covariance matrix for the fitting the fit to show properly on the graph. Or,
parameters. Select this 3×3 block and use make a new column of Hr values (below
Excel’s MMULT vector and matrix multi- those in the data table), evaluate the val-
plication function, its TRANSPOSE func- ues for mλ they produce and use these
tion and its MINVERSE function to cre- columns for the fit.
2 −1 mλ −1
ate [σb2 ] = [[Jbmλ ]T [σmλ ] [Jb ]] , where
2 −1
[σmλ ] is the N × N weighting matrix 14. Make a separate column for and plot the
and is referenced by the name given to residuals: ∆mλ = mλ − (Ds sin Hr +
it in the name manager and [Jbmλ ] is the Dc cos Hr + D0 ) vs. Hr . Again use sym-
N × M Jacobian made in Step 5. bols, no lines. Error bars of ±σmλ should
be included. Misidentified wavelengths
10. Now you can get the first set of fitting pa- or bad angular measurements should be
rameters. In the block reserved for them obvious from the residuals, which should
in Step 6, make the 3 × 1 vector of fitting show only random deviations (hopefully,
2 −1
parameters: b = [σb2 ][Jbmλ ]T [σmλ ] y, no more than a nanometers or so) cen-
where y is the vector of mλ values. tered around zero. Bad points should be
rechecked for possible errors in Hr , m or
11. The first set of parameters obtained above λ. If you cannot reconcile a bad point,
are based on an equally weighted fit be- remeasure it.
cause the weighting matrix was made us-
ing an initial column of ones for the σmλ 15. Add a cell in the top area of the
uncertainties. So now do the Copy|Paste spreadsheet for the resulting chi square.

October 14, 2022

Visible Spectrometer VIS 13

You can use the usual formula for χ2 have the same columns as for the calibra-
or use two MMULT functions to cre- tion data, but several modifications and
ate the equivalent matrix version χ2 = additions will be needed. For starters, the
2 −1
∆mλT [σmλ ] ∆mλ, where ∆mλ is the values of λ will not come from reference
column of deviations. Even though this is values. To get them, start by constructing
a single cell, you will need to end it with the column for hydrogen mλ values using
Ctrl|Shift|Enter. the calibration formula, Eq. 11, with the
three fitting parameters Dc , Ds , and D0
Do you need to take into account the un- and the hydrogen Hr values. But before
certainty in the reference wavelengths? The you do, first make [Jbmλ ] — the Jacobian
accuracy of the wavelengths given in Table 1 columns for the hydrogen Hr values as
is less than 0.02 nm. it was made for the calibration measure-
ments. You will need this Jacobian ma-
C.Q. 2 Use Equation set 13 with the fitting trix in later steps and you can use it here
parameters to determine d, Hi and Hn . (In to create the entire column of mλ val-
Jacobians and covariance matrices, this set ues by selecting this column and creating
of parameters will be abbreviated with a sub- the Excel array formula corresponding to
script or superscript d.) Use [σb2 ] and the Ja- mλ = [Jbmλ ]b where b is the 3 × 1 array
cobian for these equations to determine [σd2 ], of parameter values.
the covariance matrix for d, Hi and Hn giv-
ing these variables’ variances and covariances. 17. Create the λ column based on the mλ
Compare the fitted parameter values with the found above and the recorded order m.
manufacturer’s value for d and your directly
measured values for Hi and Hn . Hint: [σd2 ] is 18. Additional columns will be needed for
obtained from propagation of error as [σd2 ] = determining a fit of the hydrogen wave-
[Jbd ][σb2 ][Jbd ]T where [Jbd ] is the 3 × 3 set of lengths to the Rydberg formula (Eq. 3).
derivatives of the d parameters with respect to In particular, make a new column for
the b parameters according to Equation set 13. y = 1/λ and one for the values of x =
Interestingly, in this case it turns out [Jbd ] is 1/n2f − 1/n2i . For the latter, first make a
more easily determined (and less prone to mis- column for the initial state quantum num-
takes) as the inverse of [Jdb ], the set of deriva- bers ni for each measured hydrogen line.
tives of b parameters with respect to the d pa- Recall that for all Balmer lines, nf = 2.
rameters according to Equation set 12. That You will have to figure out which ni goes
is, [Jbd ] = [Jdb ]−1 . with each line. Check that a plot of y vs.
x appears to lie on a straight line through
Hydrogen Data the origin. Check for outliers or other
problems. Don’t be concerned that you
Next, you will use the calibration results to have multiple 1/λ values (from different
determine the wavelengths of hydrogen lines. orders) corresponding to each x. They
You will then analyze the wavelengths to de- will all be used for the Rydberg fit and
termine the hydrogen Rydberg constant. all will properly affect the results.

16. Make a second table a row or two below 19. For a properly weighted fit you will need
the table for the calibration data. It will 2
the (diagonal) weighting matrix [σ1/λ ]−1

October 14, 2022

VIS 14 Advanced Physics Laboratory

associated with σ1/λ , the uncertainties in fitted Rydberg constant from R =

R †
1/λ due to the uncertainty σr in Hr . The [J1/λ ] y where y is the 1 × N column vec-
uncertainty in the Rydberg due to cali- tor of y-values for 1/λ.
bration uncertainty will be addressed in
Step 23. 23. The next step is to find the uncertainty
From propagation of errors, the uncer- in the Rydberg due to the uncertainty
tainty in 1/λ is calculated from σ1/λ = in the calibrations parameters as given
|∂ (1/λ)/∂ Hr |σr . Since we already have by their covariance matrix [σb2 ]. The
2(b)
the column for |∂ mλ/∂ Hr |σr (by copying equation for this 1 × 1 matrix is [σR ] =
R † 1/λ T 1/λ R †T
the formula from the calibration data ta- [J1/λ ] [Jmλ ] [Jbmλ ][σb2 ][Jbmλ ]T [Jmλ ][J1/λ ] .
ble), the derivative above can be obtained The only new quantity in this equation is
using the chain rule: ∂(1/λ)/∂ Hr = 1/λ
[Jmλ ], the N × N diagonal Jacobian ma-
(∂(1/λ)/∂ mλ)(∂(mλ)/∂ Hr ). Show that trix obtained in Excel from the array for-
∂(1/λ)/∂ mλ = −1/mλ2 and make a mula =MUNIT(N)*array, where array is
column for these values. They will the N × 1 column of values for −1/mλ2 .
be used again shortly. Then use 1/λ
Because [Jmλ ] is diagonal, it is its own
that column to make a column for
transpose and can be used as such in the
σ1/λ = |(1/mλ2 )(∂(mλ)/∂Hr )|σr , where 2(b)
equation for [σR ] above. Recall, [σb2 ] is
σr must be in radians.
the 3 × 3 parameter covariance matrix de-
20. In the Name Manager, make the N × N termined in the calibration analysis.
2
weighting matrix [σ1/λ ]−1 from the Excel
array formula =MUNIT(N)/array^2 where 24. Make plots of the data and fit for 1/λ
N is the number of Balmer lines and array versus x. Also make a plot of residuals
is the column of σ1/λ -values created in the with error bars. Calculate the χ2 for the
previous step. fit to the Rydberg formula.

Now you are ready for a one-parameter fit to

the Rydberg formula with the y-values com-
Comprehension Questions
ing from the column of 1/λ and the x-values 3. For both the calibration fit and the Ryd-
coming from the single column of 1/n2f − 1/n2i berg fit, discuss the χ2 and residuals in re-
values. lation to your estimates of the uncertain-
2(r) ties in the angular measurements. How
21. Make a cell for [σR ]: the 1 × 1 co-
precise were your Hr values? Can you
variance matrix (whose only element is
conclude your measurements are in agree-
the variance of R due to the uncer-
2(r) ment with the calibration and Rydberg
tainties in Hr ) as given by [σR ] =
formulas?
]−1 [JR ]]−1 , where [JR ] is
1/λ 1/λ 1/λ
[[JR ]T [σ1/λ
2

the N × 1 column vector of x-values: 4. Find high-accuracy values for infinite-

1/n2f − 1/n2i . mass and hydrogen Rydberg for compar-
22. For later work, create the 1 × N row ison with your value. Does your result
R †
vector for the pseudo inverse [J1/λ ] = test the difference between RH and R∞ ?
Explain.
]−1 and then create the
2(r) 1/λ
[σR ][JR ]T [σ1/λ
2

October 14, 2022

Visible Spectrometer VIS 15

5. The calibration wavelengths in Table 1

are given in air and the calibration lines
were measured in air. We used the cal-
ibration and measured the Balmer lines
in air so their calculated values are air
wavelengths as well. The Rydberg for-
mula, however, is valid for vacuum wave-
lengths only. The relation between λm ,
an air wavelength, and λ, the correspond-
ing vacuum wavelength is: λ/λm = n
where here n is the index of refraction of
air, which differs from one by less than
one part in a thousand. So, is it an is-
sue here? Hint: Check the NIST link
at emtoolbox.nist.gov/Wavelength/Docu-
mentation.asp to see that the index of re-
fraction of air is nearly constant over the
range of wavelengths measured here and
get its value. Would correcting for the
index of refraction of air make the Ryd-
berg constant larger or smaller than the
value without the correction? How big is
the correction to the Rydberg, and is it
significant at the precision of your mea-
surements?

October 14, 2022